Public and Private Schools Revisited

Our dismay at the quality of Coleman, Hoffer and Kilgore's rejoinder to our critique of their study compels us to respond. We do so to restore our scientific credibility to readers of this journal, and to provide guidance for those who share our concern that quantitative methods be used competently and responsibly in social science. Our critique (Goldberger and Cain, 1982; henceforth GC) focused on two reports by Coleman, Hoffer and Kilgore (henceforth CHK): 1981a (henceforth DR) and 1982a (henceforth SEI). CHK's rejoinder was contained in 1982b (henceforth SE2). We will also refer to the other publications that present their analyses of the 1980 High School and Beyond sample: 1981b (henceforth HER), 1981c (henceforth FR) and 1982c (henceforth BB). The discussion that follows covers a range of issues: the reliability and validity of test scores, the appropriateness of the regression models employed and statistical inferences drawn, the extent of selectivity bias in background controls, the validity of the common hypothesis, the use of dropout rates to adjust sophomore-to-senior changes, and finally, the attribution of causality to what are regarded as school policies. To make the discussion reasonably self-contained, we will be restating the main elements of our critique and of CHK's rejoinder.