# Problem 5 A study was done to address an important issue regarding whether there are racial differences in hypertension among children. We define hypertension as being above the 95th percentile for either systolic blood pressure (SBP) or diastolic blood pressure (DBP) among children of the same age, height, and sex. Another issue is that BMI, which may vary among ethnic groups, is positively related to hypertension. Hence, a logistic regression model was run of hypertension on ethnic group **and** BMI. There were three ethnic groups considered: Caucasian (reference category), African-American, and Hispanic. The results among 27,009 boys in the Pediatric Task Force Data are given in the following table. | Variable | Regression Coefficient | *se* | |--------------------:|-------------------------:|----------:| | Intercept | -4.277 | 0.090 | | African-American | 0.009 | 0.042 | | Hispanic | 0.186 | 0.060 | | BMI (kg/$m^2$) | 0.107 | 0.004 | ## Part b (5 points) Suppose, in the dataset, we know that the BMI of African Boys is significantly higher than that for Caucasian boys. Find the p-value for each covariate effect and determine whether BMI is a confounder of the association between ethnic group and hypertension? If so, is it a **positive** or **negative** confounder? ANSWER IN RSTUDIO