CitationTeachman, Jay & Morgan, S. Philip (1990). A Brief Response to DeMaris. Journal of Marriage and the Family, 52(1), 277.
AbstractA recent issue of this journal presented a "minisymposium" on family research methods, which included an article on logistic regression (Morgan and Teachman, 1988). The authors of this article have done a very fine job of presenting, in a nontechnical fashion, what essentially is a fairly complex statistical tool. However, the article contained a misinterpretation of certain mathematical properties of the model, which could cause confusion among those who are just becoming acquainted with this statistical technique. The purpose of this critique is to offer the correct interpretation of those properties and to make a few other, related comments about differences between logit analysis for categorical predictors and logistic regression models for continuous predictors. In the spirit of the Morgan and Teachman article, this discussion will also be framed in as nontechnical a fashion as possible. Morgan and Teachman have suggested the use of a transformation for logistic regression that would supposedly facilitate interpretation of results. This transformation, however, is mathematically inappropriate, given the nature of the model, and will typically produce misleading information if used indiscriminately. To see why this is so, let us reexamine the logistic regression model, using the authors' data and equations for illustrative purposes.
Reference TypeJournal Article
Journal TitleJournal of Marriage and the Family
Morgan, S. Philip