[R] Another SEM question

Jarrett Byrnes byrnes at msi.ucsb.edu
Tue Jul 21 16:33:39 CEST 2009


You should also need coefficients for your error terms.

On Jul 21, 2009, at 12:36 AM, Stein, Luba (AIM SE) wrote:

> Hello Jarrett,
>
> Thank you very much indeed for your help. I could solve my problem  
> and you were right that I had to choose the connections in the model  
> right.
> Thus the entry
> model <- specify.model()
> Z -> M, z1, NA
> Z -> USM, z2, NA
> Z -> R, z3, 1
> M <-> M
> USM <-> USM
> R <-> R
> Z <-> Z
>
> works and gave me moreover a really good fit.
>
> So thank you for your support once again.
>
> Best wishes,
> Luba
>
>
>
>
>
>
>
> -----Urspr?ngliche Nachricht-----
> Von: r-help-bounces at r-project.org [mailto:r-help-bounces at r- 
> project.org] Im Auftrag von Stein, Luba (AIM SE)
> Gesendet: Dienstag, 21. Juli 2009 09:13
> An: Jarrett Byrnes
> Cc: r-help at r-project.org
> Betreff: Re: [R] Another SEM question
>
> Hello,
>
> Perhaps this is a good point. I use the Eclipse platform. The  
> problem was that when I first used the structure
> Z -> M, z1, NA the compiler took only the value Z ->M. Thus I erased  
> it totally.
> Maybe it is really an Eclipse problem.
>
> Do you know how to solve this difficulty?
>
> Thanks for all your support,
> Luba
>
>
>
>
>
>
> -----Urspr?ngliche Nachricht-----
> Von: Jarrett Byrnes [mailto:byrnes at msi.ucsb.edu]
> Gesendet: Dienstag, 21. Juli 2009 09:01
> An: Stein, Luba (AIM SE)
> Cc: r-help at r-project.org
> Betreff: Re: AW: [R] Another SEM question
>
> Ah, the larger problem is in how you specify your model.  You provide
> no parameter names, nor starting estimates (even an NA).  See the sem
> help file for an example.  Basically, it must look something like as
> follows
>
>
> model <- specify.model()
> Z -> M, zm, NA
> Z -> I, zi, NA
>
> etc.
>
>
> On Jul 20, 2009, at 11:37 PM, Stein, Luba (AIM SE) wrote:
>
>>
>> Hi,
>>
>>
>>             [,1]          [,2]          [,3]
>> [1,]  4.820719e-03 -5.558801e-05 -5.718939e-05
>> [2,] -5.558801e-05  4.763194e-06 -7.661872e-06
>> [3,] -5.718939e-05 -7.661872e-06  1.662150e-03
>>
>> This is mod.cov. It is the covariance matrix of (R, I, M).
>> R, I and M are vectors of length 109 which are contained in the file
>> data4.csv.
>>
>> As far as I understood the package sem. I consider R, I and M as the
>> external veriables and Z as the latent variable which I will receive
>> as an result after calculating the estimated errors and parameters.
>> This is what atually is missing in the output.
>> Moreover, the output provides the information about the quality of
>> the fitted model. I have to admit that this model does not fit quite
>> well.
>> Nevertheless, it should provide the estimated errors like it does
>> just for the first variable Z ->M.
>>
>> Thanks a lot for your help,
>> Luba
>>
>>
>> -----Urspr?ngliche Nachricht-----
>> Von: r-help-bounces at r-project.org [mailto:r-help-bounces at r-
>> project.org] Im Auftrag von Jarrett Byrnes
>> Gesendet: Dienstag, 21. Juli 2009 08:19
>> An: Stein, Luba (AIM SE)
>> Cc: r-help at r-project.org
>> Betreff: Re: [R] Another SEM question
>>
>> You don't appear to be defining Z here.
>>
>> Might that be the problem?
>>
>> Or, I, M, and R may not be defined either.  It is unclear.  What does
>> mod.cov look like?
>>
>> On Jul 20, 2009, at 11:11 PM, Stein, Luba (AIM SE) wrote:
>>
>>> Thank you for your advice. So I am sending the whole code
>>>
>>> data.dir <- file.path(home.dir, "Data")
>>> file     <- file.path(data.dir, "data4.csv")
>>>
>>> SEM <- read.csv(file)
>>> print(SEM)
>>>
>>> library(sem)
>>> SEM1 <- as.matrix(cbind(SEM$R1, SEM$I1, SEM$M1))
>>> print(SEM1)
>>> mod.cov <- cov(SEM1)
>>> print(mod.cov)
>>>
>>> I <- SEM$I1
>>> M <- SEM$M1
>>> R <- SEM$R1
>>>
>>> model <- specify.model()
>>> Z -> M
>>> Z -> I
>>> Z -> R
>>> M <-> M
>>> I <-> I
>>> R <-> R
>>> Z <-> Z
>>>
>>> sem.mod <- sem(model, mod.cov, N=109)
>>> summary(sem.mod)
>>>
>>>
>>>
>>> All vectors have a length of 109.
>>>
>>>
>>> Thank you for your help once again.
>>>
>>> Best wishes,
>>> Luba
>>>
>>>
>>>
>>> -----Urspr?ngliche Nachricht-----
>>> Von: r-help-bounces at r-project.org [mailto:r-help-bounces at r-
>>> project.org] Im Auftrag von Jarrett Byrnes
>>> Gesendet: Montag, 20. Juli 2009 18:25
>>> An: Stein, Luba (AIM SE)
>>> Cc: r-help at r-project.org
>>> Betreff: Re: [R] Another SEM question
>>>
>>> Luba,
>>>
>>> If you could provide the code you ran, perhaps the listserv can be  
>>> of
>>> help.
>>>
>>> On Jul 20, 2009, at 7:55 AM, Stein, Luba (AIM SE) wrote:
>>>
>>>> Hello,
>>>>
>>>> I use the function sem the following way
>>>> sem.mod <- sem(model, mod.cov, N=109) where the variables are
>>>> modelled:
>>>>
>>>> Z -> M
>>>> Z -> I
>>>> Z -> R
>>>> M <-> M
>>>> I <-> I
>>>> R <-> R
>>>> Z <-> Z
>>>>
>>>> The output is
>>>> ...
>>>>
>>>> Normalized Residuals
>>>> Min. 1st Qu. Median Mean 3rd Qu. Max.
>>>> -7.3300 -0.2750 -0.2670 -0.1290 -0.0369 9.0300
>>>>
>>>> Parameter Estimates
>>>> Estimate Std Error z value Pr(>|z|)
>>>> 0.0021625 0.00017037 12.693 0 M <--- Z
>>>>
>>>> Iterations = 13
>>>>
>>>>
>>>> In "Structural Equation Modeling With the sem Package in R" by John
>>>> Fox is stated that there should be an output for each external
>>>> variable.
>>>>
>>>> Where is my fault, that I receive the output only for the first
>>>> variable?
>>>>
>>>>
>>>> Thanks for your help,
>>>> Luba
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> 	[[alternative HTML version deleted]]
>>>>
>>>> ______________________________________________
>>>> R-help at 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.
>>>
>>> ______________________________________________
>>> R-help at 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.
>>
>> ______________________________________________
>> R-help at 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.
>
> ______________________________________________
> R-help at 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.




More information about the R-help mailing list