* Marriage to Divorce;
* Identify those people who changed from married to divorced
	in NLS data (mature women cohort);
* Also record the NLS wave and the year where the change was
	recorded;

* Marriage status varibles: R0002400 R0133700 R0205100 R0288200 
	R0308300 R0367600 R0455600 R0491300 R0661400 R0666600 R0721700 
	R0869900 R0997700 R1290700 R1664710 R3507200 R4278200 
	R5447500 R6516200;
* Years of the survey: 1967 1971 1972 1974 1976 1977 1979 1981 1982 
	1984 1986 1987 1989 1992 1995 1997 1999 2001 2003;

libname y "y:\sas\arrays";
libname library (y);

*Initial look at the data;
proc freq data=y.nlswomen;
	tables R0002400 R0133700 R0205100 R0288200 R0308300 R0367600 R0455600 R0491300 
		R0661400 R0666600 R0721700 R0869900 R0997700 R1290700 R1664710 R3507200
		 R4278200 R5447500 R6516200;
	run;

* A problem with comparing these variables across years is that
	the categories change.  An easy solution is to recode 1 -> 2
	in all of these variables;

*One approach to recoding;
data new;
	set y.nlswomen;
	if R0002400 eq 2 then R0002400 =1;
	if R0133700 eq 2 then R0133700 =1;
	if R0205100 eq 2 then R0205100 =1;
	* etc etc etc;
	* This gives you the correct results,
		but it is tedious to type, bulky to
		debug, and there happens to be a
		better way;
run;

proc freq data=new;
	tables R0002400 R0133700 R0205100;
	run;

*Using an array to do repetitive recoding;
data new1;
	set y.nlswomen;
	array recodes{5} R0002400 R0133700 R0205100 R0288200 R0308300;  * define an array;
	do i=1 to 5;
		if recodes{i}=2 then recodes{i}=1;  * use the array;
		end;
	run;

proc freq data=new1;
	tables R0002400 R0133700 R0205100 R0288200 R0308300;
	run;

*Extend this to include more variables;
data new2;
	set y.nlswomen;
	array recodes{*} R0002400 R0133700 R0205100 R0288200 
	R0308300 R0367600 R0455600 R0491300 R0661400 R0666600 R0721700 
	R0869900 R0997700 R1290700 R1664710 R3507200 R4278200 
	R5447500 R6516200;
		*You don't need to count the variables;
	do i=1 to dim(recodes);  * Now you use dim() to bound the loop;
		if recodes{i}=2 then recodes{i}=1;
		end;
	drop i;		* You don't need to keep i, do you?;
	run;

*Using an array to create new variables for the recodes;
data new3;
	set y.nlswomen;
	array original{*} R0002400 R0133700 R0205100 R0288200 
	R0308300 R0367600 R0455600 R0491300 R0661400 R0666600 R0721700 
	R0869900 R0997700 R1290700 R1664710 R3507200 R4278200 
	R5447500 R6516200;
	array mstatus{19};
		* Create 19 new variables, mstatus1 - mstatus19;
	/*array mstatus{*} mstatus1 - mstatus19;*/ *same thing;
	do i=1 to dim(original);
		if original{i}=2 then mstatus{i}=1;
			else mstatus{i} = original{i};
		end;
	drop i;		
	run;

proc freq data=new3;
	tables mstatus1-mstatus19;
	run;