[Mitarbeiter.zoologie] Statistical club meeting - SEM

[Panagiotis Theodorou]

Dear all,

It seems that I missed a very interesting seminar....
My view on this is somehow similar to Lefcheck. In the SEM framework the
goal is not to find the best model that fits the data. In the SEM, you test
a theory and then you consider whether or not the model is consistent with
the data. However, in this regard, do correct models really exist? Are
there any equivalent or near-equivalent meaningful models or hypotheses?
and what about parsimony?
Since I started using this framework the debate is still on in regards to
model selection and goodness-of-fit assessment in the SEM framework and
beyond..
In general someone cannot avoid subjective judgement (similar to all
statistical methods) that should have a solid theoretical basis.

All the best,

Panas


On Sat, Feb 3, 2024 at 10:34 AM Antonella Soro <
antonella.soro at zoologie.uni-halle.de> wrote:

> Dear Lucie,
>
> I thank you very much for this material and for your presentation, which
> was definitively useful in many respects. I understood things I did not
> understand before.
>
> A couple of hours after your presentation, i have attended the weekly
> meeting with richard McElreath and asked him specifically about whether it
> is appropriate to use AIC criteria when running SEM and he reiterated that
> it is not (his words: “it is wrong”).
>
> even if less categorically Jon Lefcheck says the same thing:
>
> *“*Users should take caution when exploring these techniques as to avoid
> dredging the model. SEM is a technique that relies heavily on informed
> model specification. Adding paths in that are suggested by the data but not
> anticipated by the user only in an effort to achieve adequate fit might be
> appropriate in other applications, but ignore the basic philosophy behind
> SEM that relationships are based on *a priori *knowledge and intuition*” *[italics
> is Lefcheck’s]*.*
>
> Thanks again and a good week-end to all,
>
> Antonella
>
>
>
>
>
>
> On 2. Feb 2024, at 14:19, Lucie M. Baltz <
> lucie.baltz at zoologie.uni-halle.de> wrote:
>
> Hey everyone,
>
> attached it the Presentation and some Paper that might be usefull for you.
> In my opinion the one from Shipley and the one from Garrido are the most
> helpful ones.
>
> If you want to use SEM, this link will be the most helpful!
> https://jslefche.github.io/sem_book/index.html
>
> Best,
> Lucie
>
> (she/her)
> PhD candidate
> General Zoology
> Martin-Luther-University Halle-Wittenberg
> Hoher Weg 8
> 06120 Halle (Saale)
> lucie.baltz at zoologie.uni-halle.de
>
>
> <10705511.2020.1871355.pdf><Ecology - 2013 - Shipley - The AIC model
> selection method applied to path analytic models compared using a
> d‐separation.pdf><Methods Ecol Evol - 2021 - Garrido - A model selection
> approach tostructural equation modelling  A critical evaluation and.pdf>
> <nihms661496.pdf><qt490131xj.pdf>
> <rohrer-et-al-2022-that-s-a-lot-to-process-pitfalls-of-popular-path-models.pdf>
> <s11135-017-0469-8.pdf><Structural Equation Modelling.pptx>
> _______________________________________________
> Mitarbeiter.zoologie mailing list
> Mitarbeiter.zoologie at lists.uni-halle.de
> https://lists.uni-halle.de/cgi-bin/mailman/listinfo/mitarbeiter.zoologie
>
>
> _______________________________________________
> Mitarbeiter.zoologie mailing list
> Mitarbeiter.zoologie at lists.uni-halle.de
> https://lists.uni-halle.de/cgi-bin/mailman/listinfo/mitarbeiter.zoologie
>


-- 
==========================================
Dr. Panagiotis Theodorou
General Zoology, Institute of Biology
Martin Luther University Halle-Wittenberg
Hoher Weg 8, Room 6.10
06120 Halle (Saale), Germany
Phone: +49(0)345 55 26511
Email: panatheod at gmail.com
Web: https://panatheod.wixsite.com/panagiotis-theodorou
<http://panatheod.wixsite.com/panagiotis-theodorou>
===========================================
-------------- nächster Teil --------------
Ein Dateianhang mit HTML-Daten wurde abgetrennt...
URL: <http://lists.uni-halle.de/pipermail/mitarbeiter.zoologie/attachments/20240203/bb539b6d/attachment.html>