Take a sneak peek at the new NIST.gov and let us know what you think!
(Please note: some content may not be complete on the beta site.).

View the beta site
NIST logo

Publication Citation: Evaluating Potential Bias in Non-Randomly Reported Fire Incident Data

NIST Authors in Bold

Author(s): David T. Butry; Douglas S. Thomas;
Title: Evaluating Potential Bias in Non-Randomly Reported Fire Incident Data
Published: September 26, 2012
Abstract: This analysis is a part of an effort to develop statistics and uncertainty measures for characterizing, tracking, and better understanding the root causes of the total burden of fire in the United States. They will be used to develop performance metrics, enabling comparisons between the use of new fire mitigation technologies and their impact on the U.S. fire burden, with a particular focus on residential fires involving upholstered furniture. This portion of the analysis has the following objectives: (1) to develop a statistical approach for evaluating the ,representativeness‰ of fire incident data reported in the National Fire Incident Reporting System (NFIRS) to depict fire activity in non-reporting cities; (2) to test (statistically) for differences between reporting and non-reporting cities of those factors believed correlated with fire risk and NFIRS reporting status; and (3) to discuss how the findings could be used to weight NFIRS-based statistics to produce more accurate national statistics. Results show that factors believed correlated to fire risk occur at different rates between reporting and non-reporting cities. This suggests that detailed fire statistics derived from NFIRS data may not best represent the U.S. fire problem, as these factors are also correlated with NFIRS reporting status. However, a weighting scheme, based on propensity scores, may provide a mechanism to adjust NFIRS-based fire incident statistics to provide more accurate nationwide metrics.
Citation: Technical Note (NIST TN) - 1770
Keywords: NFIRS, fire risk, statistics, propensity score matching, sample selection bias
Research Areas: Microeconomic Analysis
PDF version: PDF Document Click here to retrieve PDF version of paper (3MB)