Scholion.ai

"Clarifying some controversial issues surrounding statistical procedures for detecting moderator variables": Correction to Stone and Hollenbeck (1989)

Stone, Eugene F., Hollenbeck, John R.

Journal of Applied Psychology · 1990 · Volume 75 · Issue 1

DOI: 10.1037/h0090376

Abstract

Reports an error in "Clarifying some controversial issues surrounding statistical procedures for detecting moderator variables: Empirical evidence and related matters" by Eugene F. Stone and John R. Hollenbeck ( Journal of Applied Psychology , 1989[Feb], Vol 74[1], 3-10). There were several typographical errors in Tables 1, 2, and 3. The corrected tables are provided in the erratum. In addition, the presence of several negative values in Arnold's (1982) original data sets rendered invalid the results and conclusions that were presented on page 6 concerning the regression analyses of log-transformed values for Data Set 2. To rectify this problem, the authors reanalyzed corrected, log-transformed values of the original data. The tables shown in the erratum reflect the results of analyses for the corrected data. (The following abstract of the original article appeared in record 1989-17703-001 .) Data used by H. J. Arnold (see record 1982-20211-001 ) to demonstrate the need for separate data analytic methods to assess form of relationship and degree of relationship differences were shown to have such problems as nonhomogeneous variance and heteroscedasticity, making them unsuitable for analysis with moderated regression. After applying variance-stabilizing transformations, we analyzed the same data by using moderated regression. Results of these analyses showed that the results reported by Arnold do not serve as a viable basis for perpetuating the practice of assessing degree of relationship differences with the subgrouping method. Moreover, the moderated regression strategy is capable of detecting not only form of relationship differences but also degree of relationship differences. Implications of these findings are discussed. In addition, a number of other issues surrounding the detection and description of moderating effects are considered. (PsycInfo Database Record (c) 2025 APA, all rights reserved)

Enable JavaScript for citation tools and additional paper interactions.