The quality-only straw-man hypothesis. Finally, there is a challenging methodological complication in all these styles of analysis: without specific evidence, one cannot dismiss out of hand the possibility that helpfulness is being evaluated purely based on the textual content of the reviews, and that these non-textual factors are simply correlates of textual
quality. In other words, it could be that people who write long reviews, people who assign particular star ratings in particular situations, and people from Massachusetts all simply write reviews that are textually more helpful — and that users performing helpfulness evaluations are simply reacting to the text in ways that are indirectly reflected in these other
features. Ruling out this hypothesis requires some means of controlling for the text of reviews while allowing other features to vary, a problem that we also address below.