*! Version 1.0 12Feb2016
* Enrique.Hernandez@EUI.eu
* Predictive margins and marginal effects plots with histogram in the back after regress, logit, xtmixed and mixed.
capture program drop marhis
program define marhis
version 13.0
syntax varlist(max=1), [mar(varname)] [cate(varname)] [points(integer 15)] [percent] [discrete] [level(integer 95)] [label(integer 4)] [summaryno] [confidenceno] [atmeans]
local cmd = e(cmd)
if "`cmd'" != "logit" & "`cmd'" != "xtmixed" & "`cmd'" != "mixed" & "`cmd'" != "regress" {
display as error "`cmd' not supported. marhis can only be used after: regress, logit, xtmixed, mixed"
exit 301
}
local labell = `label'+1
if "`percent'" == "" {
local measure = "Density"
}
if "`percent'" != "" {
local measure = "Percent"
}
if "`mar'" == "" & "`cate'" == "" {
local depvar = e(depvar)
quietly {
tempvar sample
gen `sample' = e(sample)
local coeff = _b[`varlist']
local se = _se[`varlist']
local t = `coeff'/`se'
local coeff = round(`coeff', .001)
local t = round(`t', .01)
sum `varlist' if `sample' == 1
local min = r(min)
local max = r(max)
local steps = (`max'-`min')/`points'
local lab = (`max'-`min')/`labell'
}
display "Estimating margins"
quietly {
if "`atmeans'" == "" {
margins, at(`varlist'=(`min'(`steps')`max')) level(`level')
local sti = "margins, at(`varlist'=(`min'(`steps')`max')) level(`level')"
local estimation = r(predict_label)
local estimation2 = "Average Adjusted Predictions (AAPs)"
}
if "`atmeans'" != "" {
margins, atmeans at(`varlist'=(`min'(`steps')`max')) level(`level')
local sti = "margins, atmeans at(`varlist'=(`min'(`steps')`max')) level(`level')"
local estimation = r(predict_label)
local estimation2 = "Adjusted Predictions at Means (APMs)"
}
}
display "Margins obtained through: `sti'"
display "Graphing `estimation2' of `depvar' across values of `varlist'"
quietly {
matrix A = r(at)
matrix B = r(table)'
local maxat = `points' + 1
matrix D = A[1..`maxat', "`varlist'"]
matrix C = B, D
tempname tempr
svmat C, name(`tempr')
if "`cmd'" == "regress" {
if "`summaryno'" == "" {
if "`confidenceno'" == "" {
twoway (hist `varlist' if `sample' == 1, yaxis(2) `percent' `discrete' xlabel(`min'(`lab')`max') yscale(axis(2) alt) color(gs14) ytitle("`measure' `varlist'", axis(2) size(small)) xtitle(`varlist', size(small))) ///
(line `tempr'1 `tempr'10, yaxis(1) yscale(axis(1) alt) lcolor(navy) ytitle("`estimation' `depvar' with `level' percent CI", size(small))) ///
(rline `tempr'5 `tempr'6 `tempr'10, lpattern(dash) yaxis(1) lcolor(navy)) , ///
legend(off) note(Coefficient `varlist' = `coeff' t-statistic = `t' )
drop `tempr'*
exit
}
if "`confidenceno'" != "" {
twoway (hist `varlist' if `sample' == 1, yaxis(2) `percent' `discrete' xlabel(`min'(`lab')`max') yscale(axis(2) alt) color(gs14) ytitle("`measure' `varlist'", axis(2) size(small)) xtitle(`varlist', size(small))) ///
(line `tempr'1 `tempr'10, yaxis(1) yscale(axis(1) alt) lcolor(navy) ytitle("`estimation' `depvar'", size(small))), ///
legend(off) note(Coefficient `varlist' = `coeff' t-statistic = `t' )
drop `tempr'*
exit
}
}
if "`summaryno'" != "" {
if "`confidenceno'" == "" {
twoway (hist `varlist' if `sample' == 1, yaxis(2) `percent' `discrete' xlabel(`min'(`lab')`max') yscale(axis(2) alt) color(gs14) ytitle("`measure' `varlist'", axis(2) size(small)) xtitle(`varlist', size(small))) ///
(line `tempr'1 `tempr'10, yaxis(1) yscale(axis(1) alt) lcolor(navy) ytitle("`estimation' `depvar' with `level' percent CI", size(small))) ///
(rline `tempr'5 `tempr'6 `tempr'10, lpattern(dash) yaxis(1) lcolor(navy)) , ///
legend(off)
drop `tempr'*
exit
}
if "`confidenceno'" != "" {
twoway (hist `varlist' if `sample' == 1, yaxis(2) `percent' `discrete' xlabel(`min'(`lab')`max') yscale(axis(2) alt) color(gs14) ytitle("`measure' `varlist'", axis(2) size(small)) xtitle(`varlist', size(small))) ///
(line `tempr'1 `tempr'10, yaxis(1) yscale(axis(1) alt) lcolor(navy) ytitle("`estimation' `depvar'", size(small))), ///
legend(off)
drop `tempr'*
exit
}
}
}
if "`cmd'" == "logit" | "`cmd'" == "xtmixed" | "`cmd'" == "mixed" {
if "`summaryno'" == "" {
if "`confidenceno'" == "" {
twoway (hist `varlist' if `sample' == 1, yaxis(2) `percent' `discrete' xlabel(`min'(`lab')`max') yscale(axis(2) alt) color(gs14) ytitle("`measure' `varlist'", axis(2) size(small)) xtitle(`varlist', size(small))) ///
(line `tempr'1 `tempr'10, yaxis(1) yscale(axis(1) alt) lcolor(navy) ytitle("`estimation' with `level' percent CI", size(small))) ///
(rline `tempr'5 `tempr'6 `tempr'10, lpattern(dash) yaxis(1) lcolor(navy)) , ///
legend(off) note(Coefficient `varlist' = `coeff' z-statistic = `t' )
drop `tempr'*
exit
}
if "`confidenceno'" != "" {
twoway (hist `varlist' if `sample' == 1, yaxis(2) `percent' `discrete' xlabel(`min'(`lab')`max') yscale(axis(2) alt) color(gs14) ytitle("`measure' `varlist'", axis(2) size(small)) xtitle(`varlist', size(small))) ///
(line `tempr'1 `tempr'10, yaxis(1) yscale(axis(1) alt) lcolor(navy) ytitle("`estimation'", size(small))), ///
legend(off) note(Coefficient `varlist' = `coeff' z-statistic = `t' )
drop `tempr'*
exit
}
}
if "`summaryno'" != "" {
if "`confidenceno'" == "" {
twoway (hist `varlist' if `sample' == 1, yaxis(2) `percent' `discrete' xlabel(`min'(`lab')`max') yscale(axis(2) alt) color(gs14) ytitle("`measure' `varlist'", axis(2) size(small)) xtitle(`varlist', size(small))) ///
(line `tempr'1 `tempr'10, yaxis(1) yscale(axis(1) alt) lcolor(navy) ytitle("`estimation' with `level' percent CI", size(small))) ///
(rline `tempr'5 `tempr'6 `tempr'10, lpattern(dash) yaxis(1) lcolor(navy)) , ///
legend(off)
drop `tempr'*
exit
}
if "`confidenceno'" != "" {
twoway (hist `varlist' if `sample' == 1, yaxis(2) `percent' `discrete' xlabel(`min'(`lab')`max') yscale(axis(2) alt) color(gs14) ytitle("`measure' `varlist'", axis(2) size(small)) xtitle(`varlist', size(small))) ///
(line `tempr'1 `tempr'10, yaxis(1) yscale(axis(1) alt) lcolor(navy) ytitle("`estimation'", size(small))), ///
legend(off)
drop `tempr'*
exit
}
}
}
}
}
if "`mar'" != "" {
local depvar = e(depvar)
local cmd = e(cmd)
quietly {
tempvar sample case
gen `sample' = e(sample)
local coeff = _b[c.`varlist'#c.`mar']
local se = _se[c.`varlist'#c.`mar']
local t = `coeff'/`se'
local coeff = round(`coeff', .001)
local t = round(`t', .01)
sum `varlist' if `sample' == 1
local min = r(min)
local max = r(max)
local steps = (`max'-`min')/`points'
local lab = (`max'-`min')/`labell'
}
display "Estimating margins"
quietly {
if "`atmeans'" == "" {
margins, dydx(`mar') at(`varlist'=(`min'(`steps')`max')) level(`level')
local sti = "margins, dydx(`mar') at(`varlist'=(`min'(`steps')`max')) level(`level')"
local estimation = r(predict_label)
local estimation2 = "Average Marginal Effects (AMEs)"
}
if "`atmeans'" != "" {
margins, atmeans dydx(`mar') at(`varlist'=(`min'(`steps')`max')) level(`level')
local sti = "margins, atmeans dydx(`mar') at(`varlist'=(`min'(`steps')`max')) level(`level')"
local estimation = r(predict_label)
local estimation2 = "Marginal Effects at Means (MEMs)"
}
}
display "Margins obtained through: `sti'"
display "Graphing `estimation2' of `mar' on `depvar' across values of `varlist''"
if "`cmd'" == "regress" {
matrix A = r(at)
matrix B = r(table)'
local maxat = `points' + 1
matrix D = A[1..`maxat', "`varlist'"]
matrix C = B, D
tempname tempr
svmat C, name(`tempr')
quietly {
if "`summaryno'" == "" {
if "`confidenceno'" == "" {
twoway (hist `varlist' if `sample' == 1, yaxis(2) `percent' `discrete' xlabel(`min'(`lab')`max') yscale(axis(2) alt) color(gs14) ytitle("`measure' `varlist'", axis(2) size(small)) xtitle(`varlist', size(small))) ///
(function y=0 if `sample' == 1, ra(`min' `max')) ///
(line `tempr'1 `tempr'10, yaxis(1) yscale(axis(1) alt) lcolor(navy) ytitle("Effects of `mar' on `estimation' with `level' percent CI", size(small))) ///
(rline `tempr'5 `tempr'6 `tempr'10, lpattern(dash) yaxis(1) lcolor(navy)) ///
, legend(off) note(Coefficient of product term = `coeff' t-statistic = `t')
drop `tempr'*
exit
}
if "`confidenceno'" != "" {
twoway (hist `varlist' if `sample' == 1, yaxis(2) `percent' `discrete' xlabel(`min'(`lab')`max') yscale(axis(2) alt) color(gs14) ytitle("`measure' `varlist'", axis(2) size(small)) xtitle(`varlist', size(small))) ///
(function y=0 if `sample' == 1, ra(`min' `max')) ///
(line `tempr'1 `tempr'10, yaxis(1) yscale(axis(1) alt) lcolor(navy) ytitle("Effects of `mar' on `estimation' with `level' percent CI", size(small))) ///
, legend(off) note(Coefficient of product term = `coeff' t-statistic = `t')
drop `tempr'*
exit
}
}
if "`summaryno'" != "" {
if "`confidenceno'" == "" {
twoway (hist `varlist' if `sample' == 1, yaxis(2) `percent' `discrete' xlabel(`min'(`lab')`max') yscale(axis(2) alt) color(gs14) ytitle("`measure' `varlist'", axis(2) size(small)) xtitle(`varlist', size(small))) ///
(function y=0 if `sample' == 1, ra(`min' `max')) ///
(line `tempr'1 `tempr'10, yaxis(1) yscale(axis(1) alt) lcolor(navy) ytitle("Effects of `mar' on `estimation' with `level' percent CI", size(small))) ///
(rline `tempr'5 `tempr'6 `tempr'10, lpattern(dash) yaxis(1) lcolor(navy)) ///
, legend(off)
drop `tempr'*
exit
}
if "`confidenceno'" != "" {
twoway (hist `varlist' if `sample' == 1, yaxis(2) `percent' `discrete' xlabel(`min'(`lab')`max') yscale(axis(2) alt) color(gs14) ytitle("`measure' `varlist'", axis(2) size(small)) xtitle(`varlist', size(small))) ///
(function y=0 if `sample' == 1, ra(`min' `max')) ///
(line `tempr'1 `tempr'10, yaxis(1) yscale(axis(1) alt) lcolor(navy) ytitle("Effects of `mar' on `estimation' with `level' percent CI", size(small))) ///
, legend(off)
drop `tempr'*
exit
}
}
}
}
if "`cmd'" == "logit" | "`cmd'" == "xtmixed" | "`cmd'" == "mixed" {
matrix A = r(at)
matrix B = r(table)'
local maxat = `points' + 1
matrix D = A[1..`maxat', "`varlist'"]
matrix C = B, D
tempname tempr
svmat C, name(`tempr')
quietly {
if "`summaryno'" == "" {
if "`confidenceno'" == "" {
twoway (hist `varlist' if `sample' == 1, yaxis(2) `percent' `discrete' xlabel(`min'(`lab')`max') yscale(axis(2) alt) color(gs14) ytitle("`measure' `varlist'", axis(2) size(small)) xtitle(`varlist', size(small))) ///
(function y=0, ra(`min' `max')) ///
(line `tempr'1 `tempr'10, yaxis(1) yscale(axis(1) alt) lcolor(navy) ytitle("Effects of `mar' on `estimation' with `level' percent CI", size(small))) ///
(rline `tempr'5 `tempr'6 `tempr'10, lpattern(dash) yaxis(1) lcolor(navy)) ///
, legend(off) note(Coefficient of product term = `coeff' z-statistic = `t')
drop `tempr'*
exit
}
if "`confidenceno'" != "" {
twoway (hist `varlist' if `sample' == 1, yaxis(2) `percent' `discrete' xlabel(`min'(`lab')`max') yscale(axis(2) alt) color(gs14) ytitle("`measure' `varlist'", axis(2) size(small)) xtitle(`varlist', size(small))) ///
(function y=0, ra(`min' `max')) ///
(line `tempr'1 `tempr'10, yaxis(1) yscale(axis(1) alt) lcolor(navy) ytitle("Effects of `mar' on `estimation' with `level' percent CI", size(small))) ///
, legend(off) note(Coefficient of product term = `coeff' z-statistic = `t')
drop `tempr'*
exit
}
}
if "`summaryno'" != "" {
if "`confidenceno'" == "" {
twoway (hist `varlist' if `sample' == 1, yaxis(2) `percent' `discrete' xlabel(`min'(`lab')`max') yscale(axis(2) alt) color(gs14) ytitle("`measure' `varlist'", axis(2) size(small)) xtitle(`varlist', size(small))) ///
(function y=0, ra(`min' `max')) ///
(line `tempr'1 `tempr'10, yaxis(1) yscale(axis(1) alt) lcolor(navy) ytitle("Effects of `mar' on `estimation' with `level' percent CI", size(small))) ///
(rline `tempr'5 `tempr'6 `tempr'10, lpattern(dash) yaxis(1) lcolor(navy)) ///
, legend(off)
drop `tempr'*
exit
}
if "`confidenceno'" != "" {
twoway (hist `varlist' if `sample' == 1, yaxis(2) `percent' `discrete' xlabel(`min'(`lab')`max') yscale(axis(2) alt) color(gs14) ytitle("`measure' `varlist'", axis(2) size(small)) xtitle(`varlist', size(small))) ///
(function y=0, ra(`min' `max')) ///
(line `tempr'1 `tempr'10, yaxis(1) yscale(axis(1) alt) lcolor(navy) ytitle("Effects of `mar' on `estimation' with `level' percent CI", size(small))) ///
, legend(off)
drop `tempr'*
exit
}
}
}
}
}
if "`cate'" != "" {
quietly: ta `cate'
local ncat = r(r)
local depvar = e(depvar)
local labelling : value label `cate'
if "`labelling'" == "" {
display as error "Values of categorical term of the interaction (`cate') are not labelled."
exit 182
}
if `ncat' > 5 {
display as error "Number of categories of categorical term of the interaction exceeds maximum of 5"
exit 149
}
if `ncat' >= 2 {
tempvar sample x h
gen `sample' = e(sample)
quietly {
sum `varlist' if `sample' == 1
local min = r(min)
local max = r(max)
local steps = (`max'-`min')/`points'
local lab = (`max'-`min')/`labell'
}
quietly {
levelsof `cate', local(levels)
local lbe : value label `cate'
foreach l of local levels {
local f`l' : label `lbe' `l'
}
}
display "Estimating margins"
quietly {
if "`atmeans'" == "" {
margins `cate', at(`varlist'=(`min'(`steps')`max')) level(`level')
local sti = "margins `cate', at(`varlist'=(`min'(`steps')`max')) level(`level')"
local estimation = r(predict_label)
local estimation2 = "Average Adjusted Predictions (AAPs)"
}
if "`atmeans'" != "" {
margins `cate', atmeans at(`varlist'=(`min'(`steps')`max')) level(`level')
local sti = "margins `cate', atmeans at(`varlist'=(`min'(`steps')`max')) level(`level')"
local estimation = r(predict_label)
local estimation2 = "Adjusted Predictions at Means (APMs)"
}
}
display "Margins obtained through: `sti'"
display "Graphing `estimation2' of `depvar' for categories of `cate' across values of `varlist'"
if `ncat' == 2{
quietly{
matrix C = r(table)'
matrix A = r(at)
local maxat = `points' + 1
matrix A = A[1..`maxat',"`varlist'"]
levelsof `cate', local(values)
matrix input Z = (`values')
forval cat = 1/2{
local cat`cat' = Z[1,`cat']
}
forval i = 1/`maxat' {
local num = `i'
matrix b = C["`i'._at#`cat1'.`cate'", "b".."ul"]
if `i' == 1 matrix F = b
else matrix F = (F \ b)
}
forval i = 1/`maxat' {
local num = `i'
matrix b = C["`i'._at#`cat2'.`cate'", "b".."ul"]
if `i' == 1 matrix G = b
else matrix G = (G \ b)
}
matrix ca1 = A,F
matrix ca2 = A,G
tempname catt1
svmat ca1, name(`catt1')
tempname catt2
svmat ca2, name(`catt2')
if "`confidenceno'" == "" {
twoway ///
(hist `varlist' if `sample' == 1, yaxis(2) `percent' `discrete' xlabel(`min'(`lab')`max') yscale(axis(2) alt) color(gs14) ytitle("`measure' `varlist'", axis(2) size(small)) xtitle(`varlist',size(small))) ///
(line `catt1'2 `catt1'1, yaxis(1) yscale(axis(1) alt) lcolor(navy) ytitle("`estimation' with `level' percent CI", size(small))) ///
(rcap `catt1'7 `catt1'6 `catt1'1, lpattern(solid) yaxis(1) lcolor(navy)) ///
(line `catt2'2 `catt2'1, yaxis(1) lcolor(maroon) lpattern(dash)) ///
(rcap `catt2'7 `catt2'6 `catt2'1, lpattern(solid) yaxis(1) lcolor(maroon)) ///
, legend(order(2 "`f`cat1''" 4 "`f`cat2''"))
drop `catt1'* `catt2'*
exit
}
if "`confidenceno'" != "" {
twoway ///
(hist `varlist' if `sample' == 1, yaxis(2) `percent' `discrete' xlabel(`min'(`lab')`max') yscale(axis(2) alt) color(gs14) ytitle("`measure' `varlist'", axis(2) size(small)) xtitle(`varlist',size(small))) ///
(line `catt1'2 `catt1'1, yaxis(1) yscale(axis(1) alt) lcolor(navy) ytitle("`estimation'", size(small))) ///
(line `catt2'2 `catt2'1, yaxis(1) lcolor(maroon) lpattern(dash)) ///
, legend(order(2 "`f`cat1''" 3 "`f`cat2''"))
drop `catt1'* `catt2'*
exit
}
}
}
if `ncat' == 3{
quietly{
matrix C = r(table)'
matrix A = r(at)
local maxat = `points' + 1
matrix A = A[1..`maxat',"`varlist'"]
levelsof `cate', local(values)
matrix input Z = (`values')
forval cat = 1/3{
local cat`cat' = Z[1,`cat']
}
forval i = 1/`maxat' {
local num = `i'
matrix b = C["`i'._at#`cat1'.`cate'", "b".."ul"]
if `i' == 1 matrix F = b
else matrix F = (F \ b)
}
forval i = 1/`maxat' {
local num = `i'
matrix b = C["`i'._at#`cat2'.`cate'", "b".."ul"]
if `i' == 1 matrix G = b
else matrix G = (G \ b)
}
forval i = 1/`maxat' {
local num = `i'
matrix b = C["`i'._at#`cat3'.`cate'", "b".."ul"]
if `i' == 1 matrix H = b
else matrix H = (H \ b)
}
matrix ca1 = A,F
matrix ca2 = A,G
matrix ca3 = A,H
tempname catt1
svmat ca1, name(`catt1')
tempname catt2
svmat ca2, name(`catt2')
tempname catt3
svmat ca3, name(`catt3')
if "`confidenceno'" == "" {
twoway ///
(hist `varlist' if `sample' == 1, yaxis(2) `percent' `discrete' xlabel(`min'(`lab')`max') yscale(axis(2) alt) color(gs14) ytitle("`measure' `varlist'", axis(2) size(small)) xtitle(`varlist',size(small))) ///
(line `catt1'2 `catt1'1, yaxis(1) yscale(axis(1) alt) lcolor(navy) ytitle("`estimation' with `level' percent CI", size(small))) ///
(rcap `catt1'7 `catt1'6 `catt1'1, lpattern(solid) yaxis(1) lcolor(navy)) ///
(line `catt2'2 `catt2'1, yaxis(1) lcolor(maroon) lpattern(dash)) ///
(rcap `catt2'7 `catt2'6 `catt2'1, lpattern(solid) yaxis(1) lcolor(maroon)) ///
(line `catt3'2 `catt3'1, yaxis(1) lcolor(dkgreen) lpattern(dot)) ///
(rcap `catt3'7 `catt3'6 `catt3'1, lpattern(solid) yaxis(1) lcolor(dkgreen)) ///
, legend(order(2 "`f`cat1''" 4 "`f`cat2''" 6 "`f`cat3''" ))
drop `catt1'* `catt2'* `catt3'*
exit
}
if "`confidenceno'" != "" {
twoway ///
(hist `varlist' if `sample' == 1, yaxis(2) `percent' `discrete' xlabel(`min'(`lab')`max') yscale(axis(2) alt) color(gs14) ytitle("`measure' `varlist'", axis(2) size(small)) xtitle(`varlist',size(small))) ///
(line `catt1'2 `catt1'1, yaxis(1) yscale(axis(1) alt) lcolor(navy) ytitle("`estimation' `depvar'", size(small))) ///
(line `catt2'2 `catt2'1, yaxis(1) lcolor(maroon) lpattern(dash)) ///
(line `catt3'2 `catt3'1, yaxis(1) lcolor(dkgreen) lpattern(dot)) ///
, legend(order(2 "`f`cat1''" 3 "`f`cat2''" 4 "`f`cat3''" ))
drop `catt1'* `catt2'*
exit
}
}
}
if `ncat' == 4{
quietly{
matrix C = r(table)'
matrix A = r(at)
local maxat = `points' + 1
matrix A = A[1..`maxat',"`varlist'"]
levelsof `cate', local(values)
matrix input Z = (`values')
forval cat = 1/4{
local cat`cat' = Z[1,`cat']
}
forval i = 1/`maxat' {
local num = `i'
matrix b = C["`i'._at#`cat1'.`cate'", "b".."ul"]
if `i' == 1 matrix F = b
else matrix F = (F \ b)
}
forval i = 1/`maxat' {
local num = `i'
matrix b = C["`i'._at#`cat2'.`cate'", "b".."ul"]
if `i' == 1 matrix G = b
else matrix G = (G \ b)
}
forval i = 1/`maxat' {
local num = `i'
matrix b = C["`i'._at#`cat3'.`cate'", "b".."ul"]
if `i' == 1 matrix H = b
else matrix H = (H \ b)
}
forval i = 1/`maxat' {
local num = `i'
matrix b = C["`i'._at#`cat4'.`cate'", "b".."ul"]
if `i' == 1 matrix I = b
else matrix I = (I \ b)
}
matrix ca1 = A,F
matrix ca2 = A,G
matrix ca3 = A,H
matrix ca4 = A,I
tempname catt1
svmat ca1, name(`catt1')
tempname catt2
svmat ca2, name(`catt2')
tempname catt3
svmat ca3, name(`catt3')
tempname catt4
svmat ca4, name(`catt4')
if "`confidenceno'" == "" {
twoway ///
(hist `varlist' if `sample' == 1, yaxis(2) `percent' `discrete' xlabel(`min'(`lab')`max') yscale(axis(2) alt) color(gs14) ytitle("`measure' `varlist'", axis(2) size(small)) xtitle(`varlist',size(small))) ///
(line `catt1'2 `catt1'1, yaxis(1) yscale(axis(1) alt) lcolor(navy) ytitle("`estimation' with `level' percent CI", size(small))) ///
(rcap `catt1'7 `catt1'6 `catt1'1, lpattern(solid) yaxis(1) lcolor(navy)) ///
(line `catt2'2 `catt2'1, yaxis(1) lcolor(maroon)) ///
(rcap `catt2'7 `catt2'6 `catt2'1, lpattern(solid) yaxis(1) lcolor(maroon)) ///
(line `catt3'2 `catt3'1, yaxis(1) lcolor(dkgreen)) ///
(rcap `catt3'7 `catt3'6 `catt3'1, lpattern(solid) yaxis(1) lcolor(dkgreen)) ///
(line `catt4'2 `catt4'1, yaxis(1) lcolor(red)) ///
(rcap `catt4'7 `catt4'6 `catt4'1, lpattern(solid) yaxis(1) lcolor(red)) ///
, legend(order(2 "`f`cat1''" 4 "`f`cat2''" 6 "`f`cat3''" 8 "`f`cat4''"))
drop `catt1'* `catt2'* `catt3'* `catt4'*
exit
}
if "`confidenceno'" != "" {
twoway ///
(hist `varlist' if `sample' == 1, yaxis(2) `percent' `discrete' xlabel(`min'(`lab')`max') yscale(axis(2) alt) color(gs14) ytitle("`measure' `varlist'", axis(2) size(small)) xtitle(`varlist',size(small))) ///
(line `catt1'2 `catt1'1, yaxis(1) yscale(axis(1) alt) lcolor(navy) ytitle("`estimation' with `level' percent CI", size(small))) ///
(line `catt2'2 `catt2'1, yaxis(1) lcolor(maroon)) ///
(line `catt3'2 `catt3'1, yaxis(1) lcolor(dkgreen)) ///
(line `catt4'2 `catt4'1, yaxis(1) lcolor(red)) ///
, legend(order(2 "`f`cat1''" 3 "`f`cat2''" 4 "`f`cat3''" 5 "`f`cat4''"))
drop `catt1'* `catt2'* `catt3'* `catt4'*
exit
}
}
}
if `ncat' == 5{
quietly{
matrix C = r(table)'
matrix A = r(at)
local maxat = `points' + 1
matrix A = A[1..`maxat',"`varlist'"]
levelsof `cate', local(values)
matrix input Z = (`values')
forval cat = 1/5{
local cat`cat' = Z[1,`cat']
}
forval i = 1/`maxat' {
local num = `i'
matrix b = C["`i'._at#`cat1'.`cate'", "b".."ul"]
if `i' == 1 matrix F = b
else matrix F = (F \ b)
}
forval i = 1/`maxat' {
local num = `i'
matrix b = C["`i'._at#`cat2'.`cate'", "b".."ul"]
if `i' == 1 matrix G = b
else matrix G = (G \ b)
}
forval i = 1/`maxat' {
local num = `i'
matrix b = C["`i'._at#`cat3'.`cate'", "b".."ul"]
if `i' == 1 matrix H = b
else matrix H = (H \ b)
}
forval i = 1/`maxat' {
local num = `i'
matrix b = C["`i'._at#`cat4'.`cate'", "b".."ul"]
if `i' == 1 matrix I = b
else matrix I = (I \ b)
}
forval i = 1/`maxat' {
local num = `i'
matrix b = C["`i'._at#`cat5'.`cate'", "b".."ul"]
if `i' == 1 matrix J = b
else matrix J = (J \ b)
}
matrix ca1 = A,F
matrix ca2 = A,G
matrix ca3 = A,H
matrix ca4 = A,I
matrix ca5 = A,J
tempname catt1
svmat ca1, name(`catt1')
tempname catt2
svmat ca2, name(`catt2')
tempname catt3
svmat ca3, name(`catt3')
tempname catt4
svmat ca4, name(`catt4')
tempname catt5
svmat ca5, name(`catt5')
if "`confidenceno'" == "" {
twoway ///
(hist `varlist' if `sample' == 1, yaxis(2) `percent' `discrete' xlabel(`min'(`lab')`max') yscale(axis(2) alt) color(gs14) ytitle("`measure' `varlist'", axis(2) size(small)) xtitle(`varlist',size(small))) ///
(line `catt1'2 `catt1'1, yaxis(1) yscale(axis(1) alt) lcolor(navy) ytitle("`estimation' with `level' percent CI", size(small))) ///
(rcap `catt1'7 `catt1'6 `catt1'1, lpattern(solid) yaxis(1) lcolor(navy)) ///
(line `catt2'2 `catt2'1, yaxis(1) lcolor(maroon)) ///
(rcap `catt2'7 `catt2'6 `catt2'1, lpattern(solid) yaxis(1) lcolor(maroon)) ///
(line `catt3'2 `catt3'1, yaxis(1) lcolor(dkgreen)) ///
(rcap `catt3'7 `catt3'6 `catt3'1, lpattern(solid) yaxis(1) lcolor(dkgreen)) ///
(line `catt4'2 `catt4'1, yaxis(1) lcolor(red)) ///
(rcap `catt4'7 `catt4'6 `catt4'1, lpattern(solid) yaxis(1) lcolor(red)) ///
(line `catt5'2 `catt5'1, yaxis(1) lcolor(blue)) ///
(rcap `catt5'7 `catt5'6 `catt5'1, lpattern(solid) yaxis(1) lcolor(blue)) ///
, legend(order(2 "`f`cat1''" 4 "`f`cat2''" 6 "`f`cat3''" 8 "`f`cat4''" 10 "`f`cat5''"))
drop `catt1'* `catt2'* `catt3'* `catt4'* `catt5'*
exit
}
if "`confidenceno'" != "" {
twoway ///
(hist `varlist' if `sample' == 1, yaxis(2) `percent' `discrete' xlabel(`min'(`lab')`max') yscale(axis(2) alt) color(gs14) ytitle("`measure' `varlist'", axis(2) size(small)) xtitle(`varlist',size(small))) ///
(line `catt1'2 `catt1'1, yaxis(1) yscale(axis(1) alt) lcolor(navy) ytitle("`estimation' with `level' percent CI", size(small))) ///
(line `catt2'2 `catt2'1, yaxis(1) lcolor(maroon)) ///
(line `catt3'2 `catt3'1, yaxis(1) lcolor(dkgreen)) ///
(line `catt4'2 `catt4'1, yaxis(1) lcolor(red)) ///
(line `catt5'2 `catt5'1, yaxis(1) lcolor(blue)) ///
, legend(order(2 "`f`cat1''" 3 "`f`cat2''" 4 "`f`cat3''" 5 "`f`cat4''" 6 "`f`cat5''"))
drop `catt1'* `catt2'* `catt3'* `catt4'* `catt5'*
exit
}
}
}
}
}
end