Figure 1

Quantile regression models demonstrating effects of residency and demographic factors along the BMI distribution. Note: The dependent variable is continuous BMI. The vertical axis shows the associated covariates while the horizontal axis shows the continuous BMI quantiles. The dashed lines denote the OLS regression coefficients estimates for the covariate shown in each panel; the solid lines denote the quantile regression coefficient estimates; the shaded areas are the 95% confidence intervals for the quantile estimates. Take the first panel for example: the dashed line shows the OLS estimates of the BMI differences between urban and rural (it shows that on average urban population is relatively heavier than rural but it is not statistically significant); the solid lines shows the quantile regression estimates of the BMI differences between urban and rural across the distribution of the BMI (it shows that the only statistically significant urban/rural gradient exists among those who had relatively smaller BMI).