Probability and Statistics
Module 5
Hypothesis Testing I
Introduction to Testing of Hypothesis
Dr. Jisha Francis,
VIT, Vellore
Learning Goals
1. Introduction to Statistical Inference
2. What is Hypothesis Testing?
3. Null Hypothesis vs Alternative
Hypothesis
Population
● A statistical population is an aggregate of
numerical records of measurements on certain
characteristics of interest.
● It includes every possible observation about the
variable of interest.
● If we are interested in the heights of students in a
university, the population would include the
height of every student in the university.
Parameter
● Parameters are specific characteristics of the
population that are of interest to the experimenter.
● Parameters are usually denoted by Greek letters
(e.g., θ).
● Examples:
○ The mean height of all students in a university
(denoted as μ).
○ The standard deviation of exam scores for all students
in a class (denoted as σ).
Sample
● Let be a random sample drawn
from a population with distribution function F.
● Suppose is the realization of
.
● Each individual set of values is
called a sample point.
Sample
● The sample space is the collection of all possible
sample points.
● Example:
○ For a random sample of size 2 from N(μ,σ2), the
sample space is {(x,y) : x,y ∈ R}.
Statistic
● Let be a random sample drawn
from a population with distribution function F.
● A statistic is a function of the sample observations
that is independent of the unknown parameter θ.
● Formally, let be a real-valued
function. Then is a statistic.
Statistic
● Example:
○ For a sample from N(μ,1), the sample mean is a
statistic.
Parameters and Statistics
Parameter Statistic
A number that describe the A number that describe the sample
population
Unknown quantity Known quantity
Fixed quantity Changing from one sample to
another.
Parameters and Statistics
Taxonomy of Statistical Inference
Statistical Hypothesis - Definition
● A hypothesis is a statement about a population
parameter or the probability distribution of a
random variable.
● An assertion or conjecture about one or more
populations.
● Use experimental or sample data to support or
refute the hypothesis.
Statistical Hypothesis - Examples
Medical Research:
● Coffee increases cancer risk in humans.
Engineering:
● Two gauges differ in accuracy.
Sociology:
● Blood type and eye color are independent
variables.
Uncertainty in Statistical Hypotheses
● The true or false status of a hypothesis is rarely
known with certainty.
● Examining the entire population is often
impractical.
● A random sample is taken from the population to
provide evidence.
Uncertainty in Statistical Hypotheses
● Sample data may either:
○ Support the hypothesis, or
○ Show inconsistency with the hypothesis.
● Rejection of Hypothesis:
○ If the sample evidence is inconsistent, the hypothesis
is rejected.
Formal Statement of Hypotheses
● Hypothesis statements should consider the
probability of wrong conclusions.
● Example: To support a claim, state the hypothesis
in a way that leads to rejection.
○ Medical Example: Test if there is no increase in
cancer risk from coffee drinking.
○ Engineering Example: Test if there is no difference in
accuracy between two gauges.
Null and Alternative Hypotheses
Hypotheses are formulated using:
● Null Hypothesis (H0): The hypothesis we wish to
test.
● Alternative Hypothesis (H1): Represents the
question or theory to be tested.
● Symbol: H0 (Null), H1 (Alternative)
Hypothesis Testing Outcomes
Two Possible Conclusions:
● Reject H0 in favor of H1 (due to sufficient
evidence).
● Fail to Reject H0 (insufficient evidence)
● Note: "Fail to reject H0" does not mean "accept
H0."
Example: Proportion of Defectives
● Hypothesis Setup
○H0: p = 0.10 (Historical defect rate).
○ H1: p > 0.10 (Concern that defect rate has
increased).
● Conclusion
○ 12 defectives out of 100 → Fail to reject H0.
○ 20 defectives out of 100 → Reject H0 in favor of
H1.
Simple Vs. Composite Hypothesis
● A hypothesis is said to be a simple hypothesis if
that hypothesis uniquely specifies the distribution
from which the sample is taken.
○ H₀: μ = 50 Vs H1: μ = 52
● Any hypothesis that is not simple is called a
composite hypothesis.
○ H₀: μ ≥ 50.
Types of Alternative Hypotheses in
Hypothesis Testing
In practice one may have hypotheses such as
H0 : μ = μ0 against one of the following alternatives:
Test Statistic
● The test statistic is a function of a random sample.
● It serves as a tool to evaluate the validity of the
null hypothesis.
Nature of Test Statistic:
● It is a random variable whose distribution is
known under the null hypothesis.
Test Statistic
Observed Test Statistic:
● The value of the test statistic calculated from
specific sample values is referred to as the
observed test statistic or simply the test statistic.
Rejection Region (Critical Region)
● A rejection region (or a critical region) is the
region (denoted by R) that specifies the values of
the observed test statistic for which the null
hypothesis will be rejected.
● This is the range of values of the test statistic that
corresponds to the rejection of H0 at some fixed
level of significance, α (which will be explained
later).
Rejection Region (Critical Region)
● If the value of the observed test statistic falls in the
rejection region, the null hypothesis is rejected and
we will conclude that there is enough evidence to
decide that the alternative hypothesis is true.
● If the observed test statistic does not fall in the
rejection region, we conclude that we cannot reject
the null hypothesis.
Summary
● Statistical Inference Overview
● Hypothesis Testing
● Types of Hypotheses
● Key terms in Hypothesis
Testing
Abstract on Next Session
● Steps in Hypothesis Testing
Thank You!!!