The higher correlation between SRE-based LR scores and future heavy drinking compared to alcohol challenge and future heavy drinking in Table 2 is worth comment. While this was not seen once a single offspring per family was chosen (where both LR measures correlated with T2 drinking at ∼.37), it is also noteworthy that across the full sample and more restricted group, the alcohol challenge explained less of the SRE’s predictive ability. There are at least three potential explanations for the better performance of the SRE. First, it is possible that the more global evaluation of LR through the retrospective measure of what one experienced early in the drinking career might actually better capture the predictive ability of a low LR to alcohol early in life. Second, it is possible that the SRE-based measure taps into additional life attributes not picked up by the challenge, such as personality characteristics, emotional reactivity, or accuracy of recollection, and these might themselves reflect a higher risk for heavier drinking. Third, in this study the LR from the SRE is meant to capture LR very
