You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

358 lines
14 KiB
Plaintext

*! Version 5 5March2016
************************************************************************************************************
* Traces: Traces of items
* Version 5: March 5, 2016 /*numerical items*/
*
* Historic:
* Version 1 (2003-06-29): Jean-Benoit Hardouin
* Version 2 (2003-07-04): Jean-Benoit Hardouin
* version 3 (2003-07-09): Jean-Benoit Hardouin
* Version 3.1 (2005-06-07): Jean-Benoit Hardouin /*small modifications*/
* Version 3.2: May 27, 2007 /*onlyone option*/
* Version 3.3: October 16, 2012 /*minor modifications*/
* Version 4: January 22, 2015 /*ICC*/
*
* Jean-benoit Hardouin, phD, Assistant Professor
* EA4275-SPHERE "MethodS in Patients-centered outcomes and HEalth ResEarches"
* University of Nantes - Faculty of Pharmaceutical Sciences
* France
* jean-benoit.hardouin@anaqol.org
*
* News about this program :http://www.anaqol.org
* FreeIRT Project website : http://www.freeirt.org
*
* Copyright 2003, 2005, 2007, 2012, 2015, 2016 Jean-Benoit Hardouin
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*
************************************************************************************************************
program define traces, rclass
version 8.0
syntax varlist(numeric min=2) [, COMPare(varlist min=2 max=2) ICC SAVeicc noGraph Score Test Restscore Logistic CI CUMulative REPFiles(string) SCOREFiles(string) RESTSCOREFiles(string) LOGISTICFile(string) noDraw noDRAWComb REPlace ONLYone(string) THResholds(string) Black]
local nbitems : word count `varlist'
tokenize `varlist'
if "`onlyone'"!=""&"`drawcomb'"!="" {
local drawcomb
}
tempvar varscore
qui gen `varscore'=0
label variable `varscore' "Total score"
local scoremax=0
local flag=0
if "`score'"==""&"`restscore'"==""&&"`logistic'"=="" {
local score="score"
}
local modamax=-999
local modamin=999
forvalues i=1/`nbitems' {
qui replace `varscore'=`varscore'+``i''
qui su ``i''
local modamax`i'=r(max)
local modamin`i'=r(min)
if r(min)!=0 {
local flag=1
}
local scoremax=`scoremax'+`modamax`i''
if `modamax`i''!=1 {
local flagbin=0
}
if `modamax`i''>`modamax' {
local modamax=`modamax`i''
}
if `modamin`i''<`modamin' {
local modamin=`modamin`i''
}
}
/*
if `flag'==1 {
di as error "The lower modality of the item must be 0"
exit
}
*/
if "`flagbin'"!=""&"`logistic'"!="" {
di as error "The logistic option is not possible with polytomous items"
exit
}
qui su `varscore'
local maxscore=r(max)
local minscore=r(min)
forvalues i=`minscore'/`maxscore' {
qui count if `varscore'>`=`i'-0.5'&`varscore'<=`=`i'+0.5'
local nscore`i'=r(N)
}
global score
global restscore
global logistic
qui count
local N=r(N)
if `c(matsize)'<`N'&"`saveicc'"!="" {
set matsize `N'
}
if "`score'"!="" {
if "`thresholds'"!="" {
* set trace on
local nbth:word count `thresholds'
forvalues t=1/`nbth' {
local th`t':word `t' of `thresholds'
}
tempname label
local recode min/`th1'=1 `=`th`nbth''+1'/max=`=`nbth'+1'
qui label define `label' 1 "min/`th1'",add
qui label define `label' `=`nbth'+1' "`=`th`nbth''+1'/max",add
forvalues j=2/`nbth' {
local recode `recode' `=`th`=`j'-1''+1'/`th`j''=`j'
qui label define `label' `j' "`=`th`=`j'-1''+1'/`th`j''",add
}
tempname varscore2
qui gen `varscore2'=`varscore'
qui recode `varscore' `recode'
qui label values `varscore' `label'
local nbgroups=`nbth'+1
local minimum=1
}
else {
local nbgroups=`maxscore'
local minimum=0
}
local listicc
forvalues i=1/`nbitems' {
local y`i'
tempvar icc``i''
gen `icc``i'''=0
local listicc `listicc' `icc``i'''
local l=0
forvalues k=`=`modamin`i''+1'/`modamax`i'' {
tempvar propscore`i'`l' tmp
if "`cumulative'"!="" {
qui gen `tmp'=``i''>=`k'&``i''!=.
bysort `varscore' : egen `propscore`i'`l''=mean(`tmp')
label variable `propscore`i'`l'' "Item ``i''>=`k'"
}
else {
qui gen `tmp'=``i''==`k'&``i''!=.
bysort `varscore' : egen `propscore`i'`l''=mean(`tmp')
label variable `propscore`i'`l'' "Item ``i''=`k'"
*di "replace icc``i''=icc``i''+`k'*`propscore`i'`k''"
qui replace `icc``i'''=`icc``i'''+`k'*`propscore`i'`l''
}
local y`i'="`y`i'' `propscore`i'`l''"
local style="solid"
local color="black"
local width="medthick"
if `modamax`i''==1&"`ci'"!="" {
tempvar icscoreminus icscoreplus
forvalues m=1/`maxscore' {
qui count if `varscore'==`m'
local nscore`m'=r(N)
}
qui gen `icscoreminus'=`propscore`i'1'-1.96*sqrt(`propscore`i'1'*(1-`propscore`i'1')/`nscore1')
qui gen `icscoreplus'=`propscore`i'1'+1.96*sqrt(`propscore`i'1'*(1-`propscore`i'1')/`nscore1')
label variable `icscoreminus' "Lower 95% confidence interval"
label variable `icscoreplus' "Upper 95% confidence interval"
local y`i'="`icscoreminus' `icscoreplus' `propscore`i'1'"
local style="dash dash solid"
local color="red red black"
local width="thin thin medthick"
}
if `modamax`i''==1&"`test'"!="" {
qui regress `propscore`i'1' `varscore'
local p=Fden(e(df_m),e(df_r),e(F))
if `p'<0.0001 {
local note="Test: slope=0, p<0.0001"
}
else {
local p=substr("`p'",1,6)
local note="Test: slope=0, p=`p'"
}
}
local ++l
}
if ("``i''"=="`onlyone'"|"`onlyone'"=="")&"`graph'"=="" {
qui graph twoway (line `y`i'' `varscore', clpattern(`style') clcolor(`color') clwidth(`width')) if `varscore'!=0&`varscore'!=`maxscore' , note("`note'") ylabel(0(.25)1) xlabel(`minimum'(1)`nbgroups',valuelabel) name(score`i',replace) title("Trace of the item ``i'' as a function of the score") ytitle("Rate of positive response") `draw' /*areastyle(none)*/
}
if ("``i''"=="`onlyone'"|"`onlyone'"=="")&"`icc'"!=""&"`graph'"==""{
qui graph twoway (line `icc``i''' `varscore', clpattern(`style') clcolor(`color') clwidth(`width')) if `varscore'!=0&`varscore'!=`maxscore' , ylabel(0(.5)`modamax`i'') xlabel(`minimum'(1)`nbgroups',valuelabel) name(icc`i',replace) title("Approximate ICC of the item ``i'' as a function of the score") ytitle("Mean response") `draw' /*areastyle(none)*/
}
global score "$score score`i'"
if "`scorefiles'"!="" {
graph save score`i' `repfiles'\\`scorefiles'``i'' ,`replace'
}
}
if "`saveicc'"!="" {
local listicc `listicc' `varscore'
tempname matscore
mkmat `listicc',matrix(`matscore')
}
if "`thresholds'"!="" {
qui replace `varscore'=`varscore2'
}
}
if "`compare'"!=""&"`nograph'"!="" {
local tmp1 : word 1 of `compare'
local tmp2 : word 2 of `compare'
di "compare `compare' tmp1 `tmp1' tmp2 `tmp2'"
label variable `icc`tmp1'' `tmp1'
label variable `icc`tmp2'' `tmp2'
di "qui graph twoway (line `icc`tmp1'' `varscore', clpattern(`style') clcolor(`color') clwidth(`width')) (line `icc`tmp2'' `varscore', clpattern(`style') clcolor(`color') clwidth(`width')) if `varscore'!=0&`varscore'!=`maxscore' , ylabel(0(.5)`modamax') xlabel(`minimum'(1)`nbgroups',valuelabel) name(score`i',replace) title(Comparison of the ICC of the items `tmp1' and `tmp2') `draw' /*areastyle(none)*/"
qui graph twoway (line `icc`tmp1'' `varscore', clpattern(`style') clcolor(blue red) clwidth(`width')) (line `icc`tmp2'' `varscore', clpattern(`style') clcolor(`color') clwidth(`width')) if `varscore'!=0&`varscore'!=`maxscore' , ylabel(0(.5)`modamax') xlabel(`minimum'(1)`nbgroups',valuelabel) /*name(score`i',replace)*/ title("Comparison of the ICC of the items `tmp1' and `tmp2'") ytitle("Mean response") /*areastyle(none)*/
}
if "`restscore'"!="" {
forvalues i=1/`nbitems' {
local y`i'
tempvar restscore`i'
qui gen `restscore`i''=`varscore'-``i''
label variable `restscore`i'' "Rest score with respect to the item ``i''"
if "`thresholds'"!="" {
* set trace on
local nbth:word count `thresholds'
forvalues t=1/`nbth' {
local th`t':word `t' of `thresholds'
}
tempname label
local recode 0/`th1'=1 `=`th`nbth''+1'/max=`=`nbth'+1'
qui label define `label' 1 "min/`th1'",add
qui label define `label' `=`nbth'+1' "`=`th`nbth''+1'/max",add
forvalues j=2/`nbth' {
local recode `recode' `=`th`=`j'-1''+1'/`th`j''=`j'
qui label define `label' `j' "`=`th`=`j'-1''+1'/`th`j''",add
}
*di "recode `restscore`i'' `recode'"
qui recode `restscore`i'' `recode'
qui label values `restscore`i'' `label'
local nbgroups=`nbth'+1
local minimum=1
}
else {
local nbgroups=`maxscore'
local minimum=0
}
forvalues k=1/`modamax`i'' {
tempvar rtmp proprestscore`i'`k'
if "`cumulative'"!="" {
qui gen `rtmp'=``i''>=`k'&``i''!=.
bysort `restscore`i'': egen `proprestscore`i'`k''=mean(`rtmp')
label variable `proprestscore`i'`k'' "Item ``i''>=`k'"
}
else {
qui gen `rtmp'=``i''==`k'&``i''!=.
bysort `restscore`i'': egen `proprestscore`i'`k''=mean(`rtmp')
label variable `proprestscore`i'`k'' "Item ``i''=`k'"
}
local y`i'="`y`i'' `proprestscore`i'`k''"
local style="solid"
local color="black"
local width="medthick"
if `modamax`i''==1&"`ci'"!="" {
tempvar icrestscoreminus icrestscoreplus
qui su `restscore`i''
local maxrestscore=r(max)
forvalues l=1/`maxrestscore' {
qui count if `restscore`i''==`l'
local nrestscore`i'=r(N)
}
qui gen `icrestscoreminus'=`proprestscore`i'1'-1.96*sqrt(`proprestscore`i'1'*(1-`proprestscore`i'1')/`nrestscore`i'')
qui gen `icrestscoreplus'=`proprestscore`i'1'+1.96*sqrt(`proprestscore`i'1'*(1-`proprestscore`i'1')/`nrestscore`i'')
label variable `icrestscoreminus' "Lower 95% confidence interval"
label variable `icrestscoreplus' "Upper 95% confidence interval"
local y`i'="`icrestscoreminus' `icrestscoreplus' `proprestscore`i'1'"
local style="dash dash solid"
local color="red red black"
local width="thin thin medthick"
}
if `modamax`i''==1&"`test'"!="" {
qui regress `proprestscore`i'1' `varscore'
local p=Fden(e(df_m),e(df_r),e(F))
if `p'<0.0001 {
local note="Test: slope=0, p<0.0001"
}
else {
local p=substr("`p'",1,6)
local note="Test: slope=0, p=`p'"
}
}
}
local restscoremax=`scoremax'-`modamax`i''
if ("``i''"=="`onlyone'"|"`onlyone'"=="")&"`nograph'"!="" {
*tab `proprestscore`i'1' `restscore`i''
qui graph twoway (line `y`i'' `restscore`i'', clpattern(`style') clcolor(`color') clwidth(`width')), note("`note'") ylabel(0(0.25)1) xlabel(`minimum'(1)`nbgroups',valuelabel) name(restscore`i',replace) title("Trace of the item ``i'' as a function of the restscore") ytitle("Rate of positive response") `draw'
}
global restscore "$restscore restscore`i'"
if "`restscorefiles'"!="" {
graph save restscore`i' `repfiles'\\`restscorefiles'``i'' ,`replace'
}
}
}
if "`logistic'"!="" {
forvalues i=1/`nbitems' {
qui logistic ``i'' `varscore'
tempname coef
matrix `coef'=e(b)
local pente`i'=`coef'[1,1]
local intercept`i'=`coef'[1,2]
tempvar logit`i'
qui gen `logit`i''=exp(`intercept`i''+`pente`i''*`varscore')/(1+exp(`intercept`i''+`pente`i''*`varscore'))
label variable `logit`i'' "Item ``i''"
sort `varscore'
global logistic "$logistic `logit`i''"
}
}
if "`drawcomb'"!="" {
local drawcomb="nodraw"
}
*di " (`score'!=&`onlyone'==)&`nograph'!= {"
if ("`score'"!=""&"`onlyone'"=="")&"`nograph'"!="" {
graph combine $score , title("Trace of the items as a function of the score") name(score,replace) `drawcomb'
if "`scorefiles'"!="" {
graph save score `repfiles'\\`scorefiles' ,`replace'
}
}
if ("`restscore'"!=""&"`onlyone'"=="")&"`nograph'"!="" {
graph combine $restscore , title("Trace of the items as a function of the restscores") name(restscore,replace) `drawcomb'
if "`restscorefiles'"!="" {
graph save restscore `repfiles'\\`restscorefiles' ,`replace'
}
}
if ("`logistic'"!="")&"`nograph'"!="" {
graph twoway (line $logistic `varscore'), ylabel(0(0.25)1) xlabel(0(1)`nbitems') title("Logistic traces") ytitle("") name(logistic,replace) `drawcomb'
if "`logisticfile'"!="" {
graph save logistic `repfiles'\\`logisticfile' ,`replace'
}
}
if "`saveicc'"!="" {
return matrix matscore=`matscore'
}
end