Predicting community response to road traffic noise

Several problems related to identifying the potential future impacts of road traffic noise on residential areas require for their solution the ability to predict subjective response to road traffic noise. The main difficulty in using existing regression equations relating subjective response and traffic noise for such predictions is that there has been no reported test of whether or not the data used meet the assumptions of the regression model. If the assumptions are not met, the replicability of the results and hence the reliability of the predictions, as measured by confidence limits or standard errors, cannot be established, because such inference rests on the statistical assumptions. Investigation of the data collected in a traffic noise impact study in southern Ontario indicates that such data meet the assumptions necessary for inference from regression analysis. Consequently, valid estimates of the reliability of predictive equations derived from regression analysis can be made using the standard errors of the regression parameters. This stronger inferential base also permits comparisons among different noise measures oramong different response measures. It appears that several noise measures (Leq, L10, Ldn) are all equally good predictors of subjective response. It also appears that different indicators of subjective response yield significantly different regression parameters.