Use the data in file agpop.dat for this exercise. Let yi be the value of acres92 for unit i and xi be the value of acres87 for unit i. Draw an SRS of size 400. Now generate missing data from your sample by generating a standard uniform random variable Ui for each observation and deleting the observation if 16Ui ≥ ln (xi). (Sample SAS code for this is on the website.)

i. If you ignore the missing data, do you expect the mean of y to be too large or too small?

ii. Calculate the mean from the data set with missing values. Does a 95% CI, computed ignoring the non response, contain the true mean from the population?