Discrimination Learning and Discrimination Reversal Learning in Normal and Feebleminded Children

Two studies compared methods of estimating factor scores for the Child's Report of Parental Behavior Inventory. In Study I, least squares factor scores for 5th and 6th graders were computed and correlated with estimates obtained by multiple regression, weighted summation, and unweighted summation with either (a) all scales best defining each factor, (b) the three best, or (c) the two best defining scales. All three estimation methods were highly accurate, and estimates obtained with three and even two scales per factor still correlated highly with the factor scores. In Study 11, factor scores for college students were correlated with estimates obtained by multiple regression and unweighted summation of the same subsets of scales. Again both estimation methods were highly accurate, and reducing the number of scales did not markedly reduce the correspondence of estimated scores to exact scores.