Does a dose-response relationship reduce sensitivity to hidden bias?