/*******************************************************************************************************
This program creates the following education variables for survey year 2021

variable name in      variable name
the program           in the NLS
------------------------------------------------------
ENCAT                CV_ENROLLSTAT
GRSURV               CV_HGC_EVER_EDT
DEGSURV              CV_HIGHEST_DEGREE_EVER_EDT
GRSURV_CVC           CVC_HGC_EVER
DEGSURV_CVC          CVC_HIGHEST_DEGREE_EVER

 *******************************************************************************************************/
options nocenter nolabel linesize=100 pagesize=40;

libname xin20;
libname xin19;

data one;
 set xin19.out_hgcR19;
 locatorr19=locator; locar19=loca; locfr19=locf; grsurvFr19=grsurvF;
 chk31r19=chk31; chk32r19=chk32; chk33r19=chk33;  chk5r19=chk5;
 chk11r19=chk11; chk12r19=chk12; chk13r19=chk13; chk14r19=chk14; chk22r19=chk22;
 clg_okr19=clg_ok; chk1encr19=chk1enc; chk1gedr19=chk1ged; chk2r19=chk2; chk3r19=chk3;
 hgcYr19=graug;
 keep pubid he2_en he3_en he4_en he5_en he6_en he7_en he8_en he9_en he10_en he11_en he12_en he13_en he14_en he15_en he16_en he17_en he18_en he19_en
             locatorr: locar: grsurvFr: locfr:
       he2_dg he3_dg he4_dg he5_dg he6_dg he7_dg he8_dg he9_dg he10_deg he11_deg he12_deg he13_deg he14_deg he15_deg he16_deg he17_deg he18_deg he19_deg
       chk31r: chk32r: chk33r: chk5r:
       chk11r: chk12r: chk13r: chk14r: chk22r: clg_okr: chk1encr: chk1gedr: chk2r: chk3r: hgcYr17 hgcYr18 hgcYr19;

data two;
 set xin20.out_date_dgR20;
 cvc_hgcr19=hgcr19; cvc_hdgr19=CVC_HIGHEST_DEGREE_EVER_XRND;
keep  pubid
      gdr1-gdr19 hsr1-hsr19 aar3-aar19 bar4-bar19 mar6-mar19 dcr10-dcr19 pfr9-pfr19
      CDTGD CDTHS CDTAA CDTBA CDTMA CDTDC CDTPF
      scoder1_1-scoder1_2
      scoder2_1-scoder2_6
      scoder3_1-scoder3_7
      scoder4_1-scoder4_5
      scoder5_1-scoder5_6
      scoder6_1-scoder6_5
      scoder7_1-scoder7_5
      scoder8_1-scoder8_5
      scoder9_1-scoder9_4
      scoder10_1-scoder10_5
      scoder11_1-scoder11_5
      scoder12_1-scoder12_8
      scoder13_1-scoder13_3
      scoder14_1-scoder14_5
      scoder15_1-scoder15_4
      scoder16_1-scoder16_4
      scoder17_1-scoder17_4
      scoder18_1-scoder18_6
      scoder19_1-scoder19_5
      scoder20_1-scoder20_3
      stopmr20: stopyr20:
      strtmr20: strtyr20:
      e11700: e11900:
      e13300: e13500:
      e27337r20_1-e27337r20_3
      e23450r20_1-e23450r20_3
      t7000r20_1-t7000r20_6
      intdr20 intmr20 intyr20
      lftscr20_1-lftscr20_3
      doicmR: e3112r: e3113r:
      hdgr1-hdgr19 hdgYr1-hdgYr19
      encatr1-encatr19 cvc_hgcr19 cvc_hdgr19;

 data three;
  set xin20.in_C_HGCr20;
 data four;
  set xin20.in_P_Rostr_HGCr20;
 data final;
  merge one two three four;
  by pubid;

********************************************** SECTION I **********************************************;

** Initialize and define variables;

ENROLL=-3;    * Indicates if R currently enrolled;
ENCAT=-3;     * Enrollment status variable (including educational attainment);
HSD=0;        * Indicates if R has a H. S. Diploma (check previous rounds);
GED=0;        * Indicates if R has a GED (check previous rounds);
hgsch=0;      * Indicates if R is currently attending grades 1-12 & no HSD or GED;
college=0;    * Indicates if R is in college and working towards a undergrad (=1) or grad degree (=2);
attsch=0;     * Indicates if R attended elementary, middle or high school;
flagcoll=0;   * Indicates if Rs currently attending grades 1-12, no HSD or GED but enrolled in college
                and working towards a grad or undergrad degree;
flaghs=0;     * Indicates if Rs who have a H.S. Diploma and HGC=12 but still report enrollment in school;
somecoll=0;   * Indicates if Rs have attended a college (in a degree program) in current or previous
                rounds;
prevcoll=0;   * Indicates if Rs have attended a college (in a degree program) in previous rounds. This
                pertains to those actually enrolled in a degree program in college only;
currcoll=0;   * Indicates if Rs have attended a college (in a degree program) in current round. This
                pertains to those enrolled in a college and reporting to be working towards a degree
                program. They may or may not be enrolled in a degree program.;
clg_ok=-4;    * Indicates if R being in college (as reported in this round) w/o GED/HSD is OK. That is if
                R is actually enrolled in a degree program. This pertains to currcoll and not prevcoll.;
last_enr=0;   * Identifies the last round when R was enrolled;
ungraded=0;   * Indicates Rs with no H.S. Diploma or GED have HGA=95 (ungraded);
same_mth=0;   * Indicates Rs who received a degree in the same month as the interview.;
both=0;       * Indicates that R reports receiving both GED and H. S. Diploma;
rosprb1=0;    * Indicates Rs with scode(i) in (4,5) but e27337(i)=-4;
rosprb2=0;    * Indicates Rs with scode(i)=-4 but e27337(i)>-4;
chk_comm=0;   * Indicates if R reports college in this round although R does not report receiving GED or
                H.S. Diploma, by this round. Check the comments in current round to see whether they
                were actually enrolled in a degree program in a college.;
chk_commP=0;  * Indicates if ENCAT in any prior rounds shows college although R does not report
                receiving GED or HS Diploma, by those rounds. Check the comments in current and earlier
                rounds for these cases to see whether they were actually enrolled in a degree program in
                a college.;
hgc_hga=0;    * Indicates if Rs reporting HGC greater than HGA;
hgc_hga_n=-4; * Indicates if Rs reporting HGC greater than HGA, after HGC and HGA were edited for those
                never enrolled at or since DLI such that they remained the same as those at DLI.;

 **************************************************************
 ** dlicm:     date of last interview                        **
 ** dli_rnd:   the last round that R was interviewed         **
 ** doicm:     date of current interview                     **
 ** enroll_Rx: whether enrolled in earlier rounds            **
 ** encatP:    ENCAT when last interviewed                   **
 ** hgaloop:   highest grade attended (from loops)           **
 ** hgcloop:   highest grade completed (from loops)          **
 ** hgcdt:     date of completion of hgcloop                 **
 **************************************************************;

MONYR=500;
YEAR=2021;
MONTH=8;
doicm_LR=doicmr19;

*** Define arrays to be used later in the program;

array scode  (*) scoder20_1-scoder20_3;
array e27337 (*) e27337r20_1-e27337r20_3;
array e23450 (*) e23450r20_1-e23450r20_3;

array strtm  (*) strtmr20_11-strtmr20_13 strtmr20_21-strtmr20_22;
array strty  (*) strtyr20_11-strtyr20_13 strtyr20_21-strtyr20_22;
array stopm  (*) stopmr20_11-stopmr20_13 stopmr20_21-stopmr20_22;
array stopy  (*) stopyr20_11-stopyr20_13 stopyr20_21-stopyr20_22;

array degre  (*) e27337r20_1-e27337r20_3 e27337r20_1-e27337r20_2;
array codes  (*) scoder20_1-scoder20_3    scoder20_1-scoder20_2 ;
array reasn  (*) lftscr20_1-lftscr20_3    lftscr20_1-lftscr20_2 ;

array stpcm  (*) stpcm11-stpcm13 stpcm21-stpcm22;
array strcm  (*) strcm11-strcm13 strcm21-strcm22;

array gratnd   (*)  e8416r20_11 e8416r20_21 dum;

array e6784 (*) e6784r20_1-e6784r20_3;
array e7142 (*) e7142r20_11 e7142r20_21 e7142r20_31;
array e7192 (*) e7192r20_1-e7192r20_3;
array e6938 (*) e6938r20_1 dum dum;
array e6943 (*) e6943r20_11 dum dum;

array doicmR   (*) doicmr1-doicmr19;
array encatR   (*) encatr1-encatr19;
array enrollR  (*) enrollr1-enrollr19;
array graugR   (*) hgcr1 hgcYr2-hgcYr19;
array grsurvR  (*) hgcr1-hgcr18 cvc_hgcr19;
array hgaR     (*) e3500r1 e2857r2-e2857r19;
array hgcR     (*) e5000r1 e3112r2-e3112r19;
array gdR      (*) gdr1-gdr19;
array hsR      (*) hsr1-hsr19;
array locaR    (*) dum locar2-locar19;
array chk31R   (*) chk31r1-chk31r19;
array chk32R   (*) chk32r1-chk32r19;
array chk1encR   (*) chk1encr1-chk1encr19;
array chk1gedR   (*) chk1gedr1-chk1gedr19;
array chk2R   (*) chk2r1-chk2r19;
array chk3R   (*) chk3r1-chk3r19;
array chk5R   (*) chk5r1-chk5r19;

array scoder  [19,8] scoder1_1-scoder1_2 dum dum dum dum dum dum
          scoder2_1-scoder2_6 dum dum
          scoder3_1-scoder3_7 dum
          scoder4_1-scoder4_5 dum dum dum
          scoder5_1-scoder5_6 dum dum
          scoder6_1-scoder6_5 dum dum dum
          scoder7_1-scoder7_5 dum dum dum
          scoder8_1-scoder8_5 dum dum dum
          scoder9_1-scoder9_4 dum dum dum dum
          scoder10_1-scoder10_5 dum dum dum
          scoder11_1-scoder11_5 dum dum dum
          scoder12_1-scoder12_8
          scoder13_1-scoder13_3 dum dum dum dum dum
          scoder14_1-scoder14_5 dum dum dum
          scoder15_1-scoder15_4 dum dum dum dum
          scoder16_1-scoder16_4 dum dum dum dum
          scoder17_1-scoder17_4 dum dum dum dum
          scoder18_1-scoder18_6 dum dum
          scoder19_1-scoder19_5 dum dum dum;

array degR (*) hdgr1-hdgr18 cvc_hdgr19;

*** initiate the cvc_ variables;
if e3112r20=-5 then do;
  grsurv=cvc_hgcr19;
  degsurv=cvc_hdgr19;
end;

*** Create intermediate variables to be used in the program later.;

** Checking int dates;
doicmr20=-5;
if intyr20 ne -5 then doicmr20=(intyr20-1980)*12+intmr20;
if doicmr20~=-5 then doicm=doicmr20;

* Whether received GED, ever;
if CDTGD>-4 then GED=1;
* Whether received HS Diploma, ever;
if CDTHS>-4 then HSD=1;
* Whether received both GED and HS Diploma, ever;
if HSD=1 & GED=1 then both=1;

* Compute dlicm and dli_rnd;
do j=1 to dim(doicmR);
   if doicmR(j)~=-5 then do;
      dlicm=doicmR(j);
      dli_rnd=j;
      do k=1 to 8;
         if scoder(j,k) in (4,5) then ever_24yrinst=1;
      end;
   end;
end;
hsrp=hsr19; gdrp=gdr19;
if HSD=0 & hsrP>-4 then hsrP=-4;
if GED=0 & gdrP>-4 then gdrP=-4;

* Enrollment status at each interview and ENCAT at last interview;
do i=1 to dim(doicmR);
   if encatR(i) in (8,9,10,11) then enrollR(i)=1;
   if encatR(i) in (1,2,3,4,5,6,7) then enrollR(i)=0;
   if -5<encatR(i)<0 then enrollR(i)=encatR(i);
   if encatR(i)~=-5 then encatP=encatR(i);
   if encatR(i)~=-5 then enrollP=enrollR(i);
   if enrollR(i)=1 then last_enr=i;
   if hgaR(i) not in (-4, -5) then hgaP=hgaR(i); *Dec 09, to consider the situation where 2857/3112=-4, which appears starting from R12;
end;

** Create variable indicating HGC by DLI;
do i=1 to dim(graugR);
   if graugR(i)>-5  then graugP=graugR(i);
   if locaR(i)>-5 then locaP=locaR(i);
end;
grsurvP=cvc_hgcr19;
grsurvP_no3061=grsurvP;
if 0<grsurvP<graugP<95 then graugP=grsurvP;
degsurvP=cvc_hdgr19;

** Start and stop dates of enrollment in continuous month scheme;
do i=1 to dim(degre);
   if stopm(i)>0 & stopy(i)>0 then stpcm(i)=((stopy(i)-1980)*12)+stopm(i);
   if -4<stopm(i)<0|-4<stopy(i)<0 then stpcm(i)=-3;
   if strtm(i)>0 & strty(i)>0 then strcm(i)=((strty(i)-1980)*12)+strtm(i);
   if -4<strtm(i)<0|-4<strty(i)<0 then strcm(i)=-3;
end;

** Data and quex problems w/ roster information;
rosprb1=0; rosprb2=0;
do i=1 to dim(scode);
   if scode(i) in (4,5,6) & e27337(i)=-4 then rosprb1=1;
   if scode(i)=-4 & e27337(i)>-4 then rosprb2=1;
   if scode(i) in (-2,-3) then INV1=1;
   if scode(i) in (4,5,6) & e27337(i) in (-2,-3) then INV2=1;
end;
do i=1 to dim(strcm);
   if strcm(i)=-3 then INV3=1;
   if stpcm(i)=-3 then INV3=2;
   if strcm(i)=-3 & stpcm(i)=-3 then INV3=3;
end;

* Last regular school enrolled in, its start and stop date of enrollmnet, type of school and degree
  working towards & school loop number;
STOP=-4;
do i=1 to dim(degre);
   if stpcm(i)>STOP then do;
      STOP=stpcm(i);
      STRT=strcm(i);
      sch=codes(i);
      deg=degre(i);
      rosnum=i;
      STOP_loc=1;
   end;
   if stpcm(i)=STOP & sch in (4,5,6) & codes(i)=sch & deg in (1,3,4,5,6,10) & degre(i)=deg then do;
      STOP=stpcm(i);
      STRT=strcm(i);
      sch=codes(i);
      deg=degre(i);
      rosnum=i;
      STOP_loc=2;
   end;
   if 0<stpcm(i)<STOP & 0<cdths<STOP & STOP=doicm & 0<STRT<strcm(i) & 0<cdths<=strcm(i)
   &  codes(i) in (4,5,6) & sch in (1,2,3)
   &  degre(i) in (1,3,4,5,6,10) & deg not in (1,3,4,5,6,10) then do;
      STOP=stpcm(i);
      STRT=strcm(i);
      sch=codes(i);
      deg=degre(i);
      rosnum=i;
      STOP_loc=3.1;
   end;
   if 0<stpcm(i)<STOP & 0<cdtgd<STOP & STOP=doicm & 0<STRT<strcm(i) & 0<cdtgd<=strcm(i)
   &  codes(i) in (4,5,6) & sch in (1,2,3)
   &  degre(i) in (1,3,4,5,6,10) & deg not in (1,3,4,5,6,10) then do;
      STOP=stpcm(i);
      STRT=strcm(i);
      sch=codes(i);
      deg=degre(i);
      rosnum=i;
      STOP_loc=3.2;
   end;
   if stpcm(i)=STOP & codes(i) in (5,6) & sch=4 & degre(i) in (1,3,4,5,6,10) & deg in (1,3,4,5,6,10) then do;
      STOP=stpcm(i);
      STRT=strcm(i);
      sch=codes(i);
      deg=degre(i);
      rosnum=i;
      STOP_loc=4;
   end;
   if stpcm(i)=STOP & codes(i)=6 & sch=5 & degre(i) in (1,3,4,5,6,10) & deg in (1,3,4,5,6,10) then do;
      STOP=stpcm(i);
      STRT=strcm(i);
      sch=codes(i);
      deg=degre(i);
      rosnum=i;
      STOP_loc=4;
   end;
   if stpcm(i)=STOP & codes(i) in (5,6) & sch in (5,6) & degre(i) in (4,5,6,10) & deg in (1,3) then do;
      STOP=stpcm(i);
      STRT=strcm(i);
      sch=codes(i);
      deg=degre(i);
      rosnum=i;
      STOP_loc=5;
   end;
   if stpcm(i)=STOP & codes(i) in (4,5,6) & sch in (1,2,3) & (HSD=1|GED=1) then do;
      STOP=stpcm(i);
      STRT=strcm(i);
      sch=codes(i);
      deg=degre(i);
      rosnum=i;
      STOP_loc=6;
   end;
   if 0<stpcm(i)<=STOP & codes(i)in (1,2,3) & sch in (1,2,3) & STOP=doicm & strcm(i)>=STRT>0 then do;
      STOP=stpcm(i);
      STRT=strcm(i);
      sch=codes(i);
      deg=degre(i);
      rosnum=i;
      STOP_loc=7;
   end;
   if stpcm(i)=STOP & STOP=doicm & 0<STRT<strcm(i)
   &  codes(i) in (4,5,6) & sch in (4,5,6)
   &  degre(i) in (1,3,4,5,6,10) & deg not in (1,3,4,5,6,10) then do;
      STRT=strcm(i);
      sch=codes(i);
      deg=degre(i);
      rosnum=i;
      STOP_loc=8;
   end;
end;
do i=1 to dim(degre);
   if stpcm(i)=STOP & codes(i) in (1,2,3) & sch in (4,5,6) & HSD=0 & GED=0 then do;
      STOP=stpcm(i);
      STRT=strcm(i);
      sch=codes(i);
      deg=degre(i);
      rosnum=i;
      STOP_loc=9;
   end;
end;
if STOP>0 then diff=doicm-STOP;
do i=1 to dim(degre);
   if rosnum=i then do;
      if STOP=doicm & stpcm(i)=doicm & reasn(i)>-4 then diff=10000;
   end;
end;
do i=1 to dim(degre);
   if strcm(i)>=-3 & stpcm(i)>=-3 then lastloopnum=i;
end;
* Whether the last school attended was High School or College, and if the latter then for what degree?;
if sch in (-1,-2,-3) then lastsch=-3;
if sch in (1,2,3) then lastsch=1;
if sch in (4,5,6) and deg in (1,3,4,5,6,10) then lastsch=2;
if sch in (4,5,6) and deg in (2,7,8,999,11,12) then lastsch=2.1;
if sch in (4,5,6) and deg in (9, 13) then lastsch=2.2;
if sch in (4,5,6) and deg in (-1,-2,-3) then lastsch=2.3;
if sch in (4,5,6) and deg=-4 then lastsch=2.4;

* - When R attended college (in a degree program) in a previous round;
do i=1 to dim(encatR);
   if encatR(i) in (4,5,6,7,9,10,11) & degR(i)>=1 then prevcoll=1;
   if encatR(i) in (4,5,6,7,9,10,11) & degR(i)<1 then do;
      prevcoll=1;
      chk_commP=1;
      if hsr19>-4|gdr19>-4 then chk_commP=2;
   end;
end;
do i=1 to dim(encatR);
   if (e2806r20=-4|e2806r20>=13) & encatR(i) in (4,5,6,7,9,10,11) & (hsR(i)>-4|gdR(i)>-4) then prevcoll=1;
   if (e2806r20=-4|e2806r20>=13) & encatR(i) in (4,5,6,7,9,10,11) & hsR(i)=-4 & gdR(i)=-4 then do;
      prevcoll=1;
      chk_commP=1;
      if 0<CDTHS<=doicmR(i) then chk_commP=2;
      if 0<CDTGD<=doicmR(i) then chk_commP=2;
   end;
end;

* - When R attended college (in a degree program) in this round;
do i=1 to dim(scode);
   if scode(i) in (4,5,6) & e27337(i) in (1,3,4,5,6,10) then currcoll=1;
   if scode(i) in (4,5,6) & e27337(i) not in (2 9 11 13) then currcoll_1=1;
end;
if hsd=0 & ged=0 & currcoll=1             then chk_comm=1;
if hsd=0 & ged=0 & currcoll=1 & clg_ok=-4 then clg_ok=0;
if (hsd=1|ged=1) & currcoll=1             then clg_ok=1;
if prevcoll=1|currcoll=1                  then somecoll=1;

* Grade information from loops;
hgaloop=e8416r20_11;
do i=2 to dim(gratnd);
   if gratnd(i)>hgaloop then do;
      hgaloop=gratnd(i);
   end;
end;
if (1<=e2857r20<=12) & HSD=0 & GED=0 then hgsch=1;
do i=1 to dim(scode);
   if scode(i) in (1,2,3) then attsch=1;
end;
do i=1 to dim(e27337);
   if e27337(i) in (1,3) then college=1;
   if e27337(i) in (4,5,6,10) then college=2;
   if e27337(i) in (-1,-2,-3) then college=-3;
end;

if 0<=e2857r20<e3112r20<95 then hgc_hga=1;
if (0<=e2857r20<12) & e3112r20=12 then hgc_hga=2;
if 0<=e2857r20<=12<e3112r20<95 & hsd=0 & ged=0 then hgc_hga=3.1;
if 0<=e2857r20<=12<e3112r20<95 & (hsd=1|ged=1) then hgc_hga=3.2;
if 12<e2857r20<e3112r20<95 then hgc_hga=4;
********************************************************************************************************
***                       (I) CREATED VARIABLE - ENROLLMENT STATUS (ENCAT)                           ***
********************************************************************************************************;

*** [IA] Determining whether the respondent is enrolled or not;
    if diff=0 then ENROLL=1;
    if diff>0 then ENROLL=0;
    if -3<=STOP<=-1 then ENROLL=0;
    if e3877A1r20=0 then ENROLL=0;
       do i=1 to dim(scode);
          if scode(i) in (4,5,6) then chk_enr2=0;
       end;
       if HSD=1 & ENROLL=1 & chk_enr2 ne 0 then do;
          chk_enr2=1;
          ENROLL=0;
       end;

*** ENRollx;
    ** Computing ENROLL by another method - ENROLLx;
    if e3877A1r20=0 then enrollx=0;
    if e4795r20=1 & e4951r20=0 then enrollx=1;
    if e4795r20=0 then do;
       if e1605r20=-4 & e4951r20=0 then enrollx=0;
       if e1605r20=1 then do;
          if e1615r20_1=1 & e4951r20=0 then enrollx=1;
          if e1615r20_1=0 & e4951r20=0 then enrollx=0;
       end;
       if e1605r20=0 and e4951r20=0 then enrollx=0;
    end;
    if (e4951r20=1 or e3877A1r20=1) then do;
       do i=1 to dim(e6784);
          if e6784(i)=1 & e7192(i)=0 then enrollx=1;
       end;
       do i=1 to dim(e6784);
         if e6938(i)=0 & e7192(i)=0 then enrollx=0;
       end;
       do i=1 to dim(e6784);
          if e6784(i)=0 & e7192(i)=0 then do;
             if e6943(i)=1 then enrollx=1;
          end;
       end;
       do i=1 to dim(e6784);
          if e7142(i)=0 & e7192(i)=0 then enrollx=0;
       end;
    end;

*** change "enroll" after going through the cases where "enroll" disagrees with "enrollx";
  oldenroll=enroll;
  if oldenroll=0 & enrollx=1 & diff=10000 & rosnum>=4 then enroll=1;

*** (IB) Determining the sub-categories of not enrolled;
    if ENROLL=0 then do;
       if (GED<=0) & (HSD<=0) then ENCAT=1;
       if GED=1 then ENCAT=2;
       if HSD=1 then ENCAT=3;
       if prevcoll=1 then ENCAT=4;
     if currcoll=1 & ((13<=e2857r20<=20)|e3112r20>=12|HSD=1|GED=1|somecoll=1) then ENCAT=4;
       if CDTAA>-4 then ENCAT=5;
       if CDTBA>-4 then ENCAT=6;
       if CDTMA>-4|CDTDC>-4|CDTPF>-4 then ENCAT=7;
    end;

*** (IC) Determining the sub-categories of enrolled;
    if ENROLL=1 then do;
       if lastsch=1 then do;
          if CDTHS>0 & e3112r20=12 then flaghs=1;
          if (1<=e2857r20<=12) & HSD<=0 & GED<=0 then ENCAT=8;
          if e2857r20=13 & HSD=0 & GED=0 then ENCAT=8;
          if e2857r20=95 & HSD=0 & GED=0 then do;
             ENCAT=8;
             ungraded=1;
          end;
       end;
       if lastsch=2 then do;
          if sch=4 & deg in (1,3,4,5,6,10) & ((13<=e2857r20<95)|e3112r20>=12|GED=1|HSD=1|clg_ok=1)
          then ENCAT=9;
          if sch=5 & deg in (1,3,4,5,6,10) & ((13<=e2857r20<95)|e3112r20>=12|GED=1|HSD=1|clg_ok=1)
          then ENCAT=10;
          if sch in (5,6) & deg in (4,5,6,10)     & ((16<e2857r20<95)|e3112r20>=16) then ENCAT=11;
          if sch in (5,6) & deg in (4,5,6,10)     & CDTBA>-4 then ENCAT=11;
          if sch=6 and ENCAT ne 11 then gradsch=1;
      if (1<=e2857r20<=12) & HSD=0 & GED=0 & somecoll~=1 then do;
             ENCAT=8;
             flagcoll=1;
          end;
       end;
       if lastsch>2 then do;
          if GED=0 & HSD=0 & somecoll=0 then ENCAT=8;
          if currcoll=0 & prevcoll=1 then do;
             ENCAT=4;
             if CDTAA>-4 then ENCAT=5;
             if CDTBA>-4 then ENCAT=6;
             if CDTMA>-4|CDTDC>-4|CDTPF>-4 then ENCAT=7;
      end;
          if currcoll=1 & (HSD=1|GED=1|clg_ok=1) then do;
             ENCAT=4;
             if CDTAA>-4 then ENCAT=5;
             if CDTBA>-4 then ENCAT=6;
            if CDTMA>-4|CDTDC>-4|CDTPF>-4 then ENCAT=7;
          end;
       end;
       if lastsch=-3 then do;
          if (HSD=1|GED=1|clg_ok=1) & currcoll=1 then ENCAT=4;
      if prevcoll=1 then ENCAT=4;
          if CDTAA>-4               & somecoll=1 then ENCAT=5;
          if CDTBA>-4               & somecoll=1 then ENCAT=6;
          if (CDTMA>-4|CDTDC>-4|CDTPF>-4) & somecoll=1 then encat=7;
       end;
       if GED=1 & HSD=0 & somecoll=0 then ENCAT=2;
       if HSD=1 & somecoll=0 then ENCAT=3;
    end;
    if CDTHS>0 & doicm=CDTHS & ENCAT in (-3,1,2,8) then do;
       ENCAT=3;
       same_mth=1;
    end;
    if CDTGD>0 & doicm=CDTGD & ENCAT in (-3,1,8) then do;
       ENCAT=2;
       same_mth=1;
    end;
    if CDTAA>0 & doicm=CDTAA & ENCAT in (-3,1,2,8,3,4,9) then do;
       ENCAT=5;
       same_mth=1;
    end;
    if CDTBA>0 & doicm=CDTBA & ENCAT in (-3,1,2,8,3,4,5,10) then do;
       ENCAT=6;
       same_mth=1;
    end;
    if ((CDTMA>0 & doicm=CDTMA)|(CDTDC>0 & doicm=CDTDC)|(CDTPF>0 & doicm=CDTPF)) & ENCAT~=7 then do;
       ENCAT=7;
       same_mth=1;
    end;
    if ENROLL=1|(ENROLL=0 & e3877A1r20~=0) then do;
       if lastsch=-3 & ENCAT<0 then do;
          ENCAT=-3;
          reason=1;
       end;
       if INV2=1 then do;
          flagcoll=-3;
          college=-3;
          if enroll=0 & somecoll=0 & lastsch=2.3 then do;
             ENCAT=-3;
             reason=2.2;
          end;
       end;
    if (HSD=1|GED=1) & (13<=e2857r20<95) & (12<=e3112r20<95) then do;
          tmpvar1=ENCAT;
          tmpvar2=reason;
          ENCAT=-3;
          reason=3;
          do i=1 to dim(scode);
             if scode(i) in (-1,-2,-3,4,5,6) | prevcoll=1 then do;
                ENCAT=tmpvar1;
                reason=tmpvar2;
             end;
          end;
       end;
       if (13<=e2857r20<=20) & GED=0 & HSD=0 & clg_ok~=1 & currcoll=1 & prevcoll~=1 then do;
          ENCAT=-3;
          reason=5;
       end;
       if rosprb1=1 & ENCAT=-3 & reason=. then reason=6;
    end;
********************************************** SECTION II **********************************************;
   if ENCAT in (1,2,3,4,5,6,7) & ENROLL~=0 then do;
      ENROLL=0;
      enr_chg=2;
   end;
   if ENCAT in (8,9,10,11) & ENROLL~=1 then do;
      ENROLL=1;
      enr_chg=3;
   end;

subsitm=0;
subhgam=0;
subprev=0;
setto12=0;
dateitm=0;
disagre=0;
datedeg=0;

********************************************************************************************************
***            [II] CREATED VARIABLE - HIGHEST GRADE COMPLETED AS OF THE SURVEY DATE (GRSURV)        ***
********************************************************************************************************;
       GRSURV=GRSURVP; LOCATOR=1;
       GRAUG=GRAUGP; LOCA=1;
       if ENCAT in (4,5,6,7,9,10,11) & e3112r20 in (1,2,3,4,5,6) then do;
          inclg_lowHGC=1;
          e3112r20_imp=e3112r20+12;
       end;

***** For those who are asked of 3112;
  if (e3112r20>12 & (currcoll_1=1 or prevcoll=1)) or (e3112r20=12 & cdths ne hsr19) or (0<=e3112r20<12 & attsch=1) or (-4<e3112r20<0 & e3877A1r20~=0)  then do;
       grsurv=e3112r20; locator=2;
       if e3112r20<=12|e3112r20=95 then do;  GRAUG=e3112r20; loca=2; end;
       if 0<=GRSURVP<e2857r20<e3112r20 then do;
          GRSURV=grsurvP;
          LOCATOR=4.13;
          LOCA=4.13;
       end;
       if e2857r20>e3112r20+1 & e3112r20>GRSURVP>=0 then gr_ncomp1=3;
       if e2857r20>e3112r20+1 & 0<=e3112r20<GRSURVP then gr_ncomp1=4;

       if 0<=e3112r20<=grsurvP<95 then do;
          GRSURV=grsurvP;
          LOCATOR=4.3;
          LOCA=4.3;
       end;

       if ENCAT in (4,5,6,7,9,10,11) & 0<=e3112r20<12 then do;
          if e3112r20_imp>12 then GRSURV=e3112r20_imp;
          else inclg_lt12=1;
          LOCATOR=5.2;
       end;

       if currcoll=1 & 0<=grsurvP<95 & 0<=e3112r20<95 then inclgCI_grchg=e3112r20-grsurvP;
       if currcoll=0 & prevcoll=1 & encatP in (4,5,6,7,9,10,11) & 0<=grsurvP<95 & 0<=e3112r20<95
       then inclgDLI_grchg=e3112r20-grsurvP;

    ** check on disagre;
       if 0<grsurvP<95 & 0<GRSURV<grsurvP then disagre=1;
       if disagre=1 & (hsrP>-4|gdrP>-4) & GRSURV<12 & grsurvP>=12 then do;
          disagre=0.5;
          subsprev=1;
          GRSURV=grsurvP;
          LOCATOR=6.10;
       end;
       if disagre=1 then do;
        if 1<=dli_rnd<=15 then rnd_dif= YEAR-(1996 + dli_rnd)+1;
      else if dli_rnd>15 then rnd_diff= YEAR-(2011 + (dli_rnd -15)*2)+1;
          sitm_pr=e3112r20-grsurvP;
          if ENCAT in (1,8,-3) then do;
             if ENCAT=-3 then do;
                GRSURV=-3;
                if e3112r20<=12|e3112r19=95 then GRAUG=-3;
                LOCATOR=6.11;
                LOCA=6.11;
             end;
             if 0<=e3112r20<=12 & (0<=sitm_pr<=rnd_dif|grsurvP=95) then do;
                subsitm=1;
                GRSURV=e3112r20;
                LOCATOR=6.12;
                GRAUG=e3112r20;
                LOCA=6.12;
             end;
          end;
          if 12<=grsurvP<95 & ENCAT in (4,5,6,7,9,10,11) then do;
             GRSURV=grsurvP;
             subprev=1;
             LOCATOR=6.21;
           if currcoll=1 then do;
             subsitm=1;
             if 0<=sitm_pr<=rnd_dif then do;
                GRSURV=e3112r20;
                LOCATOR=6.22;
             end;
             if e3112r20_imp>grsurvP then do;
                GRSURV=e3112r20_imp;
                LOCATOR=6.23;
             end;
           end;
          end;
       end;
       if 0<grsurvP<95 & 0<GRSURV<grsurvP then do;
          disagre=3;
       end;

    ** check again;
       if (hsrP>-4|gdrP>-4) & GRSURV~=grsurvP & 0<GRSURV<12 then do;
          if attsch~=1 & grsurvP=grsurvP_no3061 then do;
             GRSURV=grsurvP;
             LOCATOR=9.1;
             LOCA=9.1;
          end;
          if attsch=1|grsurvP~=grsurvP_no3061 then do;
             LOCATOR=9.2;
          end;
       end;

    ** GRAUG;
       if GRAUG=-3 & GRSURV>0 then do;
          if 0<STOP<MONYR & sch in (1,2,3) then do;
             GRAUG=GRSURV;
             LOCA=10.4;
          end;
       end;
       if GRSURV=grsurvP & (0<=grsurvP<=12|grsurvP=95) then do;
          GRAUG=grsurvP;
          LOCA=12.1;
       end;
       if GRAUG=grsurvP & grsurvP=12 & (cdths>MONYR>0) then do;
             GRAUG=-3;
             LOCA=12.4;
       end;
       if (hsrP>-4|gdrP>-4) & 0<=grsurvP<=12 & GRAUG<grsurvP then do;
             subprev=1;
             GRAUG=grsurvP;
             LOCA=13.11;
       end;
     if GRAUG<graugP then do;
          if 0<=graugP<=12 then do;
             subprev=1;
             GRAUG=graugP;
             LOCA=13.12;
          end;
          if grsurvP<0 & graugP<0 then do;
             GRAUG=-3;
             LOCA=13.13;
          end;
       end;

    ** If R reports a HGC>12 but R is not in a college-degree program we assign the HGC in highschool
       as the HGC received.;
       if ENCAT in (1,2,3,8,-3) & (12<GRSURV<95|(12<e3112r20<95 & GRSURV=-3)) then do;
             whyover12=1;
             if (12<GRSURV<95 & 0<grsurvP<=12)|(GRSURV<grsurvP &  0<grsurvP<=12) then do;
                GRSURV=grsurvP;
                GRAUG=grsurvP;
                LOCATOR=13.42;
                LOCA=13.42;
             end;
             if (12<GRSURV<95 & 0<grsurvP_no3061<=12)|(GRSURV<grsurvP_no3061 & 0<grsurvP_no3061<=12) then do;
                GRSURV=grsurvP_no3061;
                GRAUG=grsurvP_no3061;
                LOCATOR=13.5;
                LOCA=13.5;
             end;
             if (12<GRSURV<95 & 0<e3112r20<=12)|(GRSURV<e3112r20 &  0<e3112r20<=12) then do;
                subsitm=1;
                GRSURV=e3112r20;
                GRAUG=-3;
                LOCATOR=13.6;
                LOCA=13.61;
                if GRSURV=grsurvP then do;
                   GRAUG=grsurvP;
                   LOCA=13.62;
                end;
                if GRSURV=grsurvP_no3061 then do;
                   GRAUG=grsurvP_no3061;
                   LOCA=13.63;
                end;
             end;
       end;
       if 12<GRSURV<95 & ENCAT in (1,2,3,8,-3) then LOCATOR=13.7;
       if 12<GRAUG<95 & ENCAT in (1,2,3,8,-3) then LOCA=13.7;

    ** more on GRAUG;
       if grsurvP=12 & GRSURV>=12 & GRAUG~=12 then do;
          GRAUG=12;
          LOCA=16;
       end;
       if grsurvP<=12 & GRSURV>=12 & attsch=0 & (GRAUG=.|GRAUG=-3) & ENCAT in (4,5,6,7,9,10,11) then do;
          GRAUG=grsurvP;
          LOCA=17;
       end;
       if 12<grsurvP<95 then do;
          GRAUG=graugP;
          LOCA=18;
       end;
       if graugP=12 then do;
          GRAUG=12;
          LOCA=19;
       end;

    ** attsch=0;
       if LOCATOR~=15 then do;
          if attsch=0 & e3877A1r20~=0 & 0<=e3112r20<=12 & 0<=grsurvP<GRSURV<=12 then do;
             GRSURV=grsurvP;
             GRAUG=GRSURV;
             LOCATOR=20.1;
             LOCA=20.1;
          end;
          if attsch=0 & e3877A1r20~=0 & 12<e3112r20<95 & 0<=grsurvP_no3061<GRAUG<=12 then do;
             GRAUG=grsurvP_no3061;
             LOCA=20.2;
          end;
       end;
    end;

***** for those who were HE in the DLI and not enrolled at or since DLI;
    if e3877A1r20=0 then do;
      if locatorr19=15 then do;
        grsurv=cvc_hgcr19;
        locator=15;
      end;
      if locaP=15 then do;
        graug=graugP;
        loca=15;
      end;
    end;
********************************************************************************************************
***           [IV] CREATED VARIABLE - HIGHEST DEGREE RECEIVED AS OF THE SURVEY DATE (DEGSURV)        ***
********************************************************************************************************;
    DEGSURV=0;

    ** GED;
    if GED=1 then DEGSURV=1;
    if GED=-3 then DEGSURV=-3;

    ** High School Diploma;
    if HSD=1 then DEGSURV=2;
    if HSD=-3 then DEGSURV=-3;

    ** Junior College or two-year Associate Degree;
    if CDTAA>-4 then DEGSURV=3;

    ** Bachelor's Degree;
    if CDTBA>-4 then DEGSURV=4;

    ** Master's Degree;
    if CDTMA>-4 then DEGSURV=5;

    ** Doctoral Degree;
    if CDTDC>-4 then DEGSURV=6;

    ** Professional Degree;
    if CDTPF>-4 then DEGSURV=7;

    ** If cannot determine what degree received;
    do i=1 to dim(e23450);
     if e23450(i) in (-1, -2, -3) & DEGSURV not in (4 5 6 7) then DEGSURV=-3;
    end;
    if 0<=DEGSURV<cvc_hdgr19 then prev_deg=1;

********************************************************************************************************
***          [V] CREATED VARIABLE - HIGHEST DEGREE COMPLETED AS OF AUGUST 31, 2021 (DEGAUG)           ***
********************************************************************************************************;

    DEGAUG=0;

    ** GED;
    if 0<CDTGD<=MONYR then DEGAUG=1;
    if CDTGD in (-3, -2, -1) then DEGAUG=-3;

    ** High School Diploma;
    if 0<CDTHS<=MONYR then DEGAUG=2;
    if CDTHS in (-3, -2, -1) then DEGAUG=-3;

    ** Junior College or 2-year Associate Degree;
    if 0<CDTAA<=MONYR then DEGAUG=3;
    if CDTAA in (-3, -2, -1) then DEGAUG=-3;

    ** Bachelor's Degree;
    if 0<CDTBA<=MONYR then DEGAUG=4;
    if CDTBA in (-3, -2, -1) then DEGAUG=-3;

    ** Master's Degree;
    if 0<CDTMA<=MONYR then DEGAUG=5;
    if CDTMA in (-3, -2, -1) then DEGAUG=-3;

    ** Doctoral Degree;
    if 0<CDTDC<=MONYR then DEGAUG=6;
    if CDTDC in (-3, -2, -1) then DEGAUG=-3;

    ** Professional Degree;
    if 0<CDTPF<=MONYR then DEGAUG=7;
    if CDTPF in (-3, -2, -1) then DEGAUG=-3;

    if DEGSURV=-3 then DEGAUG=-3;
    if (DEGSURV=1 & -4<CDTGD<=0)|(DEGSURV=2 & -4<CDTHS<=0)|(DEGSURV=3 & -4<CDTAA<=0)
      |(DEGSURV=4 & -4<CDTBA<=0)|(DEGSURV=5 & -4<CDTMA<=0)|(DEGSURV=6 & -4<CDTDC<=0)
      |(DEGSURV=7 & -4<CDTPF<=0) then do;
      datedeg=1;
      DEGAUG=-3;
    end;
    if DEGAUG~=DEGSURV & DEGSURV=degsurvP & DEGSURV>=0 then do;
       DEGAUG=degsurvP;
    end;

  if (cdtgd=gdr19 & cdths=hsr19 & cdtaa=aar19 & cdtba=bar19 & cdtma=mar19 & cdtpf=pfr19
     & cdtdc=dcr19) then do;
    if he19_deg ne . then do;
      he20_deg=degsurv;
      degsurv=-3;
      degaug=-3;
      if encatr19=-3 & encat in (1,2,3,5,6,7) then do;
        he20_en=encat;
        encat=-3;
      end;
    end;
    else if degsurv ne degsurvP then chk21=1;
  end;

if ENCAT    in (-1,-2) then ENCAT=-3;
if GRSURV   in (-1,-2) then GRSURV=-3;
if GRAUG   in (-1,-2) then GRAUG=-3;
if DEGSURV  in (-1,-2) then DEGSURV=-3;
if DEGAUG  in (-1,-2) then DEGAUG=-3;

***** generate GRSURVF which is used to generate HGC in LEFT_HS program;
grsurvFP=grsurvFr19;
    grsurvF=grsurvFP;
    locF=6;
     if (e3112r20>12 & (currcoll_1=1 or prevcoll=1)) or (e3112r20=12 & cdths ne hsr19) or (0<=e3112r20<12 & attsch=1) or (-4<e3112r20<0 & e3877A1r20~=0)  then do;
       if hsrP=-4 & (grsurvFP<12|grsurvFP=95) & (GRSURV<=12|GRSURV=95) then do;
          grsurvF=GRSURV;
          locF=1;
       end;
       if hsrP=-4 & (grsurvFP<12|grsurvFP=95) & 12<GRSURV<95 then do;
             grsurvF=grsurvFP;
             locF=3;
       end;
       if hsrP>-4|gdrP>-4|grsurvFP=12 then do;
          grsurvF=grsurvFP;
          locF=4.1;
          if grsurvFP=-3 then do;
           if 0<grsurv<=12 then do;
            grsurvF=grsurv;
            locF=4.21;
           end;
           else if 0<graug<=12 then do;
            locF=4.22;
           end;
          end;
          if grsurvFP<grsurv<=12 then do;
           grsurvF=grsurv;
           locF=4.3;
          end;
       end;
     if hsrP=-4 & gdrP=-4 & grsurvFP<12 & ENCAT in (1,2,3,8,-3) & (GRSURV<=12|GRSURV=95) then do; *Feb 2015;
          grsurvF=GRSURV;
          locF=5;
       end;
       if LOCATOR~=15 & attsch=0 & e3877A1r20~=0 & 0<=e3112r20<=12 & 0<=grsurvP<GRSURV<=12 & locF in (1,5) then do;
          grsurvF=GRSURVP;
          locF=7;
       end;
    end;

***** Finalize two cvc_ variables;
grsurv_cvc=grsurv;
degsurv_cvc=degsurv;

***** finalize the CV_ variables;
if e3112r20=-5 then do;
   ENROLL=-5;
   ENCAT=-5;
   GRSURV=-5;
   GRAUG=-5;
   DEGSURV=-5;
   DEGAUG=-5;
   GED=-5;
   HSD=-5;
   LOCATOR=-5;
   LOCA=-5;
end;

***** flags to check;
if ENCAT in (4,5,6,7,9,10,11) then do;
   if 0<=grsurv_cvc<12 then chk1=1;
   if grsurv_cvc=95 then chk1=2;
end;
if chk1=1 then do;
 if encat in (5,6,7) then chk1enc=1;
 if encat in (4,9,10,11) & ged=1 then chk1ged=0;
 if encat in (4,9,10,11) & ged ne 1 then chk1ged=1;
end;
if ENCAT in (1,2,3,8,-3) then do;
  if 12<grsurv_cvc<95 then chk2=1;
  if grsurv_cvc=95 then chk2=2;
end;
if degsurv_cvc=0 & ENCAT in (2,3,4,5,6,7,9,10,11) then chk3=1;
if degsurv_cvc=1 & ENCAT in (1,8,3,5,6,7) then chk3=2;
if degsurv_cvc=2 & ENCAT in (1,8,2,5,6,7) then chk3=3;
if degsurv_cvc=3 & ENCAT in (1,8,2,3,6,7,4) then chk3=4;
if degsurv_cvc=4 & ENCAT in (1,8,2,3,5,7,4) then chk3=5;
if degsurv_cvc in (5,6,7) & ENCAT in (1,8,2,3,5,6,4) then chk3=6;
if 0<=e2857r20<grsurv<95 then do;
   if 0<=e2857r20<grsurvP<95 & e3877A1r20~=0 then chk4=1.1;
   if 0<=e2857r20<e3112r20<95 then chk4=1.3;
   if hsd=1 & grsurv_cvc=12 then chk4=.;
end;
if 0<=e2857r20<6 & ENCAT in (4,5,6,7,9,10,11) then z=e2857r20+12;
if z=. then chk4z=chk4;
if 0<=z<grsurv_cvc then do;
   if 0<=z<grsurvP & e3877A1r20~=0 then chk4z=1.1;
   if 0<=z<e3112r20 then chk4z=1.3;
   if hsd=1 & grsurv_cvc=12 then chk4z=.;
end;
if ENCAT=-3 & grsurv_cvc~=-3 then chk5=1;
if ENCAT~=-3 & grsurv_cvc=-3 & LOCATOR~=15 then chk5=2;
if ENCAT in (4,5,6,7,9,10,11) & currcoll=0 & 0<=grsurvP<grsurv_cvc & ever_24yrinst~=1 then chk6=1;
if 0<grsurv_cvc<grsurvP<95 then disagre=4;
if cdths>-4 & 0<=grsurv_cvc<12 & degsurv_cvc ne -3 then chk31=1;
if cdths=-4 & grsurv_cvc=12 & degsurv_cvc ne -3 then chk32=1;
if degsurv_cvc>2 & 0<=grsurv_cvc<=12 then chk33=1;
if e3877A1r20=0 & grsurv_cvc ne cvc_hgcr19 then chk11=1;
if 0<grsurv_cvc<=12 & graug ne grsurv_cvc & graug ne -5 then chk12=1;
if 0<grsurv_cvc<=12 & grsurvF ne grsurv_cvc then chk13=1;
if (e3112r20 ne -5 & grsurvF ne graug) or (e3112r20=-5 & grsurvF ne graugP) then chk14=1;
  if (e3112r20 ne -5 & grsurvF ne graug) then chk14=1.1;
  if (e3112r20=-5 & grsurvF ne graugP) then chk14=1.2;
if degsurv=2 & degaug not in (2,-5) then chk22=1;

************************HE;

***** finalize the CV_ variables;
if e3112r20=-5 then do;
   ENROLL=-5;
   ENCAT=-5;
   GRSURV=-5;
   GRAUG=-5;
   DEGSURV=-5;
   DEGAUG=-5;
   GED=-5;
   HSD=-5;
   LOCATOR=-5;
   LOCA=-5;
end;