We used a land-use regression model to predict ambient
nitrogen dioxide (NO2) levels and sought the best socio-economic and population predictor
variables. We used a generalised least squares model that accounted for spatial correlation
in NO2 levels to examine the associations between the variables.