Alcohol abuse and its relationship with poverty and unemployment, a structural equation modelling approach
Poverty and unemployment lead to psychological stress, which results in increased alcohol abuse. Numerous studies including both cross-sectional and longitudinal designs conducted in different countries have focused on finding a relationship between unemployment and alcohol consumption. Contradictory results have been found. The researchers have not agreed on whether unemployment increases, decreases, or does not alter drinking behaviour. Most previous researchers have used correlational methods predominantly based on male samples, and measured only alcohol consumption to denote alcohol abuse. No causal relationship between unemployment and alcohol abuse was established. The present study investigated the relationship of alcohol abuse with poverty and unemployment aiming to find a causal path between them. Other criterion measures of alcohol abuse (i.e., alcohol problems and alcohol dependence) in addition to alcohol consumption were used in a sample with equal representation of men and women. Poverty was used as an independent latent variable measured by income, number of family members, education level and employment status. The latent variables of alcohol use (measured by daily ethanol consumption averaged over a week, and drinking patterns of heavy drinking occasions and maximum drinks at a sitting), alcohol problems (measured by eight social and physical problem types), and alcohol dependence (measured by DIS-III-R, SADD and MAST scales) were considered to be dependent variables. Two models were tested using a random sample of longitudinal data (N = 1257) of community residents collected in 1989 (Wave 1) and 1991 (Wave 2) by the Winnipeg Health and Drinking Survey (WHDS), (Murray, Barnes, & Patton, 1994). Model 1 hypothesized a causal relationship between poverty, alcohol use, alcohol problems and alcohol dependence in the cross-sectional data. Model 2 hypothesized an increase in alcohol use with recent unemployment and a decrease with longer unemployment. This model also tested the longitudinal effects of the hypotheses of Model 1. The models and their variants were tested using structural equation modelling (SEM). Version 5.0 of the EQS program developed by Bentler (1995) was used for this purpose. In the secondary analysis, both of these models were te ted separately on gender groups (men and women) and on age groups (younger, middle age, and older) with both Wave 1 and Wave 2 data. The results indicate that in a cross-sectional sample, (1) increased poverty causes alcohol use and alcohol problems to increase, and (2) increased alcohol use causes increased alcohol problems and increased alcohol dependence. Results from longitudinal analysis suggest that (1) recent unemployment decreases alcohol use while longer unemployment increases it, (2) prolonged poverty increases alcohol use and, there is indirect support that prolonged poverty causes increased alcohol problems, (3) prolonged alcohol use causes increased poverty, increased alcohol problems and increased alcohol dependence. Results from the secondary analysis indicate that alcohol use, alcohol problems and alcohol dependence are more prevalent in men and in younger age group. A number of recommendations were made suggesting improvement in the model and need for further study.