If your advisor or a reviewer has ever handed back your results section with red ink, it's usually one of three things — and none of them are about whether your test was 'right.'
1. You reported a p-value with no effect size.
A significant t-test or ANOVA tells a committee something happened, not how much. APA 7th edition expects Cohen's d, eta-squared, or an equivalent alongside every test statistic. If your write-up reads "t(48) = 2.31, p = .025" and stops there, that's the first thing a reviewer circles. Add the effect size and a one-line plain-language read of its magnitude (small/medium/large per convention for that test).
2. You ran multiple comparisons without correcting for them.
Running five t-tests instead of one ANOVA, or testing every pair after a significant omnibus test, inflates your false-positive rate. If you didn't apply a Bonferroni, Tukey HSD, or Holm correction (and say so explicitly in the write-up), expect a comment asking why not. This is one of the most common flags on multi-group comparisons and post-hoc analyses.
3. You didn't check — or didn't report checking — your test's assumptions.
Normality for parametric tests, homogeneity of variance for ANOVA/t-tests, linearity and independence for regression. Committees don't expect perfection; they expect a sentence: "Levene's test indicated equal variances (p = .41), so a standard independent t-test was used." Silence on assumptions reads as "didn't check," even if you did.
The fix for all three is the same: build the effect size, correction, and assumption check into your workflow before you write the results section, not after committee feedback forces a rewrite. Free calculators (t-test, ANOVA, regression, factor analysis, etc.) that surface effect sizes and assumption checks alongside the test statistic, and format the whole thing as an APA-ready sentence, save the second pass entirely — https://statmate.org
Top comments (0)