What is Endogeneity in SEM (Structural Equation Modeling)
Вставка
- Опубліковано 14 лип 2024
- This video explains what is endogeneity and the causes of endogeneity. It also explains how to address endogeneity through a two stage least squares analysis. The topic of endogeneity is addressed in a structural equation model.
If you are looking for more information about structural equation modeling, see my book titled, Applied Structural Equation Modeling using AMOS. The book can be purchased at:
Amazon:
www.amazon.com/Applied-Struct...
and
Barnes and Noble:
www.barnesandnoble.com/w/appl
Dear Dr. Collier,
My name is Haerold and I am one of the subscribers of your UA-cam Channel. We are currently publishing in an A* journal and one of the major issues raised for our revision is did we address the issue of endogeneity. Our model is a fully latent mediated model with 3 independent variables, 2 mediating variables, 1 dependent variable, 3 control variables, and a marker variable to control for CMB using the method of Williams et al. (2010). We used SPSS and AMOS as tools for analyzing our structural model.
In this video you recognize the difficulties/challenges of instrumental variable using two-stage least square for compound or serial models because of the presence of multiple DVs in which some DVs are mediating variables acting as IVs for other DVs in the model. As an alternative, you then propose an "omitted variable" as control variable to address the confounding effects in the model and hold constant some of the other possibly omitted variables that could have bias on the results. In that way, the said control variable can control for, or influence all the dependent variables in the model, hence not addressing the endogeneity problem as piecemeal but assessing it as a whole. Furthermore, you added that the use of additional omitted variables as control variable can further give validity to the results found between IVs and DVs.
As one of the world's renowned experts in SEM, do you have published procedure (either article, book, or video) on how to conduct endogeneity test and address endogeneity issues using AMOS that we can follow or cite? I checked your book and it seems that the for "omitted variable" as control variable to address endogeneity issue is not discussed. I am looking at the possibility of using one or a combination of our control variables and evaluate its merit as our instrumental variable since we can no longer collect data for a separate instrumental variable. We wanted to cite an expert like you on this regard in our ongoing article.
Hoping to hear from you soon.
Sincerely,
Haerold
Haerold, Endogeneity is not addressed in my book. My publisher put me on a page count and this got cut from the first volume of the book. I am hoping to address it when I do volume 2. In the meantime, here is a resource for you. There is an article called "endogeneity and marketing strategy research: an overview" in the Journal of the Academy of Marketing Science by Oliver Rutz and George Watson. Here is a link to the article:
link.springer.com/article/10.1007/s11747-019-00630-4
It should be a good resource for you to cite. I hope that helps.
Prof. @@joelcollier9387 I already checked the article you referred me (i.e., Rutz & Watson, 2019). Aside from it is published in a 4* & A* ranked journal, the principles and explanations that I am looking to further justify and strengthen our methodology can be found in the said article. This is really a great help for us. Looking forward to reading the Volume 2 of your book!
Sincerely,
Haerold
@@joelcollier9387 This helped me. Thank you.
Dear Sir, do we choose control variables based only on theory; What if we suspect endogeneity but we ommited in our survey to include the "right" control variable based on theory;
Yes, the control variable is based on theory or even other possible research that states a variable might be the "true" cause of the relationships. If you do not have a control variable to include, then your only option is to use existing literature and theory to justify why your chosen variables are having the influential effect.