When the covariate of an evaluation study using regression analysis is falhbly measured, the statistical test of program effectiveness is biased. In programs where the target population is "below average, " the bias tends to suppress the beneficial effects of the program, while raising them in...