D) Sample selection

1) Suppose you randomly drop 10% of a sample. What effect do you expect this to have on the estimated coefficients and their standard errors?

2)Suppose you drop the 10% of the sample that have the lowest values of a particular X-variable. What effect do you expect this to have on the estimated coefficients? Can you think of a case in which the effect on the coefficients might be large?

3)Suppose you drop the 10% of the sample with the lowest values of Y? Does this bias the coefficients? Explain.