As many demographic variables and geographic locations were significantly associates with participants' PRR of cancer and risk perception of death toward nuclear facilities, these variables were considered as potential confounding factors and controlled in the modeling.