* Solution exercise 3;

* Part 1;
data wtsbp;
 infile '~/sasbiostat/2011/data/tomhs.data' termstr=crlf ;
 input
      @1    ptid $10
      @25   group 1.
      @27   age 2.
      @30   sex 1.
      @47   marital 1.
      @85   wtbl 5.
      @91   wt6 5.
      @115  sbpbl 3.
      @119  sbp6 3.
             ;
* Part 2;
 wtdif = wt6 - wtbl;
 sbpdif = sbp6 - sbpbl;
 if group = 6 then active=2; else active=1;
RUN;
PROC MEANS;
RUN;

* Part 3;
PROC FREQ data=wtsbp;
    TABLES sex;
RUN;

* 27% of patients were women;

* Part 4;
PROC MEANS data=wtsbp ;
 VAR wtbl wt6 sbpbl sbp6 wtdif sbpdif;
RUN;
* The average weight change was -11.51 ;

* Part 5;
PROC MEANS data=wtsbp ;
 CLASS active;
 VAR wtbl wt6 sbpbl sbp6 wtdif sbpdif;
RUN;

* The average SBP change in the active group was -15.49 ; 
* The average SBP change in the placebo group was -12.55 ;

PROC TTEST data=wtsbp;
 CLASS active;
 VAR sbpdif;
RUN;
* The change in sbp was not sign between active and placebo groups;

* Part 6;
PROC SGPLOT data=wtsbp;
 SCATTER X= wtdif y=sbpdif;
 TITLE 'Plot of Blood Pressure Versus Weight Change';
RUN;

PROC CORR data=wtsbp;
 VAR wtdif sbpdif;
RUN;

* The correlation is 0.24775 ;

ODS GRAPHICS ON;
PROC REG data=wtsbp;
 MODEL sbpdif = wtdif;
RUN;
* The regression equation is -11.31610 + -0.30496 * wtdif ;

* Part 7;
ODS GRAPHICS OFF;
PROC UNIVARIATE data=wtsbp;
 VAR wtdif;
RUN;

* The 25th, 50th and 75th percentiles of weight change are;
* -16.5, -8.4 and -4.5 respectively ;

* Part 8;
PROC FREQ data=wtsbp;
  TABLES marital*sex/ CHISQ;
RUN;

* The percentage of men currently married was 87.67;
* The percentage of women that are currently married was 77.78;
* The distribution of marital status did not significantly differ ;
* between men and women (p=0.4824);