For a certain common type of experimental situation, the better the subject's performance — i.e., the more skilled he is, or the easier his task (perhaps due to good human-engineering design) — the more markedly skewed his distribution of scores tends to be and the more nonrobust parametric tests tend to be. So when performance conditions are optimal, test conditions tend to be pessimal and vice versa.
Get full access to this article
View all access options for this article.
References
1.
BradleyJ. V.The central limit effect for a variety of populations and the influence of population moments. Journal of Quality Technology, 1973, 5, 171–177.
2.
GayenA. K.Significance of difference between the means of two non-normal samples. Biometrika, 1950, 37, 399–408.