🎯 Learning goals

  • Understand why sample size matters for generalisability of regression results.
  • Learn common rules of thumb and formulas for determining minimum sample size.
  • Apply guidelines to estimate required cases for your own study.

1) Why sample size matters

Small samples may lead to results that cannot be replicated or generalised to other populations. Reliable regression analysis requires enough cases to provide stable coefficient estimates and meaningful hypothesis tests.

2) Key guidelines from research

  • Stevens (1996): For social science research, about 15 participants per predictor are needed for a reliable equation.
  • Tabachnick & Fidell (2013): Provide the formula N ≄ 50 + 8m (where m = number of predictors). Example: With 5 predictors, you need at least 90 cases.
  • Stepwise regression: Requires more cases — around 40 participants per predictor.
  • Skewed dependent variables: Larger samples are recommended to maintain reliability.

Tip: Always calculate your sample size during study planning. These rules are guidelines, but a power analysis gives a more precise estimate based on expected effect size and desired statistical power.

3) Quick check-in

  1. If you plan to use 4 predictors, how many participants would Tabachnick & Fidell’s formula suggest?
  2. Why might stepwise regression require more cases per predictor?
  3. How would you justify your chosen sample size in your thesis or research report?
Modifié le: dimanche 21 septembre 2025, 07:50