Bodvarsson and Gibson (1999) address the problems they see with existing tests of the tipping–
service relationship by asking people how much they would tip under hypothetical conditions of
“satisfactory,” “very good,” and “poor” service. However, the respondents to this survey clearly
overstated the effects of service quality on tipping behaviors. In that study, regular restaurant patrons
from St. Cloud reported that they would tip 13.3% for satisfactory service and 19.4% for very good
service. Yet in an earlier study involving post dining interviews with restaurant patrons as they were
departing the restaurant, Bodvarsson and Gibson (1994) themselves found that people from the St.
Cloud area tipped 13.9% for service rated a 3 out of 5 and tipped only 13.8% for service rated a 5 out of
5. The later of these two studies is the most trustworthy, because previous research has clearly
demonstrated that people are often poor at predicting and explaining their own behavior (Myers, 1990).
What people think and say they will do in a given situation is often different from what they actually do
(Freedman, 1969). On the other hand, people can accurately report their own attitudes and can