P-Value and Hypothesis Testing - Chapter 8 Notes
P-Value and Hypothesis Testing - Chapter 8 Notes
Key Concepts of Hypothesis Testing
- Understanding Hypothesis Testing: A statistical method using sample data to evaluate the validity of a hypothesis about a population.
- Foundational Steps:
- State null (H0) and alternative hypotheses (H1).
- Set criteria for decision based on critical regions.
- Conduct tests (e.g., z-test) on collected data.
- Make a probability-based decision to reject or retain the null hypothesis.
- Errors:
- Type I Error: Rejecting a true null hypothesis.
- Type II Error: Failing to reject a false null hypothesis.
Hypotheses
- Definition: A precise, testable prediction about an event or outcome.
- Null Hypothesis (H0): No meaningful change or effect at the population level (e.g., H0: μControl = μTreatment).
- Alternative Hypothesis (H1): States there is a significant effect/change (e.g., H1: μControl ≠ μTreatment for two-tailed tests or H1: μControl > μTreatment for one-tailed tests).
Logic of Hypothesis Testing
- State Hypothesis: Predict about the population.
- Predict Characteristics: Based on hypotheses, forecast what the sample data should look like.
- Collection of Data: Gather a random sample.
- Comparison: Evaluate the sample data against the predictions from the hypothesis.
- If consistent, retain H0; if inconsistent, reject H0.
Steps in Hypothesis Testing
- State the Hypotheses:
- Null Hypothesis (H0) and Alternative Hypothesis (H1).
- Set Criteria for Decision:
- Identify critical regions where the null is rejected via alpha levels.
- Common α-level is 0.05 (5%).
- Collect Data:
- Obtain random samples and calculate the sample statistics, especially the mean.
- Make a Decision:
- Compare the sample outcome to the critical value.
- If z-score is in the critical region, reject H0; if not, fail to reject H0.
Decision Criteria and Alpha Levels
- The Alpha Level (α): The probability threshold for rejecting the null hypothesis.
- Critical regions are defined as areas where if the sample mean falls, H0 is rejected (e.g., α = 0.05 splits into two tails for two-sided tests: 0.025 in each tail).
- Z-scores corresponding to different alpha levels are:
- α = .05: z = ±1.96
- α = .10: z = ±1.645
- α = .01: z = ±2.576
Examples of Hypothesis Testing
- Example of Violence in Media:
- H0: Children who view violent media show equal or lesser aggression than those who do not.
- H1: Children who view violent media show greater aggression.
- Example of Note-Taking Method:
- H0: μLaptop = μHand
- H1: μLaptop ≠ μHand
- Example of Antidepressants:
- H0: μNo Prozac ≤ μProzac
- H1: μNo Prozac > μProzac
Interpretation of Results
- If results fall within the critical region:
- Reject H0 (suggests the effect is statistically significant).
- If results do not fall within the critical region:
- Fail to reject H0 (no significant evidence found).
- Strive to use probability language when discussing results (e.g., p < α implies significant findings).
Summary of Hypothesis Testing Steps
- Step 1: Formulate hypotheses, clarify if one- or two-sided.
- Step 2: Determine critical regions based on chosen alpha level.
- Step 3: Collect data and compute statistics (e.g., z-scores, p-values).
- Step 4: Draw conclusions based on statistical significance to accept or reject the null hypothesis (H0).