Item functioning of the alcohol dependence scale in a high-risk sample

Area of Science:

• Psychology
• Psychometrics
• Addiction Medicine

Background:

• The Alcohol Dependence Scale (ADS) is used to assess alcohol problems.
• High-risk populations, such as those mandated to domestic violence intervention programs, require accurate assessment tools.
• Understanding the psychometric properties of assessment tools in specific populations is crucial for effective intervention.

Purpose of the Study:

• To analyze the functioning of Alcohol Dependence Scale (ADS) items in a high-risk sample.
• To identify a reliable subset of ADS items for measuring alcohol problems.
• To examine potential gender-based differential item functioning (DIF) in the ADS.

Main Methods:

• Maximum likelihood common factors analysis of the ADS.
• Nonparametric kernel smoothing to generate item characteristic curves (ICC) and option characteristic curves (OCC).
• Identification of items reliably discriminating between minimal and excessive alcohol use.

Main Results:

• A unidimensional factor structure was primarily indicated for the ADS.
• Nine out of 25 ADS items reliably discriminated between levels of alcohol problems.
• Dichotomous scoring was deemed most appropriate for the selected items.
• No differential item functioning (DIF) by gender was detected.
• The nine-item abbreviated ADS correlated highly (r(s)=0.96) with the original scale and had superior distributional properties.

Conclusions:

• An empirically-derived nine-item ADS is an efficient and effective measure for assessing excessive or hazardous drinking in high-risk populations.
• The retained items focus on hazardous drinking patterns, suggesting their utility in samples with elevated alcohol use.
• The findings support a continuum of alcohol problem severity measurable with dichotomous items.