Table 7.
OLS coefficient estimates for three models based on covariates that were selected by the spatial scale (SS) LARS/lasso model of BMI z-score: (1) when constraining all selected area-level covariates to enter at the CBG level; (2) when constraining all selected area-level covariates to enter at the CT level; and (3) when allowing all selected area-level covariates to enter at the model-selected spatial scale. The horizontal dashed line separates the individual-level variables and the area-level variables.
| Explanatory Variable | CBG | CT | SS |
|---|---|---|---|
| Visit Age | 0.080 | 0.080 | 0.081 |
| Black | 0.045 | 0.044 | 0.043 |
| Distance to Medical Center | 0.017 | 0.014 | 0.014 |
| Population Density | −0.033 | −0.033 | −0.033 t |
| % Hispanic White | 0.035 | 0.034 | 0.033 b |
| % Hispanic Black | 0.020 | 0.022 | 0.025 t |
| Median Household Income | −0.044 | −0.065 | −0.061 t |
| % Renter | −0.028 | −0.037 | −0.035 t |
| % Vacant | 0.017 | 0.014 | 0.015 t |
| Personal Crime Index | −0.018 | −0.007 | −0.013 b |
| Property Crime Index | −0.002 | −0.014 | −0.012 t |
| Park Density | 0.007 | 0.011 | 0.012 t |
| Exercise Equipment | −0.047 | −0.029 | −0.034 b |
Notes: Values marked with b denote area-level covariates selected at the CBG level, and values marked with t denote area-level covariates selected at the CT level.