/* Replication of conditional logit and predicted probabilities
from Glasgow (2005) PRQ.  Note this won't replicate exactly, as 
imputed data was used in the published article. */

#delimit;
set mem 5000k;
set more off;

log using clogit.log, replace;

use PRQ2005;

generate id = _n;          /* created a variable for ID# */

generate choice = 1;      /* to keep track of which canidate 
                             this sub-dataset is for */

generate votec=0;
replace votec=1 if vote==1;    /* =1 if vote for Bush */

generate discand = disbush;

egen mdisb = mean(disbush);
egen mdisc = mean(disclin);
egen mdisp = mean(dispero);

keep id choice votec discand unionh joblossh joblossa joblossp 
unionjlh unionjla unionjlp wage1992 wagejlh manuf manufjlh 
hisp_aff im_aff ill_im_aff perfin natecon globecon japan_unfair 
limimp pid yrseduc income minority female age mdisb mdisc mdisp;


save vbush, replace;

/* We need to do this so we can use the "append" command later */

clear;

use PRQ2005;

generate id = _n;  

generate choice = 2;

generate votec=0;
replace votec=1 if vote==2;    /* =1 if vote for Clinton */

generate discand = disclin;

egen mdisb = mean(disbush);
egen mdisc = mean(disclin);
egen mdisp = mean(dispero);

keep id choice votec discand unionh joblossh joblossa joblossp 
unionjlh unionjla unionjlp wage1992 wagejlh manuf manufjlh 
hisp_aff im_aff ill_im_aff perfin natecon globecon japan_unfair 
limimp pid yrseduc income minority female age mdisb mdisc mdisp;

save vclin, replace;

clear;

use PRQ2005;

generate id = _n;      

generate choice = 3;

generate votec=0;
replace votec=1 if vote==3;    /* =1 if vote for Perot */

generate discand = dispero;

egen mdisb = mean(disbush);
egen mdisc = mean(disclin);
egen mdisp = mean(dispero);

keep id choice votec discand unionh joblossh joblossa joblossp 
unionjlh unionjla unionjlp wage1992 wagejlh manuf manufjlh 
hisp_aff im_aff ill_im_aff perfin natecon globecon japan_unfair 
limimp pid yrseduc income minority female age mdisb mdisc mdisp;

save vpero, replace;

clear;

use vbush;
append using vclin;
append using vpero;

/* this stacks all of our sub-datasets into 1 dataset */

sort id;

generate cbush = 0;
replace cbush = 1 if choice==1;

generate cclin = 0;
replace cclin = 1 if choice==2;

/* creating data matrix, normalized wrt Perot */

generate bunion = unionh*cbush;
generate cunion = unionh*cclin;

generate bjoblh = joblossh*cbush;
generate cjoblh = joblossh*cclin;

generate bjobla = joblossa*cbush;
generate cjobla = joblossa*cclin;

generate bjoblp = joblossp*cbush;
generate cjoblp = joblossp*cclin;

generate bujlh = unionjlh * cbush;
generate cujlh = unionjlh * cclin;

generate bujla = unionjla * cbush;
generate cujla = unionjla * cclin;

generate bujlp = unionjlp * cbush;
generate cujlp = unionjlp * cclin;

generate bwage1992 = wage1992 * cbush;
generate cwage1992 = wage1992 *cclin;

generate bwagejlh = wagejlh * cbush;
generate cwagejlh = wagejlh * cclin;

generate bmanuf = manuf * cbush;
generate cmanuf = manuf * cclin;

generate bmanufjlh = manufjlh * cbush;
generate cmanufjlh = manufjlh * cclin;

generate bhisp_af = hisp_aff * cbush;
generate chisp_aff = hisp_aff * cclin;

generate bim_aff = im_aff * cbush;
generate cim_aff = im_aff * cclin;

generate bill_im_aff = ill_im_aff * cbush;
generate cill_im_aff = ill_im_aff * cclin;

generate bperfin = perfin * cbush;
generate cperfin = perfin * cclin;

generate bnatecon = natecon * cbush;
generate cnatecon = natecon * cclin;

generate bglobecon = globecon * cbush;
generate cglobecon = globecon * cclin;

generate bjapan = japan_unfair * cbush;
generate cjapan = japan_unfair * cclin;

generate blimimp = limimp * cbush;
generate climimp = limimp * cclin;

generate bpid = pid * cbush;
generate cpid = pid * cclin;

generate beduc = yrseduc * cbush;
generate ceduc = yrseduc * cclin;

generate bincome = income * cbush;
generate cincome = income * cclin;

generate bminority = minority * cbush;
generate cminority = minority * cclin;

generate bfemale = female * cbush;
generate cfemale = female * cclin;

generate bage = age * cbush;
generate cage = age * cclin;



clogit votec discand 
bunion bjoblh bjobla bjoblp 
bujlh bujla bujlp bwage1992 bwagejlh bmanuf bmanufjlh 
bhisp_af bim_aff bill_im_aff bperfin bnatecon 
bglobecon bjapan blimimp bpid beduc bincome bminority 
bfemale bage cbush
cunion cjoblh cjobla cjoblp 
cujlh cujla cujlp cwage1992 cwagejlh cmanuf cmanufjlh 
chisp_af cim_aff cill_im_aff cperfin cnatecon 
cglobecon cjapan climimp cpid ceduc cincome cminority 
cfemale cage cclin, 
group(id);

matrix coeffs = e(b);
matrix covmat = e(V);

drawnorm b1-b53, means(coeffs) cov(covmat) double;

egen minc = mean(income);
egen meduc = mean(yrseduc);
egen mwage = mean(wage1992);
egen mhisp = mean(hisp_aff);
egen mim = mean(im_aff);
egen mill = mean(ill_im_aff);



generate bjl0u0 = .;
generate cjl0u0 = .;
generate pjl0u0 = .;
generate bjl3u0 = .;
generate cjl3u0 = .;
generate pjl3u0 = .;
generate bjl0u1 = .;
generate cjl0u1 = .;
generate pjl0u1 = .;
generate bjl3u1 = .;
generate cjl3u1 = .;
generate pjl3u1 = .;


local a=1;
while `a'<101{;

qui generate ebjl0u0 = exp(mdisb*b1[`a'] + 0*b2[`a'] + 0*b3[`a'] + 1*b4[`a'] + 0*b5[`a'] + 
0*b6[`a'] + 0*b7[`a'] + 0*b8[`a'] + mwage*b9[`a'] + 0*b10[`a'] + 0*b11[`a'] + 0*b12[`a'] + 
mhisp*b13[`a'] + mim*b14[`a'] + mill*b15[`a'] + (-1*b16[`a']) + (-1*b17[`a']) + (-1*b18[`a']) 
+ 1*b19[`a'] + 1*b20[`a'] + 2*b21[`a'] + meduc*b22[`a'] + minc*b23[`a'] + 0*b24[`a'] + 0*b25[`a'] + 
46*b26[`a'] + b27[`a']);

qui generate ecjl0u0 = exp(mdisc*b1[`a'] + 0*b28[`a'] + 0*b29[`a'] + 1*b30[`a'] + 0*b31[`a'] + 
0*b32[`a'] + 0*b33[`a'] + 0*b34[`a'] + mwage*b35[`a'] + 0*b36[`a'] + 0*b37[`a'] + 0*b38[`a'] + 
mhisp*b39[`a'] + mim*b40[`a'] + mill*b41[`a'] + (-1*b42[`a']) + (-1*b43[`a']) + (-1*b44[`a']) + 
1*b45[`a'] + 1*b46[`a'] + 2*b47[`a'] + meduc*b48[`a'] + minc*b49[`a'] + 0*b50[`a'] + 0*b51[`a'] + 
46*b52[`a'] + b53[`a']);

qui generate epjl0u0 = exp(mdisp*b1[`a']);

qui generate sumjl0u0 = ebjl0u0 + ecjl0u0 + epjl0u0;

qui replace bjl0u0 = (ebjl0u0/sumjl0u0)*100 if _n==`a';
qui replace cjl0u0 = (ecjl0u0/sumjl0u0)*100 if _n==`a';
qui replace pjl0u0 = (epjl0u0/sumjl0u0)*100 if _n==`a';

drop ebjl0u0 ecjl0u0 epjl0u0 sumjl0u0;

qui generate ebjl3u0 = exp(mdisb*b1[`a'] + 0*b2[`a'] + 3*b3[`a'] + 1*b4[`a'] + 0*b5[`a'] + 
0*b6[`a'] + 0*b7[`a'] + 0*b8[`a'] + mwage*b9[`a'] + (mwage*3*b10[`a']) + 0*b11[`a'] + 0*b12[`a'] + 
mhisp*b13[`a'] + mim*b14[`a'] + mill*b15[`a'] + (-1*b16[`a']) + (-1*b17[`a']) + (-1*b18[`a']) 
+ 1*b19[`a'] + 1*b20[`a'] + 2*b21[`a'] + meduc*b22[`a'] + minc*b23[`a'] + 0*b24[`a'] + 0*b25[`a'] + 
46*b26[`a'] + b27[`a']);

qui generate ecjl3u0 = exp(mdisc*b1[`a'] + 0*b28[`a'] + 3*b29[`a'] + 1*b30[`a'] + 0*b31[`a'] + 
0*b32[`a'] + 0*b33[`a'] + 0*b34[`a'] + mwage*b35[`a'] + (mwage*3*b36[`a']) + 0*b37[`a'] + 0*b38[`a'] + 
mhisp*b39[`a'] + mim*b40[`a'] + mill*b41[`a'] + (-1*b42[`a']) + (-1*b43[`a']) + (-1*b44[`a']) + 
1*b45[`a'] + 1*b46[`a'] + 2*b47[`a'] + meduc*b48[`a'] + minc*b49[`a'] + 0*b50[`a'] + 0*b51[`a'] + 
46*b52[`a'] + b53[`a']);

qui generate epjl3u0 = exp(mdisp*b1[`a']);

qui generate sumjl3u0 = ebjl3u0 + ecjl3u0 + epjl3u0;

qui replace bjl3u0 = (ebjl3u0/sumjl3u0)*100 if _n==`a';
qui replace cjl3u0 = (ecjl3u0/sumjl3u0)*100 if _n==`a';
qui replace pjl3u0 = (epjl3u0/sumjl3u0)*100 if _n==`a';

drop ebjl3u0 ecjl3u0 epjl3u0 sumjl3u0;

qui generate ebjl0u1 = exp(mdisb*b1[`a'] + 1*b2[`a'] + 0*b3[`a'] + 1*b4[`a'] + 0*b5[`a'] + 
0*b6[`a'] + 0*b7[`a'] + 0*b8[`a'] + mwage*b9[`a'] + 0*b10[`a'] + 0*b11[`a'] + 0*b12[`a'] + 
mhisp*b13[`a'] + mim*b14[`a'] + mill*b15[`a'] + (-1*b16[`a']) + (-1*b17[`a']) + (-1*b18[`a']) 
+ 1*b19[`a'] + 1*b20[`a'] + 2*b21[`a'] + meduc*b22[`a'] + minc*b23[`a'] + 0*b24[`a'] + 0*b25[`a'] + 
46*b26[`a'] + b27[`a']);

qui generate ecjl0u1 = exp(mdisc*b1[`a'] + 1*b28[`a'] + 0*b29[`a'] + 1*b30[`a'] + 0*b31[`a'] + 
0*b32[`a'] + 0*b33[`a'] + 0*b34[`a'] + mwage*b35[`a'] + 0*b36[`a'] + 0*b37[`a'] + 0*b38[`a'] + 
mhisp*b39[`a'] + mim*b40[`a'] + mill*b41[`a'] + (-1*b42[`a']) + (-1*b43[`a']) + (-1*b44[`a']) + 
1*b45[`a'] + 1*b46[`a'] + 2*b47[`a'] + meduc*b48[`a'] + minc*b49[`a'] + 0*b50[`a'] + 0*b51[`a'] + 
46*b52[`a'] + b53[`a']);

qui generate epjl0u1 = exp(mdisp*b1[`a']);

qui generate sumjl0u1 = ebjl0u1 + ecjl0u1 + epjl0u1;

qui replace bjl0u1 = (ebjl0u1/sumjl0u1)*100 if _n==`a';
qui replace cjl0u1 = (ecjl0u1/sumjl0u1)*100 if _n==`a';
qui replace pjl0u1 = (epjl0u1/sumjl0u1)*100 if _n==`a';

drop ebjl0u1 ecjl0u1 epjl0u1 sumjl0u1;

qui generate ebjl3u1 = exp(mdisb*b1[`a'] + 1*b2[`a'] + 3*b3[`a'] + 1*b4[`a'] + 0*b5[`a'] + 
3*b6[`a'] + 0*b7[`a'] + 0*b8[`a'] + mwage*b9[`a'] + (mwage*3*b10[`a']) + 0*b11[`a'] + 0*b12[`a'] + 
mhisp*b13[`a'] + mim*b14[`a'] + mill*b15[`a'] + (-1*b16[`a']) + (-1*b17[`a']) + (-1*b18[`a']) 
+ 1*b19[`a'] + 1*b20[`a'] + 2*b21[`a'] + meduc*b22[`a'] + minc*b23[`a'] + 0*b24[`a'] + 0*b25[`a'] + 
46*b26[`a'] + b27[`a']);

qui generate ecjl3u1 = exp(mdisc*b1[`a'] + 1*b28[`a'] + 3*b29[`a'] + 1*b30[`a'] + 0*b31[`a'] + 
3*b32[`a'] + 0*b33[`a'] + 0*b34[`a'] + mwage*b35[`a'] + (mwage*3*b36[`a']) + 0*b37[`a'] + 0*b38[`a'] + 
mhisp*b39[`a'] + mim*b40[`a'] + mill*b41[`a'] + (-1*b42[`a']) + (-1*b43[`a']) + (-1*b44[`a']) + 
1*b45[`a'] + 1*b46[`a'] + 2*b47[`a'] + meduc*b48[`a'] + minc*b49[`a'] + 0*b50[`a'] + 0*b51[`a'] + 
46*b52[`a'] + b53[`a']);

qui generate epjl3u1 = exp(mdisp*b1[`a']);

qui generate sumjl3u1 = ebjl3u1 + ecjl3u1 + epjl3u1;

qui replace bjl3u1 = (ebjl3u1/sumjl3u1)*100 if _n==`a';
qui replace cjl3u1 = (ecjl3u1/sumjl3u1)*100 if _n==`a';
qui replace pjl3u1 = (epjl3u1/sumjl3u1)*100 if _n==`a';

drop ebjl3u1 ecjl3u1 epjl3u1 sumjl3u1;

local a = `a'+1;
};

generate dbjlu0 = bjl0u0-bjl3u0;
generate dcjlu0 = cjl0u0-cjl3u0;
generate dpjlu0 = pjl0u0-cjl3u0;

generate dbjlu1 = bjl0u1-bjl3u1;
generate dcjlu1 = cjl0u1-cjl3u1;
generate dpjlu1 = pjl0u1-cjl3u1;

generate dbjl0u = bjl0u1-bjl0u0;
generate dcjl0u = cjl0u1-cjl0u0;
generate dpjl0u = pjl0u1-cjl0u0;

generate dbjl3u = bjl3u1-bjl3u0;
generate dcjl3u = cjl3u1-cjl3u0;
generate dpjl3u = pjl3u1-cjl3u0;

generate diffb = dbjl0u-dbjl3u;
generate diffc = dcjl0u-dcjl3u;
generate diffp = dpjl0u-dpjl3u;

summarize bjl0u0 cjl0u0 pjl0u0 
bjl3u0 cjl3u0 pjl3u0 bjl0u1 cjl0u1 pjl0u1 
bjl3u1 cjl3u1 pjl3u1 dbjlu0 dcjlu0 dpjlu0 
dbjlu1 dcjlu1 dpjlu1 dbjl0u dcjl0u dpjl0u 
dbjl3u dcjl3u dpjl3u diffb diffc diffp;

log close;