Statistical analysis
Of the 968 participants, we sequentially excluded 106 persons
with missing values on personal characteristics, 7 with no data on
weekly fast-food consumption, 5 lacking perception data and 1
missing geospatial data. After excluding another 11 participants with
neutral responses to the perceived fast-food availability question,
838 participants remained for analysis. Analyses were performed
using the PROC LOGISTIC statement in SAS v9.2. (Cary, NC). To
address our research question, unadjusted and adjusted multinomial
logistic regression models were constructed to test the
association between weekly fast-food consumption and both the
perceived availability of fast food and the GIS-based presence of fastfood
outlets within a person’s neighborhood.