Using HLM, we estimated the rate of change in BMI for all participants who also had at least one personality assessment. In this sample, there were 14,531 valid assessments of height and weight, with an average of 7.31 assessments per participant (SD = 5.85, range = 1 to 32). Both the linear slope and the quadratic slope were significant, indicating non-linear changes in BMI across the adult lifespan (see Table 4). Specifically, BMI increases most during young and middle adulthood. The estimated trajectories of BMI, plotted separately for men and women, are shown in Figure 1. The intercept and slope were correlated .44 (p < .01), which indicates that those who weighed more around age 60 increased more on weight over time. Multi-level modeling analyses, such as HLM, control for the correlation between the intercept and the slope (Verbeke & Molenberghs, 2000).
