Homework Problem 21, Due Friday April 9.
----------------------------------------

Hand in edited SAS code and portions of outputs (but 
NOT the data!) showing that you have accomplished the 
following tasks in SAS.

(i) Input and create and save a SAS dataset from the 
(compressed) ASCII dataset in the Data directory of the 
course web-page

http://www.math.umd.edu/~evs/s798c/Data/pbcdata.asc.gz

All of the data can be treated as numeric after the 
original header column, which you might delete before 
saving and ftp'ing the file OR which you might read with 
conditional INPUT statements as in the class handout. However, 
the TREATGP, CIRRH, and CHOL columns are binary indicators.

(ii) Find the average AGEVAR of individuals in the TREATED 
and UNTREATED groups (defined by TREATGP=1, 0). Do this first 
with a SAS PROC (eg MEANS or UNIVARIATE or TABULATE), and 
then, separately, by coding directly within a SAS data-step 
consisting of only two lines, corresponding to TREATGP=1,0, 
in which you number of data-records in each TREATGP along 
with the mean AGEVAR for that group are the only variables. 
Display the resulting file using PROC PRINT.

(iii) Obtain summary statistics (a few, the min and max 
and quartiles and mean and median and standard deviation) 
for LOGBILI and ALBUMIN. Which do you find more useful for 
this limited purpose, PROC MEANS or PROC UNIVARIATE ?

(iv) Print the minimum and maximum LOGBILI value within 
the 16 groups of patients defined by all combinations of 
the 4 binary variables DTH, TREATGP, CIRRH, and CCHOL.