The reliability of multiple regression and an alternative method for extracting task-specific exposure estimates from...

Area of Science:

• Occupational hygiene
• Exposure assessment
• Statistical modeling

Background:

• Evaluating task-specific exposures from time-weighted average data is crucial for occupational health.
• Multiple regression analysis and P-screen are common methods for this assessment.
• Understanding the reliability of these methods is essential for accurate exposure determination.

Purpose of the Study:

• To compare the reliability of multiple regression analysis against the P-screen method for determining task-specific exposures.
• To evaluate method performance using simulated multi-task time-weighted average data.
• To assess the impact of using dichotomous task measures versus task durations.

Main Methods:

• Simulated multi-task time-weighted average data were generated from lognormal distributions with random task time-weights.
• Monte Carlo trials were used to test the performance of multiple regression and P-screen methods.
• Data sets were structured to meet P-screen requirements, including task absence in samples.

Main Results:

• Both multiple regression and P-screen demonstrated equal reliability when using dichotomous task measures (occurrence/nonoccurrence).
• Combining the two methods did not significantly enhance reliability.
• Multiple regression analysis proved more reliable than P-screen when task durations were used as regressors.

Conclusions:

• Multiple regression and P-screen methods are comparable when relying on dichotomous task presence/absence data.
• Task duration data enhances the reliability of multiple regression for exposure assessment.
• Multivariate analysis of multi-task samples remains an imperfect substitute for dedicated single-task sampling.