CHAPTER 1: INTRODUCTION

i. Definition of Statistics

The word statistics is defined in different ways depending on its use in the plural and singular
sense.

In the plural sense: - statistics is defined as the collection of numerical facts or figures ( or the raw
data themselves).

Eg. 1. Vital statistics (numerical data on marriage, births, deaths, etc).

2. The average mark of statistics course for students is 70% would be considered as a
statistics whereas Abebe has got 90% in statistics course is not statistics.
Remark: statistics are aggregate of facts. Single and isolated figures are not statistics as they cannot be
compared and are unrelated.

In its singular sense:- the word Statistics is the subject that deals with the methods of
collecting, organizing, presenting, analyzing and interpreting statistical data.

Classification of Statistics
Statistics is broadly divided into two categories based on how the collected data are used.
Descriptive Statistics: deals with describing the data collected without going further conclusion
Example 1.1: Suppose that the mark of 6 students in Statistics course for Biology section A is
given as 40, 45, 50, 60, 70 and 80. The average mark of the 6 students is 57.5 and it is considered
as descriptive statistics.
Inferential Statistics: It deals with making inferences and/or conclusions about a population
based on data obtained from a sample of observations. It consists of performing hypothesis
testing, determining relationships among variables and making predictions.
Example 1.2: In the above example, if we say that the average mark in Statistics course for
Biology section A students is 57.5, then we talk about inferential statistics (draw conclusion
based on the sample observation).

1.2 Stages in Statistical Investigation

The area of statistics points out the following five stages. These are collection, organization,
presentation, analysis and interpretation of data.

1
i. Collection of data: This is the process of obtaining measurements or counts or obtaining raw
data.
Data can be collected in a variety of ways; one of the most common methods is through the use
of sample or census survey. Survey can also be done in different methods, three of the most
common methods are:
 Telephone survey
 Mailed questionnaire
 Personal interview.
ii. Organization of data: Data collected from published sources are generally in organized form.
However if an investigator has collected data through a survey, it is necessary to edit these data
in order to correct any apparent inconsistencies, ambiguities, and recording errors.
This phase also includes correcting the data for errors, grouping data into classes and tabulating.
iii. Presentation of data: After the data have been collected and organized they can be presented
in the form of tables, charts, diagrams and graphs. This presentation in an orderly manner
facilitates the understanding as well as analysis of data.
iv. Analysis of data: the basic purpose of data analysis is to dig out useful information for
decision making. This analysis may simply be a critical observation of data to draw some
meaningful conclusions about it or it may involve highly complex and sophisticated
mathematical techniques.
v. Interpretation of data: Interpretation means drawing conclusions from the data collected and
analyzed. Correct interpretation will lead to a valid conclusion of the study & thus can aid in
decision making.
1.3 Definition of some terms
Population: It is the totality of objects under study. The population represents the target of an
investigation, and the objective of the investigation is to draw conclusions about the population
hence we sometimes call it target population. The word population doesn’t necessarily refer to
people.
Example 1.3:
 All patients of hospital suffering from TB and treated with a new drug
 All clients of Telephone Company
 All students of Debre Markos University (DMU)
 Population of families, etc
The population could be finite or infinite (an imaginary collection of units).

2
Survey: - is an investigation of a certain population to assess its characteristics. It may be census
or sample.
Sample:- is part or subset of population under study.

Sampling frame:- is the list of all possible units of the population that the sample can be drawn
from it.

Eg. List of all students of DMU, List of all residential houses in Debre Markos town, etc

Census survey: a complete enumeration of the population under study.

Sample survey: the process of collecting data covering a representative part or portion of a
population.
Parameter: is a statistical measure of a population, or summary value calculated from a
population. Examples: Average, Range, proportion, variance, etc.
Statistic: is a descriptive measure of a sample, or it is a summary value calculated from a
sample.
Sampling: The process or method of sample selection from the population.
Sample size: The number of elements or observation to be included in the sample.
An element: is a member of sample or population. It is specific subject or object (for example a
person, firm, item, etc.) about which the information is collected.
Variable: It is an item of interest that can take numerical or non-numerical values for different
elements. It may be qualitative or quantitative. Example: age, weight, sex, marital status, etc.
Observation (measurement): is the value of a variable for an element.
Qualitative variables: - are variables that assume non-numerical values. They can be
categorized and they are usually called attributes. Eg. Sex, marital status, ID number, etc

Quantitative variables: - are variables which assume numerical values. eg. Age, weight, etc.

1.4 Applications, uses and limitation of Biostatistics/Statistics

Application of Biostatistics
Application of Biostatistics is not restricted to certain experiments but is used in a wide
variety of contexts. Some of these applications are as follows:
1. Genetical statistics

3
 - In classical or Mendelian genetics-the focus of interest is centered on the inheritance
of qualitative characteristics. The statistical methods generally applied are binomial
or chi-square tests.
- Population genetics:- is concerned with studying genetic structure of populations and
changes occurring in it over generations. The frequencies of different genes & their
changes due to the effect of some forces can be estimated with the application of
different statistical methods.
2. Numerical taxonomy:- deals with grouping of taxonomic units into taxa by numerical
methods.
3. Statistical ecology:- deal with the study of temporal and spatial patterns of populations of
organisms.
4. Forest measurement:- measurement of tree length, area, volume, weight and the like.
5. Demography:- the study of measurement of human population. It measures birth, death,
marriage, morbidity, migration, etc.
6. Medical sciences:- health managers are expected to take some decisions on the basis of
whatever bit of information is available.

Applications of Statistics
Statistics can be applied in any field of study which seeks quantitative evidence. For instance,
engineering, economics, natural science, etc.
a) Engineering: Statistics have wide application in engineering.
 To compare the breaking strength of two types of materials.
 To determine the probability of reliability of a product.
 To control the quality of products in a given production process.
 To compare the improvement of yield due to certain additives such as fertilizer,
herbicides, e t c
b) Economics: Statistics are widely used in economics study and research.
 To measure and forecast Gross National Product (GNP).
 Statistical analyses of population growth, inflation rate, poverty, unemployment figures,
rural or urban population shifts and so on influence much of the economic policy making.
 Financial statistics are necessary in the fields of money and banking including consumer
savings and credit availability.

4
c) Statistics and research: there is hardly any advanced research going on without the use of
statistics in form or another. Statistics are used extensively in medical, pharmaceutical and
agricultural research.
Function/Uses of Statistics: Today the field of statistics is recognized as a highly useful tool to
making decision process by managers of modern business, industry, frequently changing
technology. It has a lot of functions in every day activities. The following are some uses of
statistics:
• It condenses and summarizes a mass of data: the original set of data (raw data) is normally
voluminous and disorganized unless it is summarized and expressed in few presentable,
understandable & precise figures.
• Statistics facilitates comparison of data: measures obtained from different set of data can be
compared to draw conclusion about those sets. Statistical values such as averages, percentages,
ratios, rates, coefficients, etc, are the tools that can be used for the purpose of comparing sets of
data.
• Statistics helps to predict future trends: statistics is very useful for analyzing the past and
present data and forecasting future events.
• Statistics helps to formulate & review policies: Statistics provide the basic material for
framing suitable policies. Statistical study results in the areas of taxation, on unemployment rate,
on inflation, on the performance of every sort of military equipment, etc, may convince a
government to review its policies and plans with the view to meet national needs and aspirations.
• Formulating and testing hypothesis: Statistical methods are extremely useful in formulating
and testing hypothesis and to develop new theories.
Limitations of Statistics

The field of statistics, though widely used in all areas of human knowledge and widely applied in
a variety of disciplines such as engineering, economics and research, has its own limitations.
Some of these limitations are:
a) It does not deal with individual values: as discussed earlier, statistics deals with aggregate of
facts. For example, wage earned by an individual worker at any one time, taken by itself is not a
statistics.
b) It does not deal with qualitative characteristics directly: statistics is not applicable to
qualitative characteristics such as beauty, honesty, poverty, standard of living and so on since
these cannot be expressed in quantitative terms. These characteristics, however, can be
statistically dealt with if some quantitative values can be assigned to these with logical criterion.

5
For example, intelligence may be compared to some degree by comparing IQs or some other
scores in certain intelligence tests.
c) Statistical conclusions are not universally true: since statistics is not an exact science, as is
the case with natural sciences, the statistical conclusions are true only under certain assumptions.
d) It can be misused: statistics cannot be used to full advantage in the absence of proper
understanding of the subject matter.
1.5 Scale of Measurement

Proper knowledge about the nature and type of data to be dealt with is essential in order to
specify and apply the proper statistical method for their analysis and inferences.
Scale Types
Measurement is the assignment of values to objects or events in a systematic fashion. Four levels
of measurement scales are commonly distinguished: nominal, ordinal, interval, and ratio and
each possessed different properties of measurement systems. The first two are qualitative while
the last two are quantitative.
Nominal scale: The values of a nominal attribute are just different names, i.e., nominal
attributes provide only enough information to distinguish one object from another. Qualities with
no ranking or ordering; no numerical or quantitative value. Data consists of names, labels and
categories. This is a scale for grouping individuals into different categories.
Example1.4: type of tree, type of insect, eye color: brown, black, . . ., sex: male, female.
 In this scale, one is different from the other
 Arithmetic operations (+, -, *, ÷) are not applicable, comparison (<, >, etc) is impossible
Ordinal scale: - defined as nominal data that can be ordered or ranked.
 Can be arranged in some order, but the differences between the data values are
meaningless.
 Data consisting of an ordering of ranking of measurements are said to be on an ordinal
scale of measurements. That is, the values of an ordinal scale provide enough information
to order objects.
 One is different from and greater /better/ less than the other
 Arithmetic operations (+, -, *, ÷) are impossible, comparison (<, >, etc) is possible.
Example1.5: Letter grading (A, B, C, D, F), rating scales (excellent, very good, good, fair,
poor), military status (general, colonel, lieutenant etc).

6
Interval Level: data are defined as ordinal data and the differences between data values are
meaningful. However, there is no true zero, or starting point, and the ratio of data values are
meaningless.
Note: Celsius & Fahrenheit temperature readings have no meaningful zero and ratios are
meaningless.
In this measurement scale:-
 One is different, better/greater and by a certain amount of difference than another
 Possible to add and subtract. For example; 800c – 500c = 300c, 700c – 400c = 300c
 Multiplication and division are not possible. For example; 60 0c = 3(200c). But this does
not imply that an object which is 600c is three times as hot as an object which is 200c.
Most common examples are: IQ, temperature.
Ratio scale: Similar to interval, except there is a true zero (absolute absence), or starting point,
and the ratios of data values have meaning.
 Arithmetic operations (+, -, *, ÷) are applicable. For ratio variables, both differences and
ratios are meaningful.
 One is different/larger /taller/ better/ less by a certain amount of difference and so much
times than the other.
 This measurement scale provides better information than interval scale of measurement
Example1.6: weight, age, number of trees, etc.
Exercise 1

1. A popular radio program, asked its listener to respond either ‘yes’ or ‘No’ to the question;
‘Are you concerned about spread of AIDS through unprotected sex?’ with in 1 minute,
91 callers expressed their views.
a/ what is the population of this study?
b/ what is the sample the radio station used to gauge the public mood?
c/ Is this sample scientific? Explain.
2. To study the average effect of fish dirt on human cholesterol level in blood, a researcher
randomly selects 500 males of 25 years of age who have never to taken fish more than
once a week, and measure their cholesterol level. The researcher then serves all the
individuals 8 ounces of fish everyday for one year. After one year the researcher
measures the cholesterol level of each individual again and calculates the difference with
the year before value (difference = pre-diet level minus post-diet level).
Determine:

7
 a/ the population
b/ the Sample
c/ the variable under study
d/ the parameter of interest and
e/ Is the variable qualitative or quantitative
3. A survey has been conducted on a sample of 350 Patients suffering from a Particular
type of headache.
Determine a/the population
b/ the sample

4. Classify the variable as qualitative or quantitative and separate which are nominal,
ordinal, interval & ratio

a/ Hair color , f/ white blood count,

b/ Type of tree, g/ cancer staging,

c / age, h/ WHO HIV stage,

d/ Grams of fat in herbages, i/ blood group

e/ types of surreal procedures,

CHAPTER 2: METHODS OF DATA COLLECTION AND

PRESENTATION

The second unit of this module introduces the methods of data collection and presentation. This
unit will deal how to collect and present the data you have collected so that they can be of use.
Thus the collected data also known as raw data are always in an unorganized form and need to be
organized and presented in a meaningful and readily comprehensible form in order to facilitate
further statistical analysis.
Objectives
At the end of this chapter students will be able to:
 Arrange raw data in an array and then classified data to construct a frequency table and a
cumulative frequency table.
 To organize data using frequency distribution.
 To present data using suitable graphs or diagrams.
2.1 Methods of Data Collection

8
 Data: is the raw material of statistics. It can be obtained either by measurement or counting.
When we determine that the appropriate approach to seeking an answer to a question will require
the use of statistics, we begin to search for suitable data to serve as the raw material for our
investigation.

Sources of data

The statistical data may be classified under two categories depending up on the sources.
1. Primary data: - Data collected by the investigator himself for the purpose of a specific
inquiry or study. Such data are original in character & are mostly generated by surveys
conducted by individuals or research institutions.
It is more reliable & accurate since the investigator can extract the correct information by
removing doubts, if any, in the minds of the respondents regarding certain questions.
2. Secondary data: - When an investigator uses data, which have already been collected by
others, such data are called secondary data. Such data are primary data for the agency that
collected them, and become secondary for some one else who uses these data for his own
purposes. Sources of secondary data are books, journals, reports, etc.
When our source is secondary data check that:
 The type and objective of the situations.
 The purpose for which the data are collected and compatible with the
present problem.
 The nature and classification of data is appropriate to our problem.
 There are no biases and misreporting in the published data.
Note: Data which are primary for one may be secondary for the other.
2.2 Methods of Data Presentation

Having collected and edited the data, the next important step is to organize it. That is to present it
in a readily comprehensible condensed form that aids in order to draw inferences from it. It is
also necessary that the like be separated from the unlike ones.
The presentation of data is broadly classified in to the following two categories:
 Tabular presentation
 Diagrammatic and Graphic presentation.
The process of arranging data in to classes or categories according to similarities technically is
called classification. It eliminates inconsistency and also brings out the points of similarity
and/or dissimilarity of collected items/data.

9
 Classification is necessary because it would not be possible to draw inferences and conclusions if
we have a large set of collected [raw] data.
2.2.1 Frequency distribution
Frequency: is the number of times a certain value or class of values occurs.
Frequency distribution (FD): is the organization of raw data in table form using classes and
frequency.
There are three types of FD and there are specific procedures for constructing each type.
The three types are:-
I. Categorical FD
II. Ungrouped FD and
III. Grouped FD
I. Categorical FD: Used for data that can be placed in specific categories; such as nominal, ordinal
level of data.
Example 2.1: Twenty five patients were given a blood test to determine their blood type. The
data is as shown below: A B B AB O A O O B AB B B B O A O O O AB AB A O O B A.
Solution: Since the data are categorical by taking the four blood types as classes we can
construct a FD as shown below.
Step 1: Make a table as shown below
CLASS TALLY FREQUANCY PERCENRT
A
B
AB
O

Step 2: Tally data and place the result under the column Tally
Step 3: Count the tallies and place the result under the column frequency.
Step 4: find the percentage of values in each class by the formula (%= f/n * 100%; f= frequency, n
total number of observation.)
CLASS TALLY FREQUANCY PERCENRT
A //// 5 5/25* 100 = 20%
B //// // 7 28%
AB //// 4 16%
O //// //// 9 9/25*100 = 36%

10
II. Ungrouped Frequency Distribution (UFD)
UFD us a table of all potential raw score values each times each actually could possibly occur in the
data along with the number of times each actually could occur.
UFD is often constructed for small set of data or data of discrete variable.

Constructing ungrouped frequency distribution:

 First find the smallest and largest raw score in the collected data.
 Arrange the data in order of magnitude and count the frequency.
 To facilitate counting one may include a column of tallies.

Example 2.2: The following data represent the number of days of sick leave taken by each of 50
workers of a company over the last 6 weeks.

2 0 0 5 8 3 4 1 0 0 7 1
7 1 5 4 0 4 0 1 8 9 7 0
1 7 2 5 5 4 3 3 0 0 2 5
1 3 0 2 4 5 0 5 7 5 1 1
0 2

i. Construct ungrouped frequency distribution

ii. How many workers had at least 1 day of sick leave?
iii. How many workers had between 2 and 6 days of sick leave?

Solution:

i. Since this data set contains only a relatively small number of distinct or different values,
it is convenient to represent it in a frequency table which presents each distinct value
along with its frequency of occurrence.

11
 Class Frequenc Cumulative
y frequency
0 12 12
1 8 20
2 5 25
3 4 29
4 5 34
5 8 42
7 5 47
8 2 49
9 1 50

ii. Since 12 of the 50 workers had no days of sick leave, the answer is 50-12=38
iii. The answer is the sum of the frequencies for values 3, 4 and 5 that is 4+5+8=17.

3. Grouped Frequency Distribution (GFD).

When the range of the data is large the data must be grouped in to classes that are more than one unit
in width.
Definition of some basic terms
 Grouped frequency distribution: is a FD when several numbers are grouped into one
class.
 Class limits (CL): It separate one class from another. The limits could actually appear in
the data and have gaps between the upper limits of one class and the lower limit of the next
class.
 Unit of measure (U): This is the possible difference between successive values. E.g. 1,
0.1, 0.01, 0.001, etc
 Class boundaries: Separate one class in a grouped frequency distribution from the other.
The boundary has one more decimal place than the raw data. There is no gap between the
upper boundaries of one class and the lower boundaries of the succeeding class. Lower
class boundary is found by subtracting half of the unit of measure from the lower class
limit and upper class boundary is found by adding half unit measure to the upper class
limit.

12
  Class width (W): The difference between the upper and lower boundaries of any
consecutive class. The class width is also the difference between the lower limit or upper
limits of two consecutive class.
 Class mark (Midpoint): It is found by adding the lower and upper class limit (boundaries)
and divided the sum by two.
 Cumulative frequency: It is the number of observation less than or greater than the upper
class boundary of class.
 CF (Less than type): it is the number of values less than the upper class boundary of a
given class.
 CF (Greater than type): it is the number of values greater than the lower class boundary
of a given class.
 Relative frequency (Rf ):The frequency divided by the total frequency. This gives the
present of values falling in that class.

Rfi = fi/n= fi/ ∑fi

 Relative cumulative frequency (RCf): The running total of the relative frequencies or the
cumulative frequency divided by the total frequency gives the present of the values which
are less than the upper class boundary or the reverse.

CRfi = Cfi/n= Cfi/∑fi

STEPS IN CONSTRUCTING A GFD

1. Find the highest and the smallest value
2. compute the range; R = H – L
3. Select the number of class desired (K)
I. Choose arbitrary between 5 and 15.
II. Using struggles formula
K= 1 + 3.322Log n; n= Total frequency
4. Find the class width (W) by dividing the range by the number of classes and round to the
nearest integer the result you get.
W = R/K
5. Identify the unit of measure usually as 1, 0.1, 0.01,

13
 6. Pick a suitable starting point less than or equal to the minimum value. Your starting point
is lower limit of the first class.
- Then continue to add the class width to get the rest lower class limits.
7. Find the upper class limits UCLi = LCLi-U. then continue to add width to get the rest
upper class limits
8. find class boundaries
LCBi = LCLi – ½ U
UCBi = UCLi + ½ U
9. Find class mark
CMi = (UCLi + LCLi )/ 2 or CMi = (UCBi + LCBi )/ 2.
10. Tally the data
11. Find the frequencies
12. Find the cumulative frequencies. Depending on what you are trying to accomplish, it may
be necessary to find the cumulative frequency.
13. If necessary find Rf and RCf.
When grouping data the following rules are important:
 The groups must not overlap, otherwise there is confusion concerning in which group a
measurement belongs.
  There must be continuity from one group to the next, which means that there must be no
gaps. Otherwise some measurements may not fit in a group.
  The groups must range from the lowest measurement to the highest measurement so that all
of the measurements have a group to which they can be assigned.
  The groups should normally be of an equal width, so that the counts in different groups can
easily be compared.
Example 2.3: The blood glucose level, in milligrams per deciliter, for 60 patients is shown
below. Construct a grouped frequency distribution for the data set.

55 70 85 90 93 86 103 74 92

63 101 83 82 100 97 97 109 84

84 75 92 68 114 84 101 81 91

82 115 86 69 59 56 84 77 90

77 97 80 101 61 74 87 80

14
 58 81 78 88 86 59 82 83

59 78 116 72 62 105 65 78

Solution:-
1) Highest value = 116, Lowest value = 55
2) Range = 116 – 55 = 61
3) K = 1+ 3.322Log60 = 1 + 3.322(1.78) = 6.9 ≈ 7
4) W = R / K = 61/7 = 8.7 ≈ 9
5) U = 1
6) LCL1=55
7) Find the upper class limits.
8) Find class boundaries
9) Find class mark

Class limit Frequency Class boundary Class CF(<) CF(>) Rf %Rf

Mark
55 – 63 9 54.5 – 63.5 59 9 60 0.15 15
64 – 72 5 63.5 – 72.5 68 14 51 0.08 8
73 – 81 12 72.5 – 81.5 77 26 46 0.2 20
82 – 90 17 81.5 – 90.5 86 43 34 0.28 28
91 – 99 7 90.5 – 99.5 95 50 17 0.12 12
100 – 108 6 99.5 – 108.5 104 56 10 0.1 10
109 –117 4 108.5 – 117.5 113 60 4 0.07 7

2.2.2 Diagrammatic presentation of data: Bar charts, Pie-chart, Cartograms

The most convenient and popular way of describing data is using graphical presentation. It is
easier to understand and interpret data when they are presented graphically than using words or a
frequency table. A graph can present data in a simple and clear way. Also it can illustrate the
important aspects of the data. This leads to better analysis and presentation of the data. In this
article, we discuss the approach for the most commonly used diagrammatic or graphical methods
such as bar charts, pie chart, histograms, frequency polygons and cumulative frequency
polygons.

15
The three most commonly used diagrammatic presentation for discrete as well as qualitative data
are:

 Pie charts
 Bar charts
 Pictogram
A) Pie chart

A pie chart is a circle that is divided in to sections or wedges according to the percentage of
frequencies in each category of the distribution. The angle of the sector is obtained using:

Value of the part

Angle of a sector= ∗360 %
The whole quantity

Example 2.4: Using the immunization status of children in certain area given in example 2.5,
draw the pie chart.

Bar

not immunized

37% 36%
partially immunized

fully immunized

27%

Charts

 Used to represent & compare the frequency distribution of discrete variables and
attributes or categorical series.
 Bars can be drawn either vertically or horizontally.

In presenting data using bar diagram,

 All bars must have equal width and the distance between bars must be equal.

16
  The height or length of each bar indicates the size (frequency) of the figure represented.

There are different types of bar charts. The most common being:

 Simple bar chart

 Component or sub divided bar chart.
 Multiple bar charts.
I. Simple bar chart
 Are used to display data on one variable.
 They are thick lines (narrow rectangles) having the same breadth. The magnitude of a
quantity is represented by the height /length of the bar.

Example 2.5 Consider the immunization status of children in certain area;

Immunization status Number/ Relative frequency in Percent
(Class) frequency
Not immunized 75 35.7 %
partially immunized 57 27.2 %
fully immunized 78 37.1 %
total 210 100.0 %
Draw a simple bar chart of the immunization status of children.

Solution:
number of children

100
80
60
40
20
0
Not immunized partially fully im-
immunized munized

immunization status

II. Component Bar chart

 When there is a desire to show how a total (or aggregate) is divided in to its component
parts, we use component bar chart.

17
  The bars represent total value of a variable with each total broken in to its component
parts and different colors or designs are used for identifications

Draw a component bar chart to represent the sales by product from 1957 to 1959.

18
Example 2.6: Consider data on immunization status of women by marital status

Immunization Status

Marital Status
Immunized Non immunized

No % No % Total
Single
58 24.7 177 75.3 235

Married 156 34.7 294 65.3 450

Divorced 10 35.7 18 64.3 28

Widowed 7 50.0 7 50.0 14

Total 231 31.0 496 68.2 727

Draw a component (sub-divided) bar chart of the immunization status of women by marital
status

Solution:

500
450
number of w om w n

400
non
350 immunized
300
250 immunized
200
150
100
50
0
Single Married Divorced Widowed
marital status

III. Multiple Bar charts

 These are used to display data on more than one variable.
 They are used for comparing different variables at the same time.

19
 Example 2.7: Draw a multiple bar chart to represent the immunization status of women by
marital status given in Example 2.6.

Solution:

350
300
250
number of womens

200 Immunized
150
Non-Immunized
100
50
0

marital status

B) Pictograph

In this diagram, we represent data by means of some picture symbols. We decide about a suitable
picture to represent a definite number of units in which the variable is measured.

2.2.4 Graphical Presentation of data

The histogram, frequency polygon and cumulative frequency graph or ogive is most commonly
applied graphical representation for continuous data.

Procedures for constructing statistical graphs:

 Draw and label the x and y axes.

 Choose a suitable scale for the frequencies or cumulative frequencies and label it on the y-
axes.
 Represent the class boundaries for the histogram or ogive or the mid points for the frequency
polygon on the x-axes.
 Plot the points.
 Draw the bars or lines to connect the points.
Histogram
A graph which displays the data by using vertical bars of various heights to represent
frequencies. Class boundaries are placed along the horizontal axes. Class marks and class limits

20
are sometimes used as quantity on the x-axis.
Example 2.8: Construct a histogram to represent the blood glucose level for 60 patients given in
example 2.3.

Solution:

Histogram
Frequency 18 17
16
14 12
12
10 9
8 7
6
6 5
4
4
2
0
1
Class boundary
l

Frequency polygon
If we join the mid-points of the tops of the adjacent rectangles of the histogram with line
segments a frequency polygon is obtained. When the polygon is continued to the x-axis just
outside the range of the lengths the total area under the polygon will be equal to the total area
under the histogram.
Example 2.9: Construct a Frequency polygon to represent the following data.
Class Frequency Class Class R.F. % R.F. Less than More than
limits marks boundarie C.F. C. F.
(percent)
s

15 - 24 3 19.5 14.5 - 24.5 0.06 6% 3 50

25 – 34 4 29.5 24.5 - 34.5 0.08 8% 7 47

35 - 44 10 39.5 34.5 - 44.5 0.20 20% 17 43

45 - 54 15 49.5 44.5 - 54.5 0.30 30% 32 33

55 - 64 12 59.5 54.5 - 64.5 0.24 24% 44 18

21
65 - 74 4 69.5 64.5 - 74.5 0.08 8% 48 6

75 - 84 2 79.5 74.5 - 84.5 0.04 4% 50 2

Solution:
Adding two class marks with f i = 0, we have 9.5 at the beginning, and 89.5 at the end, the
following frequency polygon is plotted.

Frequency Polygon
16
14
F 12
r
e 10
q 8
u
e 6
n 4
c
y 2
0
9.5 19.5 29.5 39.5 49.5 59.5 69.5 79.5 89.5

Class mark

Ogive (cumulative frequency polygon)

An Ogive (pronounced as “oh-jive”) is a line that depicts cumulative frequencies, just as the
cumulative frequency distribution lists cumulative frequencies. Note that the Ogive uses class
boundaries along the horizontal scale, and graph begins with the lower boundary of the first class
and ends with the upper boundary of the last class. Ogive is useful for determining the number of
values below some particular value. There are two type of Ogive namely less than Ogive and
more than Ogive. The difference is that less than Ogive uses less than cumulative frequency and
more than Ogive uses more than cumulative frequency on y-axis.

Example 2.10: i) Draw a less than Ogive for data of blood glucose level of the 60 patients given
in Example 2.3.

22
 The less than Ogive

70
60
ts
50
ien
40
pat
of 30
er 20
mb 10
Nu
0
54.5 63.5 72.55 81.5 90.5 99.5 108.5 117.5

blood glucose level ( upper class boundary)

ii) Draw a more than Ogive for the F.D. of Example 2.9.

The More than Ogive

uenc
Cum

Freq
ulati
ve

60
y

50
40
30
20
10
0
14.5 24.5 34.5 44.5 54.5 64.5 74.5 84.5
Class Boundaries

Note: For both ogives, one class with frequency zero is added for similar reason with the
frequency polygon.

Exercise 2

1. Classify the following as discrete or continuous variable.

a)Temperature; b) number of courses offered in DMU; c) rain fall; d) age.

2. Distinguish between primary and secondary data. What precautions should be taken
before using secondary data?
3. Construct a frequency distribution for a survey taken at a hotel, that 40 tourists arrived by
the following means of transportation:
car car bus plane plane car plane plane bus car plane car car car
plane bus car bus car plane car car car bus car bus bus
plane plane plane car plane plane plane bus bus car car plane car
23
 4. The weight of 30 students on a College of Natural and Computational Science was
recorded as follows:
143 151 104 99 121 126
138 119 104 112 132 123
121 133 137 132 126 107
139 122 127 90 129 134
133 136 113 112 140 123
Construct a frequency distribution with intervals of 7 classes.

5. Given the following frequency distribution:

Class limits 0-1 2-3 4-5 6-7 8-9

Frequency 16 25 13 4 2

Find a) the class marks; b) the class boundaries; c) the relative frequencies

24
 CHAPTER 3: MEASURES OF CENTRAL TENDENCY
Objectives
At the end of this chapter students will be able to:
 Identify measure of central tendency
 understand properties of arithmetic mean
 Summarize an aggregate of statistical data by using single measure
 Define and calculate the mean, mode and median.
 Measure the position of data using quartiles, deciles and percentiles with their
interpretation.
3.1 The Summation Notation (¿

Statistical Symbols: Let a data set consists of a number of observations, represents by x 1 , x 2

, … , x n where n (the last subscript) denotes the number of observations in the data and x i is the ith
observation. Then the sum of all numbers ( x ¿¿ i' s)¿where i goes from 1 up to n is symbolically
n
given by ∑ x i∨∑ x i ¿ ∑ x that is
i=1

∑ x i = x 1 + x 2 +…+ xn
x - whole set of numbers
x i - specific score in a set of numbers
n - total number of observations.
Example 3.1: For instance a data set consisting of six measurements 2, 3, 9, 10, 8 and -2 is
represented by x 1 , x 2 , … , x 6 where x 1=2, x 2 =3, x 3=¿ 9, x 4 = 10, x 5= 8 and x 6=-2 Their sum
6
becomes ∑ x i = x 1 + x 2 +…+ x6 = 2+3+9+10+8+ (-2) = 30
i=1

Some Properties of the Summation Notation

25
 n
1. ∑ c = n.c, where c is a constant number.
i=1

n n
2. ∑ bxi = b∑ x i where b is a constant number
i=1 i=1

n n
3. ∑ (a+ bx¿¿ i)¿ = n.a + b∑ x i
i=1 i=1

n n n
4. ∑ (x ¿¿ i± y ¿¿i) ¿¿ = ∑ x i ± ∑ yi
i=1 i=1 i=1

n n n
5. ∑ xi yi ≠ ∑ xi ∑ yi
i=1 i=1 i=1

Example 3.2:
7 7 7 7

∑ x i = 20 , ∑ yi = 30, ∑ x 2i = 420, ∑ y2i =280

i=1 i=1 i=1 i=1

7 7 7
Find i/ ∑ ¿¿ + 4 y i ¿ = 6 ∑ x i + 4∑ yi = 6.20 + 4.30 = 240
i=1 i=1 i=1
7 7
ii/ 3∑ x i −2 ∑ y i = 3.420 – 2.280 = 700
2 2

i=1 i=1

3.2 Properties of measures of central tendency

A good average should be:

1. Rigidly defined (unique).

2. Based on all observation under investigation.

3. Easily understood.

4. Simple to compute.

5. Suitable for further mathematical treatment.

6. Little affected by fluctuations of sampling.

7. Not highly affected by extreme values.

26
3.3 Types of Measures of Central Tendency
Measures of Central Tendency: give us information about the location of the center of the
distribution of data values. A single value that describes the characteristics of the entire mass of
data is called measures of central tendency. We will discuss briefly the three measures of central
tendency: mean, median and mode in this unit.
The following are types of central tendency which are suitable for a particular type of data.
These are
 Arithmetic Mean
- Weighted Arithmetic Mean
- Combined mean
 Geometric Mean
 Harmonic Mean
 Median
 Mode or modal value
3.3.1 Arithmetic Mean
Arithmetic mean:- is defined as the sum of the measurements of the items divided by the total
number of items. It is usually denoted by x .
Arithmetic Mean for individual series
Suppose x 1 , x 2 , … , x n are observed values in a sample of size n from a population of size N,
n<N then the arithmetic mean of the sample, denoted by x is given by
n

x = x 1+
x 2+¿ …+ x
n
¿=
∑ xi
i=1
n
n
If we take an entire population the Mean is denoted by μ and is given by:
N

μ = X1+
X 2 +¿…+ X N
¿=
∑ Xi
i=1
N
N
Where N stands for the total number of observations in the population.
Example 3.3: The number of flowers per plant is given below. Find the mean.
i. 5 12 9 6
ii. 6 8 6 7 8
Find the arithmetic mean
Solution:
i. The sample values are: 5 12 9 6

27
 n

x= ∑ xi = 5+12+ 9+6 = 32 = 8
i=1
4 4
n

The arithmetic mean for sample value is 8.

ii. The sample values are: 6 8 6 7 8
n

x=∑ i =
x 6+8+6 +7+8 35
= =7
i=1
5 5
n
The arithmetic mean for sample value is 7.

Arithmetic mean for discrete data arranged in frequency distribution

When the numbers x 1 , x 2 , … , x k occur with frequencies f 1 , f 2 , … , f k, respectively, then the

mean can be expressed in a more compact form as:
k

x1 f 1 + x 2 f 2 +…+ x k f k
∑ xi f i
i=1
x= = k
f 1 + f 2+ …+f k
∑ fi
i=1

Example 3.4
Calculate the arithmetic mean of the pulse rates (beats per minute) of eleven students:
60 60 71 68 71 72 71 76 72 80 80
n

x = ∑ xi = 60+60+…+ 80 = 781 = 71
i=1
11 11
n
In this case there are two 60’s, one 68, three 71’s, two 72’s, one 76, and two 80’s. The number of
times each number occurs is called its frequency and the frequency is usually denoted by f. The
information in the sentence above can be written in a table, as follows.
Value, xi 60 68 71 72 76 80 Total
Frequency, fi 2 1 3 2 1 2 11
xi fi 120 68 213 144 76 160 781

The formula for the arithmetic mean for data of this type is
k

x1 f 1 + x 2 f 2 +…+ x k f k
∑ xi f i
i=1
x= = k
f 1 + f 2+ …+f k
∑ fi
i=1

In this case we have:

28
 781
x= = 71.
11
The mean pulse rate (beats per minute) of the eleven students is 71.

Arithmetic Mean for Grouped Continuous Frequency Distribution

If data are given in the form of continuous frequency distribution, the sample mean can be
computed as

∑ xi f i x 1 f 1 + x 2 f 2+ …+ x k f k
i=1
x= = where x i is the class mark of the ith class; i=1, 2, . . . , k
k
f 1+ f 2 +…+ f k
∑ fi
i=1

f i is the frequency of the ith class and k is the number of classes

k
Note that ∑ f i = n = the total number of observations.
i=1

Example 3.5

The following frequency table gives the height (in inches) of 100 students in a college.
Class boundary 60-62 62-64 64-66 66-68 68-70 70-72 Total
Frequency (fi) 5 18 42 20 8 7 100
Calculate the mean
Solution:
The formula to be used for the mean is as follows:
k

∑ xi f i
i=1
x= k

∑ fi
i=1

Let us calculate these values and make a table for these values for the sake of convenience.

Class boundary Frequency Mid-Point ( x i f i xi

)
29
 (fi)
60 - 62 5 61 305
62 – 64 18 63 1134
64 – 66 42 65 2730
66 – 68 20 67 1340
68 – 70 8 69 552
70 – 72 7 71 497
Total 100 6558

6
Substituting these values with ∑ f i = 100, we get
i=1

∑ xi f i 6558
i=1
x= =x= = 65.58
k
100
∑ fi
i=1

The mean height of students is 65.58.

Properties of the Arithmetic Mean
 The algebraic sum of the deviations of a set of numbers x 1 , x 2 , … , x nfrom their mean x is
always zero. i.e.

2
σ
 The sum of squares of deviations from the mean is the least. That is,
μ is

minimum when ( ).
PQ N −n
n N−1

 If the mean of x 1 , x 2 , … , x n is x , then

a) The mean of x 1 ± k, x 2± k ,..., x n± k will be x ± k
b) The mean of kx 1 , kx 2 , … , kx n will be k x .
Merits of Arithmetic Mean
 Arithmetic mean has a rigidly defined mathematical formula so that its value is always
definite or unique. It can be calculated for any set of numerical data.
 It is calculated based on all observations.
 Arithmetic mean is simple to calculate and easy to understand.
 It doesn’t need arrangement of data in increasing or decreasing order.

30
  Arithmetic mean of many samples from the same population does not fluctuate
considerably.
 It affords a good standard of comparison.
Demerits of Arithmetic Mean
 It can’t be calculated for data which are not quantifiable.
 It is highly affected by extreme (abnormal) values in the series.
 It can be a number which does not exist in the series.
 It can’t be calculated for grouped continuous open-ended classes.
Weighted Arithmetic Mean
While calculating simple arithmetic mean, all items were assumed to be of equally importance
(each value in the data set has equal weight). When the observations have different weight, we
use weighted average. Weights are assigned to each item in proportion to its relative importance.
If x 1 , x 2 , … , x n represent values of the items and w 1 , w 2 , … , w nare the corresponding weights,
then the weighted mean, ( x w ) is given by

PQ
n ( N −n
N −1 )
Example 3.6
A student’s final mark in Mathematics, Physics, Chemistry and Biology are respectively A, B, D
and C. If the respective credits received for these courses are 4, 4, 3 and 2, determine the
approximate average mark the student has got for the course.
Solution
We use a weighted arithmetic mean, weight associated with each course being taken as the
number of credits received for the corresponding course.
xi 4 3 1 2 Total
wi 4 4 3 2 13
x i wi 16 12 3 4 35

pq
n−1 ( N −n
N )
= 16+12+3+4 = 35 = 2.69
13 13

31
Average mark of the student is approximately 2.69.

Combined mean:-When a set of observations is divided into k groups and x 1is the mean of n1
observations of group 1, x 2 is the mean of n2 observations of group2, …, x k is the mean of nk
observations of group k , then the combined mean ,denoted by x c, of all observations taken
together is given by

x x 1 n1+ x 2 n2 +…+ xk nk
c=¿ ¿
n1+n2+ …+ nk

This is a special case of the weighted mean. In this case the sample sizes are the weights.

Example 3.7
In the Previous year there were two sections taking Statistics course. At the end of the semester,
the two sections got average marks of 70 & 78. There were 45 and 50 students in each section
respectively. Find the mean mark for the entire students.
Solution:

x x1 n1 + x 2 n 2 70 x 45+78 x 50 7050
c=¿
x 1 n1+ x 2 n2 +…+ xk nk
¿ = = = = 74.21
n1+n2+ …+ nk n1 +n2 45+50 95

The combined mean of the entire students will be 74.21.

3.3.2 Geometric Mean

The geometric mean like arithmetic mean is calculated average. It is used when observed values are
measured as ratios, percentages, proportions, indices or growth rates.

Geometric mean for individual series:- The geometric mean, G.M. of an individual series of
positive numbers x 1 , x 2 , … , x n is defined as the nth root of their product.
pq 1
( 1 −f )
n−1 = antilog ( n ∑ logx i )

Example 3.8

Find the G. M of (a) 3 and 12 b) 2, 4 and 8

2
σ
; b) GM= √ 2 x 4 x 8 ¿ √ 64 = 4
3 3
Solution: a)

Geometric mean for discrete data arranged in FD:- When the numbers x 1 , x 2 , … , x k occur
with frequencies f 1 , f 2 , … , f m, respectively, then the geometric mean is obtained by

32
 npq 1
n− 1 = antilog ( n ∑ f i logx i)
Example 3.9
Compute the geometric mean of the following values: 3, 3, 4, 4, 4, 5, 6 and 6.
Solution
Values 3 4 5 6
Frequency 2 3 1 2

G.M. = √8 32 X 4 3 X 51 X 62 = 4.236
The geometric mean for the given data is 4.236.
Geometric mean for continuous grouped FD: The above formula can also be used whenever
the frequency distribution is grouped continuous, class marks of the class intervals are
considered as xi.
Properties of geometric mean
 It is less affected by extreme values.
 It takes each and every observation into consideration.
 If the value of one observation is zero its values becomes zero.
3.3.3 Harmonic Mean
It is a suitable measure of central tendency when the data pertains to speed, rate and time. The
harmonic of n values is defined as n divided by the sum of their reciprocal.
Harmonic mean for individual series:- If x 1 , x 2 , … , x n are n observations, then harmonic mean
can be represented by the following formula:

pq
n−1
Example 3.10: A car travels 25 miles at 25 mph, 25 miles at 50 mph, and 25 miles at 75 mph.
Find the harmonic mean of the three velocities.
Solution

pq 3
n−1 =
1 1 1
+ + = 40.9.
25 50 75

Harmonic mean for discrete data arranged in FD:- If the data is arranged in the form of
frequency distribution

33
 PQ N −n
n N−1
. ( ) pq (1−f )
, where n−1

Harmonic mean for continuous grouped FD: Whenever the frequency distribution are
grouped continuous, class marks of the class intervals are considered as x i and the above formula
can be used as
n (0.37)(0.63)(1−0.2)
H.M. =
n
f where 99
∑ xi
i=1 i

x i is the class mark of ith class

Properties of harmonic mean
 It is unique for a given set of data.
 It takes each and every observation into consideration.
 Difficult to calculate and understand.
 Appropriate measure of central tendency in situations where data is in ratio, speed or rate.
Relations among different means
i. If all the observations are positive we have the relationship among the three means given as:
x ≥ GM ≥HM
ii. For two observations √ x∗HM ¿ GM
iii. x = GM = HM if all observation are positive and have equal value.
3.3.4 Median
The median is as its name indicates the middle most value in the arrangement which divides the
data into two equal parts. It is obtained by arranging the data in an increasing or decreasing order
of magnitude and denoted by ~ x.
Median for individual series:-We arrange the sample in ascending order of the variable of
interest. Then the median is the middle value (if the sample size n is odd) or the average of the
two middle values (if the sample size n is even).
For individual series the median is obtained by
a/ ~
x = ¿ value if n is odd, and
n th n th
( ) value+( +1) value
b/ ~
x= 2 2 if n is even
2
Example 3.11
Find the median for the following data.

34
a/ -5 15 10 5 0 2 1 4 6 and 8
b/ 5 2 2 3 1 8 4

Solution;
i. The data in ascending order is given by:
-5 0 1 2 4 5 6 8 10 15
n=10 n is even. The two middle values are 5th and 6th observations. So the median
is,
10 th 10 th

~ )( value+( + 1) value 5th value+ 6th value 4+5

x= 2 2 = = =4.5.
2 2
2
ii. The data in ascending order is given by:
1 2 2 3 4 5 8
The middle value is the 4th observation. So the median is 3.
Note: The median is easy to calculate for small samples and is not affected by an "outlier".
Median for Discrete data arranged in a frequency distribution:- In this case also, the median
is obtained by the above formula. After arranging the value in an increasing order find the
smallest CF greater than or equal to that value obtained by formula and the corresponding value
is the median.

Median for grouped continuous data:-For continuous data, the median is obtained by the
following formula.

1−α=0 . 95
Where: L= the lower class boundary of the median class; w = the class width of the median class;

α/2=0.005 n=15, σ=0.27 ounce , x̄=9.87 ounce

= the frequency of the median class; and the cum. freq. corresponding to the class
preceding the median class. That is, the sums of the frequencies of all classes lower than the
median class. Where the median class is the class which contains the (n/2) th observation whether n
is odd or even, since the items have already lost their originality once they are grouped in to
continuous classes.

Example 3.12: Water percentage in the body of species of Fish is given below. Calculate the
median.

35
 C.I 15-24 25-34 35-44 45-54 55-64 Total

Freq. 7 17 16 6 4 50

Solution: Construct the less than cumulative frequency distribution, then:

C.I 15-24 25-34 35-44 45-54 55-64 Total

Freq. 7 17 16 6 4 50

Cuml. Freq. 7 24 40 46 50

Since n = 50, 50/2 = 25, and the smallest CF greater than or equal to 25 is 40; thus, the median class
P(−Z0.025<Z<Z0.025)=0.95
is the third class. And for this class, L = 34.5, w = 10, =16, CF = 24. Then applying the
formula, we get:

±Zα/2=±Z0. 25=±1.96
=34.5+ (25-24)*10/16 = 35.1.

Merits of median

• It is less affected by extreme values.

• Median can be calculated even in case of open-ended intervals.
• It can be computed for ratio, interval, and ordinal level of data.
Demerits of median
• Its value is not determined by each & every observation.
• It is not a good representative of the data if the number of items (data) is small.
• The arrangement of items in order of magnitude is sometimes very tedious process if the
number of items is very large.
3.3.5 The Mode or modal value
The mode or the modal value is the value with the highest frequency and denoted by x^ . A data
set may not have a mode or may have more than one mode. A distribution is called a bimodal
distribution if it has two data values that appear with the greatest frequency. If a distribution has

36
more than two modes, then the distribution is multimodal. If a distribution has no modes, then
the distribution is non-modal.

Mode of individual series: The mode or the modal value of individual series (raw data) is simply
obtained by locating the observation with the maximum frequency.

Example 3.13
Consider the following data:
a. 30 45 69 70 32 18 32. The Mode ( ^x ) = 32.
b. 10 20 30 10 40 30. The Mode ( ^x ) = 10 and 30.
c. 10 40 30 20 50 60. No Mode.
Note that in some samples there may be more than one mode or there may not be a mode. The
mode is not a suitable measure of central tendency in these cases. We use the mode as a measure
of central tendency if we require a measure that takes on one of the sample values. The mode can
be used for variables that are measured on a category (nominal) scale. e.g. the most popular
computer type.

Mode for discrete data arranged in a frequency distribution:-In the case of discrete grouped
data, the mode is determined just by looking to that value (s) having the highest frequency.

Mode for Grouped Continuous Frequency Distribution

For grouped data, the mode is found by the following formula. In such cases, one can only
determine the modal class easily i.e. the class with the highest frequency. After locating this
class, the mode is interpolated using:
X̄ − μ
Z=
σ / √ n , where L = the lower class boundary of the modal class;
x̄−Z α/2×
σ
√n
<μ< x̄+Z α /2 ×
σ
√n , σ , w = the common class width, Z0.025=1.96 = frequency of the class

immediately preceding the modal class; E=0.25 = frequency of the class immediately succeeding the
modal class; and fmode = frequency of the modal class.
Example 3.14
Calculate the mode for the frequency distribution (water percentage) of the data on example
3.12.

Solution: By inspection, the mode lies in the second class, where L =24.5, fmod = 17, f1= 7, f2=16, w
= 10. Using the formula, the mode is:

37
 σ =1. 50 = 24.5 + (17-7)*10/[(17-7)+(17-16)] = 33.59.

Merits of mode

 Mode is not affected by extreme values.

 We can change the size of the observations without changing the mode.
 It can be computed for all level of data. i.e., ratio, interval, ordinal or nominal.

Demerits of mode
 It may not exist
 It does not take every value into consideration.
 Mode may not exist in the series and if it exists it may not be unique
3.4. The Relationship of the Mean, Median and Mode
Comparing the Mean, Median, and the Mode
 If the data is skewed, avoid the mean.
 If there is high gap around the middle, avoid the median.
 A measure is a resistant measure if its value is not affected by an outlier or an extreme
data value.
 The mean is not a resistant measure of central tendency because it is not resistant to the
influence of the extreme data values or outliers.
 The median is resistant to the influence of extreme data values or outliers and its value
does not respond strongly to the changes of a few extreme data values regardless of how
large the change may be.
 The mode has an advantage over both the mean and the median when the data is
categorical since it is not possible to calculate the mean or median for this type of data.
Also, the mode usually indicates the location within a large distribution where the data
values are concentrated. However, the mode can not always be calculated because if a
distribution has all different data values, then the distribution is non-modal.
 In the case of symmetrical distribution; mean, median and mode coincide. That is,
mean=median = mode. However, for a moderately asymmetrical (non symmetrical)

38
 distribution, mean and mode lie on the two ends and median lies between them and they
have the following important empirical relationship, which is
(Mean – Mode) = 3(Mean - Median).
Example 3.15
In a moderately asymmetrical distribution, the mean and the mode are 30 and 42 respectively.
What is the median of the distribution?
Solution:
Median = (2mean + Mode)/2 = (2*30 + 42)/3 = 34
Hence the median of the distribution is 34.

Which of the Three Measures is the Best?

At this stage, one may ask as to which of these three measure of central tendency is the best.
There is no simple answer to this question. It is because these three measures are based upon
different concepts. The arithmetic mean is the sum of the values divided by the total number of
observations in the series. The median is the value of the middle observations tend to
concentrate, As such; the use of a particular measure will largely depend on the purpose of the
study and the nature of the data. For example, when we are interested in knowing the consumers’
preferences for different brands of television sets or kinds of advertising, the choice should go in
favor of mode. The use of mean and median would not be proper. However, the median can
sometimes be used in the case of qualitative data when such data can be arranged in an ascending
or descending order. Let us take another example. Suppose we invite applications for a certain
vacancy in our University. A large number of candidates apply for that post. We are now
interested to know as to which age or age group has the largest concentration of applicants. Her,
obviously the mode will be the most appropriate choice. The arithmetic mean may not be
appropriate as it may be influenced by some extreme values.
3.5 Measures of Non-central Locations
Median is the value of the middle item which divides the data in to two equal parts and found by
arranging the data in an increasing or decreasing order of magnitude, where as quintiles are
measures which divides a given set of data in to approximately equal subdivision and are
obtained by the same procedure to that of median. They are averages of position (non-central
tendency). Some of these are quartiles, deciles and percentiles.

39
 Quartiles: are values which divide the data set in to approximately four equal parts, denoted by
Q 1 , Q2∧Q3 . The first quartile (Q1) is also called the lower quartile and the third quartile ( Q 3) is
the upper quartile. The second quartile (Q2) is the median.
• Quartiles for Individual series:

n n
Let be n ordered observations. The ith quartile is the value of the item
corresponding with the [i(n+1)/4]th position, i = 1, 2, 3.

That is, after arranging the data in ascending order, Q1, Q2, & Q3 are, obtained by:

( ) ( ) ( ) value.
th th th
1( n+1) 2(n+1) 3(n+1)
Q 1= value , Q2= value and Q3=
4 4 4

• Quartiles for discrete data arranged in a frequency distribution:-Arranged in a frequency

distribution this case also, we will follow the same procedure as the median. That is, we construct
the less than cumulative frequency distribution and apply the formula of quartile for Individual
series.

• Quartiles in continuous data:- For continuous data, use the following formula:

=138.30≈139
Where i = 1, 2, 3, and L, w ,fQi and CF are defined in the same way as the median.

i.e. Q1 = L +
w n
f Q1 4 (
−CF , ) Q2 = L + (
w 2n
f Q2 4 )
−CF ∧¿ Q3 = L +
w 3n
f Q3 4
−CF ( )
The class under question is the one including (ixn/4)th value. That is, the class with the minimum
cumulative frequency greater than or equal to (ixn/4) th is the class of the ith quartile.

Deciles: are values dividing the data approximately in to ten equal parts, denoted by D 1 , D2 , …, D 9
.
• Deciles for Individual series:

α (D¿¿ i)¿
Let be n ordered observations. The ith decile is the value of the item
corresponding with the [i(n+1)/10]th position, i = 1, 2, . . . ,9.

That is, after arranging the data in ascending order, D1, D2, . . . & D9 are, obtained by:

40
 ( ) ( ) ( ) value.
th th th
1(n+1) 2(n+1) 9(n+ 1)
D 1= value , D2= value . . . and D9=
10 10 10

• Deciles for Discrete data arranged in a frequency distribution:-Arranged in a frequency

distribution this case also, we will follow the same procedure as the median. That is, we construct
the less than cumulative frequency distribution and apply the formula of deciles for individual
series.

• Deciles for continuous data: Apply the following formula and follow the procedures of quartile
for continuous data.

w ¿
D i=L+
f D 10
i
( )
−CF , i = 1, 2,...,9 . Then

Define the symbols similar ways as we did in the case of quartiles for continuous data.
Percentiles: are values which divide the data approximately in to one hundred equal parts, and
denoted by P1 , P2 , …, P99 .

• Percentiles for Individual series:

(α=0 . 05) (P¿¿ i) ¿

Let be n ordered observations. The ith percentile is the value of the item
corresponding with the [i(n+1)/100]th position, i = 1, 2, . . . ,99.

That is, after arranging the data in ascending order, P1, P2, . . . & P99 are, obtained by:

( ) ( ) ( ) value.
th th th
1(n+1) 2(n+1) 99(n+1)
P1= value , P2= value . . . and P99=
100 100 100

• Percentiles for Discrete data arranged in a frequency distribution:-Arranged in a frequency

distribution this case also, we will follow the same procedure as the median. That is, we construct
the less than cumulative frequency distribution and apply the formula of percentile for individual
series.

Percentiles for continuous data: Apply the following formula

❑❑
•

w ¿
Pi=L+
f P 100
i
(,
)
−CF i = 1, 2,...,99.

41
 Define the symbols similar ways as we did in the case of quartiles or deciles for continuous data.
Interpretations
1. Qi is the value below which ( i × 25) percent of the observations in the series are found (where i
= 1, 2,3). For instance Q3 means the value below which 75 percent of observations in the given
series are found.
2. Di is the value below which ( i ×10) percent of the observations in the series are found (where i =
1, 2,...,9 ). For instance D 4 is the value below which 40 percent of the values are found in the
series.
3. Pi is the value below which i percent of the total observations are found (where i = 1, 2,3,...,99 ).
For example 60 percent of the observations in a given series are below P60.
Example 3.16: Calculate Q1 , Q2 ,Q3 , D4 , D9 , P40∧P 90for the following tables.
x 10 11 12 13 14 15 16 17 18
f 2 8 25 48 65 40 20 9 2

Solution: The given data is measured and it is arranged in an increasing order. So we need to
construct only the cumulative frequency table before calculating the required values.

x 10 11 12 13 14 15 16 17 18
f 2 8 25 48 65 40 20 9 2
Cum. 2 10 35 83 148 188 208 217 219
Freq.

The total number of observations is 219 which is odd. Clearly then the median is 14 because
~
x = ¿ = ¿value = 110th value = 14

( ) (
) value = 55 value = 13
th th
1 ( n+1 ) 1 ( 219+1 )
Q 1= value = th
4 4

¿ Q =( ) value = ( ) value = 110 value = 14 = ~x

th th
2( n+1) 2 (219+ 1 ) th
2
4 4

Q =( ) value = ( ) value = 165 value = 15

th th
3(n+1) 3 ( 219+1 ) th
3
4 4

D =( ) value = ( ) value = 88 value = 14

th th
4(n+ 1) 4 (219+1) th
4
10 10

42
 ( ) ( ) value = 198 value = 16
th th
9(n+ 1) 9(219+1)
D 9= value = th
10 10

P =( ) value = ( ) value = 88 value = 14

th th
40( n+1) 40( 219+1) th
40
100 100

P =( ) value = ( ) value = 198 value = 16.

th th
90(n+1) 90(219+1) th
90
100 100
Example 3.17: Values of fecundity (rate of reproduction) of 50 Fish of a species of Fish is given
below. Based on the data find Q 1, D4 ∧P7.
rate of 1-10 11-20 21-30 31-40 41-50 51-60 61-70 71-80 Total
reproduction
f 3 11 7 4 15 0 7 3 50
Solution:- first find the class boundaries and cumulative frequency distributions.
rate of 1-10 11-20 21-30 31-40 41-50 51-60 61-70 71-80 Total
reproduction
f 3 11 7 4 15 0 7 3 50
Cumulative 3 14 21 25 40 40 47 50
freq. dist.

Q1 Measure of (n/4)th value = 12.5th value which lies in group 10.5 – 20.5

Q1 = L +
w n
f Q1 4 ( 10
)
−CF = 10.5 + ( 12.5−3 ) = 19.1
11
D4 Measure of (4n/10)th value = 20th value which lies in group 20.5 – 30.5.

D4 = L +
w
f D4 ( 410n −CF ) = 20.5 + 107 ( 20−14) = 29.1
P7 Measure of (7n/100)th value = 3.5th value which lies in group 10.5 – 20.5

P7 = L +
w 7n
(
f P 7 100
10
)
−CF = 10.5 + ( 3.5−3 ) = 11.
11
Exercise- 3

1. A random sample of 10 days gave the following information about the total number of
people treated per day at a community hospital emergency room (ER)
40,8,26,20,30,27,60,13,119,40
Find a/ the mean b/ the mode C/ the median
2. The circumferences (distance around) of 100 oak trees selected at random in a state park
were measured. The FD is given below
43
 Circumference (inch) No of trees (f)
10-24 8
25-39 22
40-54 53
55-69 17
Find the a/ mean, b/ mode, c/ median, d/ 1st quartile, e/ P2, f/ D8 , g/ D9

3. The Frequency of flowers on 30 plants is given below

Number of flowers 1 2 3 4 5 6 7 Total

Fi 1 4 12 9 2 1 1 30

Find the a/ mode, b/ median, c/ mean, d/ Q2, e/D6 , f/ P25,

4. The nose length of forty students in a class (measured) to the nearest millimeter) are given
below.
40 43 51 49 50 43 41 43 46 45 40 38 45 46 40 38 40 43 50 35 41 32 40
45 62 39 46 45 44 50 52 39 45 35 51 48 45 55 32 45
Make a frequency table with width five to organize the above data set.
5. The following table contains number of leaves on 20 plants.

No of leaves 5 12 15 20 22 Total
Fi 3 5 7 3 2 20

Find the a/ mean, b/mode, c/ median, d/D7, e/P80

6. A random sample of 10 insects had the following weight (in mlgms)

70 120 110 101 88 83 95 98 107 100
Find the a/ man b/ median c/ mode

7. The following data represent the diagnosis of patients admitted to a hospital.

Cancer Heart failure Diabetics Accident Gunshot
wound Diabetics Gunshot wound diabetics Gunshot
wound Accident Diabetics Cancer
Cancer Diabetics TB Diabetics TB
Determine the mode diagnosis.
8. The blood type of patients is given below

O A O AB AB O A B AB A

44
 A O B O O O A B AB B

Determine the mode of blood type.

45
 CHAPTER 4: MEASURES OF DISPERSION (VARIATION)
4.1 Introduction
Just as central tendency can be measured by a number in the form of an average, the amount of
variation (dispersion, spread, or scatter) among the values in the data set can also be measured.
The measures of central tendency describe that the major part of values in the data set appears to
concentrate around a central value called average with the remaining values scattered
(distributed) on either sides of that value. But these measures do not reveal how these values are
dispersed (spread or scatter) on each side of the central value. The dispersion of values is
indicated by the extent to which these values tend to spread over an interval rather than cluster
closely around an average.
The term dispersion is generally used in two senses. Firstly, dispersion refers to the variations of
the items among themselves. If the value of all the items of a series is the same, there will be no
variation among different items of a series. Secondly, dispersion refers to the variation of the
items around an average. If the difference between the value of items and the average is large,
the dispersion will be high and on the other hand if the difference between the value of the items
and averaging is small, the dispersion will be low. Thus, dispersion is defined as scatteredness or
spreadness of the individual items in a given series.

After studying this chapter, you should be able to:

 Explain the meaning of measures of dispersion

 Compare two or more sets of data using relative measures of dispersion.

 Apply the Z-score to find out the relative standing of values.
 Explain measures of skewness and kurtosis.
Objectives of measuring Variation:
 To judge the reliability of measures of central tendency
 To control variability itself.
 To compare two or more groups of numbers in terms of their variability.
 To make further statistical analysis.
4.2 Absolute and Relative Measures of Dispersion
Absolute measures of dispersion: Absolute measure is expressed in the same
statistical unit in which the original data are given such as kilograms, tones etc. These
measures are suitable for comparing the variability in two distributions having variables
expressed in the same units and of the same averaging size. These measures are not

46
 suitable for comparing the variability in two distributions having variables expressed in
different units.

Absolute
measure of
dispersion

Based on Based on all

selected items items

Mean deviation
Range & Inter-
& Standard
quartile range deviation

Relative measures of dispersion: A relative measure of dispersion is the ratio of a measure of

absolute dispersion to an appropriate average or the selected items of the data.

Relative
measure of
dispersion

Based on selected
items Based on
all items

Coefficient of range Coefficient of mean

and coefficient of deviation
quartile deviation &coefficient of
standard deviation
or coefficient of
variation

47
4.3 Types of Measures of Variation

4.3.1 The Range and Relative Range

Range is the simplest measures of dispersion. It is defined as the difference between the largest
and smallest value in a given set of data. Its formula is:

R=L−S

Where R=Range, L= Largest value in a given set of data, S= smallest value in a given set of data

For a continuous grouped distribution, the range may be obtained as:

 The difference between upper class limit of the last class and the lower class limit of the
first class, or

 The difference between the largest class mark and the smallest class mark, or

 The difference between the upper class boundary of the last class and the lower class
boundary of the first class.

The range is used in describing like the maximum change in daily temperature, rainfall, etc.
When the sample size is small, it can be an adequate measure of variation. It is commonly used
in quality control.

The relative measures of range, also called coefficient of range, is defined as

Relative Range(RR )=¿

Example 4.1: Five students obtained the following marks in statistics:20 , 35 ,25 , 30 , 15. Find
the range and relative range

Solution: Here, L=35 ,∧S=15

Range=L−S=35−15=20

α
RR
= 2

Example 4.2: Find out range and relative range of the following given data.

48
 Size 5-10 11-15 16-20 21-25 26-30

Frequency 4 9 15 30 40

Solution: Here, L = Upper class limit of the largest class = 30, S = lower class limit of the
smallest class = 5

Range = 30 – 5 = 25

RR=¿ .

Merits of the Range

 It is well-defined, easy to compute and simple to understand.

 It helps in giving an idea about the variation, just by giving the lowest value and the
greatest value of variable.

Demerits of the Range

 It is not based on all observations of the series.

 It can’t be calculated in case of open-ended distribution.
 It is affected by sampling fluctuation.
 It is affected by extreme values in the series.

4.3.2 The Quartile Deviation and Coefficient of Quartile Deviation

Inter-quartile range and quartile deviation are other measures of dispersion. The difference
between the upper quartile ( Q3 ) and lower quartile ( Q1 ) is called inter-quartile range.
Symbolically,

I nter Quartile R ange(IQR)=Q 3−Q 1.

The inter-quartile ranges covers dispersion of middle 50% of the items of the series. Quartile
deviation, also called semi-inter-quartile range, is half of the difference between the upper and
lower quartile. That is, half of the inter-quartile range. Its formula is

49
 Q3 −Q1
Quartile Deviation ( QD ) =
2

The relative measure of quartile deviation also called the coefficient of quartile deviation (CQD)
is defined as:

Q3 −Q1
CQD=
Q3 +Q1

Example 4.3: Find inter-quartile range, quartile deviation and coefficient of quartile
deviation from the following age of patients.

18, 59, 24, 42, 21, 23, 24, 32

Solution: First arrange the data in ascending order. 18, 21, 23, 24, 24, 32, 42, 59

( ) ( ) item = (2.25)
th th
n+1 8+ 1
Q1=¿ item=¿ th
item= 2 nd item + 0.25(3 rd item - 2 nd item) = 21 +
4 4
0.25(23 - 21) = 21.5

( ) ( ) item = (6.75)
th th
n+ 1 8+1
Q 3=¿ 3 item=¿ 3 th
item = 6 th item + 0.75(7 th item - 6 th item) = 32 +
4 4
0.75(42 - 32) = 39.5

Q3−Q1 39.5−21.5
IQR=Q3 −Q1=39.5−21.5=18, QD= = =9
2 2

Q 3 −Q 1 39.5−21.5 18
CQD= = = =0.295.
Q 3 +Q 1 39.5+21.5 61

Example 4.4: Find inter-quartile range, quartile deviation and coefficient of quartile
deviation from the following data.

Marks 2 3 4 5 6 7 8 9
No. Of students 10 11 12 13 5 12 7 5

Solution:

50
 Marks No. Of students CF

2 10 10
3 11 21
4 12 33
5 13 46
6 5 51
7 12 63
8 7 70
9 5 75=N

Q 1= ( N4+ 1 )= 75+1
4
th
=19 item=3

Q3=3 ( N4+1 )=3( 75+1

4 )
=57

IQR=Q3 −Q1=7−3=4

Q3−Q1 7−3
QD= = =2
2 2

Q 3 −Q 1 7−3
CQD= = =0.4 .
Q 3 +Q 1 7+ 3

Remark: Q.D or CQD includes only the middle 50% of the observation.

Merits of QD

 It is well-defined, easy to compute and simple to understand.

 It helps in studying the middle 50% item in the series.
 It is not affected by the extreme items.
 It is useful in measuring variations in the case of open-ended distributions.

Demerits of QD

 It is not based on all the items (it ignores 50% items, i.e., the first 25% and the last
25%).
 It is greatly influenced by sampling fluctuations.

51
  It is not amenable to algebraic manipulations.

4.3.3 The Mean Deviation and Coefficient of Mean Deviation

The mean deviation (MD) measures the average deviation of a set of observations about their
central value, generally the mean or the median, ignoring the plus/minus sign of the deviations.
In other words the mean deviation of a set of items is defined as the arithmetic mean of the
values of the absolute deviations from a given average. Depending up on the type of averages
used we have different mean deviations.
 The mean deviation of a sample of n observations x1, x2, . . ., xn is given as

MD=
∑ |X i− A|
n

Where | X i− A| denotes the absolute value of the deviation. Generally, arithmetic mean and
median are used in calculating mean deviation. So, A stands for the average used for calculating
MD . That is, A=median (~X ) ∨A=mean( X).

 In case of grouped data, the formula for MD becomes

MD=
∑ f i|X i −A| , where X i is the class mark of the ith class, f i is the frequency of the
n
ith class and
n = ∑ f i.
1. The mean deviation about the arithmetic mean is, therefore, given by

MD ( X ) =
∑| X i− X| ,for ungrouped data.
n

MD( X )=
∑ f i|X i −X| , for discrete data arranged in FD & for grouped frequency
n
distribution; where X i is the value or class mark of the ith class, f i is the frequency of the
ith class and n = ∑ f i .
Steps to calculate M.D for ( X )
 Find the arithmetic mean, X
 Find the deviations of each reading from X
 Find the arithmetic mean of the deviations, ignoring sign.
2. The mean deviation about the median is also given by

52
 ~
MD ( X )=
∑ |X i−~x|
, f or ungrouped data.
n
~
~ ∑ f i|X i − x| , for discrete data arranged in FD & for grouped frequency
MD( X )=
n
distribution; where X i is value or the class mark of the ith class, f i is the frequency of the
ith class and n = ∑ f i .
~
Steps to calculate M.D ( X )
 Find the median, ~ X
~
 Find the deviations of each reading from X
 Find the arithmetic mean of the deviations, ignoring sign.
3. The mean deviation about the mode is also given by

MD ( ^x )=
∑ |X i −^x| ,for ungrouped data.
n

MD( x^ )=
∑ f i|X i−^x| , for discrete data arranged in FD & for grouped frequency
n
distribution; where X i is value or the class mark of the ith class, f i is the frequency of the
ith class and n = ∑ f i .
Steps to calculate M.D ( ^x )
 Find the mode, ^x
 Find the deviations of each reading from ^x
 Find the arithmetic mean of the deviations, ignoring sign.
Example 4.5: The following are the number of visit made by ten mothers to the local doctor’s
surgery. 8, 6, 5, 5, 7, 4, 5, 9, 7, 4. Find mean deviation about mean, median and mode.
Solution:
First calculate the three averages
X =6, ~ X =5.5, ^x =5
Then take the deviations of each observation from these averages.
xi 4 4 5 5 5 6 7 7 8 9 Total
| X i−X| 2 2 1 1 1 0 1 1 2 3 14

| X i−~x| 1.5 1.5 0.5 0.5 0.5 0.5 1.5 1.5 2.5 3.5 14

|X i− ^X| 1 1 0 0 0 1 2 2 3 4 14

53
Since the distribution is ungrouped the mean deviation about mean, median and mode:

MD ( X ) =
∑| X i− X|= 14 =1.4
n 10

~ ∑ |X i−~x| 14
MD ( X )= = =1.4
n 10

MD ( ^x )=
∑ |X i −^x|= 14 =1.4.
n 10

Merits of MD

 It is well-defined, easy to compute and simple to understand.

 It is based on all observations
 It is not greatly affected by the extreme items
 It can be calculated by using any average

Demerit of MD

 It does not take in to account the signs of the deviations of items from the average

Remark: Of all the mean deviations taken about different averages or any arbitrary value, the
mean deviation about the median has the smallest value.

Coefficient of mean deviation (CMD):

The relative measure of mean deviation, also called the coefficient of mean deviation is obtained
by dividing mean deviation by the particular average used in computing mean deviation. Thus,

 CMD about the arithmetic mean is given by:

MD( X )
CMD( X )= , where MD is the mean deviation calculated about the arithmetic
X
mean.
 CMD about the median is given by:

~
~ MD( X )
CMD( X )= ~ In which case MD is calculated about the median of the
X
observations.

 CMD about the mode is given by:

MD ( ^x )
CMD( ^x )= in which case MD is calculated about the mode of the observations.
^x

54
Example 4.6
Calculate the coefficient of mean deviation about the mean, median and mode for the data in
Example 4.5 above.
Solution:
MD ( X ) 1.4
CMD ( X )= = =0.23
X 6
~
MD ( X ) 1.4
CMD (~
X )= ~ = =0.25
X 5.5
MD ( x^ ) 1.4
CMD ( ^x ) = = =0.28
x^ 5

4.3.4 The Variance, Standard Deviation and Coefficient of Variation

Variance and Standard Deviation

Like the mean deviation, the variance is also based on all observations in a set of data. But
the variance is the average of squared deviations from the mean. Recall that the sum of squared
deviations is minimum only when taken from the mean. Squared deviations are mathematically
manipulated than absolute deviations. Thus, if we averaged the squared deviations from the
mean and take the square root of the result (to compensate for the fact that the deviations were
squared), we obtain the standard deviation. This overcomes the limitation of the mean deviation.

Population Variance (σ 2)
If we divide the squared variation by the number of values in the population, we get something
called the population variance. This variance is the "average squared deviation from the mean".
 For ungrouped data
N

∑ ( X i −μ)2
[∑ ]
N
2 i=1 1 , where μ is the population arithmetic mean and N is
σ = = X i2−N μ2
N N i =1

the total number of observations in the population.

55
 For discrete data arranged in FD and for continuous grouped data

σ
2
=
∑ 2
f i ( X i−μ) 1
= [∑ f X i −N μ ]where μ is the population arithmetic mean, X i is the value
2 2
i
N N
or class mark of the ith class, f i is the frequency of the ithclass and N=∑ f i
Sample Variance ( S2)
One would expect the sample variance to simply be the population variance with the population
mean replaced by the sample mean. However, one of the major uses of statistics is to estimate
the corresponding parameter. This formula has the problem that the estimated value isn't the
same as the parameter. To offset this, the sum of the squares of the deviations is divided by one
less than the sample size.
 For ungrouped data
n

∑ ( xi −x)2
[ ]
n
1 where x is the sample arithmetic mean and n is the
2
S=
i=1
n−1
=
n−1
∑ xi2−n x2
i=1

total number of observations in the sample.

Zα
2
If the values have frequencies fi (i=1,2,…,m), then the sample variance is given by:

S=
∑2f i ( x i−x)
2

n−1
=
1
n−1
[ ∑ f i x i2−n x 2 ]
or
α
 For discrete data arranged in FD and for grouped data

S=
∑2f i ( x i−x)2
=
1
[ ∑ f i x i −n x ]where x is the sample arithmetic mean, x i is the value or
2 2
n−1 n−1
class mark of the ith class, f i is the frequency of the ithclass and n=∑ f i.

The Standard Deviation

There is a problem with variances. Recall that the deviations were squared. That means that the
units were also squared. To get the units back the same as the original data values, the square
root must be taken.
 Population Standard Deviation (σ )
σ =√ σ 2 where σ is the population variance.
2

 Sample Standard Deviation ( S )

56
 S= √ S2 where S is the sample variance.
2

Example 4.7: Find the sample variance and standard deviation for frequency distribution of height
in cms of students in a DMU given below.

Heights in cms 150 152 154 156 158 160 162 164 166

Number of students 28 40 52 100 60 48 32 20 7

Solution: Prepare the following table:

xi fi fixi xi2 fixi2

150 28 4200 22500 630000

924160
152 40 6080 23104
1233232
154 52 8008 23716
2433600
156 100 15600 24336
1497840
158 60 9480 24964
1228800
160 48 7680 25600
839808
162 32 5184 26244
537920
164 20 3280 26896
192892
166 7 1162 27556
224916 9518252
Sum 387 60674

Thus, n=∑ f i=387 , ∑ f i x i=60674 , ∑ f i x i =9518252 , ∑ x i =224916 .

2 2

1
S2=
n−1
[ ∑ f i x i2−n x 2 ]

[ ( ) ]= 3861 ( 5760.54 )=14.92 ,∧¿

2
1 60674
= 9518252−387
386 387

S= √14.92=3.86

57
 Example 4.8: Calculate the sample variance and standard deviation of the blood glucose level,
in milligrams per deciliter, for 60 patients shown below.

Class limit 55 – 63 64 – 72 73 – 81 82 – 90 91 – 99 100 – 108 109 –117

Frequency 9 5 12 17 7 6 4

Solution: In a continuous F.D., xi is the class mark representing the ith class.

Class limit α α −Z α f i xi
2

55 – 63 59 9 531 31329

64 – 72 68 5 340 23120

73 – 81 77 12 924 71148

82 – 90 86 17 1462 125732

91 – 99 95 7 665 63175

100 – 108 104 6 624 64896

109 –117 113 4 452 51076

Total 60 4998 430476

Where, n=∑ f i=60 , x =

∑ f i x i = 4998 =83.3 ,
n 60
∑ f i xi2 =430476, so that

58
 1 1 14142.6
2
S=
n−1
[ ∑ f i x i −n x ] =
2 2
59
[ 430476−60 ( 83.3 ) ]=
2
59
=239.71 ,

S= √239.71 = 15.48

Properties of Variance & Standard Deviation

1. If a constant is added to (or subtracted from) all the values, the variance remains the
H 0 : μ=μ 0
same; i.e., for any constant k, .

Example 4.9 Consider the 6 sample values xi: 54, 52,53,50,51, and 52.

μ≠μ 0
The sample variance is 2 = . Now, subtract 50 from each value to get:

−Zα/2<ZC<Zα/2 Z C >Z α/2 or Z C <−Z α/2

: 4, 2, 3, 0, 1, 2; and, the variance of this new series is 2. i.e. .

1. If each and every value is multiplied by a non-zero constant (k), the standard deviation is
ZC= Zα /2
multiplied by /k/ and the variance is multiplied by k2 ; i.e., .
2. Both the variance and the standard deviation are give more weight to extreme values and
less to those which are near to the mean.

Coefficient of Variation
The standard deviation is an absolute measure of dispersion. The corresponding relative measure
is known as the coefficient of variation (CV).
Of course, standard deviation is an absolute measure of dispersion that expresses the variation in
the same unit as the original data but it can not be the sole basis for comparing two distributions.
For instance, if we have a standard deviation of 10 and a mean of 5, the values vary by an
amount twice as large as the mean itself. If, on the other hand, we have a standard deviation of
10 and a mean of 5000, the variation relative to the mean is significant. Therefore, we cannot
know the dispersion of a set of data until we know the standard deviation, the mean, and how the
standard deviation compares with the mean.
Coefficient of variation is used in such problems where we want to compare the variability of
two or more different series. Coefficient of variation is the ratio of the standard deviation to the
arithmetic mean, usually expressed in percent.
Standard deviation
CV = ×100 %
mean
For population data:

59
 σ
CV = × 100
μ
Where σ is the population standard deviation and μ is population mean.
For sample data:
S
CV = ×100
x
Where S is the sample standard deviation and x is sample mean.
Remark: A distribution having less coefficient of variation is said to be less variable or more
consistent or more uniform or more homogeneous.
Example 4.10: One patient’s blood pressure, measured daily over several weeks, averaged 182
with a standard deviation of 12.6, while that of another patient averaged 124 with a standard
deviation of 9.4. Which patient’s blood pressure is relatively more variable?

Solution:

Given: S1=12.6 orZC=−Zα/2 =182 S2=9.4

μ>μ 0 = 124

S1 12.6
CV1 = ×100% = ×100 % = 6.923
x1 182

S2 9.4
CV2 = ×100% = ×100 % = 7.58
x2 124

Blood pressure of the second patient is relatively more variable.

4.4 Standard Scores (Z-Scores)

A standard score for sample value in a data set is obtained by subtracting the mean of the data
set from the value and dividing the result by the standard deviation of the data set. Basically, the
standard score (z-score) tells us how many standard deviations a specific value is above or below
the mean value of the data set. That is, the z-score is the number of standard deviations the data
value falls above (positive z-score) or below (negative z-score) the mean for the data set.

Z-score computed from the population

60
 X −μ
Z−score=
σ

Z-score computed from the sample

X −X
Z−score=
S

Example 4.11: What is the Z-score for the value of 14 in the following sample data set?

3 8 6 14 4 12 7 10

Solution:

14−8
X = 8, S = 3.8173 thus, Z = ≈ 1.57
3.8173

 The data value of 14 is located 1.57 standard deviations above the mean 8 because the
z-score is positive.

Example 4.12: Suppose that a student scored 66 in Statistics and 80 in Biology. The score of the
summary of the courses is given below.
Course Average score Standard deviation of the score
Statistics 51 12
Biology 72 16

In which course did the student scored better as compared to his classmates?
Solution:
X−x 66−51 15
Z-score of student in Statistics: Z= = = =1.25
S 12 12

X−x 80−72 8
Z-score of student in Biology: Z= = = =0.5
S 16 16

From these two standard scores, we can conclude that the student has scored better in Statistics
course relative to his classmates than in Biology course.

4.5 Moments, Skewness and Kurtosis

61
The measures of central tendency and variation discussed in previous one do not reveal the entire
story about a frequency distribution. Two distributions may have the same mean and standard
deviation but may differ in their shape of the distribution. Further description of their
characteristics is necessary that is provided by measures of skewness and kurtosis.

4.5.1 Moments

Moments are statistical tools used in statistical investigation. The moments of a distribution are
the arithmetic mean of the various powers of the deviations of items from some number. In our
course, we shall use it in the study of Skewness and Kurtosis of statistical distribution.

Moments about the origin

M=
∑ Xi
r

r
n

Where r =0 ,1 , 2 ,3 , …

Moments about the origin for grouped frequency distribution and for ungrouped frequency
distribution is

M=
∑ X ir
r
n

Where f i is the frequency of X i . X i is the midpoint in the case of grouped frequency distribution
or class value in the case of ungrouped frequency distribution.

Note that: M 1=X , M 0=1

Moments about the Mean (Central Moments)

M=
'∑ ( X i−X )
r

r
n

Moments about the mean for grouped frequency distribution and for ungrouped frequency
distribution

62
 M=
∑ f i ( X i− X )
'
r

r
n

Where f i is the frequency of X i . X i is the midpoint in the case of grouped frequency distribution
or class value in the case of ungrouped frequency distribution.

Note that: M '2=SD 2 if it is assumed n=n−1

Moments about any arbitrary constant A

M=
∑ ( X i− A )
'
r

r
n

Moments about any arbitrary constant A for grouped frequency distribution and for ungrouped
frequency distribution

M=
∑ f i ( X i− A )r
'
.
r
n

Example 4.13: Find the first four moments about the mean for the following individual
series

X i: 3 6 8 10 18

Solution: n=5,

S.No Xi ( X i−X ) ( X i−X )

2
( X i−X )
3
( X i−X )
4

1 3 -6 36 -216 1296
2 6 -3 9 -27 81
3 8 -1 1 -1 1
4 10 1 1 1 1
5 18 9 81 729 6561
Total ∑ X=45 ∑ ( X−X )=0 ∑ ( X−X )2=128 ∑ ( X−X )3=486∑ ( X−X )4=7940

63
Thus,

45
X = =9 , M '1=
∑ ( X i−9 ) =0 , M ' = ∑ ( X i−9 ) = 128 =25.6
1 2

2
5 5 5 5

'
M =
∑ 3
( X i −9 ) 486
=
'
=97.2 , M =
∑ 4
( X i−9 ) 7940
= =1588 .
3 4
5 5 5 5

4.5.2 Skewness

Skewness refers to lack of symmetry (or departure from symmetry) in a distribution.

 A skewed frequency distribution is one that is not symmetrical.

 Skewness is concerned with the shape of the curve not size.
A distribution is said to symmetrical when the value is uniformly distributed around the mean
(distribution of the data bellow the mean and above the mean are equal). In a symmetrical
distribution, the mean, median and mode coincide (i.e., mean = median = mode).
Positively skewed distribution:- if the value of mean is greater than the mode, skewness is said
to be positive. In a positively skewed distribution mean is greater than the mode and the median
lies somewhere in between mean and mode. A positively skewed distribution contains some
values that are much larger than the majority of other observations.
Negatively Skewed distribution:- if the value of mode is greater than the mean, skewness is
said to be negative. In a negatively skewed distribution mode is greater than the mean and the
median lies in between mean and mode. The mean is pulled towards the low-valued item (that
is, to the left). A negatively skewed distribution contains some values that are much smaller than
the majority of observations.

64
Note that: In moderately skewed distributions the averages have the following relationship.

(Mean – Mode) = 3(Mean - Median).

How to check the presence of skewness in a distribution?

Skewness present in the data if:

i) the graph not symmetrical.

ii) the mean, median and mode do not coincide
iii) the sum of positive and negative deviations from the median is not zero
iv) the frequencies are not similarly distributed on either side of the mode.

Measures of skewness (α 3)

A measure of skewness gives a numerical expression for and the direction of asymmetry in a
distribution. It gives information about the shape of the distribution and the degree of variation
on either side of the central value. The three most commonly used measures of skewness are
Pearson’s coefficient skewness, Bowley’s coefficient of skewness and coefficient of skewness
based on moments.

1. Pearson’s coefficient skewness (Pearsonian coefficient of skewness)

The skewness of the distribution can be measured by Pearson’s Coefficient of Skewness (
α 3 ), for which the formula is given below:
Mean−Mode
α 3=
Standard deviation
2. Bowley’s Coefficient of Skewness

65
 Bowley’s coefficient of skewness is based on quartiles. The formula for calculating
coefficient of skewness is:
(Q ¿ ¿ 3−Q2)−(Q2−Q1 ) Q +Q −2Q 2
α 3= ¿= 3 1
Q3−Q 1 Q 3−Q 1
3. Moment Coefficient of Skewness
Moment coefficient of skewness is based on moments. The formula for calculating
coefficient of skewness is:

M '3 M '3
α 3= 3/ 2 = 3
M' 2 σ
n
Where, M'r = ∑ (x i−x) /n
r

i=1

The shape of the curve is determined by the value of α 3

α 3> 0, the distribution is positively skewed/skewed to the right, i.e mode < median <mean.

smaller observations are more frequent than larger observations. i.e., the majority of

the observations have a value below an average.

α 3 = 0, the distribution is symmetric, i.e. mean = mode = median.

α 3 < 0, the distribution is negatively skewed/skewed to the left. i.e., mean < median < mode.

smaller observations are less frequent than larger observations. i.e., the majority of

the observations have a value above an average.

4.5.3 Kurtosis

Kurtosis is a measure of peakedness of a distribution. The degree of kurtosis of a

distribution is measured relative to the peakedness of a normal curve. If a curve is more
peaked than the normal curve it is called ‘leptokurtic’; if it is more or flat-topped than the
normal curve it is called ‘platykurtic’ or flat-topped. The normal curve itself is known as
‘mesokurtic’.

Measures of Kurtosis ( )
α4

66
The moment coefficient of kurtosis:
M '4 M '4
α 4= 2 = 4
M' 2 σ
The peakedness depends on the value of α 4
 α 4 > 3  the curve is leptokurtic,

 α 4 = 3  the curve is mesokurtic,

 α 4 < 3  the curve is platykurtic.

Example 4:14: Based on the following data:

M ' 0 = 1, M ' 1 = -0.6, M ' 2 = 1.6, M ' 3 = -2.4, M ' 4 = 5.8
a/ Find the coefficient of skewness and discuss the distribution type.
b/ Find the coefficient of kurtosis and discuss the distribution type.
Solution:
M '3 −2.4
a/ α 3= 3/ 2 = 3 /2 = -1.19 < 0, the distribution is negatively skewed.
M' 2 1.6
M '4 5.8
b/ α 4 = 2 = = 2.26 < 3, the distribution is platykurtic.
M' 2 1.62
Example 4.15: Find the coefficient of skewness and the coefficient of kurtosis for the
above example 4.13.

Solution:
M '3 97.2
i) α 3= =
M' 3/ 2 ¿¿
2

the distribution is positively skewed.

M '4 1588
ii) α 4= 2 = 2
=2.42
M' 2 25.6
the distribution is platykurtic.
Exercise 4

67
1. Calculate the mean deviation about the mean, median and mode, and their coefficients and
also variance and standard deviation for the following data.

Size of shoes 3 6 11 2 4 1 5 7 8 9
0
No. of pairs sold 10 1 25 6 4 3 2 8 9 4
5

2. Last semester, the students of department Biology section A and B look Fundamentals of
Biostatistics course. At the end of the semester, the following information was recorded.
Section A Section B
Mean score 79 64
Standard deviation 23 11

Compare the relative dispersion of the two sections’ scores using appropriate way.
3. A meteorologist interested in the consistency of temperatures in three cities during a given
week collected the following data. The temperatures for the five days of the week in the three
cities were:
City 1: 25, 24, 23, 26, 17
City 2: 22, 21, 24, 22, 20
City 3: 32, 27, 35, 24, 28
Which city have the most consistent temperature, based on these data?
4. The median and the mode of a mesokurtic distribution are 32 and 34 respectively. The
4thmoment about the mean is 243. Compute the Pearsonian coefficient of skewness and
identify the type of skewness. Assume (n-1 = n).
5. If the standard deviation of a symmetric distribution is 10, what should be the value of the
fourth moment so that the distribution is mesokurtic?

68
 CHAPTER 5: ELEMENTARY PROBABLITY
Objectives
After studying this chapter, you should be able to:
 Understand the fundamental concepts of probability
 Apply the principle of counting techniques to solve real problem.
 Define some basic terms of probability.
5.1 Definition of some probability terms
• Experiment:- Any process of observation or measurement or any process which generates well
defined outcome.
•Random experiment: it is an experiment which can be repeated any number of times under the
same conditions, but does not give unique results. The result will be any one of several possible
outcomes, but for each trial, the result will not be known in advance. A Random experiment is
also called a trial & the outcomes are called events.
• Sample space: - is the collection of all possible out comes or sample points of a random
experiment.
•Sample point: -Each element of sample space is called Sample point.
• Event: - is a subset of a sample space i.e. an event is a collection of sample points.
•Impossible event:- this is an event which will never occur.

Example 5.1: In an experiment of tossing a coin three times, S = {HHH, HHT, HTH, HTT,
THH, THT, TTH, TTT}, each sample point is an equally likely out come. It is possible to define
many events on this sample space as follows:

A = {HHH} - the event of getting only head.

B = {HHH, HHT} - the event of getting head on the first two tosses.

C = {HHT, HTH, THH} - the event of getting exactly two heads.

D = the event of getting number 9 is an impossible event.

Example 5.2: If we toss a coin the sample space (S) of this experiment, S = {head, tail} where
head and tail are two faces of a coin. If we are interested the outcome of head will turn up then
the event E= {head}.
Example 5.3
Find the sample space rolling a fair die.
S= {1, 2, 3, 4, 5, 6}

69
• Mutually exclusive event: - two events A and B are said to be mutually exclusive if there is no
sample point which is common to A and B. i.e. A ∩ B = ∅
• Independent event: two or more events are said to be independent if the occurrence or non-
occurrence of an event does not affect the occurrence or non-occurrence of the other.
• Dependent Events: Two events are dependent if the first event affects the outcome or
occurrence of the second event in a way the probability is changed.
• Complement of an Event: the complement of an event A means nonoccurrence of A and is
denoted by A', or Ac contains those points of the sample space which don’t belong to A.
• Equally likely outcomes: if each outcome in a sample space has the same chance to be
occurred.
Example 5.4: Casting a fair die all possible outcomes are equally likely.
5.2 Counting rules: addition, multiplication, Permutation & Combination rule
In order to calculate probabilities, we have to know
• The number of elements of an event
• The number of elements of the sample space.
That is in order to judge what is probable, we have to know what is possible.
In order to determine the number of out comes one can use several rules of counting:
1. Addition rule
2. Multiplication rule
3. Permutation rule
4. Combination rule.
1. The addition Rule
Suppose that a procedure, designated by 1, can be done in n 1 ways. Assume that a second
procedure designated by 2, can be done in n2 ways. Suppose furthermore, that it is not possible
that both 1 and 2 done together. Then, the number of ways in which we can do1 or 2 is n1 +n 2
ways.
Example 5.5: Suppose we are planning a trip to some place. If there are 3 bus routs & two train
routs that we can take, then there are 3 + 2 = 5 different routs that we can take.
2. Multiplication rule: If an operation consists of k steps and the 1st step can be done in n 1
ways, the 2nd step can be done in n2 ways (regardless of how the 1st step was performed), the kth
step can be done in nk ways, (regardless of how the preceding steps were performed), then the
entire operation can be performed in n1 · n2 ·… · nk ways.

70
Example 5.6: Suppose that a person has 2 different pairs of trousers and 3 shirts. In how many
ways can he wear his trousers and shirts?

Solution: He can choose the trousers Z C < Zα ways, and shirts in Z C >Z α ways. Therefore, he can

wear in Z C = Z α possible ways.

3. Permutation:-An arrangement of objects with attention given to order of arrangement is

called permutation. The number of permutation of n different objects taken r at a time is obtained

by:
μ < μ0
Permutation Rule:
a) The number of permutations of n objects taken all together is n!

i.e., n! = n*(n-1)*(n-2)*…*3*2*1 =
Z C >−Z α
Note: By definition 0! = 1.
b) The arrangement of n distinct objects in a specific order using r objects at a time is called the
permutation of n objects taken r objects at a time. It is written as nPr and the formula is

Z C <−Z α
c) The number of distinct permutation of n objects in which k1 are alike, k2 are alike,

kn are alike , etc. is

Z C =−Z α for n = n1 + n2 + n3 + …+ nk.
Example 5.7: Find number of permutations of the letters in the word ‘‘statistics’’.
Solution:
There are 3 s’s, 3t’s, 1a, 2i’s and 1c. i.e. n1 =3 ,n 2=3,n3 =1,n 4=2∧n5=1

Therefore, α=0.05 = 50,400.

Example 5.8
A photographer wants to arrange 3 persons in a row for photograph. How many different types
of photographs are possible?
Solution:
Assume 3 persons Aster (A), lemma (L), Yared (Y) and n=3
Since n! =3! = 3*2! = 6, there are 6 possible arrangement ALY, AYL, LAY, LYA, YLA and
YAL.

71
Example 5.9
Suppose we have a letters A, B, C, D,E
a) How many permutations are there taking all the four?
b) How many permutations are there taking two letters at a time?
Solution:
a) Here n = 5, there are four distinct object.
There are 5! = 120 permutations.
b) Here n = 5, r = 2
There are 5P2 = 5!/(5-2)! = 120/6 = 20 permutations.
Example 5.10
Fifteen Ethiopian athletes were entered to the race. In how many different ways could prizes for
the first, the second and the third place be awarded?
Solution
15 objects taken 3 at a time 15P3=15!/(15-3)! = 2730 ways.
4. Combination-A selection of objects considered without regard to order in which they occur is
called Combination. The number of combination of n different objects taking r of them at a time

is
x̄=8 . 4 , for σ =3 .2 .
Example 5.11
Given the letters A, B, C, and D list the permutation and combination for selecting two letters.
Solution:
Permutation Combination
AB BA CA DA AB BC
AC BC CB DB AC BD
AD BD CD DC AD DC
Note that in permutation AB is different from BA but in combination AB is the same as BA.

Example 5.12
In a club containing 7 members a committee of 3 people is to be formed. In how many ways can
the committee be formed?

Solution: 7C3 =
H 0 : μ=10 
H 1 : μ<10 = 35.

72
Example 5.13
How many four-digit numbers can be formed with the 10 digits 0, 1, 2, .. . , 9 if
a/ repetitions are allowed
b/ repetitions are allowed, and
c/ the last digit must be zero & repetitions are not allowed.
Solution:
a/ the first digit can be any one of 9 (since 0 is not allowed). The second, third and fourth digits
can be any one of 10. Then 9.10.10.10=9000 numbers can be formed.

b/ the first digit can be any one of 9 & the remaining three can be chosen in α=0.05 ways.
Thus, 9. Zα=Z0.05=1.645 = 4563 numbers can be formed.

ways. Thus, 9. Zc=−2<−Zα=−1.645 = 504

x̄−μ 8. 4−10
ZC= 0 = =−2
c/ the first digit can be chosen in 9 ways & the next two digits in σ/√n 3.2/4

numbers can be formed.

5.3 Probability of an event
Definition: Probability is a numerical measure of the chance or likelihood that a particular event
will occur & it lies in the range from 0-1, inclusive. Probability is a building block of inferential
statistics.
Definition: Let E be an experiment. Let S be a sample space associated with E. With each event
A in S we associate a real number designated by P (A) and called the probability of A.
Generally probability can be divided into two
i) Subjective probability: - probability determined based on individual’s own judgment,
experience, information, belief, etc is called Subjective probability.
ii) Objective probability: - the probability of an event in a certain experiment based on
experimental evidence.
Basic approaches to probability
There are three different conceptual approaches to the study of probability theory.
These are:
1. The classical approach.
2. The frequentist approach.
3. The axiomatic approach.
1. Classical approach:

73
Definition: If there are n equally likely outcomes of an experiment, and out of the n outcomes
event A occur only k times, the probability of the event A is denoted by P (A) is defined as
¿ n(A) k
p(A) = Number of outcomes favorable ¿ event A Total number of outcomes = =
n( S) n
Note: Classical approach of measuring probability fails to answer for the following conditions:
• If total number of outcomes is infinite or if it is not possible to enumerate all elements of the
sample space.
• If each outcome is not equally likely
Example 5.14
Compute a/ the probability of having two boys & one girl is a three child family using the
classical method, assuming boys & girls are equally likely.

b/ using (a) compute the probability of having three boys in a three-child family.
c/ using (a) compute the probability of having three girls in a three –child family.
d/ using (a) compute the probability of having two girls & one boy in three child family.
Solution

The sample space S or the experiment is S= {BBB, BBG, BGB, BGG, GBB, GBG,
GGB, GGG}
So n(S) = 8.
a/ For the event A= ‘ two boys & a girl’ = {BBG,BGB,GBB} , we have n(A)=3, Since the outcome are
equally likely , the probability of A is P(A)= n(A)/n(S)=3/8 =0.375.

b/ Compute the probability of having three boys in a three-child family.

For the event B= ‘ three boys’ = {BBB} , we have n(B)=1,Since the outcome are equally likely , the
probability of B is P(B)= n(B)/n(S)=1/8 .

c/ compute the probability of having three girls in a three –child family.

For the event C= ‘ three girls’ = {GGG} , we have n(C)=1.Since the outcome are equally likely , the
probability of C is P(C)= n(C)/n(S)=1/8

d/ Compute the probability of having two girls & one boy in three child family.

For the event D= ''two girls & one boy'' = {BGG, GBG,GGB} , we have n(A)=3, Since the outcome are
equally likely , the probability of D is P(D) = n(D)/n(S) =3/8 =0.375.

74
Example 5.15: A box of 80 candles consists of 30 defective and 50 non defective candles. If 10
of these candles are selected at random with out replacement, what is the probability?
a) all will be defective?
b) 6 will be non-defective?
c) all will be non-defective?
Solution

Total Selection:
H 0 :μ=130
a) Let A be the event that all will be defective.

Total way in which A occurs =

H1:μ≠130 α=0.05 n (A)

α/2=0.025
x̄−μ0 132−130 2
n( A) Z C= = = =1. 6 .
P (A) ) = n (S) = σ/ √n 10/√64 1. 25 0.00001825.

b) Let A be the event that 6 will be non defective.

α α
Total way in which A occur = 2 2 n (A)

n( A)
P (A) = n (S) =
Z 0 .025 Z 0 .025
=1.6 0.265

c) Let A be the event that all will be non defective.

Total way in which A occur =

H 0 α=0.05 n (A)
n( A)
P (A) = n (S) =
H0 :μ=50 H1:μ≠50 α=0.05 0.00624.
2. The Frequentist Approach (Empirical Probability): This approach to probability is based
on relative frequencies.
Definition: Suppose we do again and again a certain experiment n times and let A be an event of
the experiment and let k be the number of times that event A occurs. Therefore the
probability of the event A happening in the long run is given by:
Number of ×event A has occured k
P(A) = =
Total number of observations n

75
In other words given a frequency distribution, the probability of an event (A) being in a given
class is
Frequency of class A
P(A) =
Total frequency ∈the distribution
Example 5.16: The national center for health statistics reported that of every 539 deaths in
recent years, 24 resulted that from automobile accident, 182 from cancer, and 353 from other
disease. What is the probability that particular death is due to an automobile accident?

Solution:
P (automobile) = death due to automobile /total death =24/539 = 0.445.
The probability that particular death is due to an automobile accident is 0.445.
3. The axiomatic approach.
Let E be a random experiment and S be a sample space associated with E. With each event A a
real number called the probability of A satisfies the following properties called axioms of
probability or postulates of probability.
1.0≤ P(A )≤ 1
2. P(S) =1, S is the sure/certain event.
3. If A1 and A2 are mutually exclusive events, the probability that one or the other occur equals
the sum of the two probabilities. i. e. P(A1∪A2)=P(A1)+P(A2)
n
Similarly P(A1∪A2∪ . . . An) = P(A1)+P(A2) +. . . P(An) = ∑ A i
i=1

4. P (A') =1-P (A)

5. P (ø) =0, ø is the impossible event.
5.4 Some probability rules
Rule l: let A be an event and A' be the complement of A with respect to a given sample space of
an experiment, then P(A')=1-P(A)
Proof: let S be a sample space S=AUA' and, A and A' are mutually exclusive
A∩A' = ø
P(S) = P (AUA') = P (A') + P (A) and P(S) = 1
1= P (A') + P (A) => P (A') = 1-P (A)
Rule 2: let A and B are events of a sample space S, then
P (A' ∩ B) = P (B)-P (A ∩ B)
Proof: B =S ∩ B = (AUA') ∩ B = (A ∩B) U (A'∩ B)
If A∩B ≠ø , then P(B) =P (A∩B) +P (A'∩B)

76
P (A'∩B) = P(B) – P(A∩B)
Rule 3: Suppose A and B are two events of a sample space, then
P(AUB) = P(A) + P(B) – P(A∩B).
Proof:
(AUB) = AU(A'∩B), A and A'∩B are disjoint sets
∴ P(AU B) = P(A) + P(A'∩B) . . . .*
But we have already proved that P (A' ∩ B) = P (B) – P (A ∩ B)
Put this in equation *
P(A U B) = P(A) + P (B) – P (A ∩ B)
Example 5.17: A fair die is thrown twice. Calculate the probability that the sum of spots on the
face of the die that turn up is divisible by 2 or 3.
Solution
S={(1,1),(1,2),(1,3),(1,4),(1,5),(1,6),(2,1),(2,2),(2,3),(2,4),(2,5),(2,6),(3,1),(3,2),(3,3),(3,4),(3,5),
(3,6),(4,1),(4,2),(4,3),(4,4),(4,5),(4,6),(5,1),(5,2),(5,3),(5,4),(5,4),(5,5),(5,6), (6,1),(6,2),(6,3),
(6,4),(6,5),(6,6)}
This sample space has 6*6 =36 elements let A be the event that the sum of the spots on the die is
divisible by 2 and B be the event that the sum of the spots on the die is divisible by 3, then
A = {(1,1), (1,3), (1,5), (2,2), (2,4), (2,6), (3,1), (3,3), (3,5), (4,2), (4,4), (4,6), (5,1), (5,3), (5,5),
(6,2), (6,4), (6,6)}
B = {(1,2), (1,5), (2,1), (2,4), (3,3), (3,6), (4,2), (4,5), (5,1), (5,4), (6,3), (6,6)}
A∩B = {(1,5), (2,4), (3,3), (4,2), (5,1), (6,6)}
P (A or B) = P (A U B) = P (A) +P (B) – P (A ∩ B) = 18/36 + 12/36 -6/36 = 24/36 = 2/3
5.5 Conditional Probability and Independence
5.5.1 Conditional Probability
If A and B are events. Conditional probability of A given B means the probability of occurrence
of A when the event B has already happened.
It is denoted by P (A/B) and is defined by
P (A/B) = P(A∩B)/P (B), if P (B)≠0
Conditional probability of B given A means the probability of occurrence of B when the event A
has already happened. It is denoted by P (B/A) and is defined
P (B/A) = P(A∩B)/P (A), if P (A)≠0
P (A ∩ B) = P (A) P (B/A) = P (B) P (A/B)
5.5.2 Multiplication Law of Probability

77
If A and B are events in a sample space S, then
P (A ∩ B) = P (A) P (B/A), P (A) ≠ 0
= P (B) P (A/B), P (B) ≠ 0
Where P (B/A) represents the conditional probability of B given A and P (A/B) represents the
conditional probability of A given B.
Note: Extension of multiplication law of probability for ‘n’ events A1, A2, …, An we have
P (A1 ∩ A2 ∩… An) = P (A1) P (A2/A1) p (A3/A1 ∩ A2)…P(An/A1∩ A2 ∩…An-1)
Example 5.18: A coin is tossed twice. If it is already known that the first coin has thrown a head,
what is the probability of getting two heads?
Solution:
S = {HH, HT, TH, TT}, A = the first shows a head = {HH, HT}, B= two heads occur ={HH}
P (B/A) = P(A∩B)/ P(A)
But A∩B ={HH}, P(A∩B) =1/4, P(A)=1/2, therefore, P (B/A) = P(A∩B)/ P(A) = 1/2
Example 5.19: Let A and B are events such that P (A U B) = ¾, P (A n B) = ¼ and P(A' ) = 2/3.
Find P ( A'/B)
Solution:
P(A') = 2/3  P (A) = 1- P(A') = 1-2/3 = 1/3, Now, P (A U B) = P (A) + P (B) - P (A ∩ B)
3/4 = 1/3 + P (B) – ¼. P(B) = 3/4 - 1/3 + ¼ = 2/3
Therefore, P (A/B) = P (A∩B)/P(B) = 3/8  P(A'/B) =1-P (A/B) = 1-3/8 =5/8
5.5.3 Probability of Independent Event
Two events A and B are said to be independent if the occurrence of A has no bearing on
occurrence of B. That means knowledge of A has occurred given no information about the
occurrence of B. Two events, A and B, are said to be independent if P(A∩B) =P(A)P(B) .
Suppose A and B are independent events with 0<P (A) <1 and 0<P (B) <1. The following
statements true:
i. A' and B' are independent.
ii. A and B' are independent.
iii. A' and B are independent.
iv. P(B|A) = P(B).
v. P(B|A') = P(B).
Example 5.20: A box contains four black and six white balls. What is the probability of getting
two black balls in drawing one after the other under the following conditions?
a. The first ball drawn is not replaced

78
b. The first ball drawn is replaced
Solution
Let A= first drawn ball is black
B= second drawn is black
Required P (A n B)
a. P (A ∩ B) = P (B/A) P(A) = (4/10) (3/9) = 2/15
b. P (A ∩ B) = P (A) P (B) = (4/10) (4/10) = 16/100 = 4/25.
Exercise 5
1. A basket of fruit contains 3 mangoes, 2 Bananas and 7 pine apples. If a fruit is chosen at
random, what is the probability that it is either a mango or a banana?
2. Suppose a survey is conducted in which 50 families with three children are asked to
disclose the gender of their children. Based upon the results, it was found that 18 of the
families had two boys & one girl. Estimate the probability of having two boys & one girl
in a three-child family.
3. A box contains 6 red, 4 white and 5 black balls. A man draws 4 balls from the box at
random. Find the probability that among the balls drawn there is at least one ball of each
color.
4. Births exclude leap years from the following calculations, determine the probability that
a randomly selected person has a birth day.
a/ on the first day of a month, b/ on the 31st day of a month, c/ in the month of
December, d/ on November 8.

79
 CHAPTER 6: PROBABILITY DISTRIBUTION
The purpose of this unit is to introduce you with the concept of random variable and their
probability distributions. In a probability distribution, the variables are distributed according to
some definite probability function. In the previous unit we have discussed the concept of
probability. The different rules of probability and frequency distributions were also discussed. In
this unit we utilize this information to understand the discrete and continuous probability
distributions. Moreover, the concept of mathematical expectation is discussed.

After completing this unit you will be able to:

 define the term random variable

 understand discrete and continuous random variables
 define the term probability distribution
 differentiate between discrete and continuous probability distributions
 compute the expected value of a random variable
6.1 The Concept of Random Variables

Before introducing a probability distribution, we have to define what is meant by a random

variable, and provide some brief demonstrations based on finite sample spaces.

Definition: A variable whose values are determined by chance with associated probabilities is
called a random variable. It is a quantity in which different observations can assume different
values.

In any experiment of chance, the outcomes occur randomly. For example, the total score when a
pair of dice is rolled, the number of heads when a coin is tossed several times, annual household
income, and so on are examples of random variables (or stochastic variables).
Random variables are usually denoted with capital letter X, Y, Z etc, while the values taken by
them are denoted by lower case letters x, y, z etc. Thus, P (x 1  X  x2) is the probability that the
80
random variable X takes values between x1 and x2, both inclusive. A random variable can be
discrete or continuous.

6.1.1 Discrete Random Variable

If the random variable X can assume only a particular finite or countably infinite set of values, it
is said to be a discrete random variable. For example, if you throw a die, the outcome X is a
random variable, which can assume only the values 1, 2, 3, 4, 5 and 6.

Example 6.1: Consider an experiment of "flipping a fair coin 3 times". List the elements of the
sample space that are assumed to be equally likely (as this is what is meant by a fair or balanced
coin) and the corresponding values x of the r-v X, the number of heads observed.

Solution: If H stands for heads and T for tails, then the sample space corresponding to this
experiments is S = {HHH, HHT, HTH, HTT, THH, THT, TTH, TTT}.

Since X= the number of heads observed, the results are shown in the following table:

Element of sample space Probability X

HHH 1/8 3
HHT 1/8 2
HTH 1/8 2
HTT 1/8 1
THH 1/8 2
THT 1/8 1
TTH 1/8 1
TTT 1/8 0
Thus, we can write X(HHH) = 3, X(HHT) = 2, , X(TTT) = 0, and P(X = 3) = 1/8 = the
probability that the r-v X is 3, P(X= 2) = 3/8, and P(X=0)=1/8.

α /2=0.025
Note that the possible values of X are: .

6.1.2 Continuous Random Variable

A random variable X is said to be continuous if it can take all possible values (integral as well as
fractional) between certain limits. Continuous random variables occur when we deal with
quantities that are measured on a continuous scale. For instance, the life length of an electric
bulb, the speed of a car, weights, heights, and the like are continuous. In such cases, probabilities

81
are associated with intervals or regions of a continuous random variable, and not with individual
points.

Example 6.2: -The height of an individual

- The distance between Debre Markos and Addis Ababa

- The weight of an individual, etc.

6.2 Probability Distribution

A probability distribution shows the possible outcomes of an experiment and the probability of
each of these outcomes. That is, probability distribution is a complete list of all possible of
values of a random variable and their corresponding probabilities.
A formula giving the probability of the different values of the random variable X for:

 Discrete variable is the probability massy function (pmf) and is usually denoted by p(x).
If X is a discrete random variable taking at most a countably infinite number of values x 1,
x2, …, then P (xi) = P(X = xi): i= 1, 2 …is called the probability mass function of random
variable X. The set of ordered pairs {xi, P (x i)} i= 1, 2 … gives the probability
distribution of the random variable X. The numbers P (xi): i= 1, 2…must satisfy the
following conditions.
i) P(xi) ≥0

tα/2,n−1=t0.025,15=2. 131
ii) =1
 Continuous variable is the probability density function (pdf) and is usually denoted by
f(x). A random variable X, is said to be a continuous random variable if there is a non–
negative function, f,
x
F(x) = ∫ f (t)d t
t =−∞

The function f is called probability density function of X. And it satisfies the following
conditions.
i) f(x)≥0 for all x, -∞ < x < ∞
∞

ii) ∫ f ( x ) d x=1
−∞

82
Discrete probability distribution
Discrete probability distribution is a distribution whose random variable is discrete. It describes a
finite set of possible occurrences, for discrete “count data.”
Example 6.3: Consider the possible outcomes for the random experiment of tossing three coins
together once.
Sample space, S = {HHH, THH, HTH, HHT, TTH, THT, HTT,TTT}
Let X be the number of heads that will turn up when three coins tossed. The possible values of X
are 0,1,2 and 3.
P(X = 0) = P(X (TTT)) = 1/8,
P(X=1) = P(X (HTT))+P(X (THT) )+ P(X (TTH) )=1/8+1/8+1/8 = 3/8
P(X=2) = P(X (HHT)) +P(X (HTH)) +P(X (THH)) = 1/8+1/8+1/8 = 3/8,
P(X=3) = P(X (HHH)) = 1/8.
X 0 1 2 3
P(X=x) 1 3 3 1
8 8 8 8

Continuous probability distribution

Continuous probability distribution is a probability distribution whose random variable is
continuous. It describes an “unbroken” continuum of possible occurrences. Probability of a
single value is zero and probability of an interval is the area bounded by curve of probability
density function and interval on x-axis. Let a and b be any two values; a <b. The probability that
X assumes a value that lies between a and b is equal to the area under the curve a and b; that is
x̄−μ 53.8−50 3.8
tC= 0= = =2.92.
P(a  X  b) = s/√n 5.2/√16 1.3 . The integration from a to b in the case of the continuous variable is
analogous to the summation of probabilities in the discrete case.
Example 6.4: A continuous random variable X has a probability density function given by

f(x) =
tc=2.92>2.131,H0 0  X  1.
Find the probability that X lies between the interval 0 and 1.

Solution:
α =0 . 05.
6.3 Expectation and Variance of Random variable

83
The objective of this section is to introduce you with the most common parameters of probability
distributions. There are some summary measures in terms of which we can summarize the
behavior of probability distributions. The most common of these are the average called expected
value and dispersion about the average called the variance.
6.3.1 Expectation/Mean
The averaging process, when applied to a random variable is called expectation. It is denoted by
E(X) or  and is read as the expected value of X or the mean value of X.

Case 1: For discrete random variable

Suppose X is a discrete random variable which takes on values in a finite set x1, x2,…, xn with
probabilities P(xi) = P[X = xi] i= 1, 2, …n, then Expected value of X, E(X) of the discrete
random variable is given by:

E(x) =  =
^P
Case 2: For continuous random variable
If X is a continuous random variable then
∞ ∞

E(X) = ∫ x f X (x) dx provided ∫ ∣ x ∣ f X ( x ) dx< ∞ where f X ( x ) is the probability density function

−∞ −∞

of the continuous random variable X.

Case 3: Mathematical expectation of some real function h(x) of a discrete random variable is
given by:

E[h(x)] =
P0
Similarly if X is a continuous random variable, then
∞

E[h(x)] = ∫ h (x) f X (x) dx

−∞

Properties of Expectation
If X and Y are random variables and a, b are constants then:

84
1. E(k) = k, where k is any constant
2. E (kX) = k E(X), where k is any constant
3. E (X + k) =E(X) + k
4. E(X + Y) = E(X) +E(Y)
5. E(XY) = E(X) E(Y), if X, Y are independent random variables
6. E(X) ≥ 0, if X ≥ 0.
7. |E(X)| ≤ E(|X|)
8. |E(XY)2| ≤ E(X2) E(Y2).
6.3.2 Variance of a random variables
Mean of X = E(X)
Variance of X =σ x 2=E ( X 2) −[E ( X ) ]
2

= E [ X −E( X)]2
Case 1: Variance for discrete random variable
If X is a discrete random variable with expected value μ then the variance of X, denoted by Var
(X), is defined by:
σ x 2=¿ Var(X) = E(X-μ)2 = E(X2) – μ2
n
= ∑ xi P ¿
2

i=1

n
Alternatively, Var(X) = ∑ (x ¿ ¿ i−μx ) P ( x i ) ¿
2

i=1

Case 2: Variance for continuous random variable

If X is a continuous random variable, then var (X),
∞
σ x = ∫ ( x−x ) f x (x) dx
2 2

−∞

Properties of Variances
 For any random variable X and constant a, it can be shown that
- Var(aX) = a2Var(X)
- Var(X + a) = Var(X) +0 = Var(X)
 If X and Y are independent random variables, then
Var(X + Y) = Var(X) + Var(Y)
More generally if X1, X2 ……, Xk are independent random variables, then
Var (X1 +X2 + …..+ Xk) = Var (X1) +Var (X2) +…. + var (Xk)

85
 (∑ )
k k
i.e., Var x i =∑ Var ( x i )
i=1 i=1

 If X and Y are not independent, then

Var (X+Y) = Var(X) + 2Cov(X,Y) + Var(Y)
Var(X-Y) = Var(X) – 2Cov(X,Y) + Var(Y)
Example 6.5: Consider the random variable representing the number of episodes of diarrhea in
the first 2 years of life. Suppose this random variable has a probability mass function as below
X 0 1 2 3 4 5 6
P(X = x) 0.129 0.264 0.271 0.185 0.09 0.039 0.017
5
i) What is the expected number of episodes of diarrhea in the first 2 years of life?
ii) Compute the variance and SD for the random variable representing number of
episodes of diarrhea in the first 2 years of life.
Solution:

a) E (X) =
1−P0=0.10
= 0.P(X=0) +1.P (X=1) +2P(X=2) + … + 6P(X=6)
= 0 (0.129) + 1(0.264) +2(0.271) + … + 6(0.017)
= 2.038
Thus, on the average a child would be expected to have 2 episodes of diarrhea in the first 2 years
of life.
b) E (X) = μ = 2.038
E(X2) = 02P(X=1) +12.P(X=1) +22P(X=2) + … + 62P(X=6)
= 0 (0.129) + 1(0.264+4(0.271) + … +36(0.017)
= 6.12
Thus, Var(X) = E(X2) – μ2 = 6.12 – (2.038)2 = 1.967 and the SD of X is
σ = √ 1.967=1.402

Example 6.6: Compute the variance of f(x) =

α=0.05 for 0 < x < 3

86
 V(x) = E(x2) – [E(x)]2

E(x2) =
−Zα=−Z0.05
P^−p0 0.875−0.90
Zc= = =−1.67
√p (1−p )/n √0.90(0.10)/40
E(x) = 0 0

Therefore, V(x) = E(x2) – [E(x)]2

2
=
χ = 0.34

6.4 Common Discrete Probability Distributions

6.4.1 Binomial Distribution

The origin of binomial distribution is Bernoulli's trial. Bernoulli's trial is an experiment where
there are only two possible outcomes, “success" or "failure". In connection with this trial, a
success may be getting heads with a balanced coin; it may be passing an examination. Whenever
we face such experiment, we use binomial distribution under the assumptions stated below. Any
experiment can also be turned into a Bernoulli trial by defining one or more possible results
which we are interested as ‘‘Success” and all other possible results as “Failure”. For instance,
while rolling a fair die, a "success" may be defined as "getting even numbers on top" and odd
numbers as "Failure".

Generally, the sample space in a Bernoulli trial is S = {S, F}, S = Success, F = failure.

Notation: Let probability of success and failure are p and q respectively.

P (success) = P(s) = p and P (failure) = P (f) = q, where q= 1- p.

Definition: Let X be the number of success in n repeated Binomial trials with probability of
success p on each trial, then the probabilities distribution of a discrete random variable X is
called binomial distribution. Let p = the probability of success & q= 1-p= the probability of

87
failure on any given trial. A binomial random variable with parameters n and p represents the
number of r successes in n independent trials, when each trial has p probability of success.
If X is a random variable, then for r= 0, 1, 2… n
n! r n−r
P (( X=r ) )= p (1−p)
r !(n−1)!
n! r n−r
P (( X=r ) )= pq , where q = 1 – p
r !(n−r )!
A binomial experiment is a probability experiment that satisfies the following assumptions.
1. The experiment consists of n identical trials.
2. Each trial has only one of the two possible mutually exclusive outcomes, success or a failure.
3. The probability of each outcome does not change from trial to trial.
4. The trials are independent.
If X is a binomial random variable with two parameters n and p then
i) E (X) = np
ii) Var (X) = npq.
1
Example 6.7: Assume that, when a child is born, the probability it is a girl is and that the sex
2
of the child does not depend on the sex of an older sibling.
a) Find the probability distribution for the number of girls in a family with 4 children.
b) Find the mean and the SD of this distribution.
Solution:
Let X be number of girls born with possible values 0,1,2,3,4
4! 1 r
a) P (getting girl) = P (( X=r ) )= ( ) ¿, n =4
r !(4−r )! 2
r 0 1 2 3 4
P (( X=r ) ) 1 4 6 4 1
16 16 16 16 16

1
b) Mean = np =4 × =2
2
1 1
√
Standard deviation (SD) = √ npq = 4 × ( ) = 1.
2 2
6.4.2 Poisson Distribution

88
Another important discrete probability distribution is the Poisson distribution. It is a discrete
probability distribution which is used in the area of rare events. The Poisson distribution counts
the number of success in a fixed interval of time or within a specified region.
Examples of random variables that usually obey the Poisson distribution are:
 The number of car accidents in a day.
 Arrival of telephone calls over interval of times.
 The number of misprints on a typed page (a group of pages) of a book.
 Natural disasters like earth quake.
 The number of suicides reported by a particular city.
 The number of customers entering a post office on a given day.
To apply the Poisson distribution, two conditions must be met:
i) The number of success that occurs in any interval is independent of those that occur
in other non-overlapping intervals.
ii) The probability of a success in an interval is proportional to the size of the interval. In
short, the two important traits of the Poisson distribution are independence and
probability.
Let X is the number of occurrences in a Poisson process and λ be the actual average number of
occurrence of an event in a unit length of interval, the probability function for Poisson
distribution is,
λx e− λ
P (( X) ) = , x = 0,1,2, ….
x!
Remarks
 Poisson distribution possesses only one parameter λ
 If X has a Poisson distribution the parameter, then E (X) = λ and Var (X) = λ,
i.e. E (X) = Var (X) =λ ,

2
 χ
Example 6.8 In a small city, 10 accidents took place in a time of 50 days. Find the probability
that there will be a) two accidents in a day and b) three or more accidents in a day.
Solution:
There are 0.2 accidents per day.
Let X be the random variable, the number of accidents per day
X ～ poiss ( λ = 0.2), X = 0, 1, 2, ….

89
 2 −0.2
(0.2) e
P (( X=2) )= =0.0164
2!
b) P (X ≥ 3) = P(X = 3) + P(X = 4) + P(X = 5) +...
= 1- [P(X = 0) + P(X = 1) + P(X = 2)]

. . . . . . since
(n−1)
= 1- [0.8187 + 0.1637 + 0.0164]
= 0.0012.

6.5. Common Continuous Probability Distributions

6.5.1 Normal Distributions

It is the most important distribution in describing a continuous random variable and used as an
approximation of other distribution. Many variables in the practical world follow this distribution,
and hence in many ways it is the cornerstone of modern Statistical Theory. It has been noticed that
empirical distributions of various types of observations in natural and social sciences are often very
close to normal distribution. In statistical analysis the distributions of observations is frequently
assumed to be approximately normal. In statistical estimation and testing of hypotheses the normal
distribution plays an important role.

A random variable X has a normal distribution with parameters μ & σ2 and it is known as a normal
random variable iff its pdf is given by:

α
for n −1
The graph of the normal distribution is known as the normal curve, which is bell-shaped:

X
Normal probability curve

90
SOME PROPERTIES OF THE NORMAL CURVE
The following are the important properties of the normal curve:

1. The normal curve is “bell-shaped” and symmetrical about the mean. The property of
2
χ
symmetry can be shown using the pdf as: .

Which is equivalent to saying that α=0.05 , 0.025, 0.01, 0.005 .

Since this is the property of the median, it follows that, for the normal distribution,

Mean = Median = Mode.

1 , 2 , 3, …, 30
2. The height of the normal curve is at its maximum when , which means,
again, Mean = Mode = Median. This property can also be verified using the first and
P ( χ ≥ χ 2α , n−1 ) =α
2

second derivative tests; that is, .

This shows x = α may be maximum or minimum value of X, but using the second
2
α , n−1
derivative test, χ , we see that the point is the maximum value.
Therefore, by property 1 and 2, we can conclude that, the mean, median and mode
coincide for the normal curve.
3. The normal curve is asymptotic to the X- axis.
4. The first and the third quartiles are equidistant from the median,

i.e., α . Or, r×c .

5. The Probability that a random variable will have a value between any two points is equal
to the area under the curve between those points.

Standard Normal Distribution

The symmetrical property of the normal distribution provides a means that is helpful in
calculating probabilities, which is also facilitated by transforming any normal distribution with
any mean and variance to the standard normal distribution.

91
By standardization we mean that the random variable X will be transformed to another random
variable whose mean is 0 and variance is 1. The normal distribution with zero mean and standard
deviation one is known as standard normal distribution. If X has normal distribution with mean
μx and standard deviation σ , then the standard normal distribution Z is given by

x−μ
Z= , for population
σ

x−x
Z= , for sample
S

Using the properties of expectations, it is now trivial to show that r × c . The pdf of

Z is, thus, given by

r× c .

The entries in Table A of the Appendix are the values of

r× c .

That is, the table gives us the probabilities that a random variable Z having the standard normal
distribution will take on a value on the interval from 0 to z, for r× c and
3.99; due to the symmetrical property of the normal curve with respect to its mean, it is
unnecessary to extend the table for negative values of Z.

Note that H 0 : .

Table value

0 Z
HA

Tabulated areas under the standard N.D from 0 to z

That is, the arrowed region is

α .

92
Basic Properties of the standard normal Curve:
1. Total area under the standard normal curve is equal to 1.
2. The standard normal curve is asymptotic to x-axis.
3. The standard normal curve is symmetric about 0.
4. Most of the area under the standard normal curve lies between z= -3 and z = 3.
Given a normal distributed random variable X with mean μ and standard deviation σ
a−μ x−μ b−μ
P ( a< X < b )=P( < < )
σ σ σ
x−μ a−μ
P ( X <a )=P( < )
σ σ
x−μ a−μ
But, =Z standard normal random variable P(Z< )
σ σ
Note: i) P (a<x<b) = P (a ≤X<b)
= P (a<X≤ b)
=P (a ≤X≤ b)
ii) P (- ∞ < Z < ∞) = 1

Example 6.9: Find the probabilities that a random variable having the standard normal
distribution will take on a value:
a) Less than 1.72; b)Less than -0.88;
c) Between 1.30 and 1.75; d) Between -0.25 and 0.45.

Solution: By using the normal table,

2
r c
( O ij − E ij )
=∑ ∑
a) χ
c2 i =1 j= 1 E ij O ij E ij .

cj
Eij =n . P ( Ai ∩B j )=n. P( Ai ). P( B j )
r c
ri
b) n =∑
i= 1
∑
j =1
O ij P ( A i )=
n
, P ( B j )=
n
,
.

c) Eij =n. ( rn )( cn )= r n.c = ( row total)×(ncolumn total)

i j i j
E ij (ri ) Eij =
r i. c
n
j

d) r × c ¿ .

α χ
2
c
X 2α , (r−1)( c−1) .

Application of the Standard Normal Distribution

Let X  N( χc>χα,(r−1)(c−1) , α ). Suppose that we want to find the probability ( 1−α ) .

2 2

93
 2
r c
(Oij−Eij) r i .c j
Since a, b, ∑∑ , and
χ 2=
c i=1 j= 1 Eij Eij =
n are known (given), we standardize a, b and X as:

2 2 2 2
(85−71.76 ) (59−72 .24 ) (65−78.24) (92−78 . 76 )
χ 2= + + + =9 .34
c 71 .76 72. 24 78.24 78 .76

Now, we need only to get the readings from the Z- table corresponding to z1 and z2 to get the
required probabilities, as we have done in the preceding example.

Also, we can find the following one-sided probabilities:

2
χ α , (r−1)(c−1)= χ 0 .05, 1=3.84
2 χ =9 . 34
, and c2 .

We have seen that a Z- value measures the distance between a particular value of X and the mean

in units of standard deviation.

Example 6.10: If X N( χ0.05,1=3.84 ,

2
μ ). find the probabilities

a) μ ; b) μ=84 ; c) μ > 84 .

Solution: As in the case of α =0 . 01 , we simply replace a and b.

a)
H A : p ≠0 . 30
2
α =0 . 05 α =0 . 01 σ1 (See Table A)

σ2
2 or 68.28%.

b) Similarly, n1 n2 =2(0.4772) = 0.9544.

c) μ1 μ2 = 2(0.4987) = 0.9974, or
99.74%.

From which we can tell that,

2
a) About 68.30% lies in the region σ 1 (1 Standard Dev. on either side).

94
 2
b) About 95.50% lies in σ 2 (2 Standard Deviations on either side).
c) About 99.7% lies in H 0 : μ1 −μ2 =0 (3 Standard Deviations on either side).

Notation:
HA :μ1−μ2≠0 denotes the value of Z for which the area to its right is equal to H :μ −μ >0 . A1 2

This notation is useful in statistical inference, and note that finding

H A is identical with reading
anti-logarithms.

Example 6.11: Find a) μ1−μ2<0 ; b) |ZC|≥Zα/2, ZC≥Zα, and ZC≤−Zα

√
2 2
x̄1−x̄2−(μ1−μ2) σ1 σ2
ZC = ~N(0,1)andσ x1−x2= +
Solution: a) σx̄1−x̄2 n1 n2
corresponds to an entry of 0.5 - 0.01 = 0.4900.

In Table A, look for the value closest to 0.4900, which is 0.4901, and the Z value for this is Z=
σ x̄
2.33. Thus, 1
− x̄2
.

b) Again, σ 1 is obtained as 0.5 - 0.05 = 0.4500, which lies exactly between 0.4495 and 0.4505,

corresponding to Z = 1.64 and Z= 1.65. Hence, using interpolation, σ 2 .

Example 6.12: Suppose that X N (165, 9), where X = the breaking strength of cotton fabric. A
sample is defective if X<162. Find the probability that a randomly chosen fabric
will be defective.

Solution: Given that H 0 :μ1−μ2=0 and H A :μ1−μ 2≠0 ,

α =0 . 05
Z α /2 =1 . 96⇒ |Z c|≥1 . 96 (Since Z C=
x̄ 1− x̄ 2 −( μ1 − μ2 )
σ x̄ −x̄
1 2 )

n1=89,n2=107, x̄1=−0.20, x̄2=−0.26 (By symmetry)

s1=0.18 for σ 1 , s2=0.13 for σ2 μ1 −μ 2 (Table value for Z = 1)

6.5.2 Chi-Square Distribution

95
Chi-Square distribution may be derived from normal distributions, if Xi (i = 1, 2… n) are n
independent normal varieties with mean μi and variance σ i2 (i= 1, 2, … , n) then
χ2¿ is a chi-square variate with n degrees of freedom. The probability density function of the χ 2 –
distribution is given by
2
n −χ
1 ( −1 )
f ( χ )=
2
( χ ¿¿ 2) 2
e 2
¿ , 0< 2<∞ where n is the degree of freedom.
n/ 2 n χ
2 Γ( )
2
Since the Chi-square distribution arises in many important applications, its values have been
Z =2.63>1.96 for α =0.05,
extensively tabulated. Table C at the end of this module contains values of c
μ −μ
0.025, 0.01, 0.005 and n=1, 2, 3, …, 30, where 1 2 is such that the area to its right under the

√
2 2
σ 1 σ2
+
Chi-square curve with n degrees of freedom is equal to Z α /2 . That is, n1 n 2 is such that if X is a
random variable having a Chi-square distribution with n degrees of freedom, then
σ 1 σ2 is known as the level of significance. When n is greater than 30, the table
cannot be used and probabilities related to Chi-square distributions are usually approximated
with normal distributions.

n1 and n2

0
σ1

Properties of Chi-square Distribution

1. The exact shape of the distribution depends upon the number of degrees of freedom n. In
general, when n is small, the shape of the curve is skewed to the right and as n gets larger, the
distribution becomes more and more symmetrical.
2. The mean and variance of the χ 2distribution are n and 2n respectively.
3. As n → ∞ the χ 2 distribution approaches a normal distribution.
4. The sum of independent χ 2varieties is also χ 2 variety.
6.5.3 The t-distribution

96
Let X1,X2,….Xn be a random sample drawn from a normal distribution having mean μ and
standard deviation σ (unknown but estimated by S, sample standard deviation).
X−μ
t=
The statistic S has t – distribution with (n-1) degree of freedom where X is sample mean
√n
and S is standard deviation.
In view of its importance, the t distribution has been tabulated extensively. Table B at the end of

this module contains values of

σ 2 , for σ
2
= 0.10, 0.05, 0.025, 0.01, 0.005,etc and
tC=
¯x1−¯x2−(μ 1−μ 2) ¯x1−x̄2−(μ1−μ2 )

√
=
Sp2 S2p Sx1−x2
+
n1 n2 = 1, 2, 3,

…, degrees of freedom; where

n1+n2 -2 is such that the area to its right under the curve of the t
n −1 S2+ n −1 S2
distribution with (n-1) degrees of freedom is equal to .
( ) ( )
2 1 1 2 2
S p=
n1 +n2−2

Notation: tα ((n-1)) stands for a value of t with (n-1) degree of freedom the right of which an area
equal to a in reading the tabulated values.

n1+n2−2

-α 0
H0:μ1−μ2=δ

Student’s t Distribution

Note: 1. The table value does not contain values of

μ1−μ2≠δ, μ1−μ2>δ for μ1−μ2<δ > 0.50, since the curve is
symmetrical about

t=0 (like the normal distribution) we have,

|tC|>t , n1+n2−2, tC¿tα,n1+n2−2, or tC¿−tα,n1+n2−2.

α
2 =
H 0 :μ1=μ2 .
2. When (n-1) =30 or more, probabilities related to the t distribution are usually
approximated with the use of normal distributions.

Example 6.13: For a t-distribution with n=20 , find

μ1−μ2=0 values leaving an area of

a) 0.05 to the right; c) 0.10 to the left;

b) 0.975 to the right; d) half of H A :μ1≠μ2 =0.01 on either side.

97
Solution; referring to Table B with (n-1) =19 df, we have

a) α=0.05 =1.729; 2
c) n1=25, x̄1=47.5,s1=23.04 = -1.328.

( 25−1) 23.04+(26−1) 26.01

=
2
b) n2=26, x̄2=37.5,s2=26.01. 2.093; d) 25+26−2 2.861; & =24 . 56

Applications of t Distribution

The t distribution has wide applications in Statistics, only some are listed below:

a) Test of population Mean ( One-sample t-test)

When we are dealing with a random sample of size n<30, from a normal population, when
S p=4.96 is unknown, the t distribution with n-1 degrees of freedom, is used to test the hypothesis

that the population mean s x −x

1 2 equals a given value (Say, √ 11
4. 96 + =1.39
25 26 ), against the alternatives:
t α =t0 .025, 49=2.0
2
,49
,

or
tC =
47.5−37.5
1.39
=7.19.
, or
t c=7.19>2.0 .

X  (n1−1)S21+(n2−1)S22
S2p=
Then, we calculate t= S / n , which is to be compared with the table value n1+n2−2 , or α with n-1
degrees of freedom.

Note: The assumptions underlying student’s t-distribution for such tests are:

a) The parent population from which the sample is drawn is normal.

b) The sample observations are independent; that is, the sample is random.
c) The population standard deviation ( μ1−μ2 ) is unknown.
d) n is small; that is n<30.
Example: 6.14: In 16 one-hour test runs, the gasoline consumption of an engine averaged 16.4
gallons with a standard deviation of 2.1 gallons. In order to test the claim that the average
t α/2,n +n −2
gasoline consumption of this engine is 12.0 gallons per hour, calculate the t value and 1 2 , for
Sp =0.05.

x̄1−x̄2−(μ1−μ2 )

Solution: Substituting n=16, √ =12.0,

1 1
=16.4, and S=2.1 in the formula, we get
t= +
n 1 n2 Sx −x
1 2

t= S / n =
X 
~ =8.38; and the table value for n-1=15 is
S x −x
1 2 =1.753.

98
b) t-Test of the equality of two means (two-sample t-test)
In many problems of applied research, we are interested in hypotheses concerning the difference
2
sx
between the means of two population means, 1 − x2 , or the equality of two means
2
s . In such tests, the following are assumed:

I. The parent populations from which the samples have been drawn are normally
distributed;
2
II. The two population variances are equal, though unknown: s .
III. The two samples are random and independent of each other;
IV. The sample sizes are small: n1 and/or n2 are <30.
c) t-Test of correlation and regression coefficients
In a normal regression and correlation analyses, it is used to test:

2
a) if the population regression coefficient equals to a certain constant ( s 1 ); &
b) if the population correlation coefficient is significantly different from zero.
Exercise 6

1. From a lot containing 20 items, of which 5 are defective, 4 are chosen at random. Let X
be the number of defectives found.

a) Write down the pmf of X. b) Find the probability distribution of X.

c) Find E(X) and V(X).

s2
2
2. If X has a pdf of , for 0 <x <1, and o elsewhere, find

a) P(X < 0.5); b) E(X) and V(X);

c) a if s2
1 ; d) b if s 2
2 .

3. The amount of bread X ( in hundreds of kg) that a certain bakery is able to sell in a day
is found to be a continuous r-v with a pdf given as below:

99
 α
a) Find k; b) Find the probability that the amount of bread that will be sold tomorrow is
i) More than 500kg, ii) between 250 and 750 kg;

c) Find the expected amount of bread to be sold in any day.

4. Find the value of Z if the area between -Z and Z is a) 0.4038; b) 0.8812; c) 0.3410.
5. The reduction of a person's oxygen consumption during periods of deep meditation may
μ1 −μ 2
be looked up on as a random variable having the normal distribution with cc
t α /2,2(n−1)
per minute and cc per minute. Find the probabilities that during such a period a
person's oxygen consumption will be reduced by

a) at least 33.4 cc per minute; b) at most 34.7 cc per minute

6. For a Chi-square distribution with n-1 =14 degrees of freedom, find the table value such
that the area
a) to its right is 0.01;
b) to its left is 0.975.
7. What is the probability that a family with five children will have 3 boys and 2 girls?
8. In a research, rats are injected with a drug that inhibits body synthesis of protein. By the
previous research, it was found that the probability of a rat dying from the drug before the
experiment is over is 0.2. If 10 rats are used:
i) How many are expected to die before the experiments ends.
ii) What is the probability that at least eight will survive.
9. The heights of 10 males of a certain locality are found to be 70, 67, 62, 68, 61, 68, 70, 64,
64, and 66 inches. If it is desired to test if the average height is 64 inches, at √ =0.05,
s21 s 22
+
n n

a) calculate the t value;

b) find the table value

H0: μ1=μ2 or H0: μ1−μ2=0 .

100
 CHAPTER 7: SAMPLING AND SAMPLING DISTRIBUTION OF THE SAMPLE MEAN

After completing this unit, the student should be able to

 Describe the basic concepts of sampling.
 List down and explain random sampling versus non-random sampling techniques.
 Identify the causes of non-sampling error.
 Develop the sampling distribution of the mean.
 Construct the probability distribution of the mean.
 Calculate the means, standard deviation and variance of sample means and sample
Proportion.
 Differentiate the sampling distribution of the sample means (sample proportion) when
the population is normal or non-normal.
 Explain the importance of central limit theorem for statistical inference.
7.1 Basic Concepts
Population: is the complete collection of individuals, objects or measurements for which
inferences are to be made. The population represents the target of an investigation, and the
objective of the investigation is to draw conclusions about the population and it should be
defined on the basis of the objective of the study by the investigator.
Example 7.1
-All customers of electric supply company.
-All students of DMU.
-Population of farms having a certain type of natural fertility.
-Population of households in a certain village.
Sample: A sample from a population is the set of measurements that are actually collected in the
course of an investigation. It should be selected using some predefined sampling technique in
such a way that they represent the population very well.
Sampling (elementary) unit:- the ultimate unit to be sampled or elements of the population to
be sampled.
Example 7.2: If somebody studies economic status of the house holds, households is the
sampling unit. If one studies performance of freshman students in some college, the student is
the sampling unit.
Sampling frame: is the list of all elements (sampling units) in a population.
Example 7.3:
 List of households of a certain city.

101
  List of students in the registrar office of the university.
Parameter and Statistic are basic terms in sampling theory. Parameter is a value calculated
from the population. For instance population mean, population variance, population proportion is
parameters. Statistic is a value calculated from a sample. Sample mean, sample variance, sample
proportion, etc are statistics.
Sampling error: A type of error that may arise due to in appropriate sampling techniques
applied .A sampling error is the difference between a sample statistic and its corresponding
parameter. We can make probabilistic statements about this sampling error only if we have a
probability sample.
Non-sampling error: In addition to sampling error, the sample estimate may be subject to other
errors, sampling errors. Errors in observation, interview or measurement error, errors due to non-
response and errors in data processing: editing, coding, etc. The non-sampling error is likely to
increase with increase in sample size. For instance a census survey may have non-sampling
errors in large amount collected in the course of an investigation. It should be selected using
some predefined sampling technique in such a way that they represent the population very well.
7.2 Reasons for sampling
Sample survey saves money: It is possible to collect information from sample households and
obtain estimates that reasonably approximate the actual characteristics of a large population.
It obviously cheaper to gather information from 100 households rather than from 10,000
households.
Sample Survey saves time: sample survey requires a smaller scale of operations at all stage and
it reduces data collection and processing time.
Sample survey provides higher level of accuracy: This accuracy can be achieved through
more selective recruiting of interviewers and supervisors, more extensive training programs, a
closer supervision of the personnel involved and a more efficient monitoring of the field work.
Sample survey could be the only option for the study in some specialized area. For example,
there are some cases where information of technical nature requires highly trained personnel and
specialized equipment like in medical areas.
Experimentation could be destructive in nature like testing industrial products such as testing
the average duration of burning of bulbs, and testing the quality of wine, beer, etc. In this case
sampling is the only feasible means of study.

102
7.3 Sampling Techniques
The technique of selecting a sample is important in sampling theory and usually it depends upon
the nature of the investigation. The commonly used sampling techniques may be broadly
classified as: Non Probability and Probability Sampling.
A. Random Sampling or probability sampling.
Probability sampling techniques is a method of sampling in which all elements in the population
have a pre-assigned probability to be included in to the sample.
In this sub-section, four different techniques of taking a random sample are discussed.
a/ Simple random sampling, b/ Stratified random sampling, c/ Cluster sampling, d/ Systematic
sampling
a) Simple Random Sampling
In statistics, a simple random sample from a population is a sample chosen randomly, so that
each possible sample has the same probability of being chosen. One consequence is that each
member of the population has the same probability of being chosen as any other. In small
populations such sampling is typically done "without replacement", i.e., one deliberately avoids
choosing any member of the population more than once. Although simple random sampling can
be conducted with replacement instead, this is less common and would normally be described
more fully as simple random sampling with replacement. Simple random sampling is a method
of selecting n units out of a finite population of size N by giving equal probability to all units, or
a sampling procedure in which all possible combinations of n units that may be formed from the

H : μ >μ
finite population of size N units have the same probability of selection. There are A 2 1 distinct
possible samples in the case of sampling without replacement; the chance of selecting each one

of them is
H A :μ 2 <μ 1 There are N
n

possible samples in the case of sampling with replacement, the

chance of selecting each one of them is 1/ N n. Conceptually, simple random sampling is the
simplest of the probability sampling techniques. It requires a complete sampling frame, which
may not be available or feasible to construct for large populations. Even if a complete frame is
available, more efficient approaches may be possible if other useful information is available
about the units in the population.
Simple random sampling is free of classification error, and it requires minimum advance
knowledge of the population. It best suits situations where the population is fairly homogeneous
and not much information is available about the population. If these conditions are not true, some

103
other types of sampling techniques may be a better choice. Lottery method and computer
generated random numbers are used to select a random sample in simple random sampling.
i) Lottery method: This is a very common method of taking a random sample, under this
method; we label each member of the population by identifiable ticket or pieces of papers.
Tickets must be of identical size, color and shape. They are placed in the container and well
mixed before each draw and then draws may be continued until a sample of the required size is
selected. This shows that selection of items depends entirely on chance.
Example 7.4: If we want to take a sample of 25 persons out of a population of 150, the
procedure is to write the names of all the 150 persons on separate slips of papers, fold these slips,
mix them thoroughly and then make a blindfold selection of 25 slips without replacement.

ii) Table of random numbers

This is an alternative method of selecting a simple random sample. It is constructed from the
digits 0, 1, 2,…, 9. There are several tables available in standard books of Statistics.

Suppose we want to select a sample of size n, then

- Make a list of population to be sampled;

- Give a distinct code number to each unit of the population;
- Choose the direction of selection randomly;
- Take n units whose code numbers coincide with the random numbers as numbers of the
sample by omitting those random numbers which do not exist on the list and repeated
numbers if an element is not appear more than once in a sample.

104
 Table of Random Numbers

Column

Row 1 2 3 4 5 6 7 8

1 57172 42088 70098 17333 26902 29959 43909 49607

2 33883 87680 24923 15659 O9839 45817 89405 70743

3 77950 15344 35609 87119 15859 74577 42791 75889

4 11607 26596 16796 24498 17009 67119 60557 49521

5 56149 55678 38169 47228 49931 94303 67448 31286

6 80719 65101 77729 83949 83358 75230 56624 27549

7 93809 19505 82000 79068 45552 86776 48980 56684

8 40950 86216 48161 17646 24164 35513 94057 51834

9 12182 59744 83710 41125 14291 74773 66391 50031

10 13382 48076 73151 48724 35670 38453 63154 58116

11 38629 94576 48859 75654 17152 66516 78796 73099

12 60728 52063 12431 23898 23683 10853 O4038 75246

13 O1881 99056 46747 O8846 O1331 88163 74462 14551

14 23094 08831 24387 23917 O7421 97869 88092 72201

Example 7.5: Suppose that N= 40 and we want to select n=10 without replacement, starting
with the 3rd row and 2nd column by reading vertically using the above random table, we get
(ignoring the numbers greater than 40):

Solution: starting with the 3rd row and 2nd column by reading vertically we will get:

15, 26, 19, 08, 24, 35, 16, 38, 12 and 17.

b/ Stratified random sampling

In stratified sampling, the population of N units is sub-divided into k sub-populations, called

strata, so that the units in each stratum are as homogeneous as possible and the means of the

105
 different strata are as different as possible. These sub-populations should be non-overlapping
(x1 , y1 ), (x2 , y2 ), (x 3 , y3 ), ... , (xn , yn ) d1=y1−x1, d2=y2−x2,d3=y3−x3,. dn=yn−xn
so that they comprise the whole population such that , where
t = √sd̄ /n
c 2

represent the population size in the strata. Then a sample is drawn from each stratum
d

2
sd H 0 : μ1 =μ2
independently, the sample size within the stratum being such that
H A : μ2 > μ 1
. The procedure of taking samples in this way is called stratified sampling.
If the sample is taken randomly from each stratum, the procedure is known as stratified random
sampling.

Remarks:

In stratified random sampling, the following two points are equally important to ensure
accuracy:

a) proper stratification of the population into various strata, and

b) a suitable sample size from each stratum.

For example a population can be stratified based on the following variables:

 Sex (male, female); Age (under 18, 18 to 28, 29 to 39); Occupation (professional, other);
Geographical classifications; Administrative regions, etc.
c/ Cluster Sampling:
The population is divided in to non-overlapping groups called clusters. A simple random sample
of groups or cluster of elements is chosen and all the sampling units in the selected clusters will
be surveyed in the case of single stage cluster sampling. Clusters are formed in a way that
elements within a cluster are heterogeneous, i.e. observations in each cluster should be more or
less dissimilar. Cluster sampling is useful when it is difficult or costly to generate a simple
random sample. For example, to estimate the average annual household income in a large city we
use cluster sampling, because to use simple random sampling we need a complete list of
households in the city as sampling frame. To use stratified random sampling, we would again
need the list of households. A less expensive way is to let each block within the city represent a
cluster. A sample of clusters could then be randomly selected, and every household within these
clusters could be interviewed to find the average annual household income.
d/ Systematic Sampling:

106
Systematic sampling is the selection of every kth element from a sampling frame, where k, the
sampling interval and
k = population size / sample size = N/n.
Using this procedure each element in the population has a known and equal probability of
selection. This makes systematic sampling functionally similar to simple random sampling. It is
however, much more efficient and much less expensive to do. Like simple random sampling a
complete list of all elements with in the population (sampling frame) is required. The procedure
starts in determining the first element to be included in the sample. It is however, much more
efficient and much less expensive to do. Suppose that we have a complete and up-to-date list of
the N units in the population numbered from 1 to N in some order. To select a sample of size n,
if N is an integral multiple of n, N = k*n for some integer k,
k = population size / sample size = N/n.
The procedure starts in determining the first element to be included in the sample, select a unit i
randomly from the first group, i≤ k as the first element. The second unit will be (i+k)th element
from the frame. Totality we have a sample of size n from the population of size N, i th , (i+k)th ,
(i+2k)th ,… (i+(n-1)k)th element of the population are taken as a sample.
Example 7.6
Suppose that N = 20 and we want to select a sample of size 4, so that k = N/n =20/4 = 5.
The first element in the sample is selected from the first 5 units randomly, say 3 rd, which is the
random start. Then, every 5th unit is selected, and the sample contains the 3 rd, 8th, 13th and 18th
units of the population.
B. Non-Random Sampling or non-probability sampling.
It is a sampling technique in which the choice of individuals for a sample depends on the basis of
convenience, personal choice or interest.
Types of non-random sampling are:
1. Judgment sampling.
2. Convenience sampling
3. Quota Sampling.
1. Judgment Sampling
In this case, the person taking the sample has direct or indirect control over which items are
selected for the sample. This method is mainly used for opinion surveys but is not recommended
for general use, as it bias of the sampler.
2. Convenience Sampling

107
In this method, the decision maker selects a sample from the population in a manner that is
relatively easy and convenient.

3. Quota Sampling
This is a type of judgment sampling and may be the most commonly used one in the non-
probability category. In a quota sample, quotas are set up according to some specified
characteristics such as income groups, age groups, political or religious groups, etc. Within the
quota, the selection of sampling units depends up on personal judgment.
7.4 Sampling Distribution of the sample mean
Consider all possible samples of size n that can be drawn from a given population (either with or
without replacement). For each sample, we can compute a statistic (such as the mean & the
standard deviation) that will vary from sample to sample. In this manner we obtain a distribution
of the statistic that is called its sampling distribution.
Steps for the construction of Sampling Distribution of the mean
1. From a finite population of size N, randomly draw all possible samples of size n. There are N n

possible samples if sampling is with replacement and there are α=0.05 possible samples if

sampling is without replacement.

2. Calculate the mean for each sample.
3. Summarize the mean obtained in step 2 in terms of frequency distribution
Example 7.7: Suppose we have a population of size 5, consisting of the age of five children 3, 5,
7, 9, and 11. Population mean is 7 and population variance is 8. (Consider sampling without
replacement).
Take samples of size 2 and construct sampling distribution of the sample mean.
Solution:
2
Step 1: N= 5, n=2 we have
d̄=2.2, and sd=8.62 =10, possible samples.
(3,5), (3,7), (3,9), (3,11), (5,7), (5,9), (5,11), (7,9), (7,11) and (9, 11)
Step 2: Calculate the sample mean for each sample:
Means = 4, 5,6,7,6,7,8,8,9,10 respectively.
Step 3: Summarize the mean obtained in step 2 in terms of frequency distribution.
2.2
t = =2.37
√ 8 . 62/ 10
4 5 6 7 8 9 10 Total
fi 1 1 2 2 2 1 1 10
fi 4 5 12 14 16 9 10 70

108
 f (−7)
2
9 4 2 0 2 4 9 30

a) Mean of sample means , E( α ) =

∑ f i = 70/10 = 7
∑ fi
k

b) t
Variance of sample means, var( ) =
α/2,n−1
∑ (−E ())
2
= 30/10 = 3
i
k

n1 =
n2 =3

Example 7.8: Three students have taken a class test which is marked out of 10. We want to
estimate the mean mark using the sample mean as the estimate of the population mean. We take
a sample of size 2 in two cases and suppose the marks of the three students are 1, 2 and 6.
The population mean μ is (1+2+6)/3 = 3

The population variance x 1 = ∑ ¿¿ ¿ = 14/3.

i) Sampling without replacement

In this type of sampling an observation is included in the sample only once and is selected
randomly without any preference or conscious effort.
If sampling is without replacement we can take 3C2 =3 possible samples; the possibilities are
given below.

Possible sample (1,2) (1,6) (2,6)

Sample mean 1.5 3.5 4

The sample mean is a random variable, and we see that it can take three possible values. We can
now write down its probability distribution as follows
n1 1.5 3.5 4 Total

p
P( n 2 = 1 )
1/3 1/3 1/3 1

(−3)
2
2.25 0.25 1 3.5

i) Mean of sample means E( n − n )=∑ X i p ( ¿ ) =1.5(1/3) + 3.5(1/3) +4(1/3) =3 = population mean

x1 x2
1 2

i.e., Mean of sample means E( p1=p2=p ) = population mean

109
 k

ii) Variance of sample means, var( p1−p2=0 ) =

∑ (−E ())2 3.5/3 = 1.17
i
=¿
k
where k is number of sample mean.

In which if sampling without replacement,

p =
^p=
x 1+ x 2
n 1 +n2 =14/12 = 1.17.

ii) Sampling with replacement

In this type of sampling an observation has a chance to be selected at each draw.
Suppose that we take the sample with replacement, there are 32 = 9 possible samples.

(1,1) (1,2) (1,6) (2,1) (2,2) (2,6) (6,1) (6,2) (6,6)

Sample
Sample mean 1 1.5 3.5 1.5 2 4 3.5 4 6

The sample mean is a random variable & its probability distribution is:
x1 x2

1 1.5 2 3.5 4 6 Total

−
n1 n2
Z=

√ ( )
1 1
^p (1−^p ) +
n1 n2

x +x
p^ 1− ^p2
^p= 1 2 1/9 2/9 1/9 2/9 2/9 1/9 1
P( s = n +n )
^p 1− ^p2 1 2

H0: p1−p2=0 P( √ = H0:μ1−μ2=0 )

Z=
^p1−^p 2−(p1−p 2)
p^ 1(1−^p1 ) p^ 2(1−^p2 )
n1 n2
+
1/9 1/3 2/9 7/9 8/9 6/9 3

fi(−3)
2
4 4.5 1 0.50 2 9 21

i) Mean of sample means E( p2 )=∑ X i p ( ¿ )

=1(1/9) +1.5(2/9) + 2(1/9) +3.5(2/9) + 4(2/9) + 6(1/9) =3

Mean of sample means, E( α ) = population mean.

k

p −p
ii) Variance of sample means var( ) =
1 2
∑ (−E ())2
=21/9 = 2.33 i
k
Where k is number of sample means

H 0 : p1978−p1979 =0 = HA:p1978≠p1979 = 14/6 = 2.33

In which if sampling with replacement,

H A : p1978−p1979 ≠0 = α=0. 01 = 14/6 = 2.33

110
In each case the expected value of the sample mean equals the population mean. This explains
why the sample mean is a good estimate of the population mean. If we use the sample mean as
an estimate of the population mean we will sometimes overestimate it, and sometimes under-
estimate it, but “on average” we will be accurate.
The example above illustrates an important result.

Remark:
∑ fi
1. Mean of sample means= E( p1978=0.038, ) = ∑ X i p ( ¿ ) = population mean
∑ fi =

2. Variance of sample means, √ (0.0308) ( 0 .9692) (15,0001 +10,0001 ) ( if sampling is with replacement)

3. Variance of sample means

H0 ,(if sampling is with out replacement)
2 2
H
The quantity O
: σ =σO is finite population correction (fpc), and if n/N <0.05, fpc is ignored.
Note: the square root the variance of sample means is known as standard error.
The distribution of sample means depends on distribution of the population, sample size and
whether population variance is known or unknown. A sample may be from a normally
distributed population or from a non normally distributed population, from a population with
variance is known or un known and the sample size may be large or small.
Case-I: If sampling is from a normally distributed population with known variance: When
sampling is from a normally distributed population with known variance, the distribution of
2 2
sample means σ ≠σ O , is normal what ever the sample size.
Example 7.9: The average height of Christmas tree farm is normally distributed with mean 68
inches & variance 9 inches square. Find the probability that the mean height of a random sample
of 16 Christmas tree is more than 70 inches.
Solution:
Let X be the height of trees with mean 68 and variance 9.
2 2
A sample of size 16 is taken, the sample mean is a random variable ( σ >σO ),
(n−1)S 2
x
2

c= σ 2
2
σ 2 <σ 2O ~ α,ν
ν
=χ ~ , since the population is normally distributed, probability of a

sample mean is greater than 70 is P( α >70) = p(Z>

H0 ) = p(Z>2.67) = 0.0038.

111
Case-II: When sampling from a non normal population and when the sample size is large.
If sampling is from a non normal population and when the sample size is large the distribution of
2
c
χ depends on Central Limit Theorem.

The Central Limit Theorem

2
α/2,ν
If X1, X2, …, Xn is a random sample from a population with mean μ and variance χ , then as n
2
c
goes to infinity the distribution of the sample mean, χ , approximates normal distribution with
2 2
1−α/2,ν α,ν
mean μ and variance χ . In short as n gets large number, χ ~
2
c χ .

2
c
We can standardize this to get χ ~ N (0, 1) (approximately as n gets large). When
2
1−α, ν
population variance is unknown χ ~ N (0, 1) (approximately as n gets large).
Example 7.10: The mean weight of 500 male students at a certain university is 151 pounds (lb)
and the standard deviation is 15 lb. assuming that the weights are normally distributed. Suppose
that a sample of 64 students is taken, what is the probability that the weight in the sample is more
than 154.75 lb?
Solution:
As we have taken a large (n=64) sample we can use the Central Limit Theorem. This says that
the mean weight of the sample can be approximated by a normal random variable with a mean of

151 and a variance of 225. If we let α=0. 05 be the mean weight of the students, it is required to find
2 2 2
P( σ >154.75) = (2.1) ~ χ

2 2
P(x
2 (n−1)S2
17∗4. 4
c= σ2 = 6. 25 =11 .9 5 .
H :σ =4.4 > H O : σ <4.4 ) =
>154.75) = p( O P (Z>2.00) = 0.5 – 0.4772 = 0.0228.
Example 7.11: Suppose that 150 customers enter a supermarket on a given day. Each customer
spends a random amount. All they know about the distribution of these expenditures that its
mean is 7.50 birr and its standard deviation is 3.40 birr. What is the probability that a person, on
average, spent more than 8.00 birr during the day?
Solution:

112
We have n = 150 which is large enough to use the Central Limit Theorem. Mean =7.50 and
standard deviation = 3.40.
2
Let X be the amount of an individual’s expenditure during the day. χ 0.95,17 ~ N(7.50, 11.56)

Let x the average amount of an individual’s expenditure during the day, it is required to find P(
2

α >8)
2 (n−1 )S 2 2
P( α >8.00) = p( σ: 2
> Xα/2
) = p(Z > σ
) = p(Z>1.80) = 0.5 – P (0<Z<1.80)
= 0.5 – 0.4641 = 0.0359
This means there is only 0.0359 probabilities that a person will spent larger than 8.00 birr on
average.
Case-III: When sampling is from normally distributed population with unknown population
variance,
( n−1 ) S 2
2
a) If the sample size is large, X 1−α /2 ~ N (0, 1), where S is an estimate of α=0.01 .

b) If the sample size is small (n<30),

α=0.05 ~t(n-1). t has t-distribution with (n-1) degree of
freedom, where S is an estimate of α=0.01 .
7.5 Sampling Distribution of the sample Proportion
In situations where it is not possible to measure the characteristic under study, but is possible to
classify the whole population in various categories with respect to the attributes they possess,
consideration is usually given to estimating the population elements that belong to a defined
category of class. Suppose that we have two complementary and mutually exclusive class, C and
C' such that every unit in the population falls into either of them.
In order to know how many of the units fall in class C, we define a counting variable as

Xi ¿ {1,∧if theunit fall∈class C

0 ,otherwise
If the number of units falling in C is denoted by A for the population and by a for the sample,
then
N

∑ X i= A and hence the population proportion denoted by P is given by P = A/N.

i

n
a
Given a simple random sample of n units, the sample proportion denoted by p= =
∑ X i from
n i
n

the formula, we see that α=0. 05 and p are essentially identical. In fact p is special case of α=0. 05 , the case

113
where possible values of Xi are only 0 and 1.Consequently p possesses all properties of α=0. 05 . p is
an estimate of P, with variance

2 N

var(p) = var( α=0. 01 ) = σ where

=
n

( )( )
n n
n∑ xi yi − ∑ xi ∑ yi
∑xy−nx̄ȳ = i=1 i=1 i=1 = ∑ (x−x̄ )(y−ȳ )
()
∑x2−nx̄ 2 n 2 ∑(x−x̄)2
n ∑x 2− ∑ xi
i
i=1 =
∑ ¿¿ ¿ ¿
i = PQ

var(P) =
y
Where Q=1-P is proportion of units falling in class C'.

2
var(P) = x is estimated by using sample values as

^p
var( ) =
xy = y
2

Where sampling fraction, f = n/N

36343−(8)(67.125)(67.625)
∑ xy−n x̄ ȳ = 2 =
by its estimator s2 = 36105−(8)(67.125) .
=
This expression is obtained by replacing ∑ x2−nx̄2

The sampling fraction can be ignored, when N is large relative to sample size n, n/N<0.05.
28.375
^p =0.482 √❑
var( ) = 58.875 and the standard error of p is .

Sample proportion p is normally distributed with mean P and variance var (p) =
^y65.5=35.27 + 0. 482(65.5)=6 .841
Example 7.12
In a simple random sample of size 100, from a population of size 500, there are 37 employed
persons in the sample.
a) Estimate proportion of employed persons in the population.
b) Calculate the standard error of p.
Solution:
a) Population proportion P is estimated by p= a/n = 37/100 = 0.37. 37% of the population is
employed.
b) Standard error of p is√ ❑ = √ ❑ = 0.0434.

Exercise 7
1. A certain type of bacteria occurs in all raw milk. Let X denotes the bacteria count per ml
of milk. The public health dep’t has found that if the milk is not contaminated, then x-
has a normal distribution. The population distribution has a mean of 2250 with standard

114
 deviation 300. In a large commercial dairy, the health inspector takes 42 random samples
of milk produced each day. At the end of the day the bacteria count in each of the 42
samples is averaged to obtain the sample average bacteria count X .
a/ Assuming that the milk is not contaminated , what is the distribution of x ?
b/ Assuming that the milk is not contaminated, what is the probability that the sample
average bacteria count X for one day is between 2500 and 2600 bacteria/ml?
2. The diastolic blood pressure (DBP) of adult males between the age of 25&45 years in a
certain society is normally distributed with mean 78 mm and s.d. 10 mm. The DBP is
considered elevated if it exceeds 95 mm. An insurance company has 5000 adult male
policy holders b/n 25 and 45 years of age. What is the probability that 10% or less of
them have elevated DBP?
3. Determine the type of sampling used
a/ An interviewer in a mall is told to survey every fifth shopper starting with the second.
b/ A researcher randomly selects 5 of the 70 hospitals in a metropolitan area and then
surveys all of the surgical doctors in each hospital.
4. The amount of sulphur in a daily emission from a factory has a normal distribution with
mean of 134 pounds and a standard deviation of 22 pounds. For a day selected randomly,
find the probability that the mean amount of sulphur emission will be less than 130
pounds.
5. A population consists of the four numbers, 3,7,11, 13 and 15. Consider all possible
samples of size 2 drawn from this population without replacement. Find a) The sampling
distribution of sample means
b) The mean of sample means c) The standard deviation of the sample means.

115
 CHAPTER 8: STATISTICAL INFERENCES
The process of inferring information about a population from a sample is known as statistical
inference. This chapter has two major parts. The first part is statistical estimation discusses the
method of estimating a population parameter by using statistic, point estimation. It also explains
the concepts of confidence interval. The second part is hypothesis testing describes the different
techniques of testing a given tentative assumptions by applying an appropriate test statistic.
Objectives
After completing this unit, the student should be able to:
 Explain the concepts of statistical estimation and the confidence interval.
 Distinguish interval estimation from point estimation.
 Calculate and interpret point estimate of population mean and population proportion.
 Define the concept of hypothesis testing and differentiate types of tests.
 List down the basic steps in hypothesis testing.
 Follow the steps to solve problems on hypothesis testing.
 Identify the appropriate test statistics for a given practical problem.
8.1 Statistical Estimation

It is the procedure of using a sample statistic to estimate a population parameter. This is one way
of making inference about the population parameter where the investigator does not have any
prior notion about values or characteristics of the population parameter. A statistic used to
estimate a parameter is called an estimator and the value taken by the estimator is called an
estimate. Statistical estimation is divided into two main categories: Point Estimation and Interval
Estimation.
Point Estimation:- When we use a single value of a statistic to estimate the corresponding
parameter of a population, it is called point estimation. It is a common way of estimating a
parameter, where a random sample of n observations is selected from a population and the
statistic is calculated.
Examples:
 A sample mean is an estimate for population mean μ. That is, X is an estimator for
population mean μ.
 A sample variance is an estimate for population variance. That is, S2 is an
estimator for population Variance σ 2.
 A sample proportion estimate for population proportion.

116
Properties of best estimator
The following are some qualities of an estimator.
 It should be unbiased.
 It should be consistent.
 It should be relatively efficient.
To explain these properties let θ^ be an estimator of θ.
1. Unbiased Estimator: An estimator whose expected value is the value of the parameter
being estimated. i.e. E(θ^ ) = θ.
2. Consistent Estimator: An estimator which gets closer to the value of the parameter as
the sample size increases. i.e. θ^ gets closer to θ as the sample size increases.
3. Relatively Efficient Estimator: The estimator for a parameter with the smallest variance.
This actually compares two or more estimators for one parameter.
Interval estimation:- It is unlikely that any particular estimate will be exactly equal to the
population mean, surely an estimate can be greater than or less than the parameter .That is, it is
not always possible to estimate population parameter with out any error so allowance is needed
for such error .We take interval, ranges of values about an estimate in which the parameter may
lie. This procedure is Interval estimation.
It is the procedure that results in the interval of values of a parameter. Interval estimates indicate
the precision or accuracy of an estimate and are, therefore, preferable to point estimates. It deals
with identifying the upper and lower limits of a parameter. Confidence interval for the parameter
is:
Estimate ± critical value × Standard error of the estimator
Example 8.1: Confidence interval for the population mean is:
X ± Critical value × Standard error of ( X )
8.1.1 Confidence interval Estimation for population means

Although X possesses nearly all the qualities of a good estimator, because of sampling error, we
know that it's not likely that our sample statistic will be equal to the population parameter, but
instead will fall into an interval of values. We will have to be satisfied knowing that the statistic
is "close to" the parameter. That leads to the obvious question, what is "close"?

We can phrase the latter question differently: How confident can we be that the value of the
statistic falls within a certain "distance" of the parameter? Or, what is the probability that the
parameter's value is within a certain range of the statistic's value? This range is the confidence
interval.

117
The confidence level is the probability that the value of the parameter falls within the range
specified by the confidence interval surrounding the statistic. There are different cases to be
considered to construct confidence intervals.

Case-I: Population variance (σ2) is known and parent population is normal.

The sampling distribution of the sample mean is normal with mean μ and variance σ 2 /n, that is,
X−μ
X ~ N(μ, σ 2 /n) . We can standardize this to get Z= ~ N (0, 1).
σ /√ n
From the standard normal distribution, we have
P (−Z α /2 < Z< Z α /2 )=1−α
Where α is risk probability and 1- α confidence level. The confidence level is the probability that
the value of the parameter falls within the range specified by the confidence interval surrounding
the statistic. σ / √ n is the standard error of the statistic . Standard error is the square root of
variance where Var ( X ) = σ 2 /n.
Using the standardized form of the sampling distribution of the sample mean in the above
probability statement, we get the limits of the confidence interval as follows:

(
P −Z α / 2<
X−μ
σ /√n )
<Z α /2 =1−α

P (−Z α / 2 σ / √ n< X – μ< Z α / 2 σ / √ n ) =1−α

P (−Z α / 2 σ / √ n−X ← μ← X + Z α/ 2 σ / √ n )=1−α

P ( X −Z α /2 σ / √ n< μ< X +Z α /2 σ / √ n ) =1−α

The last statement clearly shows that, there is a (1- α ) 100% confidence interval for population
mean (μ) to lie in the interval
( X −Z α /2 σ / √ n , X + Z α/ 2 σ / √n ).
This interval is known as a (1- α ) 100% confidence interval for population mean (μ).

Here are the Z values corresponding to the most commonly used confidence levels.

(1- α ) 100% α α /2 Z α/2

90 0.10 0.05 1.645
95 0.05 0.025 1.96
99 0.01 0.005 2.58

118
 Example 8.2: The weights of full boxes of a certain kind of cereal are normally distributed
with a standard deviation of 0.27 ounce. If a sample of 15 randomly selected boxes produced a
mean weight of 9.87 ounce, find:

a) The 95% confidence interval for the true mean weight of boxes of this cereal,

b) The 99% confidence interval for the true mean weight of boxes of this cereal,

c) What effect does the increase in the level of confidence have on the width of the
interval?

Solution:
∑ ( x− x̄ )( y− ȳ ) ∑ xy−n x̄ ȳ
r=
∑ xy −n x̄ ȳ
r= r=
√∑ ( x−x̄ )2 ∑ ( y−ȳ )2 √ (∑ x −n x̄ )(∑ y −n ȳ
2 2 2 2
) √ (∑ x 2 −n x̄ 2 ) (∑ y 2 −n ȳ 2 )
a) Given , so that , .
=
520−( 25 )( 5)( 4 )
√ ( 650−25 (5 )( 5 ) )( 436−( 25)( 4 )( 4 ) )
r s
The 95% C.I. is and ounce

rs
Where .

6∑ d
2
r s =1 −
n ( n2 −1 )
Substituting these values in , the resulting confidence
interval is (9.73, 10.01).

b) Similarly the 99% C.I. is (9.69, 10.05).

c) The increase in the confidence level widens the length of the confidence interval.

Case-II: When sampling from a non normal population and when the sample size is large the
distribution of X depends on Central Limit Theorem (with known and unknown variance).
Recall the Central Limit Theorem, which applies to the sampling distribution of the mean of a
sample. Consider samples of size n drawn from a population, whose mean is μ and standard
deviation is σ. The population can have any frequency distribution. The sampling distribution of
σ
X will have a mean μ and standard deviation is . The sampling distribution of X is normal
√n
σ2 σ2
with a mean μ and variance as n gets large .That is X ~ N (μ, ) (as n gets large). We can
n n
X−μ X−μ
standardize this to get Z= ~ N(0,1) or Z= ~ N(0,1) when σ is unknown.
σ /√ n S/ √ n
A (1-α) 100% confidence interval for population mean (μ) is

119
 ( X −Z α /2 σ / √ n , X + Z α / 2 σ / √n ), if σ 2known
( X −Z α /2 S / √ n , X+ Z α /2 S / √n ), if σ 2 unknown
Example 8.3: An economist wants to estimate the average amount in checking accounts at banks
in given region. A random sample of 100 accounts gives X =$ 357.60 and S= $140.00. Give a
95% confidence interval for μ, the average amount in any checking account at a bank in the
given region.
Solution:
Given: n = 100, X =$ 357.60, S= $140.00 & α = 0.05
A 95% confidence interval for population mean (μ) is
( X – Z α/ 2 S / √n , X + Z α / 2 S/ √ n)
=(357.60−1.96 ( 140.00 / √100 ) , 357.60+1.96(140.00 / √100)¿
¿(330.16 ,385.04)

Case-III: When sampling is from normally distributed population with unknown population
variance and when the sample size is small (n<30).
When population variance σ2 is unknown, we estimate it by sample variance. The standardized
X−μ
distributions of the sample mean, t= is t-distribution with (n-1) degrees of freedom. From
S/ √ n
this distribution, (1-α) 100% confidence interval for population mean is
S S
( X – t α /2 ,n −1 , X +t α / 2, n−1 ) .
√n √n
Example 8.4: From a normal sample of size 25 a mean of 32 was found .Given that the standard
deviation is 4.2. Find

a) A 95% confidence interval for the population mean.

b) A 99% confidence interval for the population mean.

α
Solution: a/Given: n = 25 X =32, S = 4.2, 1-α = 0.95 ⟹ α = 0.05, =0.025
2
⟹ t 0.025,24 =2.064 ¿ table .

S S
⟹ The required interval will be ( X – t α /2 , X+ t α /2 )
√n √n
4.2
=32± t 0.025,24 ×
√ 25

120
 4.2
=32±2.064 ×
√25
= 32 ±1.73
= (30.27, 33.73)
α
b/ Given: n = 25 X =32, S = 4.2, 1-α = 0.99 ⟹ α = 0.01, =0.005
2
⟹ t 0.005,24 =2.797 ¿ table .

S S
⟹ The required interval will be ( X – t α /2 , X+ t α /2 )
√n √n
4.2
=32± t 0.005,24 ×
√ 25
4.2
=32±2.797 ×
√25
= 32 ±1.35
= (29.65, 34.35)
8.1.2 Sample size determination in estimation of population mean
In the process of estimating population mean μ using the sample mean with absolute margin of
error (d) and risk probability α, the sample size is given by:

[ ]
2
Zα σ
n=
2 where | X−μ|=d
d

Example 8.5: To determine the average amount of time students take to get from one class to
the next, how large a sample is needed with probability 0.95 that the error will be at most 0.25
rs
minutes, if is known from past experience to be 1.50 minutes?
'
x s y' s rs ∑d2=20
Solution: Using , and replacing , and in the formula for , we get
6 ( 20 )
rs 1−
10 ( 100−1)
=0 .88

(always rounded to the next integer) is required for the estimate.

8.1.3 Confidence interval for population proportion

The confidence interval for the population proportion is performed in the same manner as the
population mean. We have discussed that the sampling distribution of sample proportion is
normal. The sample estimate of population proportion P is sample proportion p and sample
pq pq
estimate of variance of sample proportion is Var ¿) ¿ for large sample Var ( ^p )= .
n−1 n

A (1-α)100% confidence interval for proportion p is given by (for large n):

121
 √ ^^
^p ± Z α/ 2 p q
n

Example 8.6: The Human Resource director of a large organization wanted to know what
proportion of all persons who had ever been interviewed for a job with his organization had been
hired. He was willing to settle for 95% confidence interval. A random sample of 500 interview
records revealed that 76 or 0.152 of the persons in the sample had been hired.

Solution:

Given: ^p=0.152, q^ =1− ^p =0.848 , n = 500, α = 0.05, Z 0.025 =1.96

The 95% confidence interval for the population proportion is given by

√ ^^
√
^p ± Z α/ 2 p q =0.152 ±1.96 0.152 × 0.848 =0.152± 0.031
n 500

= (0.121, 0.183)

Hence the required proportion varies between 0.121 and 0.183.

8.2 Statistical Hypothesis testing

In section 8.1, we have studied how to make estimations of the mean using point and interval
estimations. The other aspect of statistical inference is known as statistical test of hypothesis.
The branch of statistics which helps us in arriving at the criterion for deciding about the
characteristics of the population, a parameter, based on the information obtained from the
sample data is known as testing of hypothesis. We shall use the theoretical results presented for
the interval estimation, and hence, a test of hypothesis is highly connected with the theory of
estimation we studied before.

In this section, basically we will deal with testing hypotheses about population mean and
population proportion. While doing so, we shall define some important terminologies which we
may face and the errors we are committing in the process. We shall employ the standard normal
distribution (or Z-test), chi-square distribution and the t-distribution (or t-test), depending upon
the nature of the population sampled and the sample.

8.2.1 Hypothesis testing for population mean

8.2.1.1 Some terms in tests of hypothesis

Statistical hypothesis is defined as a statement (or an assertion) about the parameter of a

population or its distribution that may be proved or disproved. Its plausibility is to be evaluated
on the basis of information obtained by taking sample from the population.
122
Test statistic: is a statistic whose value serves to determine whether to reject or accept the
hypothesis to be tested. It is a random variable.
A given statement concerning a parameter could be true or false. Hence we have two
complementary hypotheses, namely, null hypothesis and alternative hypothesis.
a) Null hypothesis ( H0)
It is the hypothesis to be tested for possible rejection under the assumption that it is true and it is
the hypothesis of equality or the hypothesis of no difference.
b) Alternative hypothesis (H1)
It is hypothesis which is the complementary to the null hypothesis. It may be accepted if H 0 is
rejected or be rejected if H0 is accepted. It is the hypothesis of difference.
Statistical Test: is a test or procedure used to evaluate a statistical hypothesis for deciding
whether to reject the hypothesis depending on sample data. The decisions we make are of two
types: Either to reject H0 and conclude that H1 is accepted or retain H0 and conclude that we have
no enough evidence to reject H0.
Types of errors
Statistical test of hypothesis can lead to two kinds of errors. If the statistical test rejects Ho when
it is true, the error is type I error. If the test accepts H0 when it is false, the error is a type II error.
The following table gives a summary of types of errors:

Decision Ho is true Ho is false

Reject Ho Type I error Correct decision

Accept Ho Correct decision Type II error

Type I error is the error committed in rejecting the null hypothesis when it is true. Probability of
committing type I error is sometimes called level of significance and denoted by α.
Type II error is the error committed in accepting the null hypothesis when it is false. Probability
of committing type II error is denoted by β.
In both types of errors, a wrong decision has occurred. An ideal test procedure is one which is so
planned as to safeguard against both these errors. However, in practical situations an attempt to
minimize one of the errors maximizes the other. In view of this dilemma and the fact that wrong
rejection of Ho is a more serious error, we will hold ∑ d=0 at a predetermined low level, such as 0.1,

0.05, or 0.01 when choosing a rejection region. The level of significance 5% rs implies

123
that in 5 samples out of 100 we are likely to reject a correct H 0. In other words this implies that
we are 95% confident that our decision to reject H0 is correct.
General steps in hypothesis testing on population mean, μ
Step-1 The first step in hypothesis testing is to specify the null hypothesis (H 0) and the
alternative hypothesis (H1). Suppose the assumed or hypothesized value of μ is denoted by μ o,
then one can formulate two sided and one sided hypothesis as follows:
1. H0: μ = μo versus H1: μ  μo (two sided test)
2. H0: μ = μo versus H1: μ < μo (one sided test or left sided-test)
3. H0: μ = μo versus H1: μ > μo  (one sided test or right sided-test)
Step-2: Specify a significance level of α.
Step-3 We should identify the sampling distribution of the estimator and the test statistic.
Case-I: Population variance (σ2) is known and parent population is normal.
X−μ o
¿
The test statistic is Zc σ ~ N (0, 1).
√n
Case-II: When sampling from a non-normal population and when the sample size is large the
distribution of X depends on Central Limit Theorem (with known and unknown variance).
X−μ o
¿
a) The test statistic is: Zc σ ~ N (0, 1) with known variance
√n
X−μ o
¿
b) The test statistic is: Zc S ~ N (0, 1) with unknown variance.
√n
Case-III: When sampling is from normally distributed population with unknown population
variance.
X−μ o
Z=
i) When the sample size is large, S ~ N (0, 1)
√n
X−μ o
t=
ii) When the sample size is small (n<30), S ~ t(n-1).
√n
Step-4. The value of the test statistic can be calculated as follows:
X−μ o
a) Zc = σ with known variance.
√n

124
 X−μ o
b) Zc = S with unknown variance.
√n
X−μ o
c) tc = S with unknown variance and small sample size.
√n
where X is the sample mean and μo the parameter specified by the null hypothesis.
Step-5: Identify the critical (rejection) region or put the decision rule.
a) For two sided test H0: μ = μo versus H1: μ  μo , reject H0 if
Zc > Z α/2 or Zc < −Z α /2 .

Graphically, the rejection and acceptance regions are:

Rejection Region Acceptance Region Rejection Region


2 
2

r =+1⇒
-s
rs=−1⇒
b) For one sided test H0: μ = μo versus H1: μ > μo reject H0 if Zcalculated > Z α. Graphically, the
rejection and acceptance regions are

¿cf
Acceptance Region Rejection Region ( )

¿cf
Z

125
c) For one sided test H0: μ = μo versus H1: μ < μo reject H0 if Zc < −Z α. Graphically, the rejection
and acceptance regions are

d)

Rejection Region Acceptance Region

f (x )=
(5x)(154−x) , x=0,1,2,3,4
(204 )

Step 6: Summarization the result and put the conclusion

Decision Table

f (0),f(1),f(2),f(3)
To test against the three alternatives, the rules are summarized as:

Alternative Accept H0 if Reject H0 if Inconclusive if

Hypothesis

f (4 ) E ( X )=1 E ( X )=3 / 4 V (X )=3/80

x̄=11

Zc=2.96>2.33⇒Ho ⇒ ⇒ χ 2 =4.01

⇒ t=−2.55⇒H o
n n
∑ ( x i −x )=0 ∑ ( x i − A )2
i =1 i =1

126
 A=x
Example 8.7: Test at whether the mean of a random sample of size n = 16 is
"significantly less than 10" if the distribution from which the sample was taken is normal,
xw=
w1 x 1+w2 x 2+…+w n x n
=
∑ w i xi xw =
w 1 x 1+w2 x2 +…+w n xn
=
∑ w i xi
w1+w2+…+wn ∑ wi w1+w2+…+wn ∑ wi
and (known).

Solution:

G. M=√ x 1 .x 2 …xn
n
GM=√3×12=√ 36=6 √
n f f f
G.M.= x11 .x22 .….x mm
* versus,
n
H . M=
1 1 1
+ +…+
x1 x2 xn
* (critical value)
n
H . M=
1 1 1
+ +…+
x1 x2 xn
* (calculated value)
n
H . M=
f f f
* Since , the null hypothesis is rejected. That is, the population mean
1 2 m
+ + …+
x1 x2 xm

8.4 is significantly less than 10 at 5% level of significance.

Example 8.8: Assume that in a certain district the mean systolic blood pressure of persons aged
20 to 40 is 130 mm Hg with a standard deviation of 10 mm Hg. A random sample of 64 persons
aged 20 to 40 from village x of the same district has a mean systolic blood pressure of 132 mm
Hg. Does the mean systolic blood pressure of the dwellers of the village (aged 20 to 40) differ
from that of the inhabitants of the district (aged 20 to 40) in general, at a 5% level of
significance?

m m
n =∑ f k n =∑ f k Median= L + ( )
w n
f med 2
−CF =~x f med
Solution: and hence .
k= 1 k= 1

Z0.025 = 1.96

CF= f med

Critical Region Acceptance Region Critical Region


2 
2
127
 ~
x
Δ1
Mode = L+ ×w
Δ1+Δ2
- =-1.96 0 =1.96

Δ1=f mod −f 1 Δ2=f mod−f 2

Since Zc is in the acceptance region, is accepted. That is, the systolic blood

pressure of persons (aged 20 to 40) living in village x is the same as the mean systolic blood
pressure of the inhabitants (aged 20 to 40) of the district.

Example 8.9: A sample of 16 students gave an average mark of 53.8 with a standard deviation
f 1
of 5.2. Can you we that the population mean of marks is 50 at ?

f 2
Mode = L+
Δ1
Δ 1 +Δ 2
×w

Solution:

x 1 ,x 2 ,…, x n ( Qi )
and hence

Q i= L +
w
fQ ( in
4
−CF )

.
i

x 1 , x 2 ,… , x n
x 1 , x 2 ,… , x n
Since is rejected. i.e the population mean mark is significantly different
L −S
L+ S
from 50 at

8.2.2 Hypothesis testing for population proportion

Hypothesis testing for population proportion P is carried out in the same way as hypothesis
testing for population mean when large samples and normality assumptions are fulfilled.
The test statistic is:
^
P−p o
Z c=

√ p o q ~ N (0, 1) where q o=1−p o

n
o

The decision rule is:

a) For two sided test Ho: P = Po versus H1: P  Po reject Ho if
Z c > Z α/ 2 or Z calculated ←Z α /2.
b) For one sided test Ho: P = Po versus H1: P > Po reject Ho if
Z c> Z α

128
c) For one sided test Ho: P = Po versus H1: P < Po, reject Ho if
Z c ← Z α.

Example 8.10: : A pharmaceutical company claims that a drug which it manufactures relieves
cold symptoms for a period of 10 hrs in 90% of those who take it. In random sample of 400
people with colds who take the drug, 350 find relief for 10hrs. At a 0.05 level of significance,
is the manufacture’s correct?
L−S 35−15
= =0.4
L+S 35+15
350
=0.875
30−5
30+5
=0. 7143 xi xi
Solution: = 400 , = 0.9, , n = 400, , = -1.645

Ho: P = 0.90 Vs H1: P < 0.90

Using the z statistic, we have

Pˆ  p 0 0.875  0.90
Zc    1.67
p 0 (1  p 0 ) / n 0.90(0.10) / 400
Since computed value of Zc = -1.67 is less than the critical value of −Z 0.05=−1.645, therefore,
the null hypothesis is rejected. The manufacture’s claim is not upheld.

8.2.3 TEST OF ASSOCIATION OF ATTRIBUTES

In the tests of hypotheses considered so far, the Z-test and the t-test, we have assumed that the
samples were drawn from normally distributed populations, or we were considering large
samples, or more accurately, the tests were based on the assumption that the sample means
were normally distributed. Since the tests require assumptions about the type of population or
parameters, they are known as" parametric tests".

In this section, we will introduce the Chi-square test. The Chi-square test used to test
association of attributes. It is a test for nominal data. Before the test, the Chi-square
distribution will be introduced.

fi
8.2.3.1 The Chi-Square ( )distribution

f i xi
The - distribution is one of the simplest and most widely used in statistical applications.
This distribution is not defined for negative real numbers and is not applicable when
observations assume such values.

129
Unlike the normal and the t distributions, its curve is not symmetrical, it is rather positively
V(xi±k)=V(xi )
skewed. As in the case of the t distribution, the degree of freedom, is the parameter of
the Chi-square distribution. Since this distribution arises in many important applications, its
V ( xi)
values for different value of significance are tabulated as a function of its degrees of
yi V ( x ) =V ( y ) =2
freedom, . Table C at the end of this module provides the values for
V ( kx i )=k V ( x i )
2
x̄ 1
, etc and degrees of freedom such that

x̄ 2
, meaning the area to the right under the Chi-square curve with n -1
in n1=2
degrees of freedom is equal to .


8.2.3.2 A Test of Association

This is Chi-Square distribution

also known as
n2 =3
analysis of contingency table. A table of
n1×n2=2×3=6 n P r=
n!
(n−r)!
for r=0, 1, 2, ⋯, n

frequencies of order (r by c) without totals is said to be contingency table if the

row totals and column totals are random. Suppose that the frequencies of the occurrences of
two attributes, say A and B, are given in a contingency table having r rows and c columns.
n! n!
n Pn= = =n!
(n−n)! 0!
Then, the table gives a total of frequencies. We say that A has r levels (categories) and B
n!
n P r=
(n−r)!
has c levels. Based on the information provided by the contingency table, we test the
hypothesis
n!
n1 !.n2 !.⋯.nk !
The two attributes are independent, against

n=n1+n2+…+nk 10!
3!.3!.1!.2!1!
: The two attributes are dependent, at a specified level .

130
The test statistic is:
C
n r = n = n!
r r!(n−r)! () .

r=0,1,2,⋯,n n Cr= (nr)=r!(n−r)n! !

Where denotes the observed frequency for the cell in the i row and the j column, and
th th

is the expected frequency (the frequency or count which is expected of the two attributes A and
B if the null hypothesis of independence is true) of the same cell.

7 C3 = 7
3 ( ) = 3 ! (77−3
!
)! 9 P3
, where ,

9 P3 9 P2
ri= total for i row, th
cj= total for jth column.

9 P 2
From which we get, .

(8010)=N=n(S)
The expected frequency, , is obtained by multiplying the total of the row(i) to which the cell
(3010 )∗¿ ¿
belongs by the total of the column (j) to which it belongs (cj) and then dividing by the

grand total:
(50
0 )
=
.

Decision Rule
To reach a decision, we need to know the distribution of the test statistic. Under the
assumption that H0 is true, the test statistic follows a chi-square distribution with (r-1)(c-1)
(30
10 )
∗¿ ¿

degrees of freedom. That is, the degrees of freedom of an contingency table out of the
()
50 /¿ ¿

totals is given by (r-1) (c-1).

0

Note that the degrees of freedom here is based on the number of cases that can be freely
changed given that the row totals and the column totals are fixed. Can you verify?

131
 (8010 )=
The decision rule is then, we reject H0 (and accept H1) at level of significance if the

(304 )∗¿ ¿ (50

6 )
=
calculated value is larger than the table value, . That is, reject H0 if

( 30
4 )
∗¿ ¿
(506 ) /¿ ¿ (80
10 )
=
at level of significance, or with a 100% confidence.

Example 8.11: A geneticist took a random sample of 301 men to study whether there is
association between father and son regarding boldness. He obtained the following results. Using
α= 0.05 and test whether there is association between father and son regarding boldness.
Father Son Total
Bold Not bold
Bold 85 59 144
Not bold 65 92 157
Total 150 151 301

Solution: H0: The father and son are independent regarding boldness

H1: The father and son are dependent regarding boldness

The test statistic is:

( ) 30
0
∗¿ ¿

(50
10 )
=

144 ×150 144 ×151 157× 150

e11 = =71.76 e12 = =72.24 e21 = =78.24 e22 =
301 301 301
157× 151
=78.76
301

( 30
0 )∗¿ ¿

( 50
10 )
/¿ ¿
The degree of freed is (r-1)(c-1) = (2-1)(2-1) = 1.

( 80
10 )
= x i=0, 1, 2, 3
H0 is rejected in favour of H1 since >

132
Therefore, the sample indicates that father and son are dependent regarding boldness at 0.05
level of significance.

Exercise 8
1. From a normal population with the standard deviation is 4.2. A sample of size 25 is taken
with mean of 32. Find 99% confidence interval for the population mean.
2. A sample from an assumed normal distribution produced the values 9, 14, 10, 12, 7, 13, 12.
∞ b

a) What is the single best estimate of ∑ P(X=x ) ? i=1

i
b) Find an 80% C.I. for ∫ ? a
f (x)dx

3. Out of a sample of 80 customers 60 of them reply they are satisfied with the service they
received. Calculate a 95% confidence interval for the proportion of satisfied customers.
4. The mean pulse rate and standard deviation of a random sample of 9 first year male medical
students were 68.7 and 8.67 beats per minute respectively. (Assume normal distribution).
a) Find a 95% C.I. for the population mean.
b) If past experience indicates that the mean pulse rate of first year male medical students is
72 beats per minute, test the hypothesis that the above sample estimate is consistent with
the population mean at 5% level of significance.

5. According to the norms established for a reading comprehension test, students should
average 84. If 45 randomly selected students averaged 87.8 with a s.d of 8.6, test the null
1 1 1

∫( 14 x+ 12 )dx= 18 x2+ 12 x|10= 18 + 12 = 58

n
x+ , ∑ x i P ( xi )
4 2
hypothesis against the alternative , at .
0 i =1

6. In a survey of drug users in a large city, it is found that 18 out of 423 of them were HIV
positive. Can we conclude that at α=0.05 level of significance fewer than 5 percent of the
drug users in the population are HIV positive?

7. In a study of aviophobia, a psychologist claims that 30% of all women are afraid of flying. If,
in a random sample, 41 of 150 women are afraid of flying, test the null hypothesis p = 0.30
n
6
∑ h ( xi ) P ( xi ) ∑ x i P ( xi )
against , at .
i =1 i =o

8. A biostatistician intends to estimate μ, the mean blood pressure of women between the ages
of 45 and 50. She takes a random sample of 20 women and measures their blood pressure.
Based on past experience she believes the measurements will follow a N(μ, 100) distribution.
(Measurements are in mm mercury.) Suppose she discovers the sample mean is equal to
136.9 mm mercury. Find a 95% confidence interval for μ.

133
9. In a study of parents' feelings about a required course in sex education, a random sample of
360 parents are classified according to whether they have one, two, or three or more children
in the school system and also whether they feel that the course is poor, adequate, or good.
Based on the results shown in the following table, test at the 0.05 level of significance
whether there is a relationship between parents' reaction to the course and the number of
children they have in the school system.
Parents' Number of Children
feelings
1 2 3 or more

Poor 48 40 12

Adequate 55 53 29

Good 57 46 20

10. A special diet given to 8 overweight women helped them to lose 6, 7.5, 11, 9, 6.5, 11, 8 and
5kg within a period of 3 months. Assuming normal distribution, can we reject the claim that
x2
9
the diet will help an overweight woman to lose at least 10kg in 3 months, at ?

134
 CHPTER 9: TWO SAMPLES INFERENCE
9.1 Inferences about differences between means

In many applied research problems, we are interested in hypotheses concerning differences

between the mean values of two populations. For instance, we may want to decide upon the
mean step pulse of men is less than the mean step pulse of women.

() [] () []
3 2 34 5 3 2 4
∫x2 x9 dx=∫ x9 dx=19 x5 |30=257 ∫x x9 dx=19 x4 |30= 94
Case 1: when the populations are normal and their variances and are known
0 0 0

()
2
27 9
−
5 4
∑ P(xi )=1
i=0
Given independent random samples of size and from two normal populations having the
∞

{ ( )} −∞ < μ ∞ , −∞ < x ∞ ∧ σ > 0 . f (μ+c)=f (μ−c)

2
∑ P(xi )=1 1 -1 x−μ 1 2 2
f (x)= exp = e−(x−μ)/2σ
σ √2 π 2 σ σ √2 π
i=0
means and and the known variances and . To test the null hypothesis

P(X≤μ)=P(X≥μ)=0.5 X =μ= mean f ' ( x)=0⇒ x=μ μ ''

f ( μ)=
−1
σ3 √ 2 π
<0
, against the alternatives, , , or :

, the respective critical regions are

Q 3 −Q 2 =Q 2 −Q 1
, where

Q 1+ Q 3
Q 2=
2
.

E( Z)=0 and V(Z )=1

is called the standard error of the difference of the two samples.
N.B: When we deal with independent random samples from populations with unknown
variances that may not even be normal, we can still use the test described above with S 1
12 z
1 −z
f (z )= e 2 ,−∞<z<∞ P(0<Z<z)=∫ f(z)dz
√2π
substituted for and S2 substituted for as long as both samples are large enough to
0

invoke the Central Limit Theorem.

Example 9.1:- Vision, or more especially visual acuity, depends on a number of factors. A
study was undertaken in Australia to determine the effect of one of these factors: racial
variation. Visual acuity of recognition as assessed in clinical practice has a defined
normal value of 20/20 (or zero in log scale). The following summarized data on
monocular visual acuity (expressed in log scale) were obtained from two groups:

135
 Number Sample mean Sample standard deviation
observation
Race (of visual acuity)

Australian males of European origin 89 -0.20 0.18

Australian males of Aboriginal origin

107 -0.26 0.13

z=0.00, 0.01, 0.02, …, 3.98, P(Z<0)=P(Z>0)=0.5 α

Solution: a) , at

P( 0< Z< z)=∫

z

0
1
√2 π
e
1 2
− z
2
dz
Reject H0 if
P(Z<1.72)=P( Z<0)+P(0<Z<1.72)
, where
=0.5+0.4573 .

=0 . 9573 P(Z <−0 . 88 )=P(Z >0 . 88 )

Substituting ,

and
=0.5−P(0<Z<0. 88)
= 0, we get
=0.5−0.3106
=0 . 1894
Conclusion: Since , H0 must be rejected; in other words, the difference is
statistically significant.

One can also construct a 100(1-

P(1.30<Z<1.75)
)% CI for
=P(0<Z<1.75)−P(0<Z<1.30)
:
x1 x2 ±
-
=0.459 −0.4032 =0.0567
P(−0.25<Z<0.45) =P(−0.25<Z 0)+P(0<Z 0.45)
Case 2: Small samples and and unknown

=P(0<Z<0.25)+P(0<Z<0.45) =0. 987+0.1736 =0.2723

When are small (both less than 30), and and are unknown, the above test
cannot be used. In this case the t test will be used. t tests can be used either to compare two
independent groups or to compare observations from two measurement occasions for the same
group. The first kind of problem is treated using what is known as the independent samples t
test and the second using paired samples t test.

136
a) Independent-samples t test

This test is used to compare two groups of scores on the same variable which are independent
μ
random samples from normal distributions having the same unknown variance . We can
have two sub-cases under this case.

a/ when n1 ≠n2

2
σ which has a t distribution with
P(a< X<b)

degree of freedom.

Here,
μ is called the pooled variance.

Under the given assumptions and H0, t is a value of a random variable having the t distribution
σ ( )
a −μ X − μ b −μ
P(a<X<b)=P < < =P( z1 <Z<z2), say.
σσσ
with degrees of freedom. Thus, the appropriate critical regions of size for
(
P( X <b )=P Z <
b−μ
σ )
=P( Z< z 2 )
testing against the alternatives

(
P( X >a )= P Z >
a−μ
σ )
= P( Z > z 1 ) μ
Or . The rejection regions are, respectively,

2
σ
Example 9.2: In an attempt to assess the physical condition of joggers and non- joggers, a
sample was selected from each and their maximum volume of oxygen uptake (V-O 2) was
measured with the following results:

Number observation Sample mean Sample standard

deviation

Joggers 25 47.5 ml/kg 4.8 ml/kg

Non-joggers 26 37.5 ml/kg 5.1 ml/kg

137
 P(μ−σ<X<μ+σ) P(μ−2σ<X<μ+2σ)
Solution: * (or )

P(μ−3σ<X<μ+3σ)

P(a< X <b )

P( μ−σ < X < μ+ σ )=P ( μ−σσ −μ <Z < μ+σσ −μ ) = P (−1< Z <1 )
Results: , ,

So that,
=2 P(0<Z<1) =2(0 . 3413) =0.6828
or
P(μ−2σ<X<μ+2σ)=P(−2<Z<2)
, and

=2P(0<Z<2) P(μ−3σ<X<μ+3σ)=P(−3<Z<3)
= .

=2 P(0<Z<3)
* (critical value).

*
μ−σ∧μ+σ
μ−2σ∧μ+2σ
* Since , reject H0 at 5% level of significance.

μ−3σ∧μ+3σ Zα x1 x2±
α Zα Z 0 .01
One can also construct a 100(1- )% CI for : - .

b/ when n1 ¿n2=n

Z 0 . 05 Z 0 .01
t2(n-1)

Z0.01=2.33 Z 0.05 Z0.05=1.645 μ= 16 5 σ2=9 ( ) ( )

X−μ 162−μ 162−165
P(X<162)=P < =PZ<
σσ 3
to compute use the following formula =2 /n where = ( + )/2

=P(Z<−1)
is the sample variance of sample 1.
where

=0.5−P(−1<Z<0)
is the sample variance of sample 2.

138
One can also construct a 100(1-
P(Z<0)=0.5
)% CI for
=0.5−P(0<Z<1) x1 x2 ±
: -
=0.5−0.3413 =0.1587
9.2 Paired comparison

Suppose we have a random sample of n observations having pairs of measurements or a

random variable x say (x, y). In this case observations for each pair should be made under the
same conditions and the mean difference should be normally distributed. Variances of each
variable can be equal or unequal.
n
X i− μi 2
∑ (
σ
)2
χ α,n
The purpose is to test the hypothesis against or
i =1 i

α
2
α , n
χ
Let D=X-Y so that for the pair of samples we have
α

2
α,n
The test statistic is
χ where
2
P(X≥χ α,n)=α.
is the variance of the variable d. This statistic is,
under the null hypothesis, distributed as t with n-1 degrees of freedom. The decision rule is the
same with the t-tests.

Example 9.3: The weight of 10 boys before they are subjected to a change of diet and after a
lapse of 6 months is given below. Test whether there has been any gain in weight as
a result of change of diet.

Weights(lb) before, X 10 112 98 114 10 97 88 101 89 91

9 2

Weights(lb) after, Y 11 120 99 117 10 98 91 99 93 89

2 5

Solution:

α α χ
2
α,ν

To test against at

First calculate d.

139
 Weights(lb) before, X 10 112 98 114 10 97 88 101 89 91
9 2

Weights(lb) after, Y 11 120 99 117 10 98 91 99 93 89

2 5

d 3 8 1 3 3 1 3 -2 4 -2

Using the procedures for determining the test statistic given above, we have
tα , n−1

Thus,
α . Taking the level of significance to be
n
, we have
tα , n−1
. Since calculated value of the statistic is greater than the tabulated value, the null
hypothesis of no gain in weight will be rejected.

α α tα
One can also construct a 100(1- )% CI for D : ± sd/√ n.

9.3 Inferences about differences between Proportions

tα t α , n−1
If we have two populations with proportions of the required characteristic equal to and .
α tα , n−1 −tα,n−1
Then their estimate can be obtained on the basis of sample of size and . Suppose that
tα α t 0.05
and are the numbers of successes observed in trials of one kind and of another, the
−t0.10
trials are independent, and the corresponding probabilities of success are, respectively, and
t α =t0.005=
t0.975= 2 −t 0.05=−2.861.
, the sampling distribution of has mean and standard error of

2
σ .

140
 μ μO
To test the null hypothesis or , the standard error of the formula

μ≠μO
can be written as .

μ>μO μ<μO
Where is usually estimated by pooling the data and substituting for the combined sample
t α
2
proportion given by: .
Then, converting to standard units, we get the test statistic:

tα σ t α , n−1
= , with

This is approximately normal for large samples.

More generally, to test

α
, we use:
μ .

X̄
16.4−12.0
2.1/ √16
t 0.05,15 μ1−μ2=δ (μ1−μ2=0)
The test criteria are similar to those of , with and substituted for and
, respectively.

One can also construct a 100(1- )% CI for

β=β o ^p1 ^p ± p^ 1 q^ 1 + ^p 2 q^ 2
σ21=σ 22=σ2
: - 2 n1 n2 √
Example 9.4: A health officer is trying to study the malaria situation of Ethiopia. From the
records of seasonal blood survey results he came to understand that the proportion of people
having malaria in Ethiopia was 3.8% in 1978 (Eth. Cal.). The size of the sample considered
was 15,000. He also realized that during the year that followed (1979), blood samples were
taken from 10,000 randomly selected persons. The result of the 1979 seasonal blood survey
showed that 200 persons were positive for malaria help the health officer testing the

141
hypothesis that the malaria situation of 1979 did not show any significant difference from that
of 1978 (take the level of significance 0.01).

P( X >a)=0.05 P( X≤b)=P( X >b)

Solution: To test (or )

{ μ=38 . 6
kx , 0≤ x <5
f ( x )= k ( 10− x ) , 5≤ x <10
0 , otherwise σ =6 .5
against (or ) at

ν α

n1= 15,000, n2= 15,000

x1= 570, x2= 200

zt = z0.05=2.58

tα, ν N Cn
= 0.0308
Common (pooled) proportion,

1
.
N C n
And the standard error = = 0.0022

Hence, zc=(0.038-0.020)/0.0022=0.018/0.0022=8.2

N1+N2+…+Nk=N
Conclusion: Since zcal> ztab reject . Therefore, it is concluded that there was a statistically
significant difference in the proportion of malaria patients between 1978 and 1979 at a 0.01
level of significance.

9.4 Inferences concerning variances

Under the assumption of normality one can perform a test on a population variance. If we want
to test the null hypothesis

Ni i
th
i
th ni (i=1,2,…,k)
against one of the alternatives: , or ,or .

142
 n1+n2+…+nk =n
In such tests, we use the calculated value as ; this value is then compared with
N Cn 5 C2
x̄ i
the table value of , where =(n-1) and is the level of significance.

x̄ i x̄ i X̄ x̄ i X̄ x̄ i
The decision rule is to reject if > or < for two sided test, if >
X̄ ( ) (
8 5−2
)
2
σ N−n
V( x̄)=
n N−1 2 5−1
for right sided test or if < for left sided test.
Example 9.5: In a random sample the amount of time which 18 women took to complete the
written test for their driver licenses has standard deviation 2.1 minutes. Do the data give
sufficient evidence that the population variance is significantly less than 6.25 minutes at
2
σ
(Assume normality).

μ x̄ i X̄
Solution: Given that n=18, = =4.4, the computed is

x̄ i

x̄ i
versus

X̄

X̄ x̄ i X̄
=8.672<11.995= so accept the null hypothesis.

One can also construct a 100(1-

X̄
)% CI for
x̄ i X̄ 14/3 3−2
¿<¿ 2 3−1
( )
.

Exercise 9
1. Hearing levels in two groups of school children with normal hearing in the frequency of
500 cycles per second was found to be as follow.

No. of children Mean of hearing Stand. deviation

threshold

Group 1 62 15.5 decibels 6.5 decibels

Group 2 76 20 decibels 7.1 decibels

143
 Test if there is any difference between hearing levels recorded in the two groups at
x̄ i
.
2. A group of 12 patients was given a new type of sleeping pill. Another group consisting of

10 patients was given a conventional type. The number of hours of sleep of these patients

in one day are as follows:

Group 1 (new pill): 8.4, 6, 7, 8, 6.5, 7.4, 8.1, 9, 10.1, 7.2, 8, 11

Group 2 (conventional): 8, 7.2, 6.3, 6.4, 8.5, 9.2, 8, 7, 7, 9

Do the data suggest that there is any appreciable difference in the effect of the two types of
X̄
pills on patients for .

3. The mean produce of wheat from a sample of 10 fields comes to 200 kg per acre and

another sample of 150 fields gives a mean of 220 kg per acre. Assuming the standard

deviation of the yield at 11 kg for the population, test if there is a significant difference
x̄ i
between the means of the sample at .

4. A certain stimulus administered to each of 12 patients resulted in the following increase of

blood pressure: 5, 2, 8, -1, 3, 0, -2, 1, 5, 0, 4 & 6. Can it be concluded that the stimulus will,
x̄ i
in general, be accompanied by an increase in blood pressure at .

5. In a year there are 956 births in a town A of which 52.5% were male, while in towns A &B

combined, this proportion in a total of 1406 births was 0.496. Is there any significant
X̄
difference in the proportion of male births in the two towns at ?

6. The variance of the weights of 10 shipments was 31 square kilogram can we say that

variance of the distribution of weight of all shipments from which the above sample was

drawn is equal to 20 square kilogram.(assume normality)

7. Two different types of drugs A & B were tried on certain patients for increasing weight, 5

persons were given drug A & 7 persons were given drug B. The increase in weight (in

pounds) is given below.

144
 Drug A: 8 12 13 9 3

Drug B: 10 8 12 15 6 8 11

Do the two drugs differ significantly with regard to their effect in increasing weight at
x̄ i
?

8. Two samples of sizes 9 and 8 gave the sums of squares of deviation from their respective

means as equal to 50 and 25 with means 275 and 290. Do the sample means differ
x̄ i
significantly at

145
 CHAPTER 10: SIMPLE LINEAR REGRESSION AND CORRELATION
After completing the topic, the students will be able to:
 Determine the relationship between variables.
 Find the fitted regression line of the two variables.
 Draw and describe scatter diagram.
 Interpret the slope and intercept of the fitted regression line.
 Calculate and interpret the correlation coefficient.
 Find and interpret the coefficient of determination.
 Calculate and Interpret explained and unexplained variations.
 Calculate and interpret the spearman’s correlation coefficient.
10.1 Simple Linear Regression of Y on X
Under simple linear regression of Y on X, we have one independent variable which is influential
usually denoted by X and one dependent variable influenced by the independent variable which
we denote it by Y. For example in real world variables that may be related linearly are,
production/yield ( Y ) and amount of rainfall(X ), monthly income (Y ) and level of education
(X), ,where an increase in one variable is associated with an increase in the other variable.
Similar examples can also be given on the negative relation between two variables; the increase
in one is accompanied by a decrease in the other.
A simple linear regression model is given as
Y=α+ β X+∈
Where α is intercept of the regression line. It gives the value of Y whenever X is zero. If the
range of X does not include zero, α has no practical interpretation. β is the slope. It is a measure
of the rate of change. It shows by how much Y changes for every unit change in X. The sign of β
has also some significance; because it shows the direction of the relation between the two
variables. A positive sign of β shows that the two variables are positively related and a negative
sign of β shows that the two variables are negatively related.
The constants, α and β are parameters and are commonly referred to as regression coefficients.
- ∈ is a random error term. It is neither observable nor measurable. In real life problems, even
though two variables are linearly related, their relationship is not fixed as
Y=α+ β X
This is because the dependent variable, Y is the effect of many independent variables in which X
is one of them. Contribution of other independent variables not considered in the model may be
minor. However, we cannot be certain that Y depends only on X. Thus the contribution of these

146
variables not included in the model and other factors such as measurement error is
accommodated by ∈.
Mean of the values of ∈ is zero. Some of its values are positive, that is when the actual value lies
above the line Y^ = α^ + ^β Xi and some are negative in case when the actual value of Y lies below
the fitted regression line.
Assumptions:
1. The relationship between the dependent variable Y and independent variable X exist and is
linear.
2. For every value of the independent variable X, there is an expected value of the dependent
variable Y.
3. The dependent variable Y is a continuous random variable, whereas values of the independent
variable X are fixed values.
4. The sampling error,∈, associated with the expected value of the dependent variable Y is
assumed to be an independent random variable distributed normally with mean 0 and constant

variance X̄ about the regression line.

To estimate this model we take a sample of n independent observations which give rise to n pairs
(Xi , Yi) and find best estimates of the parameters or best fitted line using least square method of
estimation. A best fitting line is one for which the sum of squares of the errors, ∑  ε2i is
minimum.
In the principle of least square method, one would select a and b such that
∑  ε 2i = ∑ (Y ¿¿ i−Y^ i )2 ¿ is minimum where Y^ i = α^ + ^βXi
To minimize this function, first we take the partial derivatives of ∑  ε 2i with respect to α^ ∧ ^β
respectively .Then the partial derivatives are equated to zero separately and result in the
following normal equations respectively
n n

∑ Y i = nα^ + ^β ∑ X i
i=1 i=1

n n n

∑ X i Y i =α^ ∑ X i + ^β ∑ X 2i
i=1 i=1 i=1

Solving these normal equations simultaneously we can get the values of α^ ∧ ^β as follows.

^β
X̄ α^ Y ^β X

and = -
147
These estimates are denoted by α^ ∧ ^β the estimated (fitted) regression line is
given by
Y^ i = α^ + ^β Xi
Before estimating the regression coefficients, it would be wise to plot the observed data on a
graph known as a scatter diagram. Scatter diagram is a plot of all ordered pairs (x i ,yi ) on the
coordinate plane which helps to observe relationship between two variables. This diagram gives
a preliminary idea on the type of relationship the two variables have.
Regression analysis is useful in predicting the value of one variable from the given value of
another variable, Y^ i= α^ + ^β Xi.
Example 10.1

For the following example height of father in inch (X) and height of their sons in inch (Y):

Assuming simple linear relationship between X and Y,

a/ Draw the scatter diagram;

b/ Find the estimated regression equation of Y on X;
c/ Give the predicted value of Y for X= 65.5

solution: a/ The scatter diagram is as follows:

total

x̄ i
65 63 67 64 68 70 71 69 537 b) And the
necessary
X̄ 67 66 68 65 69 68 70 68 541 statistics are
computed
X̄ 4225 3969 4484 4096 4624 4900 5041 4761 36105 below:

x̄ i 4355 4158 4556 4160 4692 4760 4970 4692 36343

X̄ 4489 4356 4624 4225 4761 4624 4900 4624 36603

148
 2
σ 14 / 3 2
^β V ( X )=σ 2x= V ( X )=σ 2x=
σ
n 2 n and

α^ = 67.625 – 0.482(67.125) =35.27

14/3
Hence, the equation is 2 = 35.27 + 0.482x.

X̄
c) When X = 65.5,

10.2 Covariance and Simple Linear Correlation Analysis

Given the paired data (x1,y1), (x2,y2), . . ., (xn,yn) we may want to describe the type & strength of
relationship between the independent variable X and the dependent variable Y. We can give
these two by applying an index called simple correlation coefficient. The population correlation
coefficient is represented by ρ and its estimator by r. The correlation coefficient r is also called
Pearson’s correlation coefficient since it was developed by Karl Pearson. r is given as the ratio of
the covariance of the variables X and Y to the product of the standard deviations of X and Y. The
computational formula is:

x̄ i =
X̄
x̄ i
Alternatively: The correlation coefficient is given by

σ2
V( X )= σ 2x=
n
The correlation coefficient, r is always lies between –1 and +1, inclusive.
 r = -1 implies perfect negative linear relationship between the two variables.
 r = +1 implies perfect positive linear relationship between the two variables.
 r = 0 implies there is no linear relationship between the two variables. But the two
variables may have non-linear relationship between them.
 r approaches +1 indicates strong positive linear relationship between the two variables.
 r approaches -1 indicates strong negative linear relationship between the two variables.

149
  r approaches 0 indicates weak linear relationship between the two variables .

Coefficient of Determination (r2)

The square of the correlation coefficient, r2, is called the coefficient of determination. It
measures the variation in the dependent Y explained by the simple linear regression of Y on X.
1− r2 measures the proportion of variation in Y not explained by the simple linear regression of
Y on X.
Example 10.2
If r = 0.9, then r2 = 0.81 and 1- r2 =0.19. Approximately 81% of the variation in the dependent
variable, Y, is explained by the simple linear regression of Y on X fitted on sample data. The
remaining, 1-r2, 19 % of the variation in Y is unexplained by the simple linear regression of Y on
X fitted on sample data.
Example 10.3
The research director of the Saving and Loan Bank collected 25 observation of montage interest
rates X and number of house sales Y at each interest rate. The director computed that,
∑ x i=125 , ∑ y i=100 , ∑ x i yi =520 , ∑ x 2i =650 , ∑ y 2i = 436
Compute and interpret (i) Coefficient of correlation.
(ii) The coefficient of determination.
Coefficient of correlation.
Solution: i)

( ) (
2
σ N −n
V ( x̄ )=
n N −1
N −n
N −1 ) = 0.667

The two variables have positive linear relationship.

ii) Coefficient of determination, r2= (0.667)2 =0.44 this shows that 44% of the variation in the
number of house sales is due to the variation in the interest rate. 44% of the variation in the
number of house sales(Y) is explained by the simple linear regression of Y on X (interest rate).

10.3 Spearman’s Rank Correlation Coefficient

The simple correlation coefficient (r) cannot be used when we are dealing with a qualitative
data such as judgment about beauty, efficiency, honesty, etc. In such cases, the rank correlation

150
coefficient is used to explain the correlation or if there is an agreement in ranking. It is denoted

by X̄ and is defined as follows:

Definition: The coefficient of rank correlation, X̄ ,given by Spearman for n pairs, is

X , where d is the difference between the rank of x and the corresponding y.

To calculate ( n ) , we first rank the X

2
σ
N μ,
among themselves from least to best or from best to least;

then we rank the N (68,0.56 ) in the same way, find the sum of the squares of the differences, d, between
the ranks of the x's and the y's. When there are ties in rank, we assign to each of the tied
observations (having equal value) the mean of their ranks.

Example 10.4: Assume that ten girls in a beauty contest for Miss Debre Markos were ranked by
two judges as follows:

Girl Number 1 2 3 4 5 6 7 8 9 10

Judge A 4 8 6 7 1 3 2 5 10 9

Judge B 3 9 6 5 1 2 4 7 8 10

Calculate X̄ and interpret it.

Solution: Since the ranks are given, we need to find only the difference in ranks for each girl
and the square of these differences.

Girl Number 1 2 3 4 5 6 7 8 9 10 Total

d 1 -1 0 2 0 1 -2 -2 2 -1 0

d2 1 1 0 4 0 1 4 4 4 1 20

151
 2
For these n = 10 pairs,
70−68
√ 0 . 56 , and X̄ = σ , which is positive and
close to 1, showing that there is a very good agreement (or concordance) between the two
judges regarding the beauty of the girls.

N.B: X̄ provides a check in calculations.

2
σ
Like the values of r, the values of n also lie between -1 and +1, inclusive, and the

interpretations of its size and sign are analogous to those of r. X Perfect

positive agreement,
N μ ,( σ2
n ) Complete disagreement, where the two rankings go
completely in opposite direction.

Exercise 10
1. The following table show the heights to the nearest inch (in) and the weights to the
nearest pound (lb) of a sample or 12 male students drawn at random from the first year
students at a university .

Height x (in) 70 63 72 60 66 70 74 65 62 67 65 68
Weight y (lb) 155 150 180 135 156 168 178 160 132 145 139 152

a/ plot a scatter diagram of the data

b/ fit the least square equation

c/ estimate the weight of a student whose height is 63 inches.

2. The following table presented age of female patients (x) with total cholesterol (Y).

Age(X) 25 25 28 32 32 32 38 42 48 51 51 58 62 65
Total 180 195 186 180 210 197 239 183 204 221 243 208 228 269
Chol.(Y)

a/ Draw the scatter diagram

b/ Find the least square regression equation.

c/ Find the Correlation coefficient.

d/ find the coefficient of determination .

152
 3. The following data represent the number of calories per serving and the number of grams
of sugar per serving for a random sample of high- fiber cereals.

Calories(x 200 210 170 190 200 180 210 210 210 190 190 200
)
Sugar (Y) 18 23 17 20 18 19 23 16 17 12 11 11

a/ Draw the scatter diagram

b/ Determine the least square regression equation

4. Based on the following date answer the questions

n=10, ∑ XY =4174, ∑ X =139, ∑ Y =280, ∑ x 2=1985
a/ what is the linear regression equation of y on x.
b/ suppose the value of x=20,what will be the value of y.

153
ANSWER FOR EXERCISES

Exercise 1
1. a/ All potential listeners of the radio program, b/ Those 91 listeners who called with in a
minute of the question asked, c/ The sample is not random . Only those who listened to
that program and had access to phones could have responded to the question. Hence the
sample was biased, & not scientific.
2. a/ collection of all 25 – years old males who have never taken fish more than once a
week,
b/ a Sub-collection of 500 males selected randomly, c/ difference (pre-diet minus post-
diet) in Cholesterol level, d/ mean difference in cholesterol level, e/qualitative
3. a/ Collection of all patients suffering from that particular type of headache, b/ Collection
of 350 patients who have been surveyed or interviewed.
4. a/ nominal, b/nominal, c/ratio, d/ratio, e/nominal, f/ratio, g/ordinal, h/ ordinal, i/nominal

Exercise 2

1. a) Continuous ; b) discrete; c) continuous; d) continuous.

2. Primary data are data collected by the investigator for the intended purpose; while
secondary data are collected by some other agency for different or similar purpose. Secondary
data have to be checked whether or not they are suitable, adequate and reliable for the purpose
of the current study before using them.
3.
Means of bus car plane Total
transport
4. The Frequenc 17 9 14 40 frequency distribution is:
y

Monthly salary 90-98 99-107 108-116 117-125 126-134 135-143 144-152 Total

Frequency 1 4 3 6 9 6 1 30

5. a) Class marks: 0.5, 2.5, 4.5, 6.5, and 8.5;

b) C.B’s: -0.5-1.5, 1.5-3.5, 3.5-5.5, 5.5-7.5, and 7.5 -9.5;

c) R.F’s: 0.27, 0.42, 0.22, 0.07, and 0.03;

154
 d) Cumulative frequencies are:

X−μ X−μ
i) the less than Z=
σ/ √ n
ii) the ‘or more’ Z=
σ / √n
than

Less than 1.5 16 -0.5 or more 60

Less than 3.5 41 1.5 or more 44

Less than 5.5 54 3.5 or more 19

Less than 7.5 58 5.5 or more 6

Less than 9.5 60 7.5 or more 2

Exercise 3
1. a/mean=38.3, b/mode=40, c/median=28.5
2. a/mean=45.59, b/mode=46.44, c/median=40.16, d/Q1=36.09, P1=36.09, D8=53.65,
D9=60.68
3. a/mode=3, b/median = 3, c/mean=3.47, d/Q2=3, e/D6=4, f/P25=3
4.

CI 32-36 37-41 42-46 47-51 52-56 57-61 Total

f 4 11 15 7 2 1 40

5. a/mean=14.35, b/mode = 15, c/median=15, d/D7=14.1, e/P80=20

Exercise 4
1. M.D( x ) = 2.65, C.M.D( x ) = 0.36, M.D( ^x ) = 3.71, C.M.D( ^x ) = 0.34, S2 = 9.584, S = 3.096
2. Since CVA = 29.11% < CVB = 17.19%, section B scores was more variable than that of
section A scores.
3. City 1: x 1=23 , S 12=12.5 ,CV 1=15.374 %
City 2: x 2=21.8 , S 22=2.2 ,CV 2=6.803 %
City 1: x 3=29.2 , S 32=18.7 , CV 3 =14.808 %
Therefore, the city2 has the most consistent temperature.
4. α 3=−1.01. Sinceα 3 <0, the distribution is negatively skewed.
The fourth moment is 30,000.

155
 Exercise 5
1. 0.4167, 2/18/50, 3/0.5275, 4/ a/12/365, b/7/365, c/31/365, d/1/365
Exercise 6

1. a)
X̄ ; b) Find
X̄
and
X
in (a);

c) N (151 ,225/64 ) , and V(X)=12/19

2. a) 0.125; b) X̄ , and
X−μ
σ /√n ; c) (0.95)1/3; d) (0.5)1/3.
3. a) k = 1/25 ; b) i) 0.5 ; ii) ¾; c) 500kg
4. a)0.53; b)1.56; c)0.44 (5) a)0.7881; b)0.2743.
6. a) 29.141; b) 26.119; (7) 0.312. 8. i) 2 ii) 0.678 (9) a) 2; b)1.833
Exercise 7
1. a/Mean of sample mean = 2550, variance of sample mean = 46.291
b/ p=0.7198
2. p=1
3. a/ systematic random sampling, b/cluster sampling(first stage)
4. 0.4286, 5. b/9, c/ s.d.=2.64
Exercise 8

1. (29.833, 34.167)
154 .75−151
2. a) 15/8 b) (9.87, 12.13)
3. (0.654, 0.846)
4. a) The 95% C.I. for μ is (62.0, 75.4) beats per minute. b) Ho: μ = μo Vs H1: μ  μo. Since tcal.=
-1.14 ≤ ttab = -2.306, Ho is accepted.

5. X is rejected
6. Ho: P = 0.05 Vs H1: P < 0.05, Zc = -0.07 > -Zα = -1.645 X̄ Ho is not rejected.
7. Z = -0.81 X̄ H0 cannot be rejected
8. (132.517, 141.283)

9. X̄ X−μ
σ /√n The two attributes are independent.
8 . 00 − μ
10. σ /√n Cannot be rejected.
Exercise 9

156
1. zc= 3.88, reject Ho
2. x 1=8.1, x 2= 7.66, s21 = 2.07, s22 = 1.07, tc= 0.74, Accept Ho
3. zc= 14.08, reject Ho
4. D =2.58, s2D =9.538, tc=2.89, reject Ho
5. n1 = 956, n2 = 450, p1= 0.525, p2 = 0.434, p = 0.496, reject Ho
2
6. x c = 13.95, accept Ho
7. tc=0.5, accept Ho
8. tc= -13.81, reject Ho
Exercise 10
1. b/ ^y = -60.7 + 3.22x, c/ 142.16,
2. b/ ^y =151.354 + 1.399x, c/ 0.718, d/coeff. of deter. =51.5%
3. b/ ^y = 0.93 + 0.0821x
4. a/ ^y = -0.46 + 5.33x, b/ 106.14

157
 Appendix: Table-A: Area between z=0 and Z=z OR area between Z= 0 and Z≤z):

z 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09
0.0 0.0000 0.0040 0.0080 0.0120 0.0160 0.0190 0.0239 0.0279 0.0319 0.0359
0.1 0.0398 0.0438 0.0478 0.0517 0.0557 0.0596 0.0636 0.0675 0.0714 0.0753
0.2 0.0793 0.0832 0.0871 0.0910 0.0948 0.0987 0.1026 0.1064 0.1103 0.1141
0.3 0.1179 0.1217 0.1255 0.1293 0.1331 0.1368 0.1406 0.1443 0.1480 0.1517
0.4 0.1554 0.1591 0.1628 0.1664 0.1700 0.1736 0.1772 0.1808 0.1844 0.1879
0.5 0.1915 0.1950 0.1985 0.2019 0.2054 0.2088 0.2123 0.2157 0.2190 0.2224
0.6 0.2257 0.2291 0.2324 0.2357 0.2389 0.2422 0.2454 0.2486 0.2157 0.2549
0.7 0.2580 0.2611 0.2642 0.2673 0.2704 0.2734 0.2764 0.2794 0.2823 0.2852
0.8 0.2881 0.2910 0.2939 0.2969 0.2995 0.3023 0.3051 0.3078 0.3106 0.3133
0.9 0.3159 0.3186 0.3212 0.3238 0.3264 0.3289 0.3315 0.3340 0.3365 0.3389
1.0 0.3413 0.3438 0.3461 0.3485 0.3508 0.3513 0.3554 0.3577 0.3529 0.3621
1.1 0.3643 0.3665 0.3686 0.3708 0.3729 0.3749 0.3770 0.3790 0.3810 0.3830
1.2 0.3849 0.3869 0.3888 0.3907 0.3925 0.3944 0.3962 0.3980 0.3997 0.4015
1.3 0.4032 0.4049 0.4066 0.4082 0.4099 0.4115 0.4131 0.4147 0.4162 0.4177
1.4 0.4192 0.4207 0.4222 0.4236 0.4215 0.4265 0.4279 0.4292 0.4306 0.4319
1.5 0.4332 0.4345 0.4357 0.4370 0.4382 0.4394 0.4406 0.4418 0.4492 0.4441
1.6 0.4452 0.4463 0.4474 0.4484 0.4495 0.4505 0.4515 0.4525 0.4535 0.4545
1.7 0.4554 0.4564 0.4573 0.4582 0.4591 0.4599 0.4608 0.4616 0.4625 0.4633
1.8 0.4641 0.4649 0.4656 0.4664 0.4671 0.4678 0.4686 0.4693 0.4699 0.4706
1.9 0.4713 0.4719 0.4726 0.4732 0.4738 0.4744 0.4750 0.4756 0.4761 0.4767
2.0 0.4772 0.4778 0.4783 0.4788 0.4793 0.4798 0.4803 0.4808 0.4812 0.4817
2.1 0.4821 0.4826 0.4830 0.4834 0.4838 0.4842 0.4846 0.4850 0.4854 0.4857
2.2 0.4861 0.4864 0.4868 0.4871 0.4875 0.4878 0.4881 0.4884 0.4887 0.4890
2.3 0.4893 0.4896 0.4898 0.4901 0.4904 0.4906 0.4909 0.4911 0.4913 0.4916
2.4 0.4918 0.4920 0.4922 0.4925 0.4927 0.4929 0.4931 0.4932 0.4934 0.4936
2.5 0.4938 0.4940 0.4941 0.4943 0.4945 0.4946 0.4948 0.4949 0.4951 0.4952
2.6 0.4953 0.4955 0.4956 0.4957 0.4959 0.4960 0.4961 0.4962 0.4963 0.4964
2.7 0.4965 0.4966 0.4967 0.4968 0.4969 0.4970 0.4971 0.4972 0.4973 0.4974
2.8 0.4974 0.4975 0.4976 0.4977 0.4977 0.4978 0.4979 0.4979 0.4980 0.4981
2.9 0.4981 0.4982 0.4982 0.4983 0.4984 0.4984 0.4985 0.4985 0.4986 0.4986
3.0 0.4987 0.4987 0.4987 0.4988 0.4988 0.4989 0.4989 0.4989 0.4990 0.4990
3.1 0.4990 0.4991 0.4991 0.4991 0.4992 0.4992 0.4992 0.4992 0.4993 0.4993
3.2 0.4993 0.4993 0.4994 0.4994 0.4994 0.4994 0.4994 0.4995 0.4995 0.4995
3.3 0.4995 0.4995 0.4995 0.4996 0.4996 0.4996 0.4996 0.4996 0.4996 0.4997
3.4 0.4997 0.4997 0.4997 0.4997 0.4997 0.4997 0.4997 0.4997 0.4997 0.4998

158
 t α=p 0.1 0.05 0.025 0.01 0.005 0.0025 0.001 0.0005
df = 1 3.078 6.314 12.706 31.821 63.656 127.321 318.289 636.578
2 1.886 2.920 4.303 6.965 9.925 14.089 22.328 31.600
3 1.638 2.353 3.182 4.541 5.841 7.453 10.214 12.924
4 1.533 2.132 2.776 3.747 4.604 5.598 7.173 8.610
5 1.476 2.015 2.571 3.365 4.032 4.773 5.894 6.869
6 1.440 1.943 2.447 3.143 3.707 4.317 5.208 5.959
7 1.415 1.895 2.365 2.998 3.499 4.029 4.785 5.408
8 1.397 1.860 2.306 2.896 3.355 3.833 4.501 5.041
9 1.383 1.833 2.262 2.821 3.250 3.690 4.297 4.781
10 1.372 1.812 2.228 2.764 3.169 3.581 4.144 4.587
11 1.363 1.796 2.201 2.718 3.106 3.497 4.025 4.437
12 1.356 1.782 2.179 2.681 3.055 3.428 3.930 4.318
13 1.350 1.771 2.160 2.650 3.012 3.372 3.852 4.221
14 1.345 1.761 2.145 2.624 2.977 3.326 3.787 4.140
15 1.341 1.753 2.131 2.602 2.947 3.286 3.733 4.073
16 1.337 1.746 2.120 2.583 2.921 3.252 3.686 4.015
17 1.333 1.740 2.110 2.567 2.898 3.222 3.646 3.965
18 1.330 1.734 2.101 2.552 2.878 3.197 3.610 3.922
19 1.328 1.729 2.093 2.539 2.861 3.174 3.579 3.883
20 1.325 1.725 2.086 2.528 2.845 3.153 3.552 3.850
21 1.323 1.721 2.080 2.518 2.831 3.135 3.527 3.819
22 1.321 1.717 2.074 2.508 2.819 3.119 3.505 3.792
23 1.319 1.714 2.069 2.500 2.807 3.104 3.485 3.768
24 1.318 1.711 2.064 2.492 2.797 3.091 3.467 3.745
25 1.316 1.708 2.060 2.485 2.787 3.078 3.450 3.725
26 1.315 1.706 2.056 2.479 2.779 3.067 3.435 3.707
27 1.314 1.703 2.052 2.473 2.771 3.057 3.421 3.689
28 1.313 1.701 2.048 2.467 2.763 3.047 3.408 3.674
29 1.311 1.699 2.045 2.462 2.756 3.038 3.396 3.660
30 1.310 1.697 2.042 2.457 2.750 3.030 3.385 3.646
40 1.303 1.684 2.021 2.423 2.704 2.971 3.307 3.551
50 1.299 1.676 2.009 2.403 2.678 2.937 3.261 3.496
60 1.296 1.671 2.000 2.390 2.660 2.915 3.232 3.460
Infinity 1.282 1.645 1.960 2.326 2.576 2.807 3.090 3.290
Table B. t- table with right tail probabilities

Table C. Right tail areas for the Chi-square Distribution

159
df\area 0.995 0.99 0.975 0.95 0.9 0.25 0.1 0.05 0.025 0.01 0.005
1 0.000 0.000 0.001 0.004 0.016 1.323 2.706 3.841 5.024 6.635 7.879
2 0.010 0.020 0.051 0.103 0.211 2.773 4.605 5.991 7.378 9.210 10.597
3 0.072 0.115 0.216 0.352 0.584 4.108 6.251 7.815 9.348 11.345 12.838
4 0.207 0.297 0.484 0.711 1.064 5.385 7.779 9.488 11.143 13.277 14.860
5 0.412 0.554 0.831 1.145 1.610 6.626 9.236 11.071 12.833 15.086 16.750
6 0.676 0.872 1.237 1.635 2.204 7.841 10.645 12.592 14.449 16.812 18.548
7 0.989 1.239 1.690 2.167 2.833 9.037 12.017 14.067 16.013 18.475 20.278
8 1.344 1.647 2.180 2.733 3.490 10.219 13.362 15.507 17.535 20.090 21.955
9 1.735 2.088 2.700 3.325 4.168 11.389 14.684 16.919 19.023 21.666 23.589
10 2.156 2.558 3.247 3.940 4.865 12.549 15.987 18.307 20.483 23.209 25.188
11 2.603 3.053 3.816 4.575 5.578 13.701 17.275 19.675 21.920 24.725 26.757
12 3.074 3.571 4.404 5.226 6.304 14.845 18.549 21.026 23.337 26.217 28.300
13 3.565 4.107 5.009 5.892 7.042 15.984 19.812 22.362 24.736 27.688 29.819
14 4.075 4.660 5.629 6.571 7.790 17.117 21.064 23.685 26.119 29.141 31.319
15 4.601 5.229 6.262 7.261 8.547 18.245 22.307 24.996 27.488 30.578 32.801
16 5.142 5.812 6.908 7.962 9.312 19.369 23.542 26.296 28.845 32.000 34.267
17 5.697 6.408 7.564 8.672 10.085 20.489 24.769 27.587 30.191 33.409 35.718
18 6.265 7.015 8.231 9.390 10.865 21.605 25.989 28.869 31.526 34.805 37.156
19 6.844 7.633 8.907 10.117 11.651 22.718 27.204 30.144 32.852 36.191 38.582
20 7.434 8.260 9.591 10.851 12.443 23.828 28.412 31.410 34.170 37.566 39.997
21 8.034 8.897 10.283 11.591 13.240 24.935 29.615 32.671 35.479 38.932 41.401
22 8.643 9.542 10.982 12.338 14.041 26.039 30.813 33.924 36.781 40.289 42.796
23 9.260 10.196 11.689 13.091 14.848 27.141 32.007 35.172 38.076 41.638 44.181
24 9.886 10.856 12.401 13.848 15.659 28.241 33.196 36.415 39.364 42.980 45.559
25 10.520 11.524 13.120 14.611 16.473 29.339 34.382 37.652 40.646 44.314 46.928
26 11.160 12.198 13.844 15.379 17.292 30.435 35.563 38.885 41.923 45.642 48.290
27 11.808 12.879 14.573 16.151 18.114 31.528 36.741 40.113 43.195 46.963 49.645
28 12.461 13.565 15.308 16.928 18.939 32.620 37.916 41.337 44.461 48.278 50.993
29 13.121 14.256 16.047 17.708 19.768 33.711 39.087 42.557 45.722 49.588 52.336
30 13.787 14.953 16.791 18.493 20.599 34.800 40.256 43.773 46.979 50.892 53.672

160
References:

1. Eshetu Wencheko, Introduction to Statistics. April 2000, Addis Ababa University.

2. Getu Degu, Fasil Tessema, Biostatistics for Health Science students, Jan. 2005,
University of Gondar.
3. Gupta S.P., Gupta M.P., Business Statistics, 2001, Sultan chand & sons, New Delhi.
4. Monga G.S., Mathematics and Statistics for Economics (second revised edition),
2007.
5. Moorthy M.B.K., Subramani K. & Santha A. Probability and Statistics, Dec. 2007,
Scitechpublications (India) pvt. Ltd.
6. Pal Nabendu, Sarkar Sahadeb, Statistics concepts and applications, 2006, New Delhi.
7. Rastogi ,Veer Bala, Fundamentals of Biostatistics, 2008, University of Delhi
8. Spiegel Murry R. & Stephen Larry J.. Statistics-schaum’s outline, 1999,
ATAMCGraw-Hill edition, 3rd edition, New Delhi.
9. Sullivan Michael, iii, Statistics: informed decision using data: 2004, New Jersey.

161