Blanchard and Elkin employed alternative statistical procedures to those of Suinn, Jorgensen, Stewart, and McGuirk and arrived at different conclusions. The legitimacy of the original analysis is reasserted, and some problems with the alternatives are pointed out. Blanchard and Elkin wrote from a theoretical position which favors studying social psychological problems rather than clinical problems. This stance is not shared by the present author. Blanchard and Elkin (1973) have criticized the Suinn, Jorgensen, Stewart, and McGuirk (1971) research on statistical and theoretical grounds. This rejoinder systematically evaluates each of the criticisms.Blanchard and Elkin offered alternatives to the statistical analyses employed in the original research and drew alternative conclusions. Their first point was that the absolute magnitude of change in approach scores was lower for the experimental group than for the control group, C-2; presumably, this discrepancy in scores supports their conclusion that within-group differences were larger than between-group differences. This conclusion, however, is based only on inspection, not statistical testing. Reexamination of Table 1 in the original article (p. 275) shows that the more sophisticated inference on po demonstrated that statistically significant changes occurred only for the experimental group and not for the control samples.Blanchard and Elkin proceeded to cite / values for the control groups, C-2 and C-l, suggesting that both values are "significant" by a one-tailed test. Ordinarily, of course, one-tailed tests are not used in examining data on control groups where the experimenter cannot predict whether systematic, directional changes will occur. Some confusion may have been generated by an error in the original article. In Table 1 of that article, all / tests were said to be one-tailed. In fact, a onetailed test was used only for the experimental groups, as justified by the directional predictions which had been made for those groups.