Deviation experiments. We first check the prediction of the conformity hypothesis that the helpfulness ratio of a review will vary inversely with the absolute value of the difference between the review’s star rating and the computed product-average star rating— we call this difference the review’s deviation.