The Development and Validation of a New Measure of Adolescent Purpose

Abstract

This goal of this study was to develop a quantitative measure that captures the dimensions of youth purpose (intention, engagement, and prosocial reasoning; Damon, Menon, and Bronk, 2003), focusing on students’ life goals and aspirations. We tested the construct validity of our new measure using bifactor exploratory structural equation modeling (B ESEM), assuming that the construct of youth purpose may be best measured as a G factor that allows items to freely cross-load in the model (Morin, Arens, & Marsh, 2016). The G factor model was better fitting than a traditional confirmatory factor analysis or ESEM. Additionally the G factor was correlated with aspects of students’ civic engagement and mastery components of academic achievement motivation. Suggestions for future research are discussed.

Keywords:

• Cognitive processes/development
• factor analysis
• goals
• motivation
• high school

Notes

Notes

1 Although the desired value of RMSEA should be .06 or below according to Hu and Bentler (1999), more recent research indicates a lack of empirical support for the use of .06 as a universal cutoff value to determine adequate model fit; RMSEA as an index of fit is rather dependent on model specifications, degrees of freedom, and sample size (Chen, Curran, Bollen, Kirby, & Paxton, 2008). Specifically, models with any misspecifications, large degrees of freedom, and/or a sample size smaller than 800 will likely yield a higher percentage of rejected models on RMSEA values alone. Therefore, multiple indices should be us used to reach decisions of model fit (Bentler, 2007, Bollen & Long, 1993, Tanaka, 1993).

2 The fit of all five subscales of civic engagement was poor and improved somewhat when we eliminated future concern from the model, which did not correlate with the other subscales (χ² (246) = 877.636, p < .001, CFI = .934, TLI = .926, RMSEA = .096 (90% CI: .089 - .103), WRMR = 1.682. Our fit was still limited likely due to the number of parameters in the model relative to sample size.