st: advice re: "intervening" variable

[David Airey <david.airey@Vanderbilt.Edu>]

Hi Richard,

The discussion on structural equation models is conceptually  
relevant, but probably not appropriate if you have one variable  
measured per node: blood pressure -> diameter of blood vessel ->  
heart rate. A strong opinion on causality in SEM models can be read  
in Shipley's text, Cause and Correlation in Biology. But SEM often  
uses multiple measures per node. So maybe simple path analysis is  
more appropriate to your case, which SEM generalizes from (e.g., see  
http://www.ats.ucla.edu/stat/Stata/faq/pathreg.htm). Also, see  
section 4.5 in Vittinghoff's text in the Stata book store; it briefly  
covers the topic of mediating variables in regression models.

-Dave
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/