/****************************************************************************************************************************************
Imputation SAS code for Exemplar 6
tHIS ONE IMPUTES FROM THE big DATA FILE
NOTE that you need to run the program ex6formats.sas before you can open the data
 files here
ALSO to run the IVEWARE software you need to close all advanced editor windows
and use only the old program editor
CHANGE the libname to shere your data sets are stored
proc contents data=ex6.ex6DETAIL short position;run;
****************************************************************************************************************************************/
libname ex6 'C:\Documents and Settings\gillian raab\My Documents\aprojects\peaslaptop\ex6datafiles\data' ;
/*------------------------------------------------------------------
Now the analysis of detailed questions using the IVEWARE macros
 details of what the commands mean are in the user guide on the IVE
 site. Here all variables are treated as categorical and each has 8
 categories.
--------------------------------------------------------------------*/
title 'Imputation of detailed questions (subsets) with IVEWARE';
/*-------------------------------------------------------------------
This took almost a day.
And it only worked afer various preliminary investigations that
set the stepwise procedures to have a max of 10 variables and to
stop the iterations at 10
Otherwise it would have taken months
-----------------------------------------------------------------------*/
data temp;
set ex6.ex6det;
keep caseid yayars02 ybyars02 ycyars02 ydyars02 yeyars02 yfyars02 gender ;
run;
%impute(name=ivesetup3,
    dir=C:\Documents and Settings\gillian raab\My Documents\aprojects\peaslaptop\ex6datafiles\program_code
     ,setup=new);
   title Multiple imputation detailed questions (subsets) categorical;
  datain temp;
  dataout impout all ;
  default categorical;
  maxlogi 50;
iterations 5;
   multiples 1;
   transfer caseid ;
   print coef;
run ;

%include "C:\Documents and Settings\gillian raab\My Documents\aprojects\peaslaptop\ex6datafiles\program_code\checkimp_macro.sas";

%checkimp(impout,ex6.ex6det)
/*---------------------------- next group--------------------*/
data temp;
set ex6.ex6detail;
keep caseid yayshp02  ybyshp02  ycyshp02  ydyshp02  yeyshp02
 yfyshp02 yayscl02  ybyscl02  ycyscl02  ydyscl02
yayhom02 ybyhom02 ycyhom02 ydyhom02 gender ;
run;
%impute(name=ivesetup3,
    dir=C:\Documents and Settings\gillian raab\My Documents\aprojects\peaslaptop\ex6datafiles\program_code
     ,setup=new);
   title Multiple imputation prevalence only;
  datain temp;
  dataout imp2 all ;
  default categorical;
  iterations 5;
   multiples 2;
   maxlogi 50;
   print coef;
   transfer caseid ;
  seed 1769;
run ;
/*---------------------------- next group--------------------*/
data temp;
set ex6.ex6detail;
keep caseid yaycbk02 ybycbk02  ycycbk02  ydycbk02  yeycbk02  yfycbk02
 yayjrd02   ybyjrd02  ycyjrd02   ydyjrd02  yeyjrd02  yfyjrd02
 yayrob02 ybyrob02 gender ;
run;
%impute(name=ivesetup3,
    dir=C:\Documents and Settings\gillian raab\My Documents\aprojects\peaslaptop\ex6datafiles\program_code
     ,setup=new);
   title Multiple imputation prevalence only;
  datain temp;
  dataout imp3 all ;
  default categorical;
multiples 10;
print coef;
   transfer caseid ;
  seed 170;
run ;
/*---------------------------- next group--------------------*/
data temp;
set ex6.ex6detail;
keep caseid yaywep02  ybywep02  ycywep02  ydywep02  yeywep02  yfywep02
ycydrg02  ydydrg02 yeydrg02 yfydrg02
yayhit02 ybyhit02 ycyhit02 ydyhit02 yeyhit02 yfyhit02 gender ;
run;
%impute(name=ivesetup3,
    dir=C:\Documents and Settings\gillian raab\My Documents\aprojects\peaslaptop\ex6datafiles\program_code
     ,setup=new);
   title Multiple imputation prevalence only;
  datain temp;
  dataout imp4 all ;
  default categorical;
multiples 10;
print coef;
   transfer caseid ;
  seed 171;
run ;
/*---------------------------- next group--------------------*/
data temp;
set ex6.ex6detail;
keep caseid yayvnd02 ybyvnd02  ycyvnd02  ydyvnd02  yeyvnd02  yfyvnd02
 yaygrf02  ybygrf02  ycygrf02 ydygrf02  yeygrf02
yayars02 ybyars02 ycyars02 ydyars02 yeyars02 yfyars02 gender ;
run;
%impute(name=ivesetup3,
    dir=C:\Documents and Settings\gillian raab\My Documents\aprojects\peaslaptop\ex6datafiles\program_code
     ,setup=new);
   title Multiple imputation prevalence only;
  datain temp;
  dataout imp5 all ;
  default categorical;
multiples 10;
print coef;
   transfer caseid ;
  seed 1238;
run ;
/*---------------------------- next group--------------------*/
data temp;
set ex6.ex6detail;
keep caseid  ycyrab02 ydyrab02 yeyrab02 yfyrab02 gender ;
run;
%impute(name=ivesetup3,
    dir=C:\Documents and Settings\gillian raab\My Documents\aprojects\peaslaptop\ex6datafiles\program_code
     ,setup=new);
   title Multiple imputation prevalence only;
  datain temp;
  dataout imp6 all ;
  default categorical;
multiples 10;
print coef;
   transfer caseid ;
  seed 1988;
run ;
/*---------------------------- next group--------------------*/
data temp;
set ex6.ex6detail;
keep caseid   yayhbk02  ybyhbk02 ycyhbk02 ydyhbk02 yeyhbk02 yfyhbk02
 ycyrob02 ydyrob02 yeyrob02 yfyrob02 gender ;
run;
%impute(name=ivesetup3,
    dir=C:\Documents and Settings\gillian raab\My Documents\aprojects\peaslaptop\ex6datafiles\program_code
     ,setup=new);
   title Multiple imputation prevalence only;
  datain temp;
  dataout imp7 all ;
  default categorical;
multiples 10;
print coef;
   transfer caseid ;
  seed 16709;
run ;
/*---------------------------- next group--------------------*/
data temp;
set ex6.ex6detail;
keep caseid  ybypet02  ycypet02
ydypet02  yeypet02  yfypet02  gender ;
run;
%impute(name=ivesetup3,
    dir=C:\Documents and Settings\gillian raab\My Documents\aprojects\peaslaptop\ex6datafiles\program_code
     ,setup=new);
   title Multiple imputation prevalence only;
  datain temp;
  dataout imp8 all ;
  default categorical;
multiples 10;
print coef;
   transfer caseid ;
  seed 1338;
run ;
/*---------------------------- next group--------------------*/
data temp;
set ex6.ex6detail;
keep caseid yaybop02  ybybop02  ycybop02
ydybop02  yeybop02  yfybop02 gender ;
run;
%impute(name=ivesetup3,
    dir=C:\Documents and Settings\gillian raab\My Documents\aprojects\peaslaptop\ex6datafiles\program_code
     ,setup=new);
   title Multiple imputation prevalence only;
  datain temp;
  dataout imp9 all ;
  default categorical;
multiples 10;
print coef;
   transfer caseid ;
  seed 13568;
run ;


/*---------------------------- now calculate prevalence etc from the imputed data----------*/
data ex6.fromdetail (keep=dprev: dvol: dvar: caseid gender _mult_);
set ex6.detailimp;
/*---------------------------- sweep 1---------------------------------------*/
*Variety of delinquency at sweep one (n=15);
array offv1 yaybus02 yaybop02 yayshp02 yayjrd02 yayscl02 yaywep02 yaygrf02 yayrob02
  yayvnd02 yayhbk02 yayhom02 yaycbk02 yayars02 yayhit02 yayskv02;
        dprev1=0;
        dvar1=0;
        dvol1=0;
do i=1 to 15;
        if offv1[i]>0 then do dprev1=1; dvar1=dvar1+1; end;
        if offv1[i]=7 then offv1[i]=11; * code to match SPSS;
        dvol1=offv1[i]+dvol1; ;
end;
/*---------------------------- sweep 2---------------------------------------*/
*Variety of delinquency at sweep two (n=16);
array offv2 ybybus02 ybybop02 ybyshp02 ybyjrd02 ybyscl02 ybywep02 ybygrf02 ybyrob02
  ybyvnd02 ybyhbk02 ybyhom02 ybycbk02 ybyars02 ybyhit02 ybyskv02 ybypet02 ;
        dprev2=0;
        dvar2=0;
        dvol2=0;
do i=1 to 16;
        if offv2[i]>0 then do dprev2=1; dvar2=dvar2+1; end;
        if offv2[i]=7 then offv2[i]=11; * code to match SPSS;
        dvol2=offv2[i]+dvol2;
end;
/*---------------------------- sweep 3---------------------------------------*/
*Variety of delinquency at sweep three (n=18);
array offv3 ycybus02 ycybop02 ycyshp02 ycyjrd02 ycyscl02 ycywep02 ycygrf02 ycyrob02
  ycyvnd02 ycyhbk02 ycyhom02 ycycbk02 ycyars02 ycyhit02 ycyskv02 ycypet02 ycydrg02 ycyrab02 ;
        dprev3=0;
        dvar3=0;
        dvol3=0;
do i=1 to 18;
        if offv3[i]>0 then do dprev3=1; dvar3=dvar3+1; end;
        if offv3[i]=7 then offv3[i]=11; * code to match SPSS;
        dvol3=offv3[i]+dvol3;
end;
/*---------------------------- sweep 4---------------------------------------*/
*Variety of delinquency at sweep three (n=18);
array offv4 ydybus02 ydybop02 ydyshp02 ydyjrd02 ydyscl02 ydywep02 ydygrf02 ydyrob02
  ydyvnd02 ydyhbk02 ydyhom02 ydycbk02 ydyars02 ydyhit02 ydyskv02
ydypet02 ydydrg02 ydyrab02;
        dprev4=0;
        dvar4=0;
        dvol4=0;
do i=1 to 18;
        if offv4[i]>0 then do dprev4=1; dvar4=dvar4+1; end;
        if offv4[i]=7 then offv4[i]=11; * code to match SPSS;
        dvol4=offv4[i]+dvol4;
end;
/*---------------------------- sweep 5---------------------------------------*/
*Variety of delinquency at sweep five (n=17);
array offv5 yeybus02 yeybop02 yeyshp02 yeyjrd02 yeywep02 yeygrf02 yeyrob02
  yeyvnd02 yeyhbk02 yeycbk02 yeyars02 yeyhit02 yeyskv02 yeypet02
yeydrg02 yeyrab02 yeyrst02;
        dprev5=0;
        dvar5=0;
        dvol5=0;
do i=1 to 17;
        if offv5[i]>0 then do dprev5=1; dvar5=dvar5+1; end;
                if offv5[i]=7 then offv5[i]=11; * code to match SPSS;
        dvol5=offv5[i]+dvol5;
end;
/*---------------------------- sweep 6---------------------------------------*/
*Variety of delinquency at sweep six (n=18);
array offv6 yfybop02 yfyshp02 yfyjrd02 yfywep02 yfyrob02
  yfyvnd02 yfyhbk02 yfycbk02 yfyars02 yfyhit02 yfypet02 yfyskv02 yfydrg02 yfyrab02 yfyrst02 yfyrst22 yfybft02 yfyfrd02;
        dprev6=0;
        dvar6=0;
        dvol6=0;
do i=1 to 18;
        if offv6[i]>0 then do dprev6=1; dvar6=dvar6+1; end;
                if offv6[i]=7 then offv6[i]=11; * code to match SPSS;
        dvol6=offv6[i]+dvol6;
end;
run;
/*----now merge this with some variables from the original file-----*/
data temp1;
set ex6.ex6;
keep caseid sector szindep;
run;
proc sort data=ex6.fromdetail ;
by caseid;run;
proc sort data=temp1 by caseid;run;
data ex6.fromdetail;
merge ex6.fromdetail temp1;
by caseid;
rename _Mult_ =_imputation_; * to use proc MI;
run;
/*-----------------------------------------------------------------------------------
 POST IMPUTATION PROCEDURES
The imputation variable needs to be called _imputation_ renamed above
These are always 2 stage procedures
-------------------------------------------------------------------------------------*/
proc sort data=ex6.fromdetail; by _imputation_; run;
proc means data=ex6.fromdetail;
var dprev6 dvol6 dvar6;
by _imputation_;
run;
/*proc freq data=ex6.detailimp;
/*table yaywep02  ybywep02  ycywep02  ydywep02  yeywep02  yfywep02
ycydrg02  ydydrg02 yeydrg02 yfydrg02
yayhit02 ybyhit02 ycyhit02 ydyhit02 yeyhit02 yfyhit02
yayvnd02 ybyvnd02  ycyvnd02  ydyvnd02  yeyvnd02  yfyvnd02
 yaygrf02  ybygrf02  ycygrf02 ydygrf02  yeygrf02  gender
yayars02 ybyars02 ycyars02 ydyars02 yeyars02 yfyars02
 ycyrab02 ydyrab02 yeyrab02 yfyrab02
 yayhbk02  ybyhbk02 ycyhbk02 ydyhbk02 yeyhbk02 yfyhbk02
 ycyrob02 ydyrob02 yeyrob02 yfyrob02
ybypet02  ycypet02 ydypet02  yeypet02  yfypet02
yaybop02  ybybop02  ycybop02 ydybop02  yeybop02  yfybop02;
table yaybus02 ybybus02 ycybus02 ydybus02 yeybus02
yayskv02  ybyskv02 ycyskv02 ydyskv02 yeyskv02 yfyskv02
yeyrst02 yfyrst02 yfyrst22 yfybft02 yfyfrd02
 yayshp02  ybyshp02  ycyshp02  ydyshp02  yeyshp02
 yfyshp02 yayscl02  ybyscl02  ycyscl02  ydyscl02
yayhom02 ybyhom02 ycyhom02 ydyhom02
 yaycbk02 ybycbk02  ycycbk02  ydycbk02  yeycbk02  yfycbk02
 yayjrd02   ybyjrd02  ycyjrd02   ydyjrd02  yeyjrd02  yfyjrd02
 yayrob02 ybyrob02;
;
run;*/
/*---------------------------- get means using corr--------------------------------*/
PROC CORR DATA=ex6.fromdetail COV OUT=OUTCOV(TYPE=COV) NOCORR noprint ;
VAR   dprev1-dprev6  dvol1-dvol6 dvar1-dvar6 ;
BY _IMPUTATION_;
run;
PROC MIANALYZE data=outcov ;
VAR  dprev1-dprev6  dvol1-dvol6 dvar1-dvar6 ;
RUN;
/*------------------- logistic regression and MIANALYZE---------------------------*/
proc logistic data=ex6.fromdetail outest=outreg covout descending noprint;
by _imputation_;
class  gender szindep sector;
model dprev6=  gender szindep sector ;
run;

proc contents data= outreg short;run;* use this to find names of contrasts;


PROC MIANALYZE data=outreg edf=4325 ; * these are the residual degrees of freedom from the model;
 var GENDERFemale  SZINDEPManual_high_depr
sectorBehavioural sectorIndependent sectorSpecial  ;
run;
 /*--We can compare with the complete data analysis here---------*/

proc logistic data=ex6.ex6 descending ;
class  gender szindep sector;
model dprev6=  gender szindep sector ;
run;
/*---------------- now a check to compare prevalence by deprivation with original data---*/
proc means data=ex6.fromdetail;
class szindep;
var dprev6;
run;
proc means data=ex6.ex6;
class szindep;
var dprev6;
run;
/*------------------------------------------------------------
Now analyses with proc MI that assume all variables are
multivariate normal.
Upper bounds are set on all variables to prevent high values
But lower bounds are not set because experience showed that
this resulted in a downward bias that gave too few responses
in the zero category
---------------------------------------------------------------*/
proc contents short data=ex6.ex6detail; * this just to get list of variables;
run;
 data temp1;
 set ex6.ex6detail;
keep yaybus02 ybybus02 ycybus02 ydybus02 yeybus02
yayskv02  ybyskv02 ycyskv02 ydyskv02 yeyskv02 yfyskv02
yeyrst02 yfyrst02 yfyrst22 yfybft02 yfyfrd02
 yayshp02  ybyshp02  ycyshp02  ydyshp02  yeyshp02
 yfyshp02 yayscl02  ybyscl02  ycyscl02  ydyscl02
yayhom02 ybyhom02 ycyhom02 ydyhom02
 yaycbk02 ybycbk02  ycycbk02  ydycbk02  yeycbk02  yfycbk02
 yayjrd02   ybyjrd02  ycyjrd02   ydyjrd02  yeyjrd02  yfyjrd02
 yayrob02 ybyrob02
 gender;
run;
proc mi data=temp1  OUT=imp1 nimpute=10  round=1;
VAR  yaybus02 ybybus02 ycybus02 ydybus02 yeybus02
yayskv02  ybyskv02 ycyskv02 ydyskv02 yeyskv02 yfyskv02
yeyrst02 yfyrst02 yfyrst22 yfybft02 yfyfrd02
 yayshp02  ybyshp02  ycyshp02  ydyshp02  yeyshp02
 yfyshp02 yayscl02  ybyscl02  ycyscl02  ydyscl02
yayhom02 ybyhom02 ycyhom02 ydyhom02
 yaycbk02 ybycbk02  ycycbk02  ydycbk02  yeycbk02  yfycbk02
 yayjrd02   ybyjrd02  ycyjrd02   ydyjrd02  yeyjrd02  yfyjrd02
 yayrob02 ybyrob02
 gender ;
RUN;
/*---------- run this to check for extreme values--------*/
proc means data=imp1; run;
/*---- now pull values <0 or >7 into range------------*/
data imp1;
set imp1;
array vars caseid yaybus02 ybybus02 ycybus02 ydybus02 yeybus02
yayskv02  ybyskv02 ycyskv02 ydyskv02 yeyskv02 yfyskv02
yeyrst02 yfyrst02 yfyrst22 yfybft02 yfyfrd02
 yayshp02  ybyshp02  ycyshp02  ydyshp02  yeyshp02
 yfyshp02 yayscl02  ybyscl02  ycyscl02  ydyscl02
yayhom02 ybyhom02 ycyhom02 ydyhom02
 yaycbk02 ybycbk02  ycycbk02  ydycbk02  yeycbk02  yfycbk02
 yayjrd02   ybyjrd02  ycyjrd02   ydyjrd02  yeyjrd02  yfyjrd02
 yayrob02 ybyrob02
;
do over vars; if vars<0 then vars=0; if vars>7 then vars=7; end;
run;
data temp2;
 set ex6.ex6detail;
keep  caseid yaywep02  ybywep02  ycywep02  ydywep02  yeywep02  yfywep02
ycydrg02  ydydrg02 yeydrg02 yfydrg02
yayhit02 ybyhit02 ycyhit02 ydyhit02 yeyhit02 yfyhit02
yayvnd02 ybyvnd02  ycyvnd02  ydyvnd02  yeyvnd02  yfyvnd02
 yaygrf02  ybygrf02  ycygrf02 ydygrf02  yeygrf02  gender
yayars02 ybyars02 ycyars02 ydyars02 yeyars02 yfyars02
 ycyrab02 ydyrab02 yeyrab02 yfyrab02
 yayhbk02  ybyhbk02 ycyhbk02 ydyhbk02 yeyhbk02 yfyhbk02
 ycyrob02 ydyrob02 yeyrob02 yfyrob02
ybypet02  ycypet02 ydypet02  yeypet02  yfypet02
yaybop02  ybybop02  ycybop02 ydybop02  yeybop02  yfybop02
gender;
run;
proc mi data=temp2  OUT=imp2 nimpute=10  round=1;
VAR
 yaywep02  ybywep02  ycywep02  ydywep02  yeywep02  yfywep02
ycydrg02  ydydrg02 yeydrg02 yfydrg02
yayhit02 ybyhit02 ycyhit02 ydyhit02 yeyhit02 yfyhit02
yayvnd02 ybyvnd02  ycyvnd02  ydyvnd02  yeyvnd02  yfyvnd02
 yaygrf02  ybygrf02  ycygrf02 ydygrf02  yeygrf02  gender
yayars02 ybyars02 ycyars02 ydyars02 yeyars02 yfyars02
 ycyrab02 ydyrab02 yeyrab02 yfyrab02
 yayhbk02  ybyhbk02 ycyhbk02 ydyhbk02 yeyhbk02 yfyhbk02
 ycyrob02 ydyrob02 yeyrob02 yfyrob02
ybypet02  ycypet02 ydypet02  yeypet02  yfypet02
yaybop02  ybybop02  ycybop02 ydybop02  yeybop02  yfybop02
gender ;
RUN;
proc means data=imp2; run;
/*---- now pull values <0 or >7 into range------------*/
data imp2;
set imp2;
array vars
 yaywep02  ybywep02  ycywep02  ydywep02  yeywep02  yfywep02
ycydrg02  ydydrg02 yeydrg02 yfydrg02
yayhit02 ybyhit02 ycyhit02 ydyhit02 yeyhit02 yfyhit02
yayvnd02 ybyvnd02  ycyvnd02  ydyvnd02  yeyvnd02  yfyvnd02
 yaygrf02  ybygrf02  ycygrf02 ydygrf02  yeygrf02  gender
yayars02 ybyars02 ycyars02 ydyars02 yeyars02 yfyars02
 ycyrab02 ydyrab02 yeyrab02 yfyrab02
 yayhbk02  ybyhbk02 ycyhbk02 ydyhbk02 yeyhbk02 yfyhbk02
 ycyrob02 ydyrob02 yeyrob02 yfyrob02
ybypet02  ycypet02 ydypet02  yeypet02  yfypet02
yaybop02  ybybop02  ycybop02 ydybop02  yeybop02  yfybop02;
do over vars; if vars<0 then vars=0; if vars>7 then vars=7; end;
run;
proc freq data=imp2; * checking this one since it had some very high imputed values;
table _imputation_ *yeybop02/nopercent nocol;
run;
proc freq data=ex6.ex6detail;
table yeybop02;
run;
/*------- now merge two imputed data sets togther--------*/
data ex6.detailimpmi;
merge imp1 imp2;
run;
/*---------------------------- now calculate prevalence etc from the imputed data----------*/
data ex6.fromdetailmi (keep=dprev: dvol: dvar: caseid gender  _imputation_);
set ex6.detailimpmi;
/*---------------------------- sweep 1---------------------------------------*/
*Variety of delinquency at sweep one (n=15);
array offv1 yaybus02 yaybop02 yayshp02 yayjrd02 yayscl02 yaywep02 yaygrf02 yayrob02
  yayvnd02 yayhbk02 yayhom02 yaycbk02 yayars02 yayhit02 yayskv02;
        dprev1=0;
        dvar1=0;
        dvol1=0;
do i=1 to 15;
        if offv1[i]>0 then do dprev1=1; dvar1=dvar1+1; end;
        if offv1[i]=7 then offv1[i]=11; * code to match SPSS;
        dvol1=offv1[i]+dvol1; ;
end;
/*---------------------------- sweep 2---------------------------------------*/
*Variety of delinquency at sweep two (n=16);
array offv2 ybybus02 ybybop02 ybyshp02 ybyjrd02 ybyscl02 ybywep02 ybygrf02 ybyrob02
  ybyvnd02 ybyhbk02 ybyhom02 ybycbk02 ybyars02 ybyhit02 ybyskv02 ybypet02 ;
        dprev2=0;
        dvar2=0;
        dvol2=0;
do i=1 to 16;
        if offv2[i]>0 then do dprev2=1; dvar2=dvar2+1; end;
        if offv2[i]=7 then offv2[i]=11; * code to match SPSS;
        dvol2=offv2[i]+dvol2;
end;
/*---------------------------- sweep 3---------------------------------------*/
*Variety of delinquency at sweep three (n=18);
array offv3 ycybus02 ycybop02 ycyshp02 ycyjrd02 ycyscl02 ycywep02 ycygrf02 ycyrob02
  ycyvnd02 ycyhbk02 ycyhom02 ycycbk02 ycyars02 ycyhit02 ycyskv02 ycypet02 ycydrg02 ycyrab02 ;
        dprev3=0;
        dvar3=0;
        dvol3=0;
do i=1 to 18;
        if offv3[i]>0 then do dprev3=1; dvar3=dvar3+1; end;
        if offv3[i]=7 then offv3[i]=11; * code to match SPSS;
        dvol3=offv3[i]+dvol3;
end;
/*---------------------------- sweep 4---------------------------------------*/
*Variety of delinquency at sweep three (n=18);
array offv4 ydybus02 ydybop02 ydyshp02 ydyjrd02 ydyscl02 ydywep02 ydygrf02 ydyrob02
  ydyvnd02 ydyhbk02 ydyhom02 ydycbk02 ydyars02 ydyhit02 ydyskv02
ydypet02 ydydrg02 ydyrab02;
        dprev4=0;
        dvar4=0;
        dvol4=0;
do i=1 to 18;
        if offv4[i]>0 then do dprev4=1; dvar4=dvar4+1; end;
        if offv4[i]=7 then offv4[i]=11; * code to match SPSS;
        dvol4=offv4[i]+dvol4;
end;
/*---------------------------- sweep 5---------------------------------------*/
*Variety of delinquency at sweep five (n=17);
array offv5 yeybus02 yeybop02 yeyshp02 yeyjrd02 yeywep02 yeygrf02 yeyrob02
  yeyvnd02 yeyhbk02 yeycbk02 yeyars02 yeyhit02 yeyskv02 yeypet02
yeydrg02 yeyrab02 yeyrst02;
        dprev5=0;
        dvar5=0;
        dvol5=0;
do i=1 to 17;
        if offv5[i]>0 then do dprev5=1; dvar5=dvar5+1; end;
                if offv5[i]=7 then offv5[i]=11; * code to match SPSS;
        dvol5=offv5[i]+dvol5;
end;
/*---------------------------- sweep 6---------------------------------------*/
*Variety of delinquency at sweep six (n=18);
array offv6 yfybop02 yfyshp02 yfyjrd02 yfywep02 yfyrob02
  yfyvnd02 yfyhbk02 yfycbk02 yfyars02 yfyhit02 yfypet02 yfyskv02 yfydrg02 yfyrab02 yfyrst02 yfyrst22 yfybft02 yfyfrd02;
        dprev6=0;
        dvar6=0;
        dvol6=0;
do i=1 to 18;
        if offv6[i]>0 then do dprev6=1; dvar6=dvar6+1; end;
                if offv6[i]=7 then offv6[i]=11; * code to match SPSS;
        dvol6=offv6[i]+dvol6;
end;
run;
/*----now merge this with some variables from the original file-----*/
data temp1;
set ex6.ex6;
keep caseid sector szindep;
run;
proc sort data=ex6.fromdetailmi ;
by caseid;run;
proc sort data=temp1; by caseid;run;
data ex6.fromdetailmi;
merge ex6.fromdetailmi temp1;
by caseid;
run;
/*-----------------------------------------------------------------------------------
 POST IMPUTATION PROCEDURES
The imputation variable needs to be called _imputation_ renamed above
These are always 2 stage procedures
-------------------------------------------------------------------------------------*/
proc sort data=ex6.fromdetailmi; by _imputation_; run;
/*---------------------------- get means using corr--------------------------------*/
PROC CORR DATA=ex6.fromdetailmi COV OUT=OUTCOV(TYPE=COV) NOCORR noprint ;
VAR   dprev1-dprev6  dvol1-dvol6 dvar1-dvar6 ;
where gender=1; * boys only;
BY _IMPUTATION_;
run;
PROC MIANALYZE data=outcov ;
title 'MI detailed imputation boys';
VAR  dprev1-dprev6  dvol1-dvol6 dvar1-dvar6 ;
RUN;
PROC CORR DATA=ex6.fromdetailmi COV OUT=OUTCOV(TYPE=COV) NOCORR noprint ;
VAR   dprev1-dprev6  dvol1-dvol6 dvar1-dvar6 ;
where gender=2; * girls only;
BY _IMPUTATION_;
run;
PROC MIANALYZE data=outcov ;
title 'MI detailed imputation girls';
VAR  dprev1-dprev6  dvol1-dvol6 dvar1-dvar6 ;
RUN;
/*------------------- logistic regression and MIANALYZE---------------------------*/
proc logistic data=ex6.fromdetailmi outest=outreg covout descending noprint;
by _imputation_;
class  gender szindep sector;
model dprev6=  gender szindep sector ;
run;

proc contents data= outreg short;run;* use this to find names of contrasts;


PROC MIANALYZE data=outreg edf=4325 ; * these are the residual degrees of freedom from the model;
 var GENDER1  SZINDEPManual_high_depr
sectorBehavioural sectorIndependent sectorSpecial  ;
run;
 /*--We can compare with the complete data analysis here---------*/

proc logistic data=ex6.ex6 descending ;
class  gender szindep sector;
model dprev6=  gender szindep sector ;
run;
/*---------------- now a check to compare prevalence by deprivation with original data---*/
proc means data=ex6.fromdetail;
class szindep;
var dprev6;
run;
proc means data=ex6.ex6;
/****************************************************************************************************************************************
Imputation SAS code for Exemplar 6

THESE ARE TRIALS THAT DID NOT WORK
The output from the program showed this in all cases,
but it needed to be checked through carefully.

ALSO to run the IVEWARE software you need to close all advanced editor windows
and use only the old program editor
CHANGE the libname to shere your data sets are stored
proc contents data=ex6.ex6DETAIL short position;run;
****************************************************************************************************************************************/
libname ex6 'C:\Documents and Settings\gillian raab\My Documents\aprojects\peaslaptop\ex6datafiles\data' ;
/*------------------------------------------------------------------
Now the analysis of detailed questions using the IVEWARE macros
 details of what the commands mean are in the user guide on the IVE
 site.
--------------------------------------------------------------------*/
title 'Imputation of detailed questions with IVEWARE';
/*-------------------------------------------------------------------
Notes on IVEWARE trials on detailed questions that did not work

1 Defining the variables as CATEGORICAL with 8 classes (0 to 7)
  was too big a program to be be able to include all variables.
  Restricting to a maximum of 10 predictors seemed to work OK
  despite some warning messages until I noticed that some sparse questions
  had imputed to give almost all the missing data to the highest category.

2. Then tried dividing them into subgroups. But this could have just the same
    problems especially for sparse variables.

3 Defining variables as MIXED (0 plus regression for those with events)
      with no bounds gave many huge values + or -ve so all either 0 or 7 when truncated

4 MIXED variables with <=7 gave either all imputed at 7 or all at zero (once rounded etc)

5 MIXED with both bounds on gave almost everything predicted to zeros

6 Restricting to 10 predictors helped a bit but still several variables going to all zeros
  or all 1s

7 Defining a COUNT variable with no limits pulling the highest values back to 7
     gave rather too many in the highest category, though it seemed not bad.

8 Counts with limits <=7 seemed even better will try this a bit more. This is in the
   program EX6DET.SAS

The code for each of these efforts is given below
-----------------------------------------------------------------------*/
/*-------------------------------------------------------------------
Attempt 1
This took almost a day, so don't try it unless you are about to go away
for a bit.
And it only worked afer various preliminary investigations that
set the stepwise procedures to have a max of 10 variables and to
stop the iterations at 10, which is a bit dodgy. It does give warnings.
Otherwise it would have taken months
-----------------------------------------------------------------------*/
%impute(name=ivesetup3,
    dir=C:\Documents and Settings\gillian raab\My Documents\aprojects\peaslaptop\ex6datafiles\program_code
     ,setup=new);
   title Attempt one all categorical;
  datain EX6.ex6DETAIL;
  dataout imp1 all ;
  default categorical;
   transfer caseid ;
maxpred 10;
maxlogi 10;
  iterations 5;
  multiples 5;
 seed 1768;
run ;
/*----------------------------------------------------------------
Attempt 2
Smaller categorical model
This one fails too in a similar way to 1  and takes less time.
------------------------------------------------------------------*/
data temp;
set ex6.ex6det;
keep caseid yayars02 ybyars02 ycyars02 ydyars02 yeyars02 yfyars02 gender ;
run;
%impute(name=ivesetup3,
    dir=C:\Documents and Settings\gillian raab\My Documents\aprojects\peaslaptop\ex6datafiles\program_code
     ,setup=new);
   title Multiple imputation detailed questions (subsets) categorical;
  datain temp;
  dataout imp2 all ;
  default categorical;
  maxlogi 50;
iterations 5;
   multiples 1;
   transfer caseid ;
   print coef;
run ;
/*----------------------------------------------------------------
Attempt 2
Mixed (zeros and normal when non-zero) .
------------------------------------------------------------------*/
%impute(name=ivesetup3,
    dir=C:\Documents and Settings\gillian raab\My Documents\aprojects\peaslaptop\ex6datafiles\program_code
     ,setup=new);
   title Attempt one all categorical;
  datain EX6.ex6DETAIL;
  dataout imp3 all ;
  default mixed;
   transfer caseid ;
  iterations 5;
  multiples 1;
 seed 1768;
run ;
data imp3; set imp3; * need to sort out data after;
array all _numeric_;
gender=100+gender; * so that this 1/2 var is not affected;
do over all;
all=round(all);
if all<0 then all=0;
if all>7 and all<100 then all=7;
gender=gender-100;
end;
run;

%checkimp(imp3,ex6.ex6det);

%impute(name=ivesetup3,
    dir=C:\Documents and Settings\gillian raab\My Documents\aprojects\peaslaptop\ex6datafiles\program_code
     ,setup=new);
   title Multiple imputation prevalence only;
  datain ex6.ex6det;
  dataout impout all ;
  default count;
bounds YAYARS02(>0,<=7) YAYBOP02(>0,<=7)YAYBUS02(>0,<=7) YAYCBK02(>0,<=7) YAYGRF02(>0,<=7) YAYHBK02(>0,<=7) YAYHIT02(>0,<=7) YAYHOM02(>0,<=7) YAYJRD02(>0,<=7)
  YAYROB02(>0,<=7) YAYSCL02(>0,<=7) YAYSHP02(>0,<=7) YAYSKV02(>0,<=7) YAYVND02(>0,<=7) YAYWEP02(>0,<=7) YBYARS02(>0,<=7) YBYBOP02(>0,<=7) YBYBUS02(>0,<=7) YBYCBK02(>0,<=7)
   YBYGRF02(>0,<=7) YBYHBK02(>0,<=7) YBYHIT02(>0,<=7) YBYHOM02(>0,<=7) YBYJRD02(>0,<=7) YBYPET02(>0,<=7) YBYROB02(>0,<=7) YBYSCL02(>0,<=7)
  YBYSHP02(>0,<=7) YBYSKV02(>0,<=7) YBYVND02(>0,<=7) YBYWEP02(>0,<=7) YCYARS02(>0,<=7) YCYBOP02(>0,<=7) YCYBUS02(>0,<=7) YCYCBK02(>0,<=7)
 YCYDRG02(>0,<=7) YCYGRF02(>0,<=7) YCYHBK02(>0,<=7) YCYHIT02(>0,<=7) YCYHOM02(>0,<=7) YCYJRD02(>0,<=7) YCYPET02(>0,<=7) YCYRAB02(>0,<=7) YCYROB02(>0,<=7)
 YCYSCL02(>0,<=7) YCYSHP02(>0,<=7) YCYSKV02(>0,<=7) YCYVND02(>0,<=7) YCYWEP02(>0,<=7) YDYARS02(>0,<=7) YDYBOP02(>0,<=7) YDYBUS02(>0,<=7) YDYCBK02(>0,<=7) YDYDRG02(<=7) YDYGRF02(<=7)
  YDYHBK02(>0,<=7) YDYHIT02(>0,<=7)  YDYHOM02(>0,<=7) YDYJRD02(>0,<=7) YDYPET02(>0,<=7) YDYRAB02(>0,<=7) YDYROB02(>0,<=7)
 YDYSCL02(>0,<=7) YDYSHP02(>0,<=7) YDYSKV02(>0,<=7) YDYVND02(>0,<=7) YDYWEP02(>0,<=7) YEYARS02(>0,<=7) YEYBOP02(>0,<=7) YEYBUS02(>0,<=7)
 YEYCBK02(>0,<=7) YEYDRG02(>0,<=7) YEYGRF02(>0,<=7) YEYHBK02(>0,<=7) YEYHIT02(>0,<=7) YEYJRD02(>0,<=7) YEYPET02(>0,<=7) YEYRAB02(>0,<=7)
 YEYROB02(>0,<=7) YEYRST02(>0,<=7) YEYSHP02(>0,<=7) YEYSKV02(>0,<=7) YEYVND02(>0,<=7)
  YEYWEP02(>0,<=7) YFYARS02(>0,<=7) YFYBFT02(>0,<=7) YFYBOP02(>0,<=7) YFYCBK02(>0,<=7) YFYDRG02(>0,<=7) YFYFRD02(>0,<=7)
 YFYHBK02(>0,<=7) YFYHIT02(>0,<=7) YFYJRD02(>0,<=7) YFYPET02(>0,<=7) YFYRAB02(>0,<=7) YFYROB02(>0,<=7) YFYRST02(>0,<=7) YFYRST22(>0,<=7)
 YFYSHP02(>0,<=7) YFYSKV02(>0,<=7) YFYVND02(>0,<=7) YFYWEP02(>0,<=7);
  iterations 5;
   multiples 5;
   transfer caseid ;
run ;