An assumption of LPA, within-class normality, may not be met in our data. We examined the univariate distributions of class indicators within each class based on most likely class memberships, and they largely deviated from normality. Applying mixture models to true non-normally distributed indicators may result in the identification of spurious classes without meaningful interpretation (Bauer and Curran 2003; Lubke and Neale 2006). In addition, if the true distribution of the Null class is the uniform distribution across the range of class indicators, LPA with conditional normality assumption would result in underestimation of the Null classes by forcing the identification of classes with high and low mean profiles. However, it should also be noted that statistically distinguishing between single-class non-normal and multi-class normal data may not be possible (Bauer and Curran 2003; Muthén 2003). To check whether relaxing the normality assumption results in a different number of classes, we tried to fit LPA models with skewed t-distribution. However, estimating this model was extremely slow, likely due to the large number of SNPs, and the model never converged. On the other
