|
|||
|
Hi All
Did anyone of you come across a situation where (half of) a dataset is analyzed with the proc factor, exporatory factor analysis, and results subsequently automatically fed into a proc calis, confirmatory factor analysis of other half, as model specification (ie highest loading from proc factor is taken as the specification that one item is an indicator of one specific factor and not to any other factor). I am talking of a plain measurement model, not a structural model. Of course a structural model would need the input of the users theory, or desired model to test. A measurement model where factors are allowd to covary but an item is allowed to load on only one factor. I know that calis itself has some limited capabilities of doing an EFA, but withn that, no oblique rotations are provided, among other things. I played around with the outstat of proc factor and the outram/inram of proc calis, but none gave me what I wanted, although I see that a model specification, as done in lineqs, can be transferred from one proc calis to the next proc calis with the outram/inram. In proc calis, I am used to use the lineqs model specification (see below), and if this typing that specification based on the results of the EFA can be automated how I described it above, that would be really helpful. Thanks Toby proc calis data=psdata.ps2004ms; var q01prvrsc q02hirgol q03drgpol q04declev q05smpclm q06rptrpc q07viocor q08smrskp q09tmirsk q10hlpicp q11smimst q12mgmrsk q13tllmst q14acvrwd q15tllmbr q16hidmst q17comflo q18ftgeff q19smprch q20prsprb q21rptcsq q22fclcre q23cwklrn q24rvwpfm q25wtncwk q26mstdcp q27rptbhv q28cutcor q29losprs q30timtsk q31mstpun q32peprss q33witcwk q34donusf q35rwdact q36ftgerr q37tratem q38psimpr; lineqs q04declev = l11 f1 + e1, q05smpclm = l21 fa1 + e2, q08smrskp = l31 fa1 + e3, q11smimst = l41 fa1 + e4, q17comflo = l51 fa1 + e5, q19smprch = l61 fa1 + e6, q01prvrsc = l12 fa2 + e7, q29losprs = l22 fa2 + e8, q30timtsk = l32 fa2 + e9, q22fclcre = l13 fa3 + e10, q38psimpr = l23 fa3 + e11, q03drgpol = l14 fa4 + e12, q07viocor = l24 fa4 + e13, q09tmirsk = l34 fa4 + e14, q12mgmrsk = l44 fa4 + e15, q27rptbhv = l54 fa4 + e16, q32peprss = l64 fa4 + e17, q14acvrwd = l15 fa5 + e18, q24rvwpfm = l25 fa5 + e19, q35rwdact = l35 fa5 + e20, q37tratem = l45 fa5 + e21, q10hlpicp = l16 fa6 + e22, q13tllmst = l26 fa6 + e23, q15tllmbr = l36 fa6 + e24, q23cwklrn = l17 fa7 + e25, q26mstdcp = l18 fa8 + e26, q31mstpun = l28 fa8 + e27, q25wtncwk = l19 fa9 + e28, q34donusf = l29 fa9 + e29, q33witcwk = l39 fa9 + e30/*, q02hirgol = l10 fa0 + e31, q06rptrpc = l20 fa0 + e32, q16hidmst = l30 fa0 + e33, q18ftgeff = l40 fa0 + e34, q20prsprb = l50 fa0 + e35, q21rptcsq = l60 fa0 + e36, q28cutcor = l70 fa0 + e37, q36ftgerr = l80 fa0 + e38*/; std e1-e30 = ue1-ue30, fa1 = 1, fa2 = 1, fa3 = 1, fa4 = 1, fa5 = 1, fa6 = 1, fa7 = 1, fa8 = 1, fa9 = 1/*, fa0 = 1*/; run; |
|
|
||||
|
||||
|
|
![]() |
| Thread Tools | |
| Display Modes | |
|
|
Similar Threads
|
||||
| Thread | Thread Starter | Forum | Replies | Last Post |
| Re: proc surveyreg, proc calis, clustering, | Luo, Peter | Newsgroup comp.soft-sys.sas | 0 | 05-22-2007 01:58 PM |
| Re: proc surveyreg, proc calis, clustering, | David L Cassell | Newsgroup comp.soft-sys.sas | 0 | 05-22-2007 06:42 AM |
| Estimating non-linear GARCH model with 'proc model' | Melody Ghahramani | Newsgroup comp.soft-sys.sas | 0 | 09-06-2006 09:21 PM |
| Re: SAS Tips From SAS-L Archives January 1996 to May 2006 | Terjeson, Mark | Newsgroup comp.soft-sys.sas | 0 | 06-21-2006 04:00 PM |
| Re: Proc surveylogistic model selection question | David L. Cassell | Newsgroup comp.soft-sys.sas | 0 | 04-25-2005 06:56 PM |