Re: [R] Troubleshooting underidentification issues in structural equation modelling (SEM)
> I03 <--> > > >> I03 > > >> V[I04]1.461376e-01 7.255861e-03 20.140635357 3.251385e-90 > I04 <--> > > >> I04 > > >> V[I05]1.339123e-02 8.832859e-04 15.160696593 6.438285e-52 > I05 <--> > > >> I05 > > >> V[I06]8.789764e-02 4.794460e-03 18.333167786 4.499223e-75 > I06 <--> > > >> I06 > > >> V[I07]7.568474e-03 3.765280e-04 20.100692934 7.277043e-90 > I07 <--> > > >> I07 > > >> V[I08]6.587699e-02 3.167671e-03 20.79217 4.639577e-96 > I08 <--> > > >> I08 > > >> V[I09]3.217338e-03 1.517789e-04 21.197527600 1.006468e-99 > I09 <--> > > >> I09 > > >> V[I10]4.621928e-02 2.185030e-03 21.152695320 2.606174e-99 > I10 <--> > > >> I10 > > >> V[I11]1.535621e-01 7.387455e-03 20.786870576 5.690287e-96 > I11 <--> > > >> I11 > > >> V[I12]3.908344e-02 1.860301e-03 21.009196121 5.404186e-98 > I12 <--> > > >> I12 > > >> V[I13]1.983328e-02 9.856998e-04 20.121018746 4.830497e-90 > I13 <--> > > >> I13 > > >> V[I14]1.710572e-01 1.211810e-02 14.115839622 3.033809e-45 > I14 <--> > > >> I14 > > >> V[I15]1.075179e-03 5.071602e-05 21.199985035 9.552682e-100 > I15 <--> > > >> I15 > > >> V[I16]1.326202e-02 6.467196e-04 20.506601881 1.879773e-93 > I16 <--> > > >> I16 > > >> V[I17]3.265749e-02 1.988078e-03 16.426667150 1.232493e-60 > I17 <--> > > >> I17 > > >> V[I18]1.075154e-03 5.071579e-05 21.199589039 9.633394e-100 > I18 <--> > > >> I18 > > >> V[I19]4.579942e-02 2.353962e-03 19.456315348 2.576564e-84 > I19 <--> > > >> I19 > > >> V[I20]2.413742e-01 1.144346e-02 21.092761358 9.269013e-99 > I20 <--> > > >> I20 > > >> V[I21]1.269773e-02 6.009212e-04 21.130448044 4.175664e-99 > I21 <--> > > >> I21 > > >> V[I22]2.667065e-01 1.265916e-02 21.068268778 1.555139e-98 > I22 <--> > > >> I22 > > >> V[I23]1.072933e-03 5.069564e-05 21.164210344 2.041534e-99 > I23 <--> > > >> I23 > > >> V[I24]3.024220e-02 1.426452e-03 21.200993757 9.350120e-100 > I24 <--> > > >> I24 > > >> V[I25]4.271005e-02 2.065984e-03 20.672986805 6.064466e-95 > I25 <--> > > >> I25 > > >> V[I26]8.208471e-02 3.892796e-03 21.086314551 1.062215e-98 > I26 <--> > > >> I26 > > >> V[I27]3.448443e-02 1.627464e-03 21.189053796 1.204944e-99 > I27 <--> > > >> I27 > > >> V[I28]1.074072e-03 5.065613e-05 21.203199739 8.921947e-100 > I28 <--> > > >> I28 > > >> V[I29]1.388601e-02 6.548663e-04 21.204342235 8.707941e-100 > I29 <--> > > >> I29 > > >> V[I30]3.656256e-02 1.724532e-03 21.201435371 9.262794e-100 > I30 <--> > > >> I30 > > >> V[I31]1.989840e-01 9.383562e-03 21.205594692 8.479218e-100 > I31 <--> > > >> I31 > > >> V[I32]5.77e-02 2.882318e-03 19.968499245 1.035172e-88 > I32 <--> > > >> I32 > > >> V[I33]2.481455e-01 1.532786e-02 16.189179144 6.012530e-59 > I33 <--> > > >> I33 > > >> V[I34]1.484183e-02 7.26e-04 21.202534570 9.048952e-100 > I34 <--> > > >> I34 > > >> V[I35]7.415580e-03 3.516263e-04 21.089380308 9.955712e-99 > I35 <--> > > >> I35 > > >> V[I36]2.011634e-02 9.488573e-04 21.200591226 9.430434e-100 > I36 <--> > > >> I36 > > >> V[I37]1.047757e-03 5.025784e-05 20.847625170 1.601775e-96 > I37 <--> > > >> I37 > > >> V[I38]2.156861e-02 3.241426e-03 6.654050864 2.851341e-11 > I38 <--> > > >> I38 > > >> V[I39]1.265785e-01 6.238795e-03 20.288931432 1.610577e-91 > I39 <--> > > >> I39 > > >> V[I40] 2.541968e-01 1.242997e-02 20.450322391 5.967951e-93 > I40 <--> > > >> I40 > > >> V[I41]8.528364e-02 4.023849e-03 21.194542822 1.072350e-99 > I41 <--> > > >> I41 > > >> V[I42]8.216499e-02 3.888
Re: [R] Troubleshooting underidentification issues in structural equation modelling (SEM)
5 6.064466e-95 I25 <--> > >> I25 > >> V[I26]8.208471e-02 3.892796e-03 21.086314551 1.062215e-98 I26 <--> > >> I26 > >> V[I27]3.448443e-02 1.627464e-03 21.189053796 1.204944e-99 I27 <--> > >> I27 > >> V[I28]1.074072e-03 5.065613e-05 21.203199739 8.921947e-100 I28 <--> > >> I28 > >> V[I29]1.388601e-02 6.548663e-04 21.204342235 8.707941e-100 I29 <--> > >> I29 > >> V[I30]3.656256e-02 1.724532e-03 21.201435371 9.262794e-100 I30 <--> > >> I30 > >> V[I31]1.989840e-01 9.383562e-03 21.205594692 8.479218e-100 I31 <--> > >> I31 > >> V[I32]5.77e-02 2.882318e-03 19.968499245 1.035172e-88 I32 <--> > >> I32 > >> V[I33]2.481455e-01 1.532786e-02 16.189179144 6.012530e-59 I33 <--> > >> I33 > >> V[I34]1.484183e-02 7.26e-04 21.202534570 9.048952e-100 I34 <--> > >> I34 > >> V[I35]7.415580e-03 3.516263e-04 21.089380308 9.955712e-99 I35 <--> > >> I35 > >> V[I36]2.011634e-02 9.488573e-04 21.200591226 9.430434e-100 I36 <--> > >> I36 > >> V[I37]1.047757e-03 5.025784e-05 20.847625170 1.601775e-96 I37 <--> > >> I37 > >> V[I38]2.156861e-02 3.241426e-03 6.654050864 2.851341e-11 I38 <--> > >> I38 > >> V[I39]1.265785e-01 6.238795e-03 20.288931432 1.610577e-91 I39 <--> > >> I39 > >> V[I40]2.541968e-01 1.242997e-02 20.450322391 5.967951e-93 I40 <--> > >> I40 > >> V[I41]8.528364e-02 4.023849e-03 21.194542822 1.072350e-99 I41 <--> > >> I41 > >> V[I42]8.216499e-02 3.888144e-03 21.132187265 4.024656e-99 I42 <--> > >> I42 > >> V[I43]1.337408e-02 6.438437e-04 20.772251070 7.715629e-96 I43 <--> > >> I43 > >> V[I46]1.907454e-01 8.996895e-03 21.201249767 9.299396e-100 I46 <--> > >> I46 > >> V[I47]8.508783e-03 4.165525e-04 20.426677159 9.687421e-93 I47 <--> > >> I47 > >> V[I48]2.714640e-01 1.280461e-02 21.200497563 9.449220e-100 I48 <--> > >> I48 > >> V[I49]3.218862e-03 1.518230e-04 21.201415045 9.266795e-100 I49 <--> > >> I49 > >> V[I50]7.447779e-03 3.685477e-04 20.208454710 8.249036e-91 I50 <--> > >> I50 > >> V[I51]2.929982e-05 1.053218e-04 0.278193234 7.808640e-01 I51 <--> > >> I51 > >> V[I54]1.833931e-01 8.842196e-03 20.740673158 1.488283e-95 I54 <--> > >> I54 > >> V[I55]4.784306e-02 2.783744e-03 17.186584134 3.346789e-66 I55 <--> > >> I55 > >> V[I56]1.304849e-01 6.185550e-03 21.095115843 8.818929e-99 I56 <--> > >> I56 > >> V[I57]8.868251e-02 4.280267e-03 20.718917274 2.338858e-95 I57 <--> > >> I57 > >> V[I58]2.765876e-01 1.332324e-02 20.75954 1.000282e-95 I58 <--> > >> I58 > >> V[I59]1.309969e-01 6.275841e-03 20.873197799 9.384143e-97 I59 <--> > >> I59 > >> V[I60]2.844711e-02 1.341830e-03 21.200226581 9.503782e-100 I60 <--> > >> I60 > >> V[I61]3.368300e-02 1.992102e-03 16.908270471 3.910162e-64 I61 <--> > >> I61 > >> V[I62]7.504898e-03 3.540020e-04 21.200154519 9.518345e-100 I62 <--> > >> I62 > >> V[I63]7.472838e-02 3.568523e-03 20.940981942 2.267379e-97 I63 <--> > >> I63 > >> V[I64]5.371193e-03 2.533508e-04 21.200616220 9.425427e-100 I64 <--> > >> I64 > >> V[I65] -1.558692e+01 7.736661e+02 -0.020146825 9.839262e-01 I65 <--> > >> I65 > >> V[I66]6.009302e-02 2.837570e-03 21.177638375 1.535393e-99 I66 <--> > >> I66 > >> V[I67]1.075013e-03 5.220505e-05 20.592119939 3.229259e-94 I67 <--> > >> I67 > >> V[I69]8.817859e-02 5.04e-03 17.635704215 1.310532e-69 I69 <--> > >> I69 > >> V[I70]2.218392e-02 1.279170e-03 17.342438243 2.249872e-67 I70 <--> > >> I70 > >> V[I71]3.093500e-02 1.758727e-03 17.589432179 2.968370e-69 I71 <--> > >> I71 > >> > >> Iterations = 1000 > >> > >> - snip > >> > >> Several of the observed variables have R^2s that round to 0 and many more > >> are very small. > >> > >> I don't have your or
Re: [R] Troubleshooting underidentification issues in structural equation modelling (SEM)
01 8.842196e-03 20.740673158 1.488283e-95 I54 <--> >> I54 >> V[I55]4.784306e-02 2.783744e-03 17.186584134 3.346789e-66 I55 <--> >> I55 >> V[I56]1.304849e-01 6.185550e-03 21.095115843 8.818929e-99 I56 <--> >> I56 >> V[I57]8.868251e-02 4.280267e-03 20.718917274 2.338858e-95 I57 <--> >> I57 >> V[I58]2.765876e-01 1.332324e-02 20.75954 1.000282e-95 I58 <--> >> I58 >> V[I59]1.309969e-01 6.275841e-03 20.873197799 9.384143e-97 I59 <--> >> I59 >> V[I60]2.844711e-02 1.341830e-03 21.200226581 9.503782e-100 I60 <--> >> I60 >> V[I61]3.368300e-02 1.992102e-03 16.908270471 3.910162e-64 I61 <--> >> I61 >> V[I62]7.504898e-03 3.540020e-04 21.200154519 9.518345e-100 I62 <--> >> I62 >> V[I63]7.472838e-02 3.568523e-03 20.940981942 2.267379e-97 I63 <--> >> I63 >> V[I64]5.371193e-03 2.533508e-04 21.200616220 9.425427e-100 I64 <--> >> I64 >> V[I65] -1.558692e+01 7.736661e+02 -0.020146825 9.839262e-01 I65 <--> >> I65 >> V[I66]6.009302e-02 2.837570e-03 21.177638375 1.535393e-99 I66 <--> >> I66 >> V[I67]1.075013e-03 5.220505e-05 20.592119939 3.229259e-94 I67 <--> >> I67 >> V[I69]8.817859e-02 5.04e-03 17.635704215 1.310532e-69 I69 <--> >> I69 >> V[I70]2.218392e-02 1.279170e-03 17.342438243 2.249872e-67 I70 <--> >> I70 >> V[I71]3.093500e-02 1.758727e-03 17.589432179 2.968370e-69 I71 <--> >> I71 >> >> Iterations = 1000 >> >> - snip >> >> Several of the observed variables have R^2s that round to 0 and many more >> are very small. >> >> I don't have your original data, but I did look at the input covariance >> matrix. Here are the standard deviations of the observed variables: >> >> - snip >> >> > sqrt(diag(cov.mat)) >>I01I02I03I04I05I06 >> I07 >> >> 0.09794939 0.09239769 0.08647698 0.40592964 0.14988296 0.34276336 >> 0.09257290 >> >>I08I09I10I11I12I13 >> I14 >> >> 0.26288788 0.05673501 0.21562354 0.40159670 0.1190 0.14969750 >> 0.48787040 >> >>I15I16I17I18I19I20 >> I21 >> >> 0.03279129 0.11746460 0.20339207 0.03279129 0.22450179 0.49285671 >> 0.11291786 >> >>I22I23I24I25I26I27 >> I28 >> >> 0.51844236 0.03279129 0.17390500 0.20982058 0.28746674 0.18587268 >> 0.03279129 >> >>I29I30I31I32I33I34 >> I35 >> >> 0.11789736 0.19121352 0.44618622 0.24132578 0.50500808 0.12183229 >> 0.08647698 >> >>I36I37I38I39I40I41 >> I42 >> >> 0.14183651 0.03279129 0.20705800 0.36721084 0.51768833 0.29210990 >> 0.28739426 >> >>I43I45I46I47I48I49 >> I50 >> >> 0.11746460 0.13454976 0.43680464 0.09794939 0.52139099 0.05673501 >> 0.09239769 >> >>I51I54I55I56I57I58 >> I59 >> >> 0.03279129 0.43984267 0.26013269 0.36354251 0.30622933 0.53958761 >> 0.36898429 >> >>I60I61I62I63I64I65 >> I66 >> >> 0.16867489 0.22011795 0.08663745 0.27761032 0.07329198 0.52861343 >> 0.24514452 >> >>I67I68I69I70I71 >> 0.03279129 0.16616880 0.33665601 0.17020504 0.19965594 >> >> - snip >> >> Some of the standard deviations are very small, suggesting that the >> corresponding variables must have been close to invariant in your data set. >> >> If you haven't already done so, I think that you might back up and look >> more >> closely at your data, and perhaps seek some competent local help. >> >> I hope that this helps, >> John >> >> --- >> John Fox >> Senator McMaster Professor of Social Statistics >> Department of Sociology >> McMaster University >> Hamilton, Ontario, Canada >> >> >> >> > -Original Message- >> > From: r-help-boun...@r-project.org [mailto:r-help-boun...@r-project.org] >>
Re: [R] Troubleshooting underidentification issues in structural equation modelling (SEM)
I03I04I05I06 > I07 > > 0.09794939 0.09239769 0.08647698 0.40592964 0.14988296 0.34276336 > 0.09257290 > >I08I09 I10 I11I12 I13 > I14 > > 0.26288788 0.05673501 0.21562354 0.40159670 0.1190 0.14969750 > 0.48787040 > >I15I16I17I18I19I20 > I21 > > 0.03279129 0.11746460 0.20339207 0.03279129 0.22450179 0.49285671 > 0.11291786 > >I22I23I24I25I26I27 > I28 > > 0.51844236 0.03279129 0.17390500 0.20982058 0.28746674 0.18587268 > 0.03279129 > >I29I30I31I32I33I34 > I35 > > 0.11789736 0.19121352 0.44618622 0.24132578 0.50500808 0.12183229 > 0.08647698 > >I36I37I38I39I40I41 > I42 > > 0.14183651 0.03279129 0.20705800 0.36721084 0.51768833 0.29210990 > 0.28739426 > >I43I45I46I47I48I49 > I50 > > 0.11746460 0.13454976 0.43680464 0.09794939 0.52139099 0.05673501 > 0.09239769 > >I51I54I55I56I57I58 > I59 > > 0.03279129 0.43984267 0.26013269 0.36354251 0.30622933 0.53958761 > 0.36898429 > >I60I61I62I63I64I65 > I66 > > 0.16867489 0.22011795 0.08663745 0.27761032 0.07329198 0.52861343 > 0.24514452 > >I67I68I69I70I71 > 0.03279129 0.16616880 0.33665601 0.17020504 0.19965594 > > - snip > > Some of the standard deviations are very small, suggesting that the > corresponding variables must have been close to invariant in your data set. > > If you haven't already done so, I think that you might back up and look > more > closely at your data, and perhaps seek some competent local help. > > I hope that this helps, > John > > --- > John Fox > Senator McMaster Professor of Social Statistics > Department of Sociology > McMaster University > Hamilton, Ontario, Canada > > > > > -Original Message- > > From: r-help-boun...@r-project.org [mailto:r-help-boun...@r-project.org] > > On Behalf Of Ruijie > > Sent: Friday, February 08, 2013 9:56 PM > > To: r-h...@stat.math.ethz.ch > > Subject: [R] Troubleshooting underidentification issues in structural > > equation modelling (SEM) > > > > Hi all, hope someone can help me out with this. > > Background Introduction > > > > I have a data set consisting of data collected from a questionnaire that > > I > > wish to validate. I have chosen to use confirmatory factor analysis to > > analyse this data set. > > Instrument > > > > The instrument consists of 11 subscales. There is a total of 68 items in > > the 11 subscales. Each item is scored on an integer scale between 1 to > > 4. > > Confirmatory factor analysis (CFA) setup > > > > I use the sem package to conduct the CFA. My code is as below: > > > > cov.mat <- > > as.matrix(read.table("http://dl.dropbox.com/u/1445171/cov.mat.csv";, > > sep = ",", header = TRUE)) > > rownames(cov.mat) <- colnames(cov.mat) > > > > model <- cfa(file = "http://dl.dropbox.com/u/1445171/cfa.model.txt";, > > reference.indicators = FALSE) > > cfa.output <- sem(model, cov.mat, N = 900, maxiter = 8, optimizer > > = optimizerOptim) > > Warning message:In eval(expr, envir, enclos) : Negative parameter > > variances.Model may be underidentified. > > > > Straight off you might notice a few anomalies, let me explain. > > > >- Why is the optimizer chosen to be optimizerOptim? > > > > ANS: I originally stuck with the default optimizerSem but no matter how > > many iterations I run, either I run out of memory first (8GB RAM setup) > > or > > it would report no convergence Things "seemed" a little better when I > > switched to optimizerOptim where by it would conclude successfully but > > throws up the error that the model is underidentified. Upon closer > > inspection, I realise that the output shows convergence as TRUE but > > iterations is NA so I am not sure what is exactly happening. > > > >- The maxiter is too high. > > > > ANS: If I set it to a lower value, it refuses to converge, although as > > mentioned above, I doubt real convergence actually occurred. > > Problem > > > > So by now I guess that the model is really underidentified so I looked > > for > > resources to resolve this problem and found: > > > >- http://davidakenny.net/cm/identify_formal.htm > >- http://faculty.ucr.edu/~hanneman/soc203b/lectures/identify.html > > > > I followed the 2nd link quite closely and applied the t-rule: > > > >- I have 68 observed variables, providing me with 68 variances and > > 2278 > >covariances between variables = *2346 data points*. > >- I also have 68 regression coefficients, 68 error variances of > >variables, 11 factor variances and 55 factor covariances to estimate > > making > >it a total of 191 parameters. > >- Since I will be fixing the variances of the 11 latent factors to 1 > > for > >scaling, I would remove them from the parameters to estimate making > > it a > >total of *180 parameters to estimate*. > > - My degrees of freedom is therefore 2346 - 180 = 2166, making it > > an > > over identified model by the t-rule. > > > > Questions > > > >1. Is the low variance of some of my items a possible cause for the > >underidentification? I was advised previously to remove items with > > zero > >variance which led me to think about items which are very close to > > zero. > >Should they be removed too? > >2. After reading much, I think but am not sure that it might be a > > case > >of empirical underidentification. Is there a systematic way of > > diagnosing > >what kind of underidentification it is? And what are my options to > > proceed > >with my analysis? > > > > I have more questions but let's take it at these 2 for now. Thanks for > > any > > help! > > > > Regards, > > Ruijie (RJ) > > > > > > He who has a why can endure any how. > > > > ~ Friedrich Nietzsche > > > > [[alternative HTML version deleted]] > > > > __ > > R-help@r-project.org mailing list > > https://stat.ethz.ch/mailman/listinfo/r-help > > PLEASE do read the posting guide http://www.R-project.org/posting- > > guide.html > > and provide commented, minimal, self-contained, reproducible code. > > [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] Troubleshooting underidentification issues in structural equation modelling (SEM)
48]2.714640e-01 1.280461e-02 21.200497563 9.449220e-100 I48 <--> I48 V[I49]3.218862e-03 1.518230e-04 21.201415045 9.266795e-100 I49 <--> I49 V[I50]7.447779e-03 3.685477e-04 20.208454710 8.249036e-91 I50 <--> I50 V[I51]2.929982e-05 1.053218e-04 0.278193234 7.808640e-01 I51 <--> I51 V[I54]1.833931e-01 8.842196e-03 20.740673158 1.488283e-95 I54 <--> I54 V[I55]4.784306e-02 2.783744e-03 17.186584134 3.346789e-66 I55 <--> I55 V[I56]1.304849e-01 6.185550e-03 21.095115843 8.818929e-99 I56 <--> I56 V[I57]8.868251e-02 4.280267e-03 20.718917274 2.338858e-95 I57 <--> I57 V[I58]2.765876e-01 1.332324e-02 20.75954 1.000282e-95 I58 <--> I58 V[I59]1.309969e-01 6.275841e-03 20.873197799 9.384143e-97 I59 <--> I59 V[I60]2.844711e-02 1.341830e-03 21.200226581 9.503782e-100 I60 <--> I60 V[I61]3.368300e-02 1.992102e-03 16.908270471 3.910162e-64 I61 <--> I61 V[I62]7.504898e-03 3.540020e-04 21.200154519 9.518345e-100 I62 <--> I62 V[I63]7.472838e-02 3.568523e-03 20.940981942 2.267379e-97 I63 <--> I63 V[I64]5.371193e-03 2.533508e-04 21.200616220 9.425427e-100 I64 <--> I64 V[I65] -1.558692e+01 7.736661e+02 -0.020146825 9.839262e-01 I65 <--> I65 V[I66]6.009302e-02 2.837570e-03 21.177638375 1.535393e-99 I66 <--> I66 V[I67]1.075013e-03 5.220505e-05 20.592119939 3.229259e-94 I67 <--> I67 V[I69]8.817859e-02 5.04e-03 17.635704215 1.310532e-69 I69 <--> I69 V[I70]2.218392e-02 1.279170e-03 17.342438243 2.249872e-67 I70 <--> I70 V[I71]3.093500e-02 1.758727e-03 17.589432179 2.968370e-69 I71 <--> I71 Iterations = 1000 - snip Several of the observed variables have R^2s that round to 0 and many more are very small. I don't have your original data, but I did look at the input covariance matrix. Here are the standard deviations of the observed variables: - snip > sqrt(diag(cov.mat)) I01I02I03I04I05I06I07 0.09794939 0.09239769 0.08647698 0.40592964 0.14988296 0.34276336 0.09257290 I08I09I10I11I12I13I14 0.26288788 0.05673501 0.21562354 0.40159670 0.1190 0.14969750 0.48787040 I15I16I17I18I19I20I21 0.03279129 0.11746460 0.20339207 0.03279129 0.22450179 0.49285671 0.11291786 I22I23I24I25I26I27I28 0.51844236 0.03279129 0.17390500 0.20982058 0.28746674 0.18587268 0.03279129 I29I30I31I32I33I34I35 0.11789736 0.19121352 0.44618622 0.24132578 0.50500808 0.12183229 0.08647698 I36I37I38I39I40I41I42 0.14183651 0.03279129 0.20705800 0.36721084 0.51768833 0.29210990 0.28739426 I43I45I46I47I48I49I50 0.11746460 0.13454976 0.43680464 0.09794939 0.52139099 0.05673501 0.09239769 I51I54I55I56I57I58I59 0.03279129 0.43984267 0.26013269 0.36354251 0.30622933 0.53958761 0.36898429 I60I61I62I63I64I65I66 0.16867489 0.22011795 0.08663745 0.27761032 0.07329198 0.52861343 0.24514452 I67 I68I69 I70 I71 0.03279129 0.16616880 0.33665601 0.17020504 0.19965594 - snip Some of the standard deviations are very small, suggesting that the corresponding variables must have been close to invariant in your data set. If you haven't already done so, I think that you might back up and look more closely at your data, and perhaps seek some competent local help. I hope that this helps, John --- John Fox Senator McMaster Professor of Social Statistics Department of Sociology McMaster University Hamilton, Ontario, Canada > -Original Message- > From: r-help-boun...@r-project.org [mailto:r-help-boun...@r-project.org] > On Behalf Of Ruijie > Sent: Friday, February 08, 2013 9:56 PM > To: r-h...@stat.math.ethz.ch > Subject: [R] Troubleshooting underidentification issues in structural > equation modelling (SEM) > > Hi all, hope someone can help me out with this. > Background Introduction > > I have a data set consisting of data collected from a questionnaire that > I > wish to validate. I have chosen to use confirmatory factor analysis to > analyse this data set. > Instrument > > The instrument consists of 11 subscales. There is a total of 68 items in > the 11 subscales. Each item is scored on an integer scale between 1 to > 4. > Confirmato
[R] Troubleshooting underidentification issues in structural equation modelling (SEM)
Hi all, hope someone can help me out with this. Background Introduction I have a data set consisting of data collected from a questionnaire that I wish to validate. I have chosen to use confirmatory factor analysis to analyse this data set. Instrument The instrument consists of 11 subscales. There is a total of 68 items in the 11 subscales. Each item is scored on an integer scale between 1 to 4. Confirmatory factor analysis (CFA) setup I use the sem package to conduct the CFA. My code is as below: cov.mat <- as.matrix(read.table("http://dl.dropbox.com/u/1445171/cov.mat.csv";, sep = ",", header = TRUE)) rownames(cov.mat) <- colnames(cov.mat) model <- cfa(file = "http://dl.dropbox.com/u/1445171/cfa.model.txt";, reference.indicators = FALSE) cfa.output <- sem(model, cov.mat, N = 900, maxiter = 8, optimizer = optimizerOptim) Warning message:In eval(expr, envir, enclos) : Negative parameter variances.Model may be underidentified. Straight off you might notice a few anomalies, let me explain. - Why is the optimizer chosen to be optimizerOptim? ANS: I originally stuck with the default optimizerSem but no matter how many iterations I run, either I run out of memory first (8GB RAM setup) or it would report no convergence Things "seemed" a little better when I switched to optimizerOptim where by it would conclude successfully but throws up the error that the model is underidentified. Upon closer inspection, I realise that the output shows convergence as TRUE but iterations is NA so I am not sure what is exactly happening. - The maxiter is too high. ANS: If I set it to a lower value, it refuses to converge, although as mentioned above, I doubt real convergence actually occurred. Problem So by now I guess that the model is really underidentified so I looked for resources to resolve this problem and found: - http://davidakenny.net/cm/identify_formal.htm - http://faculty.ucr.edu/~hanneman/soc203b/lectures/identify.html I followed the 2nd link quite closely and applied the t-rule: - I have 68 observed variables, providing me with 68 variances and 2278 covariances between variables = *2346 data points*. - I also have 68 regression coefficients, 68 error variances of variables, 11 factor variances and 55 factor covariances to estimate making it a total of 191 parameters. - Since I will be fixing the variances of the 11 latent factors to 1 for scaling, I would remove them from the parameters to estimate making it a total of *180 parameters to estimate*. - My degrees of freedom is therefore 2346 - 180 = 2166, making it an over identified model by the t-rule. Questions 1. Is the low variance of some of my items a possible cause for the underidentification? I was advised previously to remove items with zero variance which led me to think about items which are very close to zero. Should they be removed too? 2. After reading much, I think but am not sure that it might be a case of empirical underidentification. Is there a systematic way of diagnosing what kind of underidentification it is? And what are my options to proceed with my analysis? I have more questions but let's take it at these 2 for now. Thanks for any help! Regards, Ruijie (RJ) He who has a why can endure any how. ~ Friedrich Nietzsche [[alternative HTML version deleted]] __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.