3.1 CA procedures: SPSS syntax

 
* Weight cases and, if necessary, reset the max number of SPSS cells to 
* greater than the max number of occ units squared. 
show mxcells. 
set mxcells 400000. 
show mxcells. 
weight by freq. 

3.1.1 
* 1) SPSS 'Correspondence' command (subcommands could be altered) :. 
correspondence table=hocc2(1 252) by wocc2(1 252) 
      /dimensions=2 /measures=chisq /standardize=rcmean 
     / normalization=symmetrical /print=table rpoint cpoint 
        /plot=ndim(1,max) biplot(20). 

3.1.2 
* 2) SPSS 'anacor' command :. 
anacor table=hocc2 (1,252) by wocc2(1,252) 
     /normalisation=canonical /print=scores. 

3.1.3 
* 3) SPSS 'homogeneity analysis' :. 
homals /variables=hocc2(252) wocc2(252) 
   /analysis=hocc2 wocc2 /dimension=2 
    /print freq eigen history discrim quant 
   /plot quant object discrim ndim(all,max) /maxiter=200 /convergence=0.00001. 

Return to Preliminary CA models

3.2 Production job: SPSS syntax

 * Example production job file to run multiple anacor CA models. 
get file="work.sav". 
show mxcells. 
set mxcells 400000. 
show mxcells. 
* Identify initial pseudo-dagionals ) :. 
compute pseudd=0. if ((hocc=123 & wocc=345) 
  | (hocc=456 & (wocc=456 | wocc=457)) 
      | (hempst=1 & wocc=678) ) pseudd=1. 
* [etc : adjust in repsonse to exploratory models]. 
weight by freq. 

** Each model, preceded by a table showing the cases excluded as PSD's. 
* (note the repeated change of output settings, which can make pasting of output easier). 
* 1) Title only :. 
set ovars both onums both tvars both tnumbers labels. 
tables /format blank missing ('.') /ftotal=ftot1 "Total" 
   /tables (labels) + ftot1 by pseudd > (hocc2 + wocc2) /statistics count ((F5.0) ' Cases ') 
   /title="Husband and Wife Occupational Titles". 
set ovars both onums both tvars both tnumbers both. 
temp. 
select if (pseudd=0). 
anacor table=hocc2 (1,252) by wocc2(1,252) /normalisation=canonical 
     /print=scores. 

* 2) Title-by-status : COMMENT OUT BECAUSE NUMBER OF UNITS TOO HIGH. 
*set ovars both onums both tvars both tnumbers labels. 
*tables /format blank missing ('.') /ftotal=ftot1 "Total" 
   /tables (labels) + ftot1 by pseudd > (hbst2 + wbst2) 
   /statistics count ((F5.0) ' Cases ') 
    /title="Husband and Wife Occupational Title-by-status units". 
*set ovars both onums both tvars both tnumbers both. 
*temp. 
*select if (pseudd=0). 
*anacor table=hbst2 (1,700) by wbst2(1,700) /normalisation=canonical 
    /print=scores. 
* 3) Employment status only :. 
set ovars both onums both tvars both tnumbers labels. 
tables /format blank missing ('.') /ftotal=ftot1 "Total" 
       /tables (labels) + ftot1 by pseudd > (hempst2 + wempst2) 
      /statistics count ((F5.0) ' Cases ') 
     /title="Husband and Wife Occupational Employment status". 
set ovars both onums both tvars both tnumbers both. 
temp. 
select if (pseudd=0). 
anacor table=hempst2 (1,5) by wempst2(1,5) 
    /normalisation=canonical /print=scores. 
* 4) Major group only :. 
set ovars both onums both tvars both tnumbers labels. 
tables /format blank missing ('.') /ftotal=ftot1 "Total" 
   /tables (labels) + ftot1 by pseudd > (h1gp2 + w1gp2) 
    /statistics count ((F5.0) ' Cases ') 
     /title="Husband and Wife Occupational Major Group". 
set ovars both onums both tvars both tnumbers both. 
temp. 
select if (pseudd=0). 
anacor table=h1gp2 (1,9) by w1gp2(1,9) /normalisation=canonical 
      /print=scores. 
* 5) Major group by status :. 
set ovars both onums both tvars both tnumbers labels. 
tables /format blank missing ('.') /ftotal=ftot1 "Total" 
    /tables (labels) + ftot1 by pseudd > (h1gpst2 + w1gpst2) 
       /statistics count ((F5.0) ' Cases ') 
   /title="Husband and Wife Occupational Major Group by status". 
set ovars both onums both tvars both tnumbers both. 
temp. 
select if (pseudd=0). 
anacor table=h1gpst2 (1,45) by w1gpst2(1,45) 
    /normalisation=canonical /print=scores. 
* 6) Minor group only :. 
set ovars both onums both tvars both tnumbers labels. 
tables /format blank missing ('.') /ftotal=ftot1 "Total" 
   /tables (labels) + ftot1 by pseudd > (h3gp2 + w3gp2) 
   /statistics count ((F5.0) ' Cases ') 
    /title="Husband and Wife Occupational Minor Group". 
set ovars both onums both tvars both tnumbers both. 
temp. 
select if (pseudd=0). 
anacor table=h1gp2 (1,80) by w1gp2(1,80) /normalisation=canonical 
       /print=scores. 
* 7) Minor group by status :. 
set ovars both onums both tvars both tnumbers labels. 
tables /format blank missing ('.') /ftotal=ftot1 "Total" 
   /tables (labels) + ftot1 by pseudd > (h3gpst2 + w3gpst2) /statistics count ((F5.0) ' Cases ') 
   /title="Husband and Wife Occupational Minor Group by status". 
set ovars both onums both tvars both tnumbers both. 
temp. 
select if (pseudd=0). 
anacor table=h3gpst2 (1,200) by w3gpst2(1,200) 
    /normalisation=canonical /print=scores. 
weight off. 

Return to Preliminary CA models  

---

  5.1.1 Handling pseudo-diagonals in SPSS: syntax

 
get file="rev1pseuds1.sav". 
** Create dummy indicator for all pseudo-diagonals concerned, eg :. 
compute m1prob=(diocc=1 | farm=1 | pub=1 | fambusi=1 | specpsd=1). 
weight by freq. set ovars both onums both tvars both tnumbers labels. 
tables /format blank missing ('.') /ftotal=ftot1 "Total" 
   /tables (labels) + ftot1 by m1prob > (hbst4 + wbst4) 
    /statistics count ((F5.0) ' Cases ') 
     /title="Husband and Wife title-by-status units". 
set ovars both onums both tvars both tnumbers both. 
temp. 
select if (m1prob=0). 
correspondence table=hbst4(1 560) by wbst4(1 560) /dimensions=2 
    /measures=chisq /standardize=rcmean / normalization=symmetrical 
     /print=table rpoint cpoint /plot=ndim(1,max) biplot(20). 
weight off. 

Return to Diagonals in SPSS CA