Bookmark and Share

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

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

Re: st: Factor analysis for mixed type data

From   "JVerkuilen (Gmail)" <>
Subject   Re: st: Factor analysis for mixed type data
Date   Sun, 30 Dec 2012 08:34:51 -0500

On Sat, Dec 29, 2012 at 10:52 AM, Mahbubeh Parsaeian
<> wrote:
> Hi every body
> I want to run a
> factor analysis on a set of variables which consists of both continuous and
> categorical variables.
> I have found out
> that mca (Multiple Correspondence Analysis) and pca (Principal Component Analysis)
> are suitable for categorical and continuous data types respectively.

Not to get into a nitpick about it, but MCA and PCA are principal
components analysis not "true" factor analysis methods with an
associated full probability model, although Joint Correspondence
Analysis could be considered analogous to a least squares based factor
analysis given how it removes the one-way margins.

> I was wondering if
> there is command in Stata which can consider mixed variables (continuous and
> discrete).

If you are diligent, you can fit maximum marginal likelihood factor
analysis in Stata with -gllamm-. You could also generate a polychoric
correlation matrix and use -factor- or -sem-. Regardless, you need to
specify what you want to do before I could say more.

> It is noteworthy to
> mention that R has a function called "AFDM" in package "FactoMiner"
> which can run Factor analysis for mixed data.

*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index