Aims

To support the free and open dissemination of research findings and information on alcoholism and alcohol-related problems. To encourage open access to peer-reviewed articles free for all to view.

For full versions of posted research articles readers are encouraged to email requests for "electronic reprints" (text file, PDF files, FAX copies) to the corresponding or lead author, who is highlighted in the posting.

___________________________________________

Monday, June 28, 2010

A Psychometric Validation of the Short Alcohol Withdrawal Scale (SAWS)



The study aimed to evaluate psychometrically a Danish translation of the Short Alcohol Withdrawal Scale (SAWS) in an outpatient setting in patients with Alcohol Dependence (AD) and Alcohol Withdrawal Symptoms/Syndrome (AWS).

One hundred and twenty-two patients with AD and AWS filled in a 10-item rating scale to describe their symptoms with four graduations on five physical and five psychological items.

The question
of dimensionality of the construct was addressed in three different ways. First, a scree plot was constructed based on the polychoric correlations between items. Second, promax factor loadings were calculated for a two-factor model. These two steps were based on exploratory factor analysis. Third, specific violations such as local dependence and differential item functioning were investigated under the one-factor model in a confirmatory factor analysis.

The scree plot supported one or two dimensions while the promax rotations gave little support for a two-factor model. The confirmatory analysis also supported a one-factor model.

The decomposition of the polychoric correlation matrix into eigenvalues and vectors suggested that there was most likely one factor underlying the 10 items in the SAWS. This was confirmed by a confirmative factor analysis with only one component when specific model violations such as local dependence and differential item findings were investigated. The SAWS is easy to use.



Read Full Abstract

Request Reprint E-Mail: becker@dadlnet.dk


_______________________________________