Multivariate Random-Effects Models Predicting the State-Level Obesity Prevalence (Percentage Points) Among 2- to 4-Year-Old WIC Participants, 2000–2014
| . | Model 1 . | Model 2 . |
|---|---|---|
| Intercept (95% CI) | 15.55*** (14.93 to 16.17) | 16.01*** (15.30 to 16.71) |
| Annual trenda, yrs (95% CI) | 0.26*** (0.22 to 0.31) | 0.23*** (0.17 to 0.29) |
| Change in annual trend after 2009 (95% CI) | −0.55*** (−0.66 to −0.43) | −0.57*** (−0.69 to −0.44) |
| Race and/or ethnicity, %b (95% CI) | ||
| American Indian | — | 0.07 (−0.003 to 0.14) |
| Asian American | — | 0.19* (0.01 to 0.37) |
| African American | — | 0.03 (−0.00 to 0.06) |
| Hispanic | — | 0.03 (−0.001 to 0.05) |
| Children in poverty (standardized)c, % (95% CI) | — | 0.07 (−0.25 to 0.39) |
| ICC | 0.78 | 0.75 |
| N | 294 | 294 |
| N groups | 49 | 49 |
| . | Model 1 . | Model 2 . |
|---|---|---|
| Intercept (95% CI) | 15.55*** (14.93 to 16.17) | 16.01*** (15.30 to 16.71) |
| Annual trenda, yrs (95% CI) | 0.26*** (0.22 to 0.31) | 0.23*** (0.17 to 0.29) |
| Change in annual trend after 2009 (95% CI) | −0.55*** (−0.66 to −0.43) | −0.57*** (−0.69 to −0.44) |
| Race and/or ethnicity, %b (95% CI) | ||
| American Indian | — | 0.07 (−0.003 to 0.14) |
| Asian American | — | 0.19* (0.01 to 0.37) |
| African American | — | 0.03 (−0.00 to 0.06) |
| Hispanic | — | 0.03 (−0.001 to 0.05) |
| Children in poverty (standardized)c, % (95% CI) | — | 0.07 (−0.25 to 0.39) |
| ICC | 0.78 | 0.75 |
| N | 294 | 294 |
| N groups | 49 | 49 |
Results are reported for models with state random effects; Durbin-Wu-Hausman tests fail to reject the null hypothesis of no significant differences in consistency between models with fixed and random effects. —, not applicable.
The annual trend is calculated with the year centered in 2009.
Demographic makeup is presented as the number of children with race reported in each category as a percentage of (for 2000 and 2004) all 1- to 4-y-old children enrolled in the WIC or (for 2008–2014) all 2- to 4-y-old WIC participants as a percentage of all children for whom a race in the 5 categories was reported. We further subtract the population-weighted mean demographic makeup from each of these variables so the intercept in model 2 can be interpreted as the mean overall obesity prevalence in 2009 for a state with a demographic makeup comparable to the demographic makeup of our population across all observations (1.14% American Indian, 3.24% Asian American and/or Pacific Islander, 19.95% African American, 44.16% Hispanic, and 31.5% white) and a child poverty rate equal to the average across state-year observations (14.84).
Standardized rate for children ages 0 to 17 y calculated with the anchored SPM available from the Center on Poverty and Social Policy at Columbia University for all US states.19
P < .05; *** P < .001