Readers who doubt that restaurant service is as uniformly good as suggested by the data in
Table 1 may be tempted to attribute the positive ratings to response bias and, thus, to accept a version
of Bodvarsson and Gibson’s thesis. However, if there is a positive response bias in ratings of service
quality, then ceiling effects should result in the highest ratings of service being given to a set of service
encounters with greater variability in true service quality than is true of the sets of service encounters
receiving lower ratings. This heterogeneity in the variance of true service quality at different levels of
rated service should also affect the variability of tip size if service quality is strongly related to tip size.
However, an examination of Table 1 reveals no systematic differences in the variability of tip size at
different levels of rated service quality. This suggests either that there is no sizable positive response
bias in service ratings or that there is no sizable positive relationship between tip size and true service
quality