Hypothesis Testing in Statistics
Hypothesis Testing Overview
Definition: A process of gathering evidence to support or rebut a claim (hypothesis).
Purpose: Decision-making process for evaluating claims about a population based on sample characteristics.
Types of Hypotheses
Null Hypothesis (H0): Indicates the absence of a relationship or effect; claims equality or no significant difference. Often represented with symbols like =, ≥, or ≤.
Alternative Hypothesis (Ha): Indicates the presence of a relationship or effect; claims inequality. Symbols include ≠, <, or >. Can be:
Directional: One-tailed (e.g., > or <)
Non-directional: Two-tailed (e.g., ≠)
Important Reminders
The null hypothesis asserts a specific value for the population parameter and is presumed true until evidence suggests otherwise.
The alternative hypothesis negates the null hypothesis.
Hypothesis Testing Decisions
Correct Decisions:
Rejecting a false H0 (correctly concluding there is a significant effect)
Accepting a true H0 (correctly concluding there is no significant effect)
Errors:
Type I Error: Rejecting a true H0 (false positive)
Type II Error: Accepting a false H0 (false negative)
Examples of common scenarios leading to errors:
Claiming an effect exists when it doesn’t (Type I)
Failing to detect an effect that exists (Type II)
Key Parameters in Hypothesis Testing
Population Mean (μ) and Population Proportion (p) are often tested parameters.
Common Phrases for Hypotheses
Examples of phrases used in formulating H0 and Ha:
H0: "is equal to" vs. Ha: "is not equal to"
H0: "is less than or equal to" vs. Ha: "is greater than"
H0: "is at least" vs. Ha: "is less than"
Example Cases and Formulations
Case A:
Statement: Average BMI of pupils in a feeding program
H0: Average BMI is not different from 18.2 kg
Ha: Average BMI is different from 18.2 kg
Type I Error Probability: 0.05
Conclusion when H0 rejected: Average BMI is different from 18.2 kg
Conclusion when H0 accepted: Average BMI is not different from 18.2 kg
Case B:
H0: Average content of soda X is 330 ml
Ha: Average content of soda X is less than 330 ml
Type I Error Probability: 0.01
Common confidence levels used in testing (e.g., 90%, 95%, 99%) for different claims.
Directionality of Tests
One-Tailed Tests: Hypothesis has a directional claim (e.g., mean is greater than or less than a value).
Two-Tailed Tests: Hypothesis tests whether there is a difference without specifying direction (e.g., mean is not equal to a value).
Exercises and Applications
Formulate H0 and Ha based on various claims, deciding on the one-tailed or two-tailed nature of the hypothesis.
Practice writing hypotheses for different scenarios, such as average prices or student study times, and determine the testing parameters and potential errors involved.