SEM: Mediation adjusted to confounders

126 views
Skip to first unread message

Vinícius Lionel Mateus

unread,
Jul 26, 2015, 3:40:43 PM7/26/15
to lavaan
Dear all,

The classical mediation does not account for confounders, but I need to test in my models.

So far, my literature research drove me to Coffman, Struct Equ Modeling. 2011 Jan 1; 18(3): 357–369.
He suggests the use of Propensity Scores.

Does anyone know how to implement this or another approach to adjust mediator and outcome for confounders?


Thanks in advance,

Best regards,
Vinícius

Terrence Jorgensen

unread,
Jul 28, 2015, 6:16:03 AM7/28/15
to lavaan, vinyne...@gmail.com
Propensity scores are calculated by predicting group membership using nonrandomized covariates (e.g., fitting a binary logistic regression to a binary grouping variable, then saving the predicted/fitted values, using them as propensity scores).  I'm not sure how this applies to your study... Do you have a categorical X or M?  

Regardless, this sounds like a general SEM question more than a lavaan programming question.  Perhaps you could find more feedback on SEMNET:


Terry

Vinícius Lionel Mateus

unread,
Jul 28, 2015, 7:17:28 AM7/28/15
to Terrence Jorgensen, lavaan
Hi Terry,

Thanks a lot for the ideas.
I found a way diving into the internet.

Besides the work of Coffman et al. 2011, I found really useful the work
of Butera et al. 2014.
My M is categorical, and I will made my X categorical splitting in the
median.

I am already in SEMNET, and I will consider this option if I got stuck
again.

Many thanks.

Best regards,
Vinícius
--
Best regards,

Vinícius Lionel Mateus, MSc (http://lattes.cnpq.br/6501001637020665)
Chemistry PhD Student
Atmospheric Chemistry Laboratory - Dep. Chemistry
PUC-Rio - Pontifical Catholic University of Rio de Janeiro
Marquês de São Vicente, 225, Gávea - Rio de Janeiro, RJ - Brazil CEP: 22451-900
Phone: (+45) 26 28 28 51
(+55) (21) 3527-1327
(+55) (21) 993 - 588 - 051
Skype: vinicius.lionel
http://www.qui.puc-rio.br/index.html

Terrence Jorgensen

unread,
Jul 28, 2015, 8:29:26 AM7/28/15
to Vinícius Lionel Mateus, lavaan
I found a way diving into the internet.  Besides the work of Coffman et al. 2011, I found really useful the work of Butera et al. 2014.

Good to hear.

My M is categorical, and I will made my X categorical splitting in the median.

If X is continuous, you would lose information by creating arbitrary groups, often yielding spurious results.  Consider reading this article on the topic of dichotomizing continuous variables:


Terry

Vinícius Lionel Mateus

unread,
Jul 29, 2015, 3:37:04 AM7/29/15
to Terrence Jorgensen, lavaan

> If X is continuous, you would lose information by creating arbitrary
> groups, often yielding spurious results. Consider reading this article
> on the topic of dichotomizing continuous variables:
Many thanks for the reference, Terry.

I am not so happy about this decision.
I will follow your first advice and post on SEMNET.

Best regards,
Vinícius

Vinícius Lionel Mateus

unread,
Jul 29, 2015, 1:30:58 PM7/29/15
to Terrence Jorgensen, lavaan
Dear Terry and all lavaan friends,

So far, no one has answered my question in SEMNET.

Then, I would propose a question that better fits the group.

Does any of you perform formal mediation in lavaan adjusting mediator and outcome for a latent confounder?

Where:
X =  continuous latent variable
M = dichotomous latent variable (adjusted for a latent confounder)
Y = continuous latent variable (adjusted for a latent confounder)

Best regards,
Vinícius
Reply all
Reply to author
Forward
0 new messages