Fig. 3

Summary of n = 1,000 simulations from ARIES data (two measurements, age 7 and 15 or 17) [15]. Rows show the distribution (median, 25th and 75th percentile, outliers) of effect size estimates derived from different models and time variables included in those models, respectively. The two columns differentiate between estimates of the interaction term as well as the fixed effect. Simulated effect sizes are marked in red (interaction = 0.1; fixed effect = 2.0). Time variables are chronological age (Age), years between measures (Years), number of measure (Timepoint, i.e., 1, 2), factorized measure (Timefactor, i.e., F07, F15). Models are linear mixed effect models (LME), generalized estimating equations (GEE), and regression on difference between two epigenetic age (EA) measures (Δ aging). All models contained Horvath clock derived EA as outcome [3]