A Pilot Study of Unknown Cause Fire Incidents in National Fire Incident Reporting System (NFIRS) Data

摘要

This pilot study explored methods of collecting data on the likely causes of fires recorded as unknown cause in the National Fire Incident Reporting System. Following a preliminary pilot test of several alternative reporting forms, data were collected in 37 fire departments in 8 states. Respondents were requested to indicate whether each of 10 causes was a strong possibility, possible, or impossible/very unlikely. In order to improve FEMA's national estimates of fire cause frequencies, several different computing algorithms were employed to assign probability values to each of the 10 causes. The principal conclusions of the study were: (1) Collecting additional information on unknown cause fires is feasible and fairly straight forward; (2) In residences, unknown cause fires are more likely to arise from nonequipment 'people-caused' fires than from equipment fires; and (3) Computation of proability assessments of cause are relatively insensitive to different assumptions.