A QUANTITATIVE TEST OF THE PREDICTIVE VALIDITY OF HEINRICH’S ACCIDENT PYRAMID

*Onwuka, C. | Department of Occupational Safety and Health, Columbia Southern University (CSU), Orange Beach, Alabama, USA.

Abstract

Herbert W. Heinrich, a pioneer researcher in the field of occupational safety, published a book on occupational accidents in 1931. In the publication, Heinrich presented an Accident Pyramid, which affirmed that one workplace fatality would occur after 300 no-injury accidents have been recorded in a workplace. Heinrich’s Accident Pyramid has since been adopted as a standard for predicting the likelihood that fatalities will occur. Moreover, Heinrich’s claim on occupational accidents was adopted as a standard because of his reputation as a pioneer of occupational safety research. This study seeks to test the statistical hypothesis put forward by Heinrich in reference to General Aviation (GA) accidents and does so with a random subsample of the National Transportation Safety Board (NTSB) data. Retrospective data published for public use by NTSB was collected from 2004 to 2013 for 15,119 GA accidents (with a mean fatal accident count of ≈ 282). The pyramid was initially recommended for use in predicting all forms of accidents, including GA accidents. Notwithstanding, the analysis of an independent variable x, the number of total accidents, and a dependent variable y, the number of fatalities, produced a p-value of approximately 0.821, which indicates no correlation between the variables. There was no statistically significant relationship found between the total number of accidents and fatal accidents, as asserted by Heinrich. This study, therefore, demonstrates that Heinrich’s Accident Pyramid is statistically invalid. Consequently, this means that a fatality may still occur even if the total number of accidents is reduced.