Confirmation that text quality is not the (only) explanatory factor. Since for a given pair of “plagiarized” reviews the text quality of the two copies should be essentially the same, a statistically significant difference between the helpfulness ratios of the members of such pairs is a strong indicator of the influence of a non-textual factor on the helpfulness evaluators.
An initial test of the data reveals that the mean difference in helpfulness ratio between “plagiarized” copies is very close to zero. rather, the algorithm makes mistakes because reviews are more complex in such situations and the classifier uses relatively shallow textual features