Citation

Chi CF, Chen CL. Safety Sci. 2003; 41(8): 681-700.

Copyright

(Copyright © 2003, Elsevier Publishing)

DOI

unavailable

PMID

unavailable

Abstract

This research reanalyzed 1230 work-related single fatalities from the years 1989, 1990, and 1992 with a model free approach. Each single fatality was analyzed in terms of the victim's age, gender, industry, and the accident type. The standardized mortality ratio (SMR) was calculated to compare the SMR among various demographic groups and to cross validate the results obtained from our previous work using analysis of variance (ANOVA). The Chi-square Automatic Interaction Detector (CHAID) was used to divide the working population into segments based on whether the worker was a victim in any fatality. Predictor variables mainly focus on the demographic variables including the victim's age, gender and industry. But different from our previous finding [Safety Science 27 (1997) 1] of an insignificant gender effect, gender was found to be the second critical factor and it was nested under the most critical variable, type of industry. Age is the least critical variable and it is nested under the gender effect. Five different interactive patterns were identified correlating the age, gender, and industries of the victims. A comparison among ANOVA, SMR and CHAID proved that CHAID outperforms other methods for it not only reveals the complex and hierarchical interactive patterns but also prioritizes the importance of the predictor variables.