What does a between-subject (unpaired) design compare?
Values recorded from different individuals.
What does a within-subject (paired) design compare?
Values recorded from the same individual.
Why should testing order be counterbalanced?
To control for the confounding effect of order.
What is a confounding variable?
An external factor that affects both independent and dependent variables.
Why can graphs of means ± SEM be misleading for paired data?
Because identical SEMs can hide differences in individual change consistency.
How should paired data be presented?
Show individual change lines or pre–post plots, not just group means.
What does a paired t-test compare?
The mean of the differences between two groups of paired data.
What does an unpaired t-test compare?
The difference between means of two unrelated groups.
What assumptions must be met for an unpaired t-test?
Normal distribution, statistical independence, and equal variance (homogeneity).
What test should be used if variances are unequal?
Welch’s corrected t-test.
Why use ANOVA instead of multiple t-tests?
Multiple t-tests increase Type I error risk.
What does ANOVA test for?
Differences in means between two or more unmatched groups.
What are the null and alternative hypotheses for ANOVA?
H₀: all group means are equal; H₁: at least one mean differs.
What are ANOVA’s main assumptions?
Normal residuals, independence, and homogeneity of variance.
What is the nonparametric alternative to ANOVA?
Kruskal–Wallis test.
What do post hoc tests do?
Identify which groups differ after rejecting the null hypothesis.
What are common post hoc tests and their uses?
Bonferroni: strict, prevents Type I errors; Tukey: balanced; Dunnett’s: all vs control.
What is a false discovery rate (FDR)?
A correction used for massive multiple comparisons (e.g. gene arrays).
When are nonparametric tests used?
When parametric test assumptions fail (e.g. non-normal data).
