The latent structure of alcohol dependence in female federal prisoners

Abstract

Results obtained from a structured interview of substance abuse diagnoses were subjected to taxometric analysis in a group of 459 low and minimum security female federal prisoners applying for admission to a comprehensive drug treatment program. Drawing indicators from a DSM-IV diagnosis of alcohol dependence (tolerance/withdrawal, loss of control, negative social/psychological consequences) the authors conducted a taxometric analysis using the following procedures: mean above minus below a cut (MAMBAC), maximum eigenvalue (MAXEIG), and latent mode factor analysis (L-Mode). Results were generally consistent with taxonic (categorical) latent structure for a DSM-IV diagnosis of alcohol dependence. The implications and limitations of this study are examined and recommendations for future research are offered.

• Alcohol dependence
• females
• prison inmates
• taxometric

Notes

Notes

1. The small range of scores for several of the indicators occasionally generated too little variance to produce a taxometric solution and so a small amount of random error (M = 0, SD = 0.000001) was added to each score so that the scores were less likely to obscure one another and the simulated curves could be calculated.

2. A MAXEIG analysis with 50 windows produced an averaged curve with three different peaks and an indeterminate CCFI (0.430–0.478). Indicators of limited range (4 values in the present study) can produce “bumps” in which low points are created by windows with little or no variation when analyzed with 50 windows (Ruscio et al. 2006). Under such circumstances it is normally advisable to reduce the number of windows to determine whether a better taxon-dimension contrast can be achieved (Ruscio & Marcus 2007). Reducing the number of windows in five window increments and stopping when the taxonic-dimensional contrast was greatest (i.e., CCFI deviation from 0.50) and the individual curves still retained their anticipated shape (peaks for taxonic, straight or bowed lines for dimensional) resulted in a solution with 15 windows.

3. Attempts were made to include the alcohol abuse items (Q4a, Q4b, Q6b, Q6d, Q6e) in the present analyses as had been done in the Walters (in press) investigation. However, the nuisance covariance was too large and so the analyses were confined to the DSM-IV criteria for alcohol dependence.