Analyses of survey data may be confounded if people use response scales differently. We formalize and generalize the theory behind using “calibration questions,” such as vignettes, to correct for ...