⚖️

week 8: Parametric vs Nonparametric Tests m7

Jul 10, 2025

Overview

This lecture explains the differences between parametric and nonparametric statistical tests, their assumptions, examples, and guidelines for choosing the appropriate test in counseling research.

Parametric Tests

  • Parametric tests include independent samples t-test, paired/dependent samples t-test, and one-way ANOVA.
  • These tests require random, independent samples and interval or ratio (scale) level data.
  • The data should be normally distributed, with no significant outliers and homogeneity of variance.
  • Larger sample sizes are typically needed for parametric tests.
  • Parametric tests are more powerful, meaning they are more likely to detect real differences (lower risk of type 2 errors).

Nonparametric Tests

  • Nonparametric test examples: Mann-Whitney, Wilcoxon signed-rank, Kruskal-Wallis, and Friedman's ANOVA.
  • Used if parametric test assumptions are not met; still require random, independent samples.
  • Specific tests have unique assumptions, such as equal shape or symmetric distributions.
  • Nonparametric tests are more conservative and have less statistical power, increasing the risk of type 2 errors.

Choosing Tests & Corresponding Equivalents

  • Check parametric test assumptions first; use nonparametric if assumptions fail.
  • If independent samples t-test is not suitable, use Mann-Whitney.
  • Wilcoxon signed-rank is the nonparametric equivalent of the paired samples t-test.
  • Kruskal-Wallis corresponds to one-way ANOVA.
  • Friedman's ANOVA matches to one-way repeated-measures ANOVA.
  • Always verify assumptions for any statistical test before use.

Key Terms & Definitions

  • Parametric Test — Statistical test requiring specific assumptions about data distribution and measurement.
  • Nonparametric Test — Test with fewer, less strict assumptions about data distribution.
  • Type 1 Error — Incorrectly rejecting a true null hypothesis.
  • Type 2 Error — Failing to reject a false null hypothesis.
  • Statistical Power — Probability that a test will detect a true effect.
  • Homogeneity of Variance — Assumption that different groups have similar variances.
  • Interval/Ratio (Scale) Data — Numeric data with equal intervals; ratio data also has a true zero.

Action Items / Next Steps

  • Review the assumptions for each statistical test before analysis.
  • Match your data and research design to the appropriate statistical test.
  • Prepare to check sample size, measurement level, and distribution before selecting a test.