Skip to content

Structural equation modelling in R compared with Amos

4 messages · Ravi Kulkarni, Tóth Dénes, Joshua Wiley

#
Hi Ravi,

Look at openmx, it uses R to do matrix optimization especially for SEM (though more general too).  It does not have the draw paths interactively feature, but it is extremely powerful and flexible.  I have used Amos, EQS, Mplus, Lisrel, and OpenMx and I believe OpenMx is competitive against any of those if you know what you are doing.  I have found Amos to be particularly limited in its ability to handle complex models.  If you are looking for a simpler interface to SEM in R for some basic models, he k out the SEM package by John Fox.

HTH,

Josh
On Oct 20, 2011, at 4:56, Ravi Kulkarni <ravi.kulk at gmail.com> wrote:

            
#
Dear Ravi,

I would also suggest you to have a look at 'lavaan' (www.lavaan.org). It
has an extremely straightforward yet flexible model syntax and very good
documentation. Although its statistical capabilities are not comparable to
those of Mplus (which is the most powerful SEM-software I think), it is
getting more and more closer to it.
Personally I prefer lavaan over OpenMX, but it is also a matter of taste.

HTH,
  Denes
#
Lavaan is a nice package too, thanks for mentioning it.  Given sufficient understanding of matrices, I think OpenMx is arguably as or more powerful than Mplus---it runs circles around Mplus in terms of flexibility, although Mplus does what it does well, with (IMO) fairly sensible defaults, and simpler syntax.
On Oct 20, 2011, at 7:40, "D?nes T?TH" <tdenes at cogpsyphy.hu> wrote: