Confirmatory Factor Analysis of an Abbreviated Abstinence Self-Efficacy Measure: The Drug and Alchohol Abstinence Self-Efficacy Scale (DAASES)

Background & Purpose: Abstinence self-efficacy (ASE) is a powerful predictor of substance use outcomes. The Drug and Alcohol Abstinence Self-Efficacy Scale (DAASES) is a 20-item measure of confidence to abstain from alcohol or drugs across four relapse risk sub-scales: negative affect; social interactions; physical discomfort; and withdrawal. Examining the structure and psychometric properties of abbreviated forms of the DAASES could be useful for assessment and research—reducing administration time, measurement error and missing data. The purpose of the present study is to evaluate the factor structure and to assess the internal consistency and criterion validity of the 20-item DAASES, as well as two abbreviated forms (a 12-tem and a 4-item) among women receiving substance abuse treatment.

Methods: This analysis included 377 participants in a NIDA-funded longitudinal study examining personal support networks among substance using women. The DAASES is a Likert-type scale with response options ranging from 1 = Not at all confident thru 5 = Extremely confident, assessing participants’ confidence to abstain from drinking or using drugs within relapse scenarios, such as “If you were being offered drugs or a drink in a social situation,” or “If you had the urge to try using just once to see what happens.” Based on factor loadings among the sub-scale indicators and previous research, a 12-item and 4-item scale were examined, along with the 20-item DAASES. Confirmatory Factor Analysis (CFA) was performed for each version, testing the goodness-of-fit of both the multidimensional model comprised of the four sub-scales, and a unidimensional construct of ASE. Fit indices followed commonly accepted guidelines: Chi square (χ2); Tucker Lewis Index (TLI) and Comparative Fit Index (CFI) >.09; and a Root Mean Squared Error of Approximation (RMSEA) <.06. Internal consistency for all three forms expected an α >.80. Relationships between substance use outcomes at 6-months post-treatment intake with the multiple DAASES measures were examined to ascertain criterion validity.

Results: Over 60% of the sample was African American, had a mean age of 36.5 (SD=10.3), and had never been married (65%). Approximately 73% of the women were reliant on government assistance, with 40% reporting less than a high school education. A majority of women indicated receiving previous treatment for substance abuse (73%). CFA supported a unidimensional construct of ASE for each version of the DAASES, with the 12-item and 4-item forms demonstrating best fit: χ2(54)=327.37, p<.001, TLI=.92, CFI=.91, RMSEA=.04; and χ2(2)=9.35, p=.009, TLI=.97, CFI=.99, RMSEA < .01, respectively. Scores for the three measures were highly correlated, >.90. Reliability for the 20-item and 12-item instruments reported an α>.95, and α=.88 for the 4-item form. All versions demonstrated comparable correlations with substance use outcomes (p<.05).

Implications: Results support the use of the 12-item or 4-item versions of the DAASE for brief assessment in clinical settings. Implications of this study suggest that substance use treatment programs for women focus on developing ASE within all four relapse domains in order to promote enduring recovery.