Imputing the usual hours based on actual hours underestimates usual hours during reference weeks with holidays.
May I can use the person's location on a cdf of actual hours. If not, just drop those variables from the regression. Or dummy variables for cases where the person is in the top 20 percent for actual hours or bottom 20 percent.
Imputing the usual hours based on actual hours underestimates usual hours during reference weeks with holidays.
May I can use the person's location on a cdf of actual hours. If not, just drop those variables from the regression. Or dummy variables for cases where the person is in the top 20 percent for actual hours or bottom 20 percent.