Body Mass Index, Diabetes, and Neighborhood Characteristics: Assessing the Role of Selection by Studying Movers and Stayers

Ken R. Smith, University of Utah
Heidi Hanson, Huntsman Cancer Institute
Barbara Brown, University of Utah
Cathleen Zick, University of Utah
Jessie X. Fan, University of Utah

In this paper we test two hypotheses: 1. Walkable neighborhoods are associated with lower levels of individual BMI and lower prevalence ralotes of diabetes. 2. Individuals moving from lower (higher) BMI neighborhoods to higher (lower) BMI neighborhoods will have higher (lower) BMI values and will be more likely to be diabetic *before they make the move* relative to stayers. We find that women in the leanest neighborhoods who will move to the heaviest neighborhoods but have not yet made the move are among the heaviest residents in their neighborhood. Women in the heaviest neighborhoods who will move to the leanest neighborhoods but have not yet made the move are among the leanest members of their neighborhoods. Some evidence of non-random selection based on diabetes is also found. This suggests that there is non-random selection into different types of neighborhoods by BMI and presence of diabetes.