In a recent JSSR article, Tamney, Johnson, and Burton (1992) reported on a valuable set of data regarding attitudes toward abortion and other issues and demographic characteristics of residents in the Muncie SMSA. They applied LISREL to the responses and presented their conclusions in the form of two path diagrams. LISREL is a complex type of regression analysis involving a priori hypotheses. However, the more complex types of statistical analysis involve more assumptions, and the results might be adversely affected to a greater extent than usual if these assumptions are invalid. In fact, path analysis has been criticized; the entire Summer 1987 issue of the Journal of Educational Statistics was devoted to critique and rebuttal by path analysts. Let us assume that one is in the position of a lecturer trying to explain the Tamney et al. results as conveyed by their two path diagrams to an audience of college graduates with no training in statistical analysis. The two diagrams seem to give different answers to the question of which factors influence attitude toward abortion. In Diagram 2B only two paths enter Abortion (AB), from AGE and Education (EDU). However, in Diagram 2A arrows enter AB from Church Influence (CI), New Life (NLI), Social Traditionalism (ST), Protect Animals (PCT), and PRI (called Privacy Organization on the diagrams but Privacy Orientation elsewhere). All four of these attitudinal variables appear in Diagram 2B but without arrows emanating from them. After examining the intercorrelations in Table 1 of the article in question, I detected a cluster of rather closely related attitudinal variables: NLI, CI, PC (Political Conservatism), and ST in one direction; and AB and PRI in the opposite direction. I wondered how a unit-weighted composite of the z scores on these six attitudinal variables would relate to the demographic variables. Formula 16.20 in Guilford's and Fruchter's (1978:394) textbook seemed appropriate. In applying this formula, the direction of AB and PRI was reflected by changing the algebraic signs of all correlations involving these two variables (except the correlation with each other). The resulting composite score