Assistant Professor Elizabeth Bowen, PhD alum Carol Scott, PhD student Andrew Irish and Research Professor Thomas Nochajski publish "Psychometric properties of the Assessment of Recovery Capital (ARC) instrument in a diverse low-income sample"
Published December 31, 2019
Congratulations to Assistant Professor Elizabeth Bowen, PhD graduate Carol Scott, PhD student Andrew Irish and Research Professor Thomas Nochajski on the publication of their article "Psychometric properties of the Assessment of Recovery Capital (ARC) instrument in a diverse low-income sample."
Bowen, E. A., Scott, C. F., Irish, A., & Nochajski, T. H. (2019). Psychometric properties of the Assessment of Recovery Capital (ARC) instrument in a diverse low-income sample. Substance Use and Misuse.
Abstract
Background: Recovery capital is a theoretical construct elucidating the resources that support recovery from addiction. The 50-item Assessment of Recovery Capital (ARC) instrument and related brief-format versions are the predominant measures of this construct. However, some of the ARC’s psychometric properties are not well-established, particularly in racially and economically diverse populations. Objectives: We aimed to determine if the ARC is a valid and reliable measure of recovery capital in a diverse sample. Methods: Paper-and-pencil survey data were collected between March 2017 and May 2018 from a low-income, racially diverse sample of adults in recovery (N = 273). Participants were recruited from nontreatment community settings throughout a mid-sized northeastern U.S. city. They completed the ARC and sociodemographic questions. To determine the ARC’s reliability and factor structure, we used item-level analyses and Cronbach’s alpha, followed by confirmatory and exploratory factor analyses. Results: Several items performed poorly, having means close to response extremes and problematically small variances. Cronbach’s alpha for the full measure was α = .92; however, alphas for the majority of subscales were below .70. The a priori 10-factor model solution failed, preventing interpretation of the confirmatory factor analysis results. Exploratory factor analysis revealed that although the 10-factor model marginally fit the data, items did not load together as proposed. Not once did all five subscale items load highly on the same factor. Conclusions/Importance: The ARC has substantial weaknesses in its theoretical alignment, item performance, and psychometric properties with diverse populations. We recommend the development of a new multidimensional, theory-aligned measure, following a rigorous measurement development protocol.
