A company uses a series of selection measures to try and predict future job performance.

The company gives the test to all applicants and then checks new hires’ job performance levels 12 months later. Theoretically, there should be a correlation between the test scores and job performance. The problem is that the first cohort of employees show a marked over-estimation of job performance in selection compared to the results a year later. There is little correlation between the two sets of scores and this leads the company to assume that the selection process test is somehow flawed. What could be the problem? How could this problem be sorted out? What can the company do about the seemingly under-performing employees after one year of employment with the company?