Today’s Wall Street Journal (April 2-3,2011) has a wonderful article on the subject of work measurement (Ch.10). The question of how to deal with employees who know they are being observed for a time study arises in 
every class I teach. About 30 years ago, Prof. Paul Sackett, at the U. of Minnesota, began timing the speed of supermarket cashiers ringing up a few dozen items. He found (unsurprisingly) that some were faster than others. All knew they were being observed. Today, with electronic scanners recording the pace of workers over long periods of time, it is once again clear that productivity varies greatly.
Sackett assumed these separate measures (short-time observed and longer-period with no observer) would generate similar rankings. But they yielded very weak correlations. This led him to believe that there are highly motivated, focused people who perform well when they know they are being tested for “maximum performance”. On the other hand, “typical performance” (how well one does over the long haul) is influenced by a whole different set of character traits.
Think of the SAT,LSAT, or GMAT. They each take just a few hours (under maximum performance) and are supposed to give a reading of an individual’s talent. Though the SAT does a modest job of predicting grades for freshmen (R-squared=.12), it doesn’t do well at predicting post-college achievement. (Same for the LSAT).
“Even the NFL Combine is a big waste of time”, says the Journal. There is no consistent statistical relationship between how well players do at the Combine and how they succeed in the NFL. Grit, on the other hand, which reflects a person’s commitment to a long-term goal, does predict levels of achievement. (It seems like one needs grit to get a PhD) . But workers, football players, and students do not reveal their levels of grit when taking a brief test. What really matters, as we all know, is what happens after the test is over.