INFERENTIAL STATISTICS

Inferential Statistics concerns with making decisions not knowing the absolute
truth that is which of the two decisions is most probable. It is just choosing what is
most likely true.
Inferential tests are either conducted on a distribution of normal shape or on a
distribution- free environment testing differences, similarities, congruencies,
relationships, associations and the like.
In inferring what is most likely true, an indirect proof is In courts of law, a person is
used. Assume one of the statements is true, and then try to assumed innocent and only
reject it. This line of reasoning is used in several fields other reject that assumption if the
than statistics. evidence is so great that
there is essentially no doubt
that the assumption of
In statistics, these conflicting statements are called innocence is wrong.
hypotheses, and their truth is tested using hypothesis test.

 The Null Hypothesis stating no effect or difference and is symbolized by H0.

 The Alternative Hypothesis that specifies an effect or A good rule is to always write
a difference and is symbolized by H1. the alternative hypothesis
first in the directional
 These hypotheses can take in either directional or non - situation.
directional forms.
 As a general rule, unless there are very strong
theoretical reasons usually non-directional hypotheses In a sense, researchers only
are preferred. However, if directional hypotheses can make a decision when they
be generated, they allow more powerful tests. reject the null hypothesis. If
they fail to reject the null
 In real life, non - directional tests are usually used. hypothesis, then they have
not really made a decision
Statistical testing involves determining whether H0 can be because they never accept
the null hypothesis, only fail
rejected. This rejection is based on a conditional probability
to reject it.

[p (result | H0 is true)].
Every experiment may be
 Reject the null hypothesis if this probability is a small said to exist only in order to
value (unlikely to get result given the null hypothesis is give the facts a chance of
true). disproving the null
 Only fail to reject the null hypothesis if this probability is hypothesis. (Ronald Fisher)
not a small value (these results could happen given that
the null hypothesis is true).

Notes in Statistics Chapter 4

39
Compiled by: Nolasco K. Malabago One Sample Z and t - Test
The Decision Box

 Type 1 error is incorrectly rejecting a true null hypothesis [symbolized with the
Greek letter alpha ()]. The probability of such an error is the alpha risk commonly
set to 5%.
 Type 2 error is not rejecting a false null hypothesis [symbolized with the Greek letter
beta ()]. The probability of such an error is called the beta risk.
 One of the important properties about
This is a very important advantage of
making decisions using inferential
statistical decision making. A researcher
statistical procedures is determining the using statistical techniques gets to decide
level of type 1 error. Knowing how often how often they will be wrong when they
the null hypothesis is rejected when it is make a decision.
true. This type 1 error is often called the
alpha level or alpha risk, or significance level, or p-level or simply the
probability.
 The most typical values for type 1 errors are 0.05, 0.01, and 0.001. These are very
low levels. At the most, one is only going to make a type 1 error 5 out of every 100
times (.05).

Try to consider the Legal Decision Box:

The person is guilty The person is not guilty
Incorrect Decision
Place the person in jail Correct Decision
Type 1 error 

Incorrect Decision
Allow the person to go free Correct Decision
Type 2 error 

Legal Decision Box

 In courts of law, an innocent person is protected not to be sent to jail by setting type 1
errors very low. The legal system has been designed to allow type 2 errors to be
higher. The ways that juries are instructed mandates that their decisions will more
likely let the guilty go free (type 2 error) than convict the innocent (type 1 error).

 This is also the case in statistics.

40 Probability Summer 2014

Critical Z values
The largest and most frequently set value for the alpha level (type 1 error) is 0.05. This
value sets up regions on the normal curve that reject values given the null hypothesis is
true 5% of the time or less. Therefore, if values are around the mean 95% of the time, fail
to reject the null hypothesis.

From the Empirical Rule:

 Recall that the mean ±2z contain about 95%
of the scores. Using the z table, instead of +2
and -2 the correct value is ±1.96z.
 Assuming a normal distribution, and that the
null hypothesis is true, then any z value that is
calculated between -1.96 and +1.96 happens
95% and z values outside of ±1.96 happen 5%
of the time or less. These are rare values and,
if found, reject the null hypothesis.
 If alpha is set to 0.05, the rare z values of
±1.96 are called critical values (zcv).

Alpha .05, Non - directional Test

If the type 1 error value was set at 5%, and the test is non - directional, then 2.5% of the
normal distribution would be in each tail.
 The first tail area of 0.025 will allow calculating the critical z value when 2.5% of the
distribution is below it.
 The second tail area of 0.975 will allow calculating the critical z value when 2.5% of
the distribution is above it. Adding both these tail areas together produces the type 1
error of 5%.

Power
The power of a statistical test is its ability to detect a difference Factors that affect the
when one exists, or the probability of rejecting Ho when Ho is false. power in an experiment:
 This is a conditional probability, the probability of 1. As alpha gets smaller,
rejecting Ho, given that Ho is false. Since type 2 error is the power decreases.
probability of failing to reject Ho when it is false, power 2. As the difference between
must be 1 - beta ( as illustrated on the table ). H0 and H1 increases, power
increases.
Ho is False 3. As the sample size in the
experiment increases, power
Reject Ho Correct Decision (1 -  ) increases.
Fail to reject Ho Incorrect Decision () 4. Directional hypotheses lead
to more power
Calculating Power

Notes in Statistics Chapter 4

41
Compiled by: Nolasco K. Malabago One Sample Z and t - Test
Like every statistic calculated, there are six steps involved.

Step 1. Specify the null and alternative hypotheses

Step 2. Set the probability level

Step 3. Collect the data

Step 4. Calculate the statistic

Step 5. Make a decision concerning the null hypothesis

Step 6. Write a summary statement

TEST STATISTICS FOR COMPARISON OF MEANS

Horizontal axis value – Distribution mean
General Equation =
Distribution standard deviation

Calculation Distribution Formula

x
z score population Population

xx
z score sample Sample
s
x  x x  x
Sampling distribution of or
one - sample z test sx 
means ( is known)
n
x  x x  x
Sampling distribution of or
one sample t test sx s
means ( is unknown)
n

Two independent samples Sampling distribution of

t 
x 1  
 x 2  1   2 
n ≤ 30 means ( is unknown) sx
1 x2

Two related samples n ≤ 30

Sampling distribution of D  D
means ( is unknown) t
sD

Two independent samples

n > 30
Sampling distribution of
means ( is unknown) z
x 1
 
 x 2  1   2 
s x1  x2
Sampling distribution of MStreatment
Multiple samples F
means (2 is unknown) MS error

42 Probability Summer 2014

The Z statistic
The one-sample z test is used when:
1. Comparing a sample mean to a population mean.
2. The population is normally distributed (or a large sample).
3. The population standard deviation () is known.

Problem
The average psychology test score administered to students entering the School Psychology
Program at State University is 32, with a standard deviation of 6.2. Is a score of 42 so different
that it wouldn't be expected to occur by chance? This is a non - directional test with alpha set at
0.05.

Solution
Using the six steps:

1. Specify the null and alternative hypotheses

Ho: µ = 32 The score of 42 is not different from the average psychology test score
administered to entrant students of the State University
Ha: µ ≠ 32 The score of 42 is different from the average psychology test score
administered to entrant students of the State University

2. Set the significance/probability level :  = 0.05

3. Collect the data: x = 42, µ = 32, σ =6.2

4. Calculate the statistic:

x  42  32
z    1.613
 6.2

5. Make a decision concerning the null

hypothesis
@  = 0.05, two tailed then ztv = ±1.96.
Since the computed value is less than the
table value (zcv = 1.613 < ttv = 1.96) the
null hypothesis is failed to be rejected.

6. Write a summary statement

Thus, it can be inferred that the score of
42 of an entrant student is not different
from the average psychology test score
administered by the State University and
can be attributed to occur by chance.

Notes in Statistics Chapter 4

43
Compiled by: Nolasco K. Malabago One Sample Z and t - Test
Three Distributions
To understand better the one-sample Z test, one must consider three separate distributions,
the population, sample and sampling distributions.

 Population Distributions are usually having the actual population distribution that
is unknown, and is represented by theoretical (open-ended) frequency polygons. The
most famous of these theoretical distributions is the normal curve.
 Sample Distribution is selected from the population of a certain size and constructs
a frequency polygon from the single selected sample. Calculate the mean, and
standard deviation of this sample.

 Constructing the population, sample, and sampling distributions needs to consider

what score is placed along the x-axis and what the y-axis measured.
a. Population. The scores are placed along the x-axis, and the y-axis indicates the
frequency of each score. This distribution have means symbolized by  and
standard deviations by .
b. Sample. The scores are placed along the x-axis, and the y-axis also indicates the
frequency of those scores. The sample mean is symbolized by x and the sample
standard deviation by s.
c. The Sampling Distribution. The sample mean is placed along the x-axis with the
frequency counts on the y-axis. The symbol for the mean of the sampling
distribution of means is  x and the symbol for the standard error of the mean

is x .

Sampling Distribution of the Mean Characteristics

 The mean of a sampling distribution of the mean  x is always equal to the population
mean.

 The standard error of the mean  x is always equal to the population standard
deviation divided by the square root of the sample size. The two equations below give
those relationships.

 x   and  x 
n
The Central Limit Theorem
 As the sample size increases, the shape of the sampling distribution of the mean
approximates the shape of a normal z - score distribution irrespective of the shape of the
original population distribution.
 As the degree of skewness increases, there is a need to increase the sample size to assure
that the sampling distribution of the mean approximates the normal z - score distribution.

 Thumb Rule: When  is known, a sample size of 20-30 is adequate to guarantee a normal
distribution of the sampling distribution of the mean.

44 Probability Summer 2014

Problem:
In a population of third graders, the mean of a reading test is 100 and the standard
deviation is 15. Drawing a sample of 100 children, gives a new reading program mean
of 92 and a standard deviation is 14.14. Did the new reading program change their
scores from the population?

Solution
Using the six step:

1. State the null and alternative hypotheses.

Ho:  = 100 The sample mean is not different from that of the population mean.
Ha: µ ≠ 100 The sample mean is different from the population mean.

2. Decide how often you want to be wrong when you reject the null hypothesis. This is
also called setting the alpha level: @alpha = 0.05 then the z critical values is ±
1.96.

3. Collect the data : The population mean = 100, the population standard deviation = 15, the
sample mean = 92, and the sample size = 100.

4. Calculate the statistic

x  x 92  100
z    5.33
x 1.5

Step 5. Make a decision

Since the z value of -5.33 is well outside of
the z critical values reject the null
hypothesis and accept the alternative.

Step 6. Write a summary statement

In this case, "This group of readers were
different than the normal comparison
group (z = -5.33, p < .05).

Table for Critical z values

Significance Level
Direction
0.10 0.05 0.025 0.01
One - tailed ±1.28 ±1.645 ±1.96 ±2.33
Two - tailed ±1.645 ±1.96 ±2.33 ±2.58

Notes in Statistics Chapter 4

45
Compiled by: Nolasco K. Malabago One Sample Z and t - Test
One-Sample t-Test

 Use to calculate the standard error of the mean using the

standard deviation of the sample when the population The one-sample t and z
standard deviation is not known. test follows exactly the
 The calculation of the standard error of the mean same general equation.
changes and the sampling distribution is t distributed
instead of normal.
 The degrees of freedom is equal to the sample size minus 1 ( df = n - 1 ).

s  Notice that there are no Greek letters in this formula. Statistics do not
sx  use them when population parameters are being calculated.
n  Unfortunately, when estimating the standard error of the mean using the
sample standard deviation, the shape of the sampling distribution of the
mean is not normal.

William Sealy Gossett ( pen name Student ) discovered that the shape of the sampling
distribution under this circumstance was slightly more leptokurtic than a normal distribution
and called the shape of this distribution a t-distribution.

 When the sample size is large, the t distribution approaches the shape of the normal
or z distribution.
 As the sample size, and degrees of freedom increase, the t-distribution becomes taller,
and the tails become thinner - looking more and more like a normal distribution.

Rules
 To determine if a sample mean is different than a population mean when  is
known, use the z test.

 To determine if a sample mean is different than a population mean when  is

unknown, use the t test.
x  x x = 
 The t formula: t  where
sx sx =
s
n
Note:
1. t critical values change depending upon the size of the sample used to
estimate the standard error of the mean (there is a family of t
distributions).
2. These critical values are not looked up using the sample size; rather they are
found using degrees of freedom (df).

46 Probability Summer 2014

Problem:
A physician is interested in evaluating the effect of a new treatment for Corona Virus.
She hypothesizes that the average length of life is 15 days after the onset of the
disease. A sample of 20 patients were taken and treated. She then waits and finds
their mean life span is 20 days with a standard deviation of 4 days. Did the treatment
change their life expectancy?

Solution
Using the six steps

Step 1. State Ho and H1

Ho:  = 15 The treatment does not change the life expectancy.
Ha: µ ≠ 15 The treatment changes the life expectancy.

Step 2. Set the alpha level and find the t critical values
This is a non - directional or two-tailed test. Using an alpha level of 0.05 and degrees of
freedom equal to 19 (df = 20-1), using a table tcv = ± 2.093.

Step 3. Get the Data

sample mean = 20, sample standard deviation = 4, the sample size = 20 and actual
populations mean or a hypothesized population mean = 15

Step 4. Calculate the statistic

The equation below provides the calculation answer.

x  x 20  15
t   5.59
sx 4
20
Step 5. Make a decision
Since 5.59 is not within the critical
values of ±2.093, the null hypothesis is
rejected.

Step 6 Write a Summary Statement

It can be inferred there is enough evidence
to prove that the physician's new found
treatment for Corona Virus significantly
change the life expectancy of the treated
patients,
(t = 5.59, df = 19, p < .05).

Notes in Statistics Chapter 4

47
Compiled by: Nolasco K. Malabago One Sample Z and t - Test