Figure 1

Predicted effect of precipitation on county prevalence of autism in 2005 when different levels of drinking water were derived from surface sources (very low = 3.3%, low = 15% , median =53.5 %, and high = 79.8%) using our final model (unstandardized) and maintaining all other other variables in the model constant at their mean value. Expected percent autism ^0.5 = 0.47529 - 0.000927187 *Surface drinking Water - 0.00867663 *Avg. unemployment + 0.0000236374* annual precipitation −0.0000230421* HDD + 2.53316 * EPA Neurological Disease Risk+ 0.00733637 *suicide rate + 0.00000989236 * precipitation* drinking water −0.00069674* EPA Neurological Disease Risk * HDD. The reference state (Washington), was used to create the interaction plot and data were back transformed before they were graphed.