Poisson Regression on the Association of Leading-Lagging Indicators and its Application in Health and Safety

This study was conducted to advance methodology of utilizing leading indicators in prediction of occupational health and safety losses. The qualitative association between leading and lagging indicators is not well documented in the literature. A method was proposed to link leading indicators to lagging indicators using statistical models (specifically, Poisson Regression Model) so that it can be used for forecasting the lagging indicators. The study utilized an extensive proprietary database of leading and lagging indicators collected during the 2015 to 2020 period from a large oil and gas firm in Southwest China. The results of the modelling simulations and analysis indicated model configurations with leading indicators containing descriptive information had about a 7% better accuracy of predicting the lagging indicator than models containing leading indicators with no additional information. The overall findings of the study suggested the selection of appropriate indicators aided by additional information will likely offer better accuracy and better predictive performance. Second, using properly selected leading indicators should result in a proper relationship between leading and lagging indicators; this relationship could be either positive or negative