Samuel H. Rikoon

Using Latent Profile Analysis to Identify Noncognitive Skill Profiles Among College Students

ABSTRACT Most research examining the relationship between noncognitive variables and academic performance has tended to focus on developing linear prediction models. However, a more holistic understanding of academic success can be achieved by examining the noncognitive profiles of students and their relationship with academic outcomes. In the present study, we used latent profile analysis to explore the noncognitive profiles of 5,120 college students. To better understand the nature of the profiles, the effect of covariates on class membership was examined. Results indicated that there were 6 latent profiles underlying the noncognitive skills of college students in our sample and that profile membership was associated with different collegiate grade point average levels. Our results have practical implications for the development of interventions intended to improve academic performance, which may be customized toward the specific needs of students presenting with diverse profiles.

Assessing Measurement Invariance in Multiple-Group Latent Profile Analysis

The study of measurement invariance in latent profile analysis (LPA) indicates whether the latent profiles differ across known subgroups (e.g., gender). The purpose of the present study was to examine the impact of noninvariance on the relative bias of LPA parameter estimates and on the ability of the likelihood ratio test (LRT) and information criteria statistics to reject the hypothesis of invariance. A Monte Carlo simulation study was conducted in which noninvariance was defined as known group differences in the indicator means in each profile. Results indicated that parameter estimates were biased in conditions with medium and large noninvariance. The LRT and AIC detected noninvariance in most conditions with small sample sizes, while the BIC and adjusted BIC needed larger sample sizes to detect noninvariance. Implications of the results are discussed along with recommendations for future research.

Bias, Type I Error Rates, and Statistical Power of a Latent Mediation Model in the Presence of Violations of Invariance:

When testing a statistical mediation model, it is assumed that factorial measurement invariance holds for the mediating construct across levels of the independent variable X. The consequences of failing to address the violations of measurement invariance in mediation models are largely unknown. The purpose of the present study was to systematically examine the impact of mediator noninvariance on the Type I error rates, statistical power, and relative bias in parameter estimates of the mediated effect in the single mediator model. The results of a large simulation study indicated that, in general, the mediated effect was robust to violations of invariance in loadings. In contrast, most conditions with violations of intercept invariance exhibited severely positively biased mediated effects, Type I error rates above acceptable levels, and statistical power larger than in the invariant conditions. The implications of these results are discussed and recommendations are offered.