First, restrictions in the variability of service ratings affect the size of the tipping– service
relationship only as measured in terms of variance explained. The absolute difference in mean tip size at
different levels of service quality provides a measure of effect size that is independent of variability in
service ratings. Examining the mean tip size when service is rated a 3 out of 5 with that when service is
rated a 5 out of 5 in Table 1 supports the idea that service quality has only modest effects on tipping.
The differences in these means range from -0.1% to 2.1% of bill size in the three studies with sufficient
sample sizes to make this comparison meaningful. This belies Bodvarsson and Gibson’s (1999) argument
that the weakness of the tipping–service relationship in the existing literature is primarily due to a lack
of variability in service ratings.