This paper shows the effects of spatially aggregating data in an analysis of fishing site choice among Atlantic and Gulf of Mexico longline fishers. Parameter estimates of expected utility, measures of risk, and estimates of welfare losses from area closures are presented. The estimated parameters and the measures of risk aversion indicate some spatial variation. However, the welfare measures from the area closure vary widely between a spatially aggregated model and a disaggregated model. The reason appears to arise from the economic behavior of fishers in New Jersey, where the expected utility model performs poorly.