Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: power analysis for CFA


From   William Buchanan <[email protected]>
To   [email protected]
Subject   Re: st: power analysis for CFA
Date   Thu, 16 Jan 2014 07:46:50 -0600

Hi Giuseppina,

I've typically been advised to set up a simulation to deal with statistical power calculations involving SEM.  The developers of Mplus have a good example of the code (in Mplus) and explanation of how to do this in that platform here (http://www.statmodel.com/power.shtml).  The _chicrit_ value is the critical Chi-squared value for a corresponding Chi-squared test with the same number of degrees of freedom and the _ncp_ is the non-centrality parameter (Lambda in the website above).  

HTH,

Billy

On Jan 15, 2014, at 3:24 PM, Giuseppina Chiri <[email protected]> wrote:

> Dear stata list members,
> I am writing to ask for help carrying out a power analysis for CFA using
> STATA. I read in Timothy Brown's book "Confirmatory Factor Analysis for
> Applied Research" that is possible to carry out power analysis using the
> Satorra-Saris method. He gives an example using LISREL, SPSS and SAS. From
> the provided covariance matrix, I was able to obtain a dataset of 500
> observations and 6 variables and to successfully replicate his results. 
> 
> I first create the dataset
> 
> clear 
> matrix input m = (1, .4550, .4680, .1365, .1592, .1479 \ .455, 1, .504,
> .147, .1715, .1592 \ .468, .504, 1, .1512, .1764, .1638 \ .1365, .147,
> .1512, 1, .4200, .3900 \ .1592, .1715, .1764, .4200, 1, .4550 \ .1479,
> .1592, .1638, .3900, .4550, 1)
> matrix list m
> 
> set seed 12345
> set obs 500
> corr2data x1 x2 x3 x4 x5 x6, cov(m)
> 
> list in 1/10, sep(0)
> corr x1-x6
> 
> Then I run SEM based on the model in the book
> 
> sem (L1 -> x1, ) (L1 -> x2, ) (L1 -> x3, ) (L2 -> x4, ) (L2 -> x5, ) (L2 ->
> x6, ), covstruct(_lexogenous, diagonal) standardized latent(L1 L2 ) cov(
> L1*L2) nocapslatent standardized
> 
> My results match those in the book.   
> 
> Brown also provides the SPSS and SAS to carry out the power analysis, which
> I translated into STATA as: 
> 
> gen power = 1 - nchi2(df,ncp,chicrit)
> list n ncp power, noobs clean table
> 
> Using the df ncp and chicrit values provided in the book, I was able to
> obtain the power estimate.
> 
> What I am having difficulties with is figuring out where the two values
> chicrit = 3.841459 and ncp = 6.3724 are coming from. Looking at the STATA
> output I don't see these numbers, not even after asking for gof stats.
> What am I missing? 
> Any suggestions would be greatly appreciated.
> Giusi
> 
> 
> 
> 
> 
> 
> 
> 
> ---
> This email is free from viruses and malware because avast! Antivirus protection is active.
> http://www.avast.com
> 
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index