Assessing the effectiveness of programmed generalization

Issues related to assessing change and retention of change were discussed. An alternative analysis was suggested for the data of a recent study by Walker and Buckley (1972). These authors had found that peer reprogramming, equating stimulus conditions, teacher training, and control groups maintained 77, 74, 69, and 67%, respectively, of appropriate behavior produced in a token economy. Their analysis made no use of baseline levels. Two analyses incorporating baseline scores were suggested. One involved change scores; the other, analysis of covariance using baselines as the covariate. Problems with the data made a clear preference difficult, but it was concluded that either analysis would have resulted in conclusions different from those of Walker and Buckley.