First, we estimated an unconditional growth model that predicted log-transformed ERO measures from baseline through the most recent assessment by age by incorporating individual participant’s age at each follow-up (Mplus option: time scores; Muthén and Muthén; https://www.statmodel.com/company.shtml). This model specifies latent variables for the random intercept, the random slope for time (rate of change in ERO value by age), and a constant or individual deviation from these mean values. This approach allowed us to simultaneously estimate the variance in ERO within and between individuals across time. The slope and residual variances were fixed to be equal across all available time points. Separate models were run for delta and theta EROs (total power) during the go and no-go conditions.
