Fig. 3
A-H Hispanic/Latino Background Differences in Predicted Weight Trajectories and Weighted Frequencies by Study Site and across Age (A-D) and Time since Immigration (E-G) for Hispanic/Latino Adults from the Hispanic Community Health Study/Study of Latinos (HCHS/SOL) Baseline Examination (2008–2011). All weight trajectories and pie chart weighted frequencies for Central Americans are shown in gold, Cubans in dark green, Dominicans in red, Mexicans in light green, Puerto Ricans in blue, and South Americans in purple, provided that each study site surveyed ≥100 individuals of each background; therefore, some background-site combinations with <100 individuals were not be displayed on each trajectory panel and were instead combined with all other less frequent background-site combinations (shown in light gray in the pie charts). Mean weight trajectories by background group-study site combination (constructed to represent combinations of more than ≥100 individuals) an across age shown in A-D represent the weighted average age at examination, proportion male, nativity/age at immigration categories, proportion born before 1980, and proportion with digit preference for self-reports ending in 0s or 5s. Within Chicago, background differences did not appear after accounting for multiple comparisons and are shown in a condensed manner (shown in dark gray in B and F, Chi-square p-value=0.025). Mean weight trajectories shown in E-G represent the weighted average at examination, proportion male, average age at immigration, proportion immigrating before 1980, and proportion with digit preference for self-reports ending in 0s or 5s. Pre/post immigration weight trajectories varied by background within each site (Chi-square p-value<0.0001). The Chi-square test of difference in the pre/post immigration slopes was significant in all cases (p<0.0001). The tips of the colored triangles signify the point in the weight trajectory where the average body mass index rises above 30 kg/m2, given the weighted mean height of that particular subgroup. Examples of how to calculate population-level weight change, or the effect of demographic or sociocultural factors, based on the final model coefficients are provided as part of Supplemental Tables 3–4