Assessing occupational exposure via the one-way random effects model with balanced data

A one-way random effects model is considered for the log-transformed shift-long personal exposure measurements, where the random effect in the model represents an effect due to the worker. Following a previous approach, we address a hypothesis-testingproblem involving the proportion of workers for whom the mean exposure exceeds the occupational exposure limit. A confidence interval is constructed for the relevant parameter of interest, following the idea of a previously presented generalized confidence interval. The confidence bound is used for the purpose of testing hypotheses, and the performance of the test is numerically in vestigated. It turns out that the test exhibits satisfactory performance regardless of the sample size, in particular, for small samples. A similar procedure is then employed for testing hypotheses concerning the overall mean exposure. The results are illustrated using examples.