Measurement Invariance and Application of an Alcohol-Related Consequence Scale for American Indian Adolescents

Abstract

American Indian (AI) adolescents experience disproportionate alcohol-related consequences. The present study evaluated the psychometric properties and application of the American Drug and Alcohol Survey (ADAS™) alcohol-related consequence scale for AI adolescents through a secondary analysis of a large population-based sample of adolescents living on or near AI reservations. We found support for the ADAS alcohol-related consequence scale as a one-factor model, invariant discretely across race, sex assigned at birth, and age, and with good internal consistency. Evidence for construct validity was found through significant positive correlations between frequency of past 12 months of drinking, frequency of past 12 months of intoxication, and lifetime alcohol-related consequences. AI adolescents were significantly more likely to report more alcohol-related consequences than their non-Hispanic White peers. Race significantly interacted with frequency of drinking in predicting alcohol-related consequences such that these associations were stronger for AI adolescents. However, race did not significantly interact with frequency of intoxication in predicting alcohol-related consequences. Results from this study demonstrate the utility of the ADAS alcohol-related consequence scale for use across demographic groups with little risk of measurement bias.

Cultural identity affiliation and alcohol use and related consequences among American Indian and White adolescents: A latent profile analysis

“…For the purposes of the present study, item responses were dichotomized such that 0 = have not experienced this problem and 1 = have ever experienced this problem to allow for the creation of a count variable reflecting how many alcohol‐related consequences participants have experienced (Range: 0–13). This scale has previously demonstrated good psychometric properties supporting its use as a single scale (Goldstein et al, 2022). Reliability in the present sample was good (Cronbach's α = 0.88).…”

“…Participants were administered the American Drug and Alcohol Survey (ADAS), a pencil‐and‐paper measure of child and adolescent substance use that has demonstrated good reliability and validity within AI populations (Goldstein et al, 2022; Oetting et al, 1985; Stanley et al, 2014). This measure was listed in the 2007 SAMHSA Measures and Instruments Resource Guide and has been widely used in large national studies to assess substance use rates and correlates.…”

“…Among American Indian (AI) adolescents in particular, rates of lifetime alcohol use are significantly higher than rates observed in national samples of non‐AI youth (i.e., 72.5% of AI 12th graders reporting lifetime alcohol use compared with 61.2% of 12th graders in Monitoring the Future; Swaim & Stanley, 2018). AI youth also tend to initiate alcohol use earlier (Spillane et al, 2015; Whitesell et al, 2012) and experience greater consequences related to their alcohol use compared with their White peers (Goldstein et al, 2022; King et al, 2014). However, other research suggests that AI communities simultaneously have higher rates of overall abstinence from alcohol compared with racial/ethnic groups (U.S. Department of Health and Human Services, 2010).…”

Assaultive trauma, alcohol use, and alcohol‐related consequences among American Indian adolescents

“…The ADAS was listed in the 2007 Substance Abuse and Mental Health Services’ Measures and Instruments Resource guide, and it is continually updated to reflect current substances used by adolescents. It has also demonstrated good reliability and validity within AI adolescent populations (Goldstein et al, under review; Oetting & Beauvais, 1990; Stanley et al, 2014).…”

“…Alcohol‐related consequences were assessed by nine different items that inquire whether drinking alcohol has ever caused different problems such as getting in trouble in school or causing a fight with others (Goldstein et al, under review). Four response options ranged from “no” to “10 or more times,” and based on the zero‐inflated nature of the variable, frequency variables were collapsed and coded as 0 = “no” and 1 = “yes,” such that a rating of 1 indicated a problem occurring one or more times.…”