Q: Can post-hoc analysis be used to compare three groups in an experiment?

Detailed Question -

When we compare three groups in an experiment can we use post-hoc analysis to interpret the data?

1 Answer to this question

Generally, for an experiment with three groups, you need to analyze the groups using ANOVA. If ANOVA results are statistically significant, that means at least two groups have significantly different means. Conducting post-hoc analyses would then tell you which of the three groups have significantly different means. If ANOVA is not significant, post hoc analyses are redundant.


Usually Least Significant Difference, Turkey's HSD, and Bonferroni's are post hoc tests of choice. Duncan's and Dunnett's are also used, but some journals don't recommend Duncan's. Please make sure you check your journal's preferences before conducting the analysis.


Related reading: