The problem of inadequate control in matched-group designs, frequently described as regression effects, is discussed in terms of the properties of partial correlations. Both correlational analysis and the comparison of matched groups suffer from unreliability and low validity in control measures, but the former does not require discard of cases that cannot be matched. Also, correlational analysis is more flexible and less susceptible to illusions of experimental control than the matching design.
Get full access to this article
View all access options for this article.
References
1.
AsherW.VockellE. L.Errors in interpreting regression toward the mean: Reply to Lytton, Croxen, and Pysh. Psychological Reports, 1974, 35, 192–194.
2.
LyttonH.CroxenM.PyshF.Regression to the mean misunderstood: A reply to Vockell and Asher. Developmental Psychology, 1973, 8, 3–5.
3.
VockellE.AsherW.Methodological inaccuracies in Croxen and Lytton's “Reading disability and difficulties in finger localization and right-left discrimination.”Developmental Psychology, 1973, 8, 1–2.
