Ryan Fehr and Michele Gelfand at the University of Maryland have drawn on research in other disciplines, including sociology and law, to explore the idea that apologies come in three forms and that their impact varies according to the character of the victim.
The three apology types or components are: compensation (e.g. I'm sorry I broke your window, I'll pay to have it repaired); empathy (e.g. I'm sorry I slept with your best friend, you must feel like you can't trust either of us ever again); and acknowledgement of violated rules/norms (e.g. I'm sorry I advised the CIA how to torture people, I've broken our profession's pledge to do no harm).
Fehr and Gelfand's hypothesis was that the effectiveness of these different styles of apology depends on how the aggrieved person sees themselves (known as ‘self-construal' in the psychological jargon). To test this, the researchers measured the way that 175 undergrad students see themselves and then had them rate different forms of apology. In a follow-up study, 171 more undergrads reported how they see themselves and then they rated their forgiveness of a fictional student who offered different forms of apology after accidentally wiping her friend's laptop hard-drive.
The researchers found that a focus on compensation was most appreciated by people who are more individualistic (e.g. those who agree with statements like ‘I have a strong need to know how I stand in comparison to my classmates or coworkers'); that empathy-based apologies are judged more effective by people who see themselves in terms of their relations with others (e.g. they agree with statements like ‘Caring deeply about another person such as a close friend is very important to me'); and finally, that the rule violation kind of apology was deemed most effective by people who see themselves as part of a larger group or collective (e.g. they agree with ‘I feel great pride when my team or work group does well' and similar statements). These patterns held regardless of the severity of the misdemeanour, as tested by using different versions of the disk-wipe scenario in which either an hour's or several weeks' worth of data were lost.