Bias in trials comparing paired continuous tests can cause researchers to choose the wrong screening modality