The disagreement is about what a tip is for, and the research suggests only one side is describing how tipping actually works. Michael Lynn's 2001 review at Cornell pooled 24 correlations from 14 separate studies of restaurant tipping and service quality. Tips did rise with perceived service quality, but the relationship was weak enough, in his words, to raise doubts about using tips to motivate servers, measure server performance, or identify dissatisfied customers. In other words the tip is largely fixed in advance by custom and by the size of the bill; the quality of the evening moves it only slightly. That leaves two coherent positions. If a tip is payment for service rendered, withholding it after bad service is the one moment the instrument does its job, and about half of Americans say that is acceptable. If a tip is in practice a near-automatic supplement to a low posted wage, then withholding it is docking someone's pay over a slow kitchen they do not control, which is roughly what the other third are objecting to. Both sides are reading the same transaction; they disagree on whether it is a rating or a wage.