Modeling HTL of industrial workers using multiple regression and path analytic techniques

This study compared path analytic with multiple regression analyses of hearing threshold levels (HTLs) on 258 adult textile workers evenly divided into low- and high-noise exposure groups. Demographic variables common in HTL studies were examined, with the addition of iris color, as well as selected two-way interactions. Variables of interest were similarly distributed in both groups. The results indicated that (1) different statistical procedures can lead to different conclusions even with the same HTL data for the same Ss; (2) conflicting conclusions may be artifacts of the analytic methodologies employed for data analysis; (3) a well-formulated theory under which path analytic techniques are employed may clarify somewhat the way a variable affects HTL values through its correlational connections with other antecedent variables included in the theoretical model; (4) multicollinearity among independent variables on which HTL is regressed usually presents a problem in unraveling exactly how each variable influences noise-induced hearing loss; and (5) because of the contradictory nature of its direct and indirect effects on HTL, iris color provides little, if any, explanatory assistance for modeling HTL.