Introduction

e testing of a statistical
hypothesis is perhaps the most
impo ant area of statistical
inference – Ronald E. Walpole
 HYPOTHESIS

statement or tentative theo which

aims to explain facts about the real
world.
e process of making inference or
generalization on population
parameters based on the results of
the study on samples.
 Why Hypothesis Testing?

Most of these hypotheses have their origin in

a question to some practical problem. In
search for an answer, “educated guesses”
and some pe inent evidences are brought
out which later on become propositions or
hypotheses.
Reliability of hypothesis testing?

e truth or falsity of a statistical

hypothesis is never known with
ce ainty unless the entire population is
examined.
 HYPOTHESIS TESTING

the process of making inference or

generalization on population
parameters based on the results
of the study on samples.
in other words, deciding between
what is reality and what is
coincidence.
STATISTICAL HYPOTHESIS

a prediction made by the researcher

regarding the possible outcome of
the study.
an asse ion or conjecture
concerning one or more populations.
 Two Kinds of Hypotheses

Null hypothesis
Alternative hypothesis
 Null Hypothesis

It is denoted by H
o

It se es as the working hypothesis

It always express the idea of
nonsigni cance of di erence
Always in the negative form
Always contain the equality sign
 Alternative Hypothesis

It is denoted by H 1

Expected to be accepted
Never contains the equality sign
Uses the < or > or the ≠ sign
Generally represents the idea which the
researcher wants to prove
 Example
Title: An evaluation of the e ectiveness of
online learning

Problem: e researcher wants to know if

online learning has increased the GWA of PUP
Engineering Students from 78%
 Hypothesis

Null Hypothesis (H ):
o

________________________________
________________________________
____
Alternative Hypothesis (H ):
1

________________________________
________________________________
____
 Null Hypothesis

Ho: μ = 78; Online learning has not

increased the GWA of the PUP
Engineering Students
 Alternative Hypothesis

H1: μ > 78; Online learning has

increased the GWA of the PUP
Engineering Students
 Example

Title: An assessment of the new method

of teaching geomet
Problem: e researcher wants to know if
the pe ormance of the students would be
di erent from the old method of teaching
geomet
 Hypothesis

Null Hypothesis (H ):
o

________________________________
________________________________
____
Alternative Hypothesis (H ):
1

________________________________
________________________________
____
Meaning of Rejection and Non-Rejection

REMEMBER:
If you reject Ho, it means it is wrong;
If you do not reject Ho, it does not mean
that it is correct – it simply means you do
not have enough evidence to reject it.
 LEVEL OF SIGNIFICANCE

e level of signi cance of a test is the

maximum value of probability of
rejecting the null hypothesis when in
fact it is true.
 What does it mean?

A 5% level of signi cance means that we can accept

about 5 chances out of 100 that we would reject the null
hypothesis when it should be accepted. A 5% level of
signi cance implies that we are 95% con dent that we
have made the right decision.
A 1% level of signi cance means that we could be wrong
with a probability of 0.01. In other words a 1% level of
signi cance implies that we are 99% con dent that we
have made the right decision.
Areas of Rejection

Two-tailed test
Areas of Rejection

One-tailed test
One-tailed and two-tailed tests

When the rejection region is located at

only one extreme of the range of values for
the test statistics, the rest is one-tailed.

When the rejection region is located at

both ends of the range of values, the test
is called a two-tailed test
One-tailed and two-tailed tests

If the test involves improvement in ce ain

aspects (test scores, pe ormance, etc.) or
reduction (errors, weight, etc.), it is a one-
tailed test.

If the test involves signi cant di erence

between two samples, it is a two-tailed
test.
 5-step approach to
hypothesis testing

1. HO: ___________
H1: ___________
2. α = ______; Critical value = ________
3. Decision Rule: Reject Ho if
│computed value│ ≥ │critical value│
4. Decision
5. Conclusion
 Normality Test
Shapiro–Wilk test,
Kolmogorov–Smirnov test,
skewness,
kurtosis,
histogram, box plot,
P–P Plot,
Q–Q Plot, and
mean with SD
 Normality Test

The Shapiro–Wilk test is more appropriate

method for small sample sizes (<50 samples)
although it can also handle larger sample size
Kolmogorov–Smirnov test is used for n ≥50.
For both of the tests, null hypothesis states that
data are taken from normal distributed
population
 Skewness and Ku osis

Skewness is a measure of symmetry, or more precisely,

the lack of symmetry of the normal distribution.
Kurtosis is a measure of the peakedness of a
distribution. The original kurtosis value is sometimes
called kurtosis (proper). “Excess” kurtosis (also called
kurtosis [excess]) obtained by subtracting 3 from the
kurtosis (proper).
Symmetric Distribution

A distribution, or data set, is symmetric if it looks the

same to the left and right of the center point.
If mean, median, and mode of a distribution coincide,
then it is called a symmetric distribution, that is,
skewness = 0, kurtosis (excess) = 0.
A distribution is called approximate normal if skewness
or kurtosis (excess) of the data are between − 1 and + 1.
 Parametric Non-Parametric
Central Tendency Mean Median
Dispersion Standard Deviation Range
Signi cant Di erence t-test (equal variances) Mann Whitney U-Test
(Independent Samples)
Signi cant Di erence t-test for paired samples Wilcoxon Rank Sign Test
(Paired Samples)

Signi cant Di erence ANOVA Kruskal Wallis H-Test

(3 or more samples)

Post-test Sche e’s Test Dunn Post Ad Hoc Test

Correlation Pearson’s r Spearman Rank
Signi cant Relationship t-test for r t-test for
 Testing the signi cance of
di erence between means

z-test if σ is known and n ≥ 30

t-test if σ is unknown and n ≤ 30
F-test if 3 or more μ
 ree types of hypotheses which can be
tested by the z-test

Population or hypothesized mean (Population

mean vs Sample mean)
Two sample means and two sample standard
deviations (sample 1 vs sample 2)
Two sample means and population standard
deviation is known (Sample mean 1 vs Sample
mean 2)
Population or hypothesized mean
Two sample means and two
sample standard deviations
 Example

e Principal of the SHS wants to know which method is

better in teaching physics. He took a random sample of 40
students handled by only one teacher in lecture and
laborato and found to have a mean of 86 with a standard
deviation of 7. Fifty students from a group handled by two
di erent teachers were randomly selected and found that
they have a mean of 90 with standard deviation of 10. Does
this indicate that the two- teacher set-up is better than a
one teacher set-up? Test at α = 0.01
 5-step solution

1. Ho: μ2 = μ1; e two-teacher set-up is not

signi cantly better than one-teacher set-up
in teaching physics.
H1: μ2 > μ1; e two-teacher set-up is
signi cantly better than one-teacher set-up
in teaching physics.
5-step solution
 5-step solution

3. Decision rule: Reject Ho if

│2.23│ ≥ │2.33│
 5-step solution

4. Decision: Accept Ho
since │2.23│ < │2.33│
 5-step solution

5. Conclusion: e two-teacher set-up

is not signi cantly better than one-
teacher set-up in teaching physics. at
0.01 level of con dence
A statistics teacher wants to know if
students without calculator got
signi cantly lower scores in statistics
midterm exam than those with calculator.
He took a sample of 40 students without
calculator with the following results:
79 80 83 90 70 65 60 71 85 89
80 87 85 85 75 73 74 71 70 68
88 78 81 85 87 91 93 90 83 84
81 80 74 73 71 66 65 60 78 78
He then took a sample of 50 students
with calculator with the following
results:
88 80 81 66 75 75 78 61 85 90
89 88 91 81 85 88 84 86 83 84
85 88 81 80 83 81 95 90 99 98
65 60 83 61 90 91 93 83 80 82
85 91 90 88 81 80 80 87 87 86

Test at α = 0.05
 5-step solution

1. Ho: μ1 = μ2; e pe ormance of students

without calculator is not signi cantly lower
compared to pe ormance of the students with
calculator
H1: μ1 < μ2: e pe ormance of the students
without calculator is signi cantly lower compared
to the pe ormance of students with calculator.
5-step solution
 5-step solution

3. Decision Rule: Reject Ho if

│─ 2.76│ ≥ │─ 1.65│
 5-step solution

4. Decision: Reject Ho because

│─ 2.76│ > │─ 1.65│
 5-step solution

5. Conclusion:
e pe ormance of the students without
calculator is signi cantly lower compared
to the pe ormance of students with
calculator at 0.05 level of signi cance
Two sample means and population
standard deviation is known
Two rival manufacturers of penlight batteries claimed that their product lasts
longer than the other. i y samples of Brand E and thi y-four of Brand N
were tested. e following are the lengths of lives of the batteries recorded in
hours.
Brand A
38 41 42 36 39 41 43 35 36 38

42 39 40 43 44 35 40 39 37 40

42 44 38 37 41 40 38 42 45 41

Brand
B
38 40 41 43 39 41 40 43 39 38

40 43 44 39 40 41 42 39 40 45

40 38 42 41 40 36 37 41 42 40

36 38 41 40

Test at 0-.05 level of signi cance if there is a signi cant di erence in the length
of life of the two brands of penlight batteries.