Student Research: Thomas Johnston
MS, , 1997
Faculty Advisor: Michael S Morgan
Demolition Worker Lead Exposure: Validation of Predictive Model Using Workplace Variables
This study validated a linear regression model describing lead exposure in the demolition industry. A previous lead study (Phase 1) of the demolition industry described lead exposure for many common demolition tasks by use of a multiple regression model. In this study 38 samples were obtained to validate the Phase 1 regression model. The 38 samples were limited to work categories and work sub-categories found in data used to form the Phase 1 regression model. Ambient lead values for the 38 samples divided into the work categories of demo & manual, cutting, and loadout were plotted against the expect lead value found from the Phase 1 linear regression model. The R-square value for the observed versus predicted lead values was 0.36. A similar analysis of tasks not involving “cutting” material improved the R-square value to 0.67. Residuals were analyzed in an attempt to explore the variability in the observed /predicted relationship. Of the variables used in the Phase 1 regression model, “fan” and “work” variables were identified as the most likely cause for the discrepancy between Phase II data and Phase 1 final predictions. The final evaluation of the Phase 1 regression model was to estimate 95% prediction interval for the mean predicted response has a range of 0.75 log-lead to 1.05 log-lead, and the 95% prediction interval for a single observation has a ranged from 0.0 log-lead to 1.8 log-lead.