Eric Schwitzgebel recently conducted an interesting study in which he sent fake e-mails, ostensibly from undergraduates, to various professors and tracked their response rates. Separately, many of these same professors filled out a survey about their attitudes and typical behavior regarding e-mail. Their findings are quite striking.
Professors claimed to respond to e-mail religiously in the survey, as half said that they respond 100% of the time, and nine-tenths said that they respond at least 90% of the time. However, their actual response rates to phony e-mails was between only 54-59% of the time. Moreover, there was no correlation between professor's survey responses about whether it is moral to respond to undergraduate's emails and whether they actually do so. Both of these divergences between beliefs and behavior are classic symptoms of the self-serving bias.