Introduction to Hypothesis Testing

Hello! Welcome to your fourth lesson in the Core Statistics module.

In our last session, we explored the Central Limit Theorem (CLT). We learned that for a large enough sample size, the distribution of sample means will be approximately normal, regardless of the original population's distribution. This powerful idea is the bridge that allows us to use sample data to make educated guesses about an entire population.

Today, we will build directly on that foundation to cover one of the most fundamental topics in statistics: hypothesis testing. You will learn the formal framework that data scientists and researchers use to test claims and make decisions based on data. This is a core competency frequently tested on the GATE DA exam.

Our learning outcome for this lesson is to describe the hypothesis testing framework, including the null and alternative hypotheses, the p-value, and the application of z-tests, t-tests, and chi-squared tests.

Let's begin by building an intuitive understanding of what hypothesis testing is all about.

1. The Core Idea: Is It a Real Effect or Just Chance?

At its heart, hypothesis testing is a formal procedure for answering the question: "Is the effect I'm seeing in my sample data real, or could it have happened by random chance?"

For a quick and highly intuitive overview of the main concepts, let's start with a video.

Statistics in 10 minutes. Hypothesis testing, the p value, t-test, chi squared, ANOVA and more

This video from Global Health with Greg Martin introduces the core logic of hypothesis testing, including the null hypothesis and the p-value, in a very accessible way.

Please watch from the beginning to 02:02. Focus on how the narrator defines: The null hypothesis as the default assumption of 'no difference'. The p-value as the probability of getting your observed result (or something more extreme) if the null hypothesis were true. The alpha value (α) as the pre-decided cutoff for making a decision.

2. The Formal Framework and Key Terminology

The video gave us the intuition. Now, let's solidify the key terms and the step-by-step process. This formal structure is what you'll need to recognize for the GATE exam.

Please read the following sections from the GeeksforGeeks article, which lays out these definitions clearly.

Hypothesis Testing - GeeksforGeeks

This article provides the formal definitions and structure you need for your notes. We will focus on the key vocabulary and the overall process.

Read the following sections to get a structured overview: 'Defining Hypotheses': Understand the roles of the Null Hypothesis (H_0) and Alternative Hypothesis (H_1). 'Key Terms of Hypothesis Testing': Pay close attention to the definitions of Significance Level (α), p-value, Test Statistic, and Critical Value. 'Types of Hypothesis Testing': Learn the difference between a one-tailed test (directional, e.g., 'greater than' or 'less than') and a two-tailed test (non-directional, e.g., 'not equal to'). 'What are Type 1 and Type 2 errors': This is critical exam vocabulary. Understand what a Type I Error (False Positive, \alpha) and a Type II Error (False Negative, eta) represent. 'How does Hypothesis Testing work?': Review the 6-step process. This is the complete framework.

Summary of the Hypothesis Testing Steps

Based on the article, the process can be summarized as follows:

Define Hypotheses: State the null hypothesis ( $H_{0}$ ), which assumes no effect, and the alternative hypothesis ( $H_{1}$ ), which is the claim you want to test.
Choose Significance Level ( $α$ ): Decide on your threshold for statistical significance. A common choice is $α = 0.05$ , which means you accept a 5% risk of making a Type I error (incorrectly rejecting a true null hypothesis).
Collect Data: Gather your sample data.
Calculate Test Statistic: Compute a single number that summarizes how far your sample data deviates from the null hypothesis. The type of test statistic depends on your data and assumptions.
Make a Decision: Compare your test statistic to a critical value or, more commonly, compare your calculated p-value to your significance level ( $α$ ).
- If p-value ≤ $α$ , you reject the null hypothesis. The result is statistically significant.
- If p-value > $α$ , you fail to reject the null hypothesis. The result is not statistically significant.
Interpret the Results: State your conclusion in the context of the problem.

3. Choosing the Right Test: Z-test vs. T-test

Step 4, "Calculate Test Statistic," is where the different tests come in. Let's start with the two most common tests for comparing means: the Z-test and the T-test. Your ability to choose the correct one is often tested.

The key difference lies in what you know about the population and your sample size. The DataCamp article below provides an excellent comparison.

T-test vs. Z-test: When to Use Each - DataCamp

This article clearly outlines when to use a t-test versus a z-test. The summary table is especially useful for quick revision.

Please read the following three sections: 'What is a t-test?': Note the conditions for its use (small sample, unknown population variance). 'What is a Z-test?': Note the conditions for its use (large sample, known population variance). 'Key Differences Between t-tests and Z-tests': Focus on the summary table. This is the main takeaway.

Decision Rule Summary:

Use a Z-test when:
- The sample size is large ( $n > 30$ ), AND
- The population standard deviation ( $σ$ ) is known.
Use a T-test when:
- The sample size is small ( $n < 30$ ), OR
- The population standard deviation ( $σ$ ) is unknown (you must estimate it using the sample standard deviation, $s$ ).

Now, let's see these tests applied in practice. This next video walks through two problems, one using a Z-test and the other a T-test.

Hypothesis Testing Problems - Z Test & T Statistics - One & Two Tailed Tests 2

This video from The Organic Chemistry Tutor provides clear, worked-out examples. Don't worry about doing the calculations yourself; focus on why a specific test was chosen and how the conclusion was reached.

Watch the two problem explanations: Problem 1 (00:01 - 05:05): This is a Z-test. Notice that the sample size is large (n=40). See how the two-tailed hypothesis ((\mu eq 80)) is set up. Problem 2 (05:58 - 13:24): This is a T-test. Notice that the sample size is small (n=10) and the population standard deviation is unknown. See how the one-tailed hypothesis (\mu < 2) is set up.

4. Testing Relationships: The Chi-Squared Test

Z-tests and T-tests are for numerical data (means). What if you want to test a relationship between two categorical variables? For example, is there an association between a person's education level (e.g., High School, Bachelor's, Master's) and their job satisfaction (e.g., Low, Medium, High)?

For this, we use the Chi-Squared ( $χ^{2}$ ) Test.

The core idea of the Chi-Squared test is to compare the observed frequencies in your data to the expected frequencies you would see if there were no relationship between the variables (i.e., if the null hypothesis were true). A large difference between observed and expected values leads to a high $χ^{2}$ statistic and a small p-value, suggesting the relationship is real.

Let's return to our first video for a simple explanation.

Statistics in 10 minutes. Hypothesis testing, the p value, t-test, chi squared, ANOVA and more

This segment explains the intuition behind the Chi-Squared test using the same hypothesis testing framework.

Watch the segment from 05:34 to 08:14. Focus on how the null hypothesis is framed as 'no relationship' or 'no association' between two categorical variables.

5. Summary: A Flowchart for Choosing Your Test

We've covered a lot of ground. This flowchart provides a simple visual guide to help you decide which statistical test to use based on the type of variables you are comparing. For the GATE exam, being able to quickly make this decision is a valuable skill.

Statistical Test Selection Flowchart with Distributions — This flowchart helps you choose a statistical test. For a **Numeric vs. Categorical** comparison, you would use a T-test or Z-test (depending on sample size and variance knowledge). For a **Categorical vs. Categorical** comparison, you would use a Chi-Squared Test. This visual map summarizes our lesson today.

Conclusion

In this lesson, we established the complete framework for hypothesis testing, a critical process for making data-driven decisions.

Key Takeaways:

Framework: Hypothesis testing is a structured 6-step process to determine if an observed effect in a sample is statistically significant or due to chance.
Core Components: The process revolves around defining a null hypothesis ( $H_{0}$ ) and an alternative hypothesis ( $H_{1}$ ), and then using the p-value to make a decision based on a pre-set significance level ( $α$ ).
Decision Rule: If p-value $\leq α$ , you reject $H_{0}$ .
Choosing a Test:
- Z-test: For means; use with large samples ( $n > 30$ ) and known population variance.
- T-test: For means; use with small samples ( $n < 30$ ) or unknown population variance.
- Chi-Squared Test: For categorical data; tests for association or independence between two variables.

Preview of the Next Lesson:
Hypothesis testing gives us a "yes/no" answer about an effect. But often, we also want to know, "What is the likely range for the true population value?" In our next and final lesson for this module, we will explore confidence intervals, which provide a way to estimate a population parameter with a certain level of confidence.

Recommended resources