In my study I have five latent variables and each one has different constructs.
I assume a reflective relationship between the constructs and its indicators, and also reflective relationship between the latents variables and its constructs.
These constructs were explained well by the latents variables as the R square ranged from 0. 628 to 0.936.
Moreover, previous studies use reflective model as well.
What is the best justification for choosing reflective or formative latent variable?
Thanks dear Abhay for your answer.
Yes I have second-order constructs.
What I mean by the latent variables is the main variable, and I use the term constructs to refer to the dimensions of the latents variables.
In my model I have five latent/main variables and 13 dimensions. Each variable have around 3-4 dimension.
Thanks again
and best regards
Dear Mohammed,
Are you referring to second degree constructs ??
Could you explain your second and third line?, its a bit confusing (there can be reflective/formative relationship between indicators and its constructs, but can't be in between construct; except second degree latent) (Also the term "hidden variable","construct"& "latent variable" presents the same thing).
For reflective & Formative Constructs:
If the indicators cause the latent variable and are not interchangeable among themselves, they are formative .
If the indicators are highly correlated and interchangeable, they are reflective.
You can also take a look of this file.
https://www.researchgate.net/post/Reflective_or_Formative_in_SEM-PLS?tpr_view=QaKhJKK6Kp2Qc6YluZbkgMzJgRaSGdWSBowd_3
Regards,Abhay
Thanks dear Abhay for your answer.
Yes I have second-order constructs.
What I mean by the latent variables is the main variable, and I use the term constructs to refer to the dimensions of the latents variables.
In my model I have five latent/main variables and 13 dimensions. Each variable have around 3-4 dimension.
Thanks again
and best regards
Not too sure of the meaning behind your question. However, by stepping back and recognizing that there are only 2 classes of variables in any system. Those being (1) independent and (2) dependent is a good place to start. The term "reflective" is also a property of a "mirror". Therein is a solution to your problem. For example: The independent variable is always the unknown (that's why it is called independent") until at which time it becomes known and thus becomes a dependent variable and a new independent variable is defined by the human. The process is best defined by a theoretical understanding of the dynamics of a circle which satisfies Newton's laws in motion. Specifically internal excitations equal external excitations.
Check this out:
https://www.smartpls.com/resources/ebook_on_pls-sem.pdf
Are you using a validated scale in your research? Take a look at published papers that have applied that scale to see what treatment (reflective/formative) they give to the variables in their model.
- Badges
- Science topic