Non-recursive model: Negative Multivariate R2

272 views
Skip to first unread message

James Patrick Cronin

unread,
Mar 28, 2013, 11:35:18 AM3/28/13
to lav...@googlegroups.com
Hi All,

Non-recursive models can produce negative multiple squared correlations (R2) for the loop's participants, which indicates either a modeling problem or that the R2 values should be adjusted (Hayduk 2006).  Does lavaan produce loop adjusted-R2 (e.g., Hayduk 2006)?

Thanks!
James

James Patrick Cronin

unread,
Mar 28, 2013, 9:18:40 PM3/28/13
to lav...@googlegroups.com

Hayduk, L. A. 2006. Blocked-Error-R 2: A Conceptually Improved Definition of the Proportion of Explained Variance in Models Containing Loops or Correlated Residuals. Quality & Quantity 40:629–649.

James Patrick Cronin

unread,
Mar 30, 2013, 2:40:09 PM3/30/13
to lav...@googlegroups.com
A follow up to my original post: As of 2011, other SEM software packages do not automatically 'loop adjust' R2 (Kline 2011).
Thus, I imagine lavaan does not. Hayduk (2006) provides the calculations in Section 6, which seem straight forward.

Kline, R.B. 2011. Principles and practice of structural equation modeling. 3rd Edition. The Guilford Press, New York.

yrosseel

unread,
Apr 2, 2013, 11:02:21 AM4/2/13
to lav...@googlegroups.com
This has been added to my TODO list. Thanks for the suggestion.

Yves.

James Patrick Cronin

unread,
Apr 2, 2013, 11:08:07 AM4/2/13
to lav...@googlegroups.com
That's great!  Thanks, Yves!

Diógenes Bido

unread,
Mar 18, 2018, 12:42:39 PM3/18/18
to lavaan
Reply all
Reply to author
Forward
0 new messages