Difference between robust.cluster and robust.cluster.sem

[h.park]

Dear experts,

I was just wondering what the difference is between the robust.cluster and the robust.cluster.sem options for producing standard errors for a model that contains clustered data? 

My syntax is as below:

sem.fit1 <- sem(mod1, data = data, se = "robust.cluster", 

                bootstrap = 10000, likelihood = "wishart", cluster = "Fam_ID")

summary(sem.fit1, standardized = TRUE, fit.measures = TRUE)

I've tried both and it doesn't seem to make a difference to the actual results. Any insight would be much appreciated, thank you!

[Yves Rosseel]

On 9/29/20 7:23 AM, h.park wrote:
> Dear experts,
>
> I was just wondering what the difference is between the
> robust.cluster and the robust.cluster.sem options for producing
> standard errors for a model that contains clustered data?

The distinction is similar to the (estimator) mlm vs mlr difference. The
former uses the classic formulas to compute 'robust' standard errors as
described in the SEM literature (robust.sem), while the latter uses the
'Huber-White' (or sandwich) formulas (robust.huber.white). Another way
of looking a this is that the default settings for specifying the
'bread' and the 'meat' of the sandwich formula is slightly different in
the two variants. But the results are often very similar (or indeed
identical).

This is (an old) document about it:

https://users.ugent.be/~yrosseel/lavaan/utrecht2010.pdf

The 'cluster' version simply extends these two variants to produce
cluster-robust standard errors.

Yves.

[Stas Kolenikov]

robust.huber.white is the right one. robust.sem makes an assumption that the model is correct, which may be too optimistic, and may lead to standard errors that are too small.

-- Stas Kolenikov, PhD, PStat (ASA, SSC)  @StatStas
-- Principal Scientist, Abt Associates @AbtDataScience
-- Opinions stated in this email are mine only, and do not reflect the position of my employer
-- http://stas.kolenikov.name
 

--
You received this message because you are subscribed to the Google Groups "lavaan" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lavaan+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/lavaan/86af3c2c-b74c-8202-d3fe-2b11b81d75f6%40gmail.com.

[Ranaivo Rasolofoson]

Dear Yves,

Could you precise which one (robust.cluster or robust.cluster.sem) corresponds to mlm? and to mlr?

Thank you.

Ranaivo

--
You received this message because you are subscribed to the Google Groups "lavaan" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lavaan+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/lavaan/86af3c2c-b74c-8202-d3fe-2b11b81d75f6%40gmail.com.

--

Ranaivo A. Rasolofoson, PhD, MS

Postdoctoral Associate

Department of Population Medicine and Diagnostic Sciences

Cornell University

https://fiorellaresearchgroup.com/people/ranaivo-a-rasolofoson/

Atkinson Research Fellow

Cornell University (https://www.atkinson.cornell.edu/)

Gund Affiliate, Gund Institute for Environment

University of Vermont  (https://www.uvm.edu/gund)

e-mail: rras...@gmail.com

[Yves Rosseel]

On 9/29/20 7:24 PM, Ranaivo Rasolofoson wrote:
> Dear Yves,
> Could you precise which one (robust.cluster or robust.cluster.sem)
> corresponds to mlm? and to mlr?

robust.cluster ~ mlr ~ robust.huber.white
robust.cluster.sem ~ mlm ~ robust.sem

(the 'robust.cluster' was first called 'robust.cluster.huber.white', but
that was too long)

Yves.

[Ranaivo Rasolofoson]

Thank you so much Yves.

Ranaivo

--
You received this message because you are subscribed to the Google Groups "lavaan" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lavaan+un...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/lavaan/484361fe-6ac4-727e-246b-2281e67a2ce6%40gmail.com.

[h.park]

Thank you for your answers, everyone!