The mantra for performance based interviewing was the following: The best predictor of future performance is past performance. Therefore, we designed interview questions that allowed candidates to describe how they had performed when facing tasks and challenges similar to those of the positions for which they were applying.
However, there is an even better predictor of future performance on a job than past performance in other jobs and life situations: Past performance on the same job. Christine Robinson, aka iMinister, well makes that point in her comment below on my last post. Her thinking is affirmed by Malcolm Gladwell in his latest book, Outliers: The Story of Success, in which he criticizes many evaluation techniques as being poor predictors of future performance. He calls this "the mismatch problem," and encourages the use of on-the-job evaluation.
This is a justice issue. One of Gladwell's examples was preference for minorities at the University of Michigan law school. Twenty years after the preferences were granted, minority attorneys who had been granted preferences were performing as well as non-minority candidates who had "higher" qualifications.
In "Minorities Will Strive to Cross a Higher Bar," David L. Kirp, professor of public policy at UC Berkeley, cites a recent study that shows that undergraduate minority students at elite private universities who receive preferential treatment perform better. People do respond to rising expectations. Kirp shows the fallacies in claims that this preferencial treatment creates another kind of mismatch problem.
How does all this translate into recommendations for examining for ministers?
- Use the best measures available and recognize their limitations.
- Take iMinister's advice and put more emphasis on on-the-job performance rather than academics.
- Provide preferential treatment when warranted and publicize your processes and rationales. This whole-person approach can consider class and other distinctions and expections that have impacted past performance.
No comments:
Post a Comment