Described is an attempt to develop an instrument which could be useful in selecting teacher education candidates. A sample of 330 beginning teachers (100 elementary, 108 language arts/social science, and 122 mathematics/science) had responded to four self-report instruments (covering values, adjustment, practices, and attitudes). Their teaching effectiveness had been evaluated using pupil and trained-observer ratings. For a random half of the sample, a factor analysis of the self-report items was followed by a stepwise regression analysis to obtain prediction equations for "effectiveness" within each group of teachers. The equations were then cross-validated on the other half of the sample. In the first half, only one prediction equation was found to be significant, and in the second half, the correlation between predicted and actual scores was very low. Thus it was not considered feasible to pursue the development of the proposed instrument further. (MM)