Hi to all,
While applying the two stage approach, shall we use the latent variable scores ( we get from the first satge ) only for the higher order constructs, or for all of the constructs and variables in the structural model? I have only one higher order construct with two lower order constructs, and the remaining constructs are all single order, and I also have control variables. Do I need to use the scores from the first step for the control variables too?
I'll be more than happy if anyone can respond and help about this issue.
Best
latent variable scores usage in second order constructs
-
Hengkov
- PLS Super-Expert
- Posts: 1599
- Joined: Sun Apr 24, 2011 10:13 am
- Real name and title: Hengky Latan
- Location: AMQ, Indonesia
- Contact:
Re: latent variable scores usage in second order constructs
Two-step approach with latent variable score is the last option. Why do not you use another approach such as repeated, hybrid and new approaches?
-
naime
- PLS Junior User
- Posts: 7
- Joined: Fri May 13, 2016 9:29 am
- Real name and title: Finance PhD student in Bilkent University, Naime Usul
Re: latent variable scores usage in second order constructs
Dear Hengkov,
The problem is that one of my first order constructs is formative and the other is reflective. In the repeated indicator approach I couldn't decide on how to handle this.
In the two stage approach, since at first stage I have the first order constructs separately, I can obtain the latent variable scores, one for formative and the other for the reflective first order construct. And then at the second stage I can use these scores and calculate the second order construct (note that both of the first order constructs are formative in the second order construct) But I am not sure whether I need to use the latent variable scores only for the first order constructs and use the original data for the rest of the variables, or I need to use the latent variable scores from the first stage for all of the constructs and variables included in the structural model.
The problem is that one of my first order constructs is formative and the other is reflective. In the repeated indicator approach I couldn't decide on how to handle this.
In the two stage approach, since at first stage I have the first order constructs separately, I can obtain the latent variable scores, one for formative and the other for the reflective first order construct. And then at the second stage I can use these scores and calculate the second order construct (note that both of the first order constructs are formative in the second order construct) But I am not sure whether I need to use the latent variable scores only for the first order constructs and use the original data for the rest of the variables, or I need to use the latent variable scores from the first stage for all of the constructs and variables included in the structural model.