Confirmatory Factor Analysis Results for a New Measure of Emerging Adults' Reasons for Substance Use

Macroeconomic conditions have resulted in an elongated adolescence and debates exist over when adolescence ends and adulthood begins.  Emerging adulthood, lasting roughly from ages 18-29, has been proposed as a new life stage (Arnett, 2000).  Mixed findings exist from studies examining whether the defining features of emerging adulthood predict substance use.  One major problem in this line of inquiry is that the scales used to measure of Arnett’s (2000) five dimensions of emerging adulthood contains subscales with low internal consistency.  Because of this, we developed a 31-item instrument called the Emerging Adult Reasons for Substance Use scale (EARS), with items mapping onto the five defining features in Arnett’s (2000) theory, including: identity exploration, Optimism, Feeling in-Between, Self-Focus, and Instability.    A sample item was “You use substances because you think substance use is a rite of passage.”  Our prior exploratory factor analysis of the EARS found support for three factors, with good discriminant validity between emerging and older adults.  Therefore, the purpose of this study was to complete a confirmatory factor analysis of the EARS, as well as identify whether the dimensions correlated with substance use among emerging adults ages 18-29.

We recruited a sample of emerging adults ages 18-29 (n=274) using Amazon’s Mechanical Turk (MTurk), an online community of workers.  This final analysis sample was after we excluded participants that did not answer validations questions correctly (i.e., please select response “5”; n=8) or those with unreasonably fast completion times (n=10).  MTurk data has been found to be reliable, accurately replicating results from in-person studies, when following such procedures.  Participants completed the EARS, as well as the AUDIT, a widely used and reliable measure of drinking behaviors.  We completed confirmatory factor analyses using MPlus to examine whether our hypothesized measurement model fit the data.  Additionally, we calculated correlations between the three hypothesized subscales of the EARS and the AUDIT.

Fit for the three factor model previously identified in EFA was adequate (RSMEA: .09, CFI: .81, TLI, .79), with three subscales that we labeled subjective invulnerability (7-items, α=.865), normative expectancies (10-items; α=.898), and developmental strain (10-items; α=.947).  Correlations (p<.05) with the AUDIT were .14, .32, and .32 for the subjective invulnerability, normative expectancies and developmental strain scales, respectively.      

Findings support the use of the EARS as a measure of developmental reasons for using substances in accordance with Arnett’s (2000) theory.  As past findings have produced mixed results, it is encouraging that we found a consistent pattern of associations between EARS subscales and risky drinking.  Future studies should determine if these developmental reasons for using substances predict actual treatment outcomes, which could help us tailor treatments to emerging adults.  Additionally, future research could determine if college status or age of onset moderates the associations between EARS subscales and risky alcohol use, which would have important implications for theory revision.