Interpreting and Reporting Post-Hoc Results
You'll learn how to extract, understand, and communicate which groups differ significantly after running post-hoc tests following ANOVA.
The Core Idea
After a significant ANOVA tells you some groups differ, post-hoc tests reveal which specific pairs differ. Reading and reporting these results properly turns your statistical findings into actionable insights. Think of ANOVA as detecting smoke in a building, and post-hoc tests as identifying which rooms have the fire—your job is to clearly communicate where the fires are.
Reading Post-Hoc Output
Post-hoc test output typically shows:
- Pairwise comparisons: which groups are being compared
- Adjusted p-values: corrected for multiple testing (Tukey, Bonferroni, etc.)
- Confidence intervals: the range of plausible differences between means
- Significance indicators: often asterisks or yes/no flags
Look for adjusted p-values below your alpha level (usually 0.05). These indicate statistically significant differences between those specific groups.
Visualizing Differences
Create compact letter displays or grouping labels where groups sharing a letter are not significantly different. For example, if groups A, B, and C receive labels "a", "b", "b" respectively, then B and C don't differ from each other, but A differs from both.
Bar charts with error bars and significance brackets connecting different groups make findings immediately clear to non-technical audiences.
Reporting in Practice
State which test you used and why: "Tukey's HSD revealed that Group A (M=15.2) scored significantly higher than Group C (M=10.4), p=.003, while Groups B and C did not differ significantly."
Always report means, the chosen correction method, and exact or bounded p-values. Include confidence intervals when space permits.
Key Takeaway: Post-hoc results translate ANOVA's "something differs" into specific, actionable "Group X exceeds Group Y"—report them with means, adjusted p-values, and clear visual aids.