Undesirable Uses of Analysis of Variance: Reporting F-values Only, comparing Cell Means in a Post-hoc manner to Interpret Interaction Effects, and Performing Nonparametric Tests for the Reason of Small Sample Size.
Kwang B. Park
(Department of Psychology, Chungbuk National University)
Jinsup Eom
(Department of Psychology, Chungbuk National University)
Korean Journal of Clinical Psychology / Korean Journal of Clinical Psychology, (E)2733-4538
2001, v.20 no.1, pp.179-194
Park,
K. B., &
Eom,
J.
(2001). Undesirable Uses of Analysis of Variance: Reporting F-values Only, comparing Cell Means in a Post-hoc manner to Interpret Interaction Effects, and Performing Nonparametric Tests for the Reason of Small Sample Size.. Korean Journal of Clinical Psychology, 20(1), 179-194.
Abstract
This paper is to remind cases in which analysis of variance is misused in psychological literature and explain why such misuses of ANOVA should be avoided The most common misuse of ANOVA is that authors report only F-values without any other results from the ANOVA. The secondly common misuse is that some authors perform unplanned comparisons of cell means to interpret significant interaction effects. The thirdly common misuse is that some authors perform nonparametric tests instead of standard parametric ANOVA for the reason of small sample sizes.
- keywords
-
ANOVA,
interaction,
post-hoc comparison,
nonparametric test
- Submission Date
- 2000-08-14
- Revised Date
- 2000-11-15
- Accepted Date
- 2000-12-21