** user=**** 
** password=**** 
** package=stata
** project=lis

#delimit;

* Ireland LIS 1994 : Example CAMSIS matching ;
* (here to person file, applies equally to household file occ vars);
* Just change the year on ie**p file to work for 1995 or 96 LIS;

use $ie94p;

* Current occ unit (isco88 two digit in four digits)=pocc; 
* valid pocc in range > 1000 and < 9400. 
* Employment status differences not applicable to Ireland;

gen isco88=-999;
replace isco88=pocc if (pocc > 0 & pocc < 9400);



* Key indexing variables is isco88;

tabulate psex if (isco88 > 0) , 
        summarize(isco88) mean standard freq ; 

sort isco88;
save cstemp, replace;
clear;
infile isco88 mcam fcam siops isei 
      using $mydata\camsis\csieis94.dat;
summarize;
sort isco88;
merge isco88 using cstemp;
gen occinfo=_merge;
tab occinfo; 
drop if (occinfo==1);
* (this removes any CAMSIS isco88 values not represented in LIS sample);
tab occinfo psex; 


* Current values are within gender; 
* also can make a cross-gender CAMSIS variable : ;

gen cgentcs=-999;
replace cgentcs=mcam if (psex==1);
replace cgentcs=fcam if (psex==2);

summarize isco88 cgentcs  ;
summarize isco88 cgentcs  if (pocc > 0 & pocc < 9400) ;

* ALL valid occ cases get camsis scores sucessfully;

***************************************;


* Quick assessment: compare values by an education recode;
gen educ=peduc; 
recode educ 1=1 2 3=2 4 5 6=3 *=-9;
label define edu  1 "Primary or less" 2 "Intermediate" 
    3 "Late secondary or higher" ; 
label values educ edu;
label variable educ "education categories";
tab educ  ;
sort psex;
by psex:
  summarize educ 
      cgentcs  if (pocc > 0 & pocc < 9400); 
by psex: 
 oneway cgentcs educ  if (pocc > 0 & pocc < 9400 & educ > 0), tabulate;



*****************************************;