Does coaching interview-takers actually increase the predictive validity of the interview? That's certainly what the results of a recent study by Maurer et al. seem to indicate.
In the April issue of the Journal of Organizational Behavior, the researchers describe a predictive study where 146 interviewees (public safety incumbents) were provided with coaching before their situational interview. Importantly, this was coaching designed around the content of the interview and helping candidates communicate during the interview--not generic strategies like "smile a lot."
Results? Predictive validity was higher in the coached sample than an uncoached sample. Why? Well, it makes sense that if candidates are better at addressing the question--both because they have more knowledge and because they're expressing themselves better--you're getting a better view of their true knowledge (i.e., true score) and less interference (i.e., error).
Implications? If you conduct interviews as part of your hiring process (and is there someone out there that doesn't?), strongly consider providing pre-interview coaching (although I might call it something else since coaching sounds a bit suspicious). It may take a bit of your time, but it will pay off in the long run by improving your ability to predict job performance AND candidates will be happier. Big win-win.
The other study to read in this issue is by Becton et al., who looked at performance during and reactions to two selection procedures among White and African-American test-takers. The two tests were a written job knowledge test and a situational interview. The candidates were competing for promotion to Sergeant positions in a police department. Results? Both groups felt the interviews were more job related than the written test. And although African-American candidates performed worse on the written test, they felt that overall both methods were more job related (compared to Whites). Why is this important? Because some have theorized that subgroup differences are related to differences in take taking motivation. This study suggests there's something else going on.