options ls=84 nocenter;

libname save '/bbkinghome/sschaner/Angrist Work/Web Papers/Angrist_2002/immig/1900/sasdata/';

/* table4.sas

   endogamy table

   8-24-00: younger cohort only
   
   6-29-00: changed dataset to data11b
   
   6-19-00: modified from rdm9c for pat

   6-19-00: add spouses age,nativity,ethnicity

   6-13-00: add (sex/age) regression-impute of occscore

   6-11-00: revised to drop Hebrews

   6-9-00: create data11b

   CREATE FAMILY OCCSCORES for 1910, 1920, 1940
   ORIGINAL ETHNICITY CODING, mother-or-both

   from rd1900m9.sas
   from rd1900m6.sas
   RESTORED (was 5-31-00 version

   save NON-SL in 1940 for family income recode;

   6-6: code FTOTINC for 1940
   6-2: correct age codings for 60, wider age groups 

   raw file is aged 18-59 SL whites 1910,1920,1940,1950,1960,1970-F2

   updated:  add wider age ranges for x-sec analyses

   notes: 1950 has no mtongue/mmtongue vars
          1960 has no mmtongue/has mtongue for FB only
	  1960 and 1970 have FBPL only for both parents FB
	  1950 poverty rates too high; data for SL/HH heads only 

   first stage in 1910, 1920, 1940  
   follow-up in 1940, 1960 */

/* notes on nativity coding:

if nativity>0 then do; fb=(nativity=5);
                       nw_bpf=(nativity=4);
                       nw_fmp=(2<=nativity<=4);
                       nw_np=(nativity=1);
                       end;

label fb='foreign born'
      nw_bpf='native white, both parents fb'
      nw_fmp='native white, foreign/mixed parents'
      nw_np='native white, native parents';
*/


data temp;
 set save.data11b;

 if ((female=1) and (18 le age le 33)) or ((female=0) and (20 le age le35));
 if marst=1 and slwt>0 and (91 le year le 92);
 
 if nativity=1 then code=-1;
 if spsnativ=1 then spscode=-1;
  
 if nativity=5 then code=code+100;


proc freq;
title 'marital status matching among ages 20-50 in 1910 and 1920';
 weight slwt;
 tables female*code*spscode;

run;