Humphreys and Drasgow (1989b) recognize two types of dependent variables: the original measure- merits collected in an experiment and mathematical variables that are subjected to statistical analysis. Overall and Woodward (1975) were explicitly con cerned with the latter, whereas Humphreys and Dras gow contend that they were concerned with reliability of the original measurements from which difference scores may be computed. These are quite different matters. Criticisms should focus on points of disagree ment, and there has never been any disagreement con cerning the importance of reliability of the original measurements. The notion that treatment effects should be considered a part of the true variance for calculation of reliability estimates is rejected as stem ming from their failure to understand the basic differ ence between reliability and validity. Index terms: control of individual differences, difference scores, measurement of change, reliability of the marginal distribution, statistical power, within-group reliabili ties.

The fact that difference scores tend to be less reli able than the original measurements from which they are calculated should not be a matter of concern in testing the significance of treatment-induced change. The reliabilities of the original measurements are im portant because unreliability attenuates correlation, and substantial correlation between prescores and post- scores is required for difference scores to be of value in controlling for individual differences. Reliability notwithstanding, difference scores provide superior control over true baseline differences in quasi-experi mental research, whereas the analysis of covariance (ANCOVA) is generally preferable for baseline control in randomized experimental designs. Index terms: analysis of covariance, baseline correction, difference scores, measurement of change, reliability.

A method for multidimensional scaling of group differences in categorical data patterns was used to investigate configural relationships among alcohol use and abuse groups. The analysis resulted in a model from which two primary etiologic concepts, plus a moderator, were derived. Exposure is the concept that summarizes demographic factors related to level of alcohol use. However, problem drinking differs from frequent drinking along a dimension in the demographic domain that is independent of the exposure dimension. Duration of frequent alcohol use is a concept that relates to age, duration, and chronicity variables. The relationship between resources and responsibilities appears to be a moderator in the dimension separating frequent and problem drinking. Low income and family responsibilities interact to make frequent alcohol use more likely to be perceived as a problem.

A convenient, two-stage general linear regression approach to analysis of variance is described for use in univariate or multivariate designs involving one repeated measurement factor and one or more independent classification factors. A brief illustrative example is provided.

d in attempting to isolate a population of depresse d patients likely to be resistant to treatment . The topic "treatment resistant depression " immediatel y raises questions of definition. What is depression! What constitutes adequate treatment"! How is resistance to treatment to be defined? Before one can begin to consider the epidemiology of "treatment resistant depression" , the nature of the entity must be given explicit operational definition.

SUBSCL is a computer program written in FORTRAN IV for the IBM 1130 disk system to accomplish multidimensional scaling of objects in terms of similarities on one set of (ordinal or interval) measurements and then to project into the resulting scale model a second set of measurement vectors to provide meaning for directions in the reduced space. The program was originally written to provide a . scale model of relationships among psychotherapeutic drugs based on analysis of animal test profiles and then to position in the resulting model symptom vectors descriptive of the clinical utility of the drugs. It should be useful in other situations where several objects are represented by two different kinds of multivariate measurement profiles, where one set of measurements is considered primary for definition of object relationships, but where the other set

Extract: Circadian pericdicity of body temperature, pulse rate, respiratory rate, systolic, and diastolic blood pressure has been studied in 143 patients. The ages of the patients varied from 3 months to 21 years. A type of power spectrum analysis was used to reduce the data to a single quantity for each of the five measurements in each patient. This quantity was examined by various methods of statistical analysis such as correlation, regression, and analysis of variance. Evidence from these ananlyses suggested that a single circadian periodicity indicator could be calculated for each patient by combining the variance in circadian periodicity for temperature, pulse, respiration, and systolic blood pressure. Analysis of this indiactor, ‘composite corcadian periodicity’, revealed highly significant correlations with age, mean levels of the physiological measurements, and duration of the study. The relation with age was best defined by a third-degree polynomial equation. The smoothed wave that was derived from the data demonstrated a peak in composite circadian periedicity at 6 years of age, with a tendency to form a plateau at about 16 years of age. Speculation: A rational extension of this study would be to apply the data to the investigation of human diease. Although ‘normal ranges’ for age are published for tempertature, pulse rate, respiratory rate, and blood pressure [1, 8], many physicians modify these intuitively (lowered body temperature during the early morning hours, higher temperatures during the late afternoon and early evening). It would seem appropriate to develop normograms for these functions that incorporate the changes in mean level that occur with age, sex, and time of the day. In this way, the increased periodicity demonstrated in the child of 5 or 6 years of age, as well as the decreased periodicity in the younger and older subjects, would be accounted for in the normograms. The practicality of these techniques could be tested in computer-oriented diagnostic evaluations.