Latent trait testosterone among 18-24 year olds: Methodological considerations and risk associations

The study investigated the relationship between latent trait testosterone (LTT) and risk-taking among 126 youth (M age = 21.34 years; 56% female; 52% African American). Latent state-trait (LST) modeling isolates observed variance of samples via their correlations into (1) a latent trait testosterone (LTT) factor capturing individual differences, and (2) a component of state testosterone factor (LST) capturing state-specific situational or environmental influences and random error variances. Participants provided four laboratory (20 min apart) and four home (waking, 20-min post-waking, noon, evening) salivary samples (later assayed for testosterone). Participants reported risk-taking tendencies and behaviors via an Audio Computer Assisted Self-Interview. Behavioral risk was measured using the Balloon Analog Risk Task. Results revealed: (1) LTT model invariance (operated similarly) for females and males; (2) LTT accounted for 18-89% (home samples) and 61-95% (lab samples) of the variance in testosterone levels, and (3) LTT was associated with risk-seeking behaviors and the strength of this association was similar across males and females. LST Modeling has potential to advance our understanding of testosterone-behavior associations to new limits by estimating stable trait-like components of the variance in testosterone levels.