Latent Variables in Structural Equation

[sing leo]

  Dear Michel,

        I noticed that in the code and explanatory documents, latent variables are defined using personal variables. However, in my model, I believe there is no relationship between personal variables and latent variables. For example, my latent variable is environmental awareness, which I measure using three related questions about environmental awareness on a Likert scale as observed variables. In this case, the observed variables are related to the latent variable. Is it possible to define a latent variable using observed variables? If I define it this way, will it affect the measurement equation?

Thank you very much for your time and insights. I look forward to hearing from you.

[Michel Bierlaire]

> On 7 Nov 2024, at 09:18, sing leo <leo594...@gmail.com> wrote:
>
> Dear Michel,
> I noticed that in the code and explanatory documents, latent variables are defined using personal variables. However, in my model, I believe there is no relationship between personal variables and latent variables. For example, my latent variable is environmental awareness, which I measure using three related questions about environmental awareness on a Likert scale as observed variables. In this case, the observed variables are related to the latent variable. Is it possible to define a latent variable using observed variables?

No, you should not do that.

The structural equations link observed variables (such as socio-economic characteristics or attributes of the alternatives) to the latent variables. These equations are essential for forecasting.

In contrast, the measurement equations connect the latent variables to responses on the Likert scale. While they are necessary for calculating the (log) likelihood function, they are not used in forecasting.

Including Likert scale responses as explanatory variables is incorrect. Not only are these responses prone to significant measurement errors, but they also cannot be incorporated into any forecasting scenarios.

> If I define it this way, will it affect the measurement equation?
> Thank you very much for your time and insights. I look forward to hearing from you.
>

> --
> You received this message because you are subscribed to the Google Groups "Biogeme" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to biogeme+u...@googlegroups.com.
> To view this discussion visit https://groups.google.com/d/msgid/biogeme/7805a36e-b5ab-42bf-817a-5c81ce02e152n%40googlegroups.com.

Michel Bierlaire
Transport and Mobility Laboratory
School of Architecture, Civil and Environmental Engineering
EPFL - Ecole Polytechnique Fédérale de Lausanne
http://transp-or.epfl.ch
http://people.epfl.ch/michel.bierlaire

[sing leo]

  I am planning to incorporate three latent variables into the choice model. How should I correctly define the structural equations—should each of the three structural equations be defined entirely by personal attributes?    We have recently published a paper on a standard discrete choice model, which I would be happy to share with you. If you have the time, I would be incredibly grateful if you could review it and provide any suggestions on implementing a mixed choice model for this context.

[Michel Bierlaire]

> On 7 Nov 2024, at 09:38, sing leo <leo594...@gmail.com> wrote:
>
> I am planning to incorporate three latent variables into the choice model. How should I correctly define the structural equations—should each of the three structural equations be defined entirely by personal attributes?

If you model attitudes, they are indeed described by socio-economic characteristics.
If you model perceptions, they may involve alternative attributes as well.

> To view this discussion visit https://groups.google.com/d/msgid/biogeme/8ab73b42-65a1-4871-9d41-c5b44abad052n%40googlegroups.com.