Hypothesis Testing Notes
Confidence Level and Alpha
- Confidence Level: A measure of how confident we are that our estimate includes the population parameter. The commonly used level is 95%.
- Alpha (α): The significance level calculated as:
- Formula: 1 - Confidence Level (in decimal)
- Example: For a 95% confidence level, α = 1 - 0.95 = 0.05
- When dealing in percentages, convert: 100% - Confidence Level\n - Example: 100 - 95 = 5% → 0.05 (as a decimal)
Degrees of Freedom
- Degrees of Freedom (df): Calculated as sample size (n) minus one.
- Example: If n = 45, df = 45 - 1 = 44.
Test Statistic and P-value Computation
- Test Statistic: For a two-sample test using t-distribution.
- T-value and P-value: Utilize calculators or statistical software to compute.
- Example: If test statistic (t) = 2.3, P-value for two-tailed test might result in approximately 0.02625.
Decision Making in Hypothesis Testing
- Null Hypothesis (H₀): Typically posits no effect or no difference.
- Rejecting or Not Rejecting:
- If the P-value < α, reject H₀.
- If P-value ≥ α, do not reject H₀.
- Critical Values
- Determine critical values based on t-distribution (for t-tests) or z-distribution (for z-tests).
- Example critical values for α = 0.05 in a two-tailed test might be ±2.015 (for df = 44).
Test Scenarios
- Scenario A (T-test with df = 44):
- Test statistic (t) is outside critical values, rejection of H₀ occurs (example: t = 2.3).
- Scenario B (Z-test with 99% confidence):
- Given a P-value of 0.035 when α is 0.01, conclusion is to reject H₀ as P-value < α.
Working with Critical Values
- Comparison of Test Statistic with Critical Values:
- If a test statistic lies between critical values, do not reject H₀.
- If it lies beyond, do reject H₀.
Hypothesis Testing Flow
- Establish Hypotheses:
- Null (H₀): mu = hypothesized value.
- Alternative (H₁): mu does not equal hypothesized value.
- Determine Confidence Level and Calculate α.
- Compute Test Statistic and P-value.
- Compare P-value with α:
- Draw conclusions based on comparison: reject or not reject.
- Confidence Intervals:
- Calculate using mean ± (critical value * standard error).
Final Thoughts
- Conclusively articulate your findings based on the statistical analysis and the context of the hypothesis test.
- Use structured statements to convey whether H₀ was rejected or not and what that indicates about the population mean.