BIS 201

Statistical Hypothesis testing

Dr. Anusha Sreejith
Assistant Professor
Dept. Of Community Medicine
August 4, 2025

[Link] COLLEGE OF MEDICINE

LEARNING OBJECTIVES
On completion of this unit, the student will be able to:

• Realize the importance of hypothesis testing in making

inferences for the study.

• State a null hypothesis and alternative hypothesis and

carry out a structured hypothesis test;
• Identify type I error, type II error and the power of the test;

• Identify different types of test.

• Identify the various tests to be used for comparison.

Statistical Inference

• Estimation
• In terms of Confidence Intervals

• Hypothesis testing
• When there is some difference between 2
comparisons
• Real or chance?
Hypothesis: Statement about one or more populations.

Research Hypothesis: Conjuncture or supposition that

motivates research

Statistical hypothesis: An assumption about the

population parameter, which may or may not be true
Hypothesis
The hypothesis may be derived from:

(a) An educated guess based on published results

(b) Preliminary observations

If the data obtained from the sample can be shown beyond

doubt to be consistent with the hypothesis, it is accepted as
an accurate statement about the population. Otherwise the
hypothesis is rejected.
• Hypothesis is classified as:
• Null hypothesis

• Alternative hypothesis

Two hypotheses should be mutually exclusive

and exhaustive
Null Hypothesis
• Statement regarding the value(s) of unknown
parameter(s).

• A hypothesis of no difference among groups (i.e., no

association between the predictor and outcome variables)
that needs to be tested.

• It is denoted by

• Null hypothesis may or may not be rejected.

• Statistician’s hypothesis
Alternative Hypothesis

• Statement contradictory to the null hypothesis.

• Hypothesis, which a researcher believes to be true.

• It is denoted as or

• cannot be tested directly, it is accepted by exclusion

if the test of significance rejects
Hypothesis testing
• Hypothesis testing is used when some comparison is to
be made
• Between single observed value Vs some hypothesized value
• Between two groups or
• Between more than two groups.

• It is the procedure of finding the probability that the

observed difference is due to chance

• Which tests to apply depends on

• The study design
• Format of the data
Types of error
Reality
true true

Measured /Perceived
Correct Type 1
False positive

Reject

Type II Correct
Do not Reject

False Negative
• Type I error
• It occurs when the null hypothesis is true but we reject it.
• It is also called false positive

• Level of significance
• Probability of type I error is called as level of significance,
denoted by
• Type II error
• It occurs when the null hypothesis is false but we do not reject
it.

• It is also called false negative.

• Probability of type II error is denoted by

• Power =

• Probability that the study will correctly identify a

significant difference/effect/association in the sample,
should one exist in the population
Example - Efficacy Test for New drug

• Drug company has new drug, wishes to compare it with

current standard treatment

• Federal regulators tell company that they must demonstrate

that new drug is better than current treatment to receive
approval

• Firm runs clinical trial where some patients receive new

drug, and others receive standard treatment

• Numeric response of therapeutic effect is obtained (higher

scores are better).
Example of Statistical
Hypotheses
• Theory: New drug- GC* will reduce joint pain for the majority
of users when compared to the older drug.

• To test this theory we need to form competing hypotheses

about the statement.

• The Null Hypothesis of no difference between the two drugs

• The Alternate Hypothesis is the opposite of the null, ‘The

research hypothesis’. New drug is better than the older drug

*GC- Glucosamine and

Chondroitin
State the Null and Alternative

The null hypothesis is denoted and the alternative hypothesis is

given by

• : There is no difference between the effects of the older drug and

newer drug GC in joint pain

• :Taking GC will reduce joint pain in the majority of users when

compared to the old drug
Based on Data, Make a Decision

• If the proportion of subjects that report less joint pain is the

same as with older drug?

• If 75% of the subjects taking G/C had pain relief and only
45% taking the older drug had pain relief?

• If the difference between G/C and the older drug was 30%?

• How large a difference in proportion is needed for you to

feel confident in rejecting the null hypothesis?
Can we make a mistake?

• Is it possible that if we concluded from our data that G/C

worked that we could be wrong?

• Is it possible that if we concluded from our data that G/C

didn’t work that we could be wrong?
 Your decision based on the data-
Sample
Accept H0 Reject H0

Null H0 true No Error Type1 Error

(No difference (α- Error)
between Drug A
The and B)

Truth
Null H0 Type 2 Error No Error
In the
false (β- Error)
Popln (There is
difference
between Drug A
and B)
 Example - Efficacy Test for New drug

• Type I error –
• Concluding that the new drug is better than the standard (H 1)
when in fact it is no better (H 0). Ineffective drug is deemed better.
• Traditionally = P(Type I error) = 0.05
(Rejecting a true null hypothesis)

• Type II error –
• Failing to conclude that the new drug is better (H 1) when in
fact it is. Effective drug is deemed to be no better.
• Traditionally a clinically important difference is assigned, and sample
sizes chosen so that:
= P(Type II error)  .20
(Accepting a false null hypothesis)
Decision Making Criteria
• Since we are dealing with sampling error there is always a
possibility that differences we find could have happened by
chance.

• We make statistical inferences based on the probability that

the results may or may not have happened by chance.

• Our model for making this decision is found on the normal

distribution
 How the decision works?
μ

Critical region Critical region

H1 H0 H1
Alternate hypothesis Null hypothesis Alternate hypothesis
Key Points

• We start by assuming that the Null Hypothesis is true - There is no

difference between the two drugs.

• If the data are so strong that we believe that they could not have happened
by chance, then we reject Ho and accept the alternate hypothesis.

• Since our decisions are based on probability theory not absolute surety, we
can make mistakes.

• The probability of concluding that the research hypothesis is correct when it

isn’t (rejecting Ho when it is true) is represented by alpha ().

• The probability of failing to find a result when there is one is a Type II error
(β).
Test statistic
• General formula for test statistic is given as:

• z=

• How many standard error is our sample estimate is away

from the population estimate?

• It compares the observed value with the expected value if

the null hypothesis is true
 Critical and non-critical region

• A critical or rejection region is a region in

normal curve where if the value of test statistic falls
the null hypothesis is rejected.
• This range of values indicate that there is significant
difference between the hypothesized value and the
corresponding point estimate for the parameter.

• A non-critical or non-rejection region is a

region where if the test statistic falls and we are not
able to reject null hypothesis
• indicates that the difference is due to chance
One-tailed test

• The region of rejection is on only one side of the

sampling distribution.

• Specifies the difference in one direction only

• Always directional
Two tailed test

• The region of rejection is on both sides of the sampling

distribution.

• Specifies the difference in either direction

• Directionless
P- Value:
• The probability which measures how likely it is that any
observed difference between groups is due to chance is
called p-value.

• p-value takes value between 0 and 1.

• p-value close to 1 suggests that there is no difference

between groups other than due to random variation
α - Value:
• Is the probability of type 1 error

• It measures the strength of evidence against

• Smaller the P value- stronger the evidence against

• Is the smallest value of α for which the null

hypothesis has been rejected
P- Value vs level:

• is chosen before the statistical test is carried out.

• p-value is calculated after the experiment is over

and sample statistic is calculated.

• [Link]
Steps in hypothesis testing
Evaluate data

Review assumptions

State hypothesis

Select test Statistics

Determine distribution of test statistic

State decision rule

Calculate test statistics

Make decision

Accept Reject
Some statistical tests to test hypothesis

There are various tests. The type of test to be applied depends on the
parameter to be tested.

• To test the difference between proportions

 Z test & Chi square test

• To test if there is association between 2 factors

 Chi square test

• To test difference between two means

 t – test

• To test difference between more than 2 means

 ANOVA
Pathway analysis of continuous variables

Normal Parametri
distribution c tests

Continuous yes
data
Non- Transform
normal to
distribution normality
Non-
No parametri
c tests
Question
A pharmaceutical company has introduced a new drug for migraine against a
standard drug. It was found that 65% of the patients who used new drug got
relief, but only 40% who used standard drug got relief.

Null hypothesis stated: No difference in pain relief for both the drugs

Alternate hypothesis: New drug is better than standard drug

The possible Type I error to be committed can be:

a) the company concluded the new drug is better, when there is no difference in
both the drugs
b) the company concluded that both drugs are same, when the new drug is better
c) the company concluded that the new drug is better, when it is actually true.
d) the company concluded that the new drug is no better that old drug, when it is
actually true.
References
• Daniel WW. Biostatistics: A Foundation for Analysis in
the Health Sciences, 11th Edition. John Wiley & Sons
Inc.; 2013. ISBN-13: 978-1119496700
• [Link]

34