Sample statistics are:

size: mean=800; SD=200
# of change requests: mean=15; SD=5
corr coeff = 0.6

1) 	What proportion of all programs had more than 20 change requests.

	Since you are interested in "all" programs, this does not require
	regression. The z score is:
			z = (20 - 15)/5 = 1
	From the standard normal table, 68.27% of the area under the
	curve is between -1 and 1. The desired proportion is therefore
	(100 - 68.27)/2 = 15.86%.

2)	Estimate the # of change requests for a program of 600 lines.

	Since you are estimating (that is predicting) # of change requests
	for a fixed size this is a regression problem and does not require
 	normal theory. Note that the run is -200 ans so:
			rise = -200(0.6(5/200)) = -3
	Hence 12 change requests can be expected.

3)	For programs of 800 lines what proportion had more than 15 change
	requests.

	Since (800, 15) is on the regression line then 15 must be the mean
	of programs of size 800 and so the proportion is 50%.

4)	For programs of 600 lines what proportion had more than 20 change
	requests.

	We already know that the mean of these programs is 12 and so we
	need to compute the SD.
			SD(y|x) = 5*sqrt(1-0.6^2) = 5*0.8 = 4
	Compute z:
			z = (20 - 12)/4 = 2
	From the standard normal table, 95.45% of the area under the
	curve is between -2 and 2. The desired proportion is therefore
	(100 - 95.45)/2 = 2.275%.