We also examined the bivariate association between current age and drinking onset age for evidence consistent with a telescoping bias. We investigated the potential for such a bias to influence our analyses related to drinking onset and SLE by testing whether current age moderated the SLE × onset age interaction in the 2 cases where we had some evidence for interactions (dSLE in women and iSLE in men). We added two 3-way interaction terms (age × SLE × ONS17 and age × SLE × ONS14) and 3 additional 2-way interaction terms (age × SLE, age × ONS14, and age × ONS17) to the models in Table 3. For dSLE in women, there was little evidence of any telescoping bias. The increase in R2 associated with the 5 additional interaction terms was <1.2%, and all the 3-way interaction terms had p > 0.55. For iSLE in men, the age × ONS14 × dSLE term was statistically significant (p = 0.054), but the regression weight was in the direction opposite to what would be predicted for a telescoping effect: older (rather than
