Richard R. Jones

Cost-Effectiveness of Teaching Family Programs for Delinquents: Results of a National Evaluation.

During the past decade the Teaching Family Model (TFM) of delinquency treatment has evolved into a national network of community-based, behaviorally oriented group homes. A longitudinal summative evaluation of 26 TFM homes and 25 comparison programs from the same or neighboring communities was mounted in 1975 and the results of a 5- year cost-effectiveness study are presented. Using a variation of output value analysis, TFM homes were found to be 7% less expensive to operate on a per diem basis and cost approximately 20% less per client. Cost-effectiveness was better for TFM programs on measures of school performance, but no different on deviant behavior or social/personality outcomes either at discharge or up to 3 years later.

Interrelationships Among Personality Scale Parameters: Item Response Stability and Scale Reliability:

FOR the assessment of individual differences in hypothetically stable personality traits, it is desirable to develop scales which will maximize both scale score variance and score retest stability. Since scales are composed of individual items, it is reasonable to expect that characteristics of scales (e.g., score variance and score stability) should be related to characteristics of their items. The cogency of this statement for scale construction is seen in the standard practice of eliminating nonvalid items from scales to increase scale validity. In a similar manner, score retest stability should be related to the stability of responses to the scale’s items. The objective of developing scales to maximize both score variance and score stability suggests two criteria for the selection of items: (a) good items should elicit responses which maximize individual differences, and (b) good items should elicit responses which are stable over time. In general, items with large variances and high retest stability should tend to maximize score variance and score stability respectively. The simultaneous satisfaction of these two

A cautionary note on predicting behavioral treatment outcome

Miller and Gottliebs (1974) Responsivity Index for Parents (RIP) as a predictor of behavioral treatment outcomes is discussed. It is argued that the 91% hit rate claimed for RIP was obtained in an inappropriately comprised cross-validation sample. However, reanalyses of the data in an appropriate cross-validation sample do provide tentative support for the validity of RIP.