Hypothesis Testing
Foundations of Hypothesis Testing
Hypothesis Testing Overview: Involves making a claim about a population parameter (e.g., mean), followed by statistical analysis to test this claim.
Types of Hypotheses:
Null Hypothesis (H0): Assumes no effect or difference; proposes a specific parameter value.
Alternate Hypothesis (H1): Contradicts the null hypothesis; indicates the presence of an effect or difference.
Testing Types:
One-Tail Test: Tests for effects in one direction.
Two-Tail Test: Tests for effects in both directions.
Errors in Hypothesis Testing:
Type I Error (⍺ Error): Rejecting H0 when it is true.
Type II Error (β Error): Failing to reject H0 when it is false.
Decision Rules:
Define a level of significance (common values: ⍺ = 0.05).
Determine the test statistic and critical value to compare with.
Test Statistics:
Z-test: Used when population variance ($\sigma$) is known and sample size (n) is large.
T-test: Used when $\sigma$ is unknown, n is small, and sample is from a normally distributed population.
Degrees of Freedom (df): Given by $df = n - 1$; one df is lost for each sample statistic used as a point estimator.
Using t-Distribution:
More spread than normal distribution, symmetrical about the mean; approaches normal as n increases.
Trade-offs in Errors: Increasing confidence (lowering ⍺) raises the risk of Type II errors (β).
Steps in Hypothesis Testing:
Formulate H0 and H1.
Specify significance level (⍺).
Select test statistic (Z, t, etc.), establish critical values.
Compute test statistic and make a decision - reject or do not reject H0.
Take action based on results (business decision).