Harvey A. Skinner

Alcohol dependence syndrome: measurement and validation.

The alcohol dependence syndrome proposed by Edwards and Gross in 1976 is a central concept in the World Health Organization classification of alcoholism. The present study focused on the measurement and validation of this syndrome using a sample of 225 individuals with alcohol-related problems. A brief 29-item Alcohol Dependence Scale was derived that exhibited substantial internal consistency reliability (.92). Scale scores conformed quite closely to a normal distribution, which supports a quantitative (existing in degrees) interpretation of the syndrome. Higher levels of alcohol dependence were associated with social consequences from drinking as well as with greater quantities of alcohol consumed. As alcohol dependence increased, clients were less likely to show up for their first treatment appointment. The degree of alcohol dependence was directly related to psychopatholo gy (thinking disorder, hypochondriasis, persecutory ideas, anxiety, depression) and also to physical symptoms of the nervous, cardiovascular, and digestive systems. Given the high correlation between alcohol dependence and adverse consequences from drinking, the Alcohol Dependence Scale provides important information for treatment planning and may be especially relevant when deciding between goals of total abstinence versus controlled drinking.

Toward the integration of classification theory and methods.

Although various criteria have been proposed for the evaluation of psychiatric classifications, there is need for an integrated paradigm. This article presents such a framework based on the principles of construct validation. Emphasis is placed on the continual interplay between theory development and empirical analyses. First, the theory formulation component involves specification of the theoretical types and their functional linkages. Then, the internal validation component entails development of an empirical typology followed by an evaluation of its reliability, coverage, homogeneity, and robustness. Finally, the external validation component evaluates predictive validity with respect to treatment outcome, clinical meaningfulness of the diagnostic categories, descriptive validity, and generalizability of the typology to other populations. This framework is used to evaluate (a) the new diagnostic system (Diagnostic and Statistical Manual of Mental Disorders-Ill) developed by the American Psychiatric Association, (b) recent classifications of psychopathology that were derived using empirical clustering methods, and (c) classifications of behavioral disorders that have an explicit theoretical basis. A key challenge for the scientific understanding of abnormal behavior is to achieve a synergism between theory formulation and empirical classification methods.

A Comparison of Cluster Analysis Techniques Withing a Sequential Validation Framework.

This study involved the comparison of 23 different methods of cluster analysis in a four-stage sequential validation design. These stages included derivation, replication, external validation, and cross-validation. Data were obtained from 750 alcohol abusers on a number of socio-behavioral variables. Results demonstrated that the solution given by Wards method of cluster analysis was particularly powerful in comparison to solutions yielded by other techniques.

Differentiating the Contribution of Elevation, Scatter and Shape in Profile Similarity

The aims of this paper are: (1) to present a similarity generating function composed of elevation, scatter and shape parameters; (2) to describe linear models for integrating these parameters either for euclidean distance or vector-product association indices; and (3) to suggest a computational strategy based upon the Eckart-Young (1936) theorem that has certain advantages for minimizing the effects of measurement error in estimating profile similarity. Given these developments, the investigator may differentiate the independent contribution of each parameter to more global indices of resemblance. A brief example from the classification of psychopathology is discussed.

Empirically Derived Classifications of Alcohol-Related Problems

In recent years a number of investigators have utilized empirical techniques to develop classification schemes for alcohol related problems. This chapter examines this literature, reviewing both dimensional and categorical models that have been proposed. Despite much consistency in the methodologies that have been used, relatively few consistent substantive findings have emerged from these studies. However, increasing sophistication in the use of these empirical techniques over the past decade has produced some interesting results. One such study which yielded a hybrid model for alcoholism incorporating advantages of both categorical and dimensional approaches, is discussed in some detail. Finally, promising directions for future research are outlined with references to the implications of such taxonomy for alcoholism treatment and course.

Modal Profile Analysis: a Computer Program for Classification Research

A FORTRAN program is described for the identification and cross-validation of relatively homogeneous subgroups within a dimensional or ordination space. This classification approach is designed for use with several samples in a replication design. The program consists of three stages: (1) identification of preliminary types within each of m samples, (2) replication of these types across samples, and (3) generalizability of the typology with new samples. The computations are based on a sequential application of the singular value decomposition. An important feature of this program is that it allows one to differentiate the independent contribution of elevation, scatter, and shape parameters in defining profile similarity.

Univocal Varimax: an Orthogonal Factor Rotation Program for Optimal Simple Structure

Univocal varimax is an orthogonal factor rotation strategy aimed at improving upon the simple structure qualities of a preliminary varimax solution. This is accomplished by targetting for patterned rotation the highest element in each row of the varimax factor loading matrix. This tends to yield a solution in which each variable in the final rotated matrix maximally loads on only one factor. Univocal varimax is particularly relevant to research problems in which each rotated factor should be marked by a relatively tight cluster of variables. A FORTRAN IV program is described for the efficient analysis of large input factor loading matrices.

Dimensional analysis of rank-order and categorical data

Skinner (1979) has described a generalized principal components model for classification research that assumes interval or quasi-interval data. First, a parsimonious set of typal dimensions is sought through a multiple replication design, and then relatively homogeneous subgroups are identified within this low dimensional space. The purpose of this paper is to describe preliminary transformations whereby the model may be extended to situations where the data are of either categorical or rank-order metric.

Explore: a Computer Program for Analyzing Two Or More Data Sets1

EXPLORE is a flexible computer program for analyzing multiple data sets. For the case of two data sets, the output includes: (1) the complete multiple correlation analysis of each variable (component) in one set regressed on all variates of the other set, (2) a canonical correlation solution, and (3) a final transformation to enhance substantive interpretation of the data. The investigator has the option of focusing upon the original variables, or of selecting a reduced rank solution where these variables are summarized by a small number of principal components. A range of transformations (rotations) is available that allows EXPLORE to be employed in both an exploratory or hypothesis testing mode.