Dear colleagues,
My sample has a total number of 260 organizations and I would like to compare two subgroups that each one includes 130 cases (i.e. the larger and the smaller ones in size). I use a mixture of the repeated indicator approach and the two-stage approach using the latent variable scores of the 1st-order variables at the second stage.
Firstly, I would like to ask if I should use the latent variable scores (of the 1st-order variables) of the total sample or I should aggregate the latent variable scores (of the 1st-order variables) of each subgroup. I suppose that the latter choice is more appropriate.
Furthermore, what would be the recommended minimum sample size of the subgroups in my case? The largest number of indicators for a first-order variable is 8 (regarding formative indicators the largest number is 6, perhaps I may reduce them to 5-6 for both reflective and formative indicators) and for a second-order variable is 13 (regarding the repeated indicator approach). Also, there is a maximum of 4 1st-order variables for a 2nd-order variable and a maximum of 4 predictors for an endogenous variable at the structural model. I note that the percentage of participation in the survey is 80% as 260 organizations have answered in a total population of 325.
Finally, about the MICOM analysis that I have to run before the MGA, at the 2nd-stage should I consider the Permutation p-values for the total of variables or only for the variables with multiple items (i.e. for the second-order variables)? Besides, when a cell of this column is blank does it mean that the Permutation p-value is 0.000? Also, should I run a MICOM analysis at the 1st-stage too? I suppose not...
I would appreciate a relatively quick response. Thank you in advance!
Best Regards,
Panagiotis
Multi-Group Analysis for a 2nd-order Model
-
PanagPanag
- PLS Junior User
- Posts: 2
- Joined: Fri Sep 16, 2016 1:49 pm
- Real name and title: Panagiotis Panagiotopoulos, PhD Student
-
PanagPanag
- PLS Junior User
- Posts: 2
- Joined: Fri Sep 16, 2016 1:49 pm
- Real name and title: Panagiotis Panagiotopoulos, PhD Student
Re: Multi-Group Analysis for a 2nd-order Model
Dear all,
I would like to add that probably I will use the following categorisation:
A subgroup that contains 147 observations and a second one with 113 observations (and not two subgroups both having 130 observations as I reffered at the previous post).
Thank you again,
Panagiotis
I would like to add that probably I will use the following categorisation:
A subgroup that contains 147 observations and a second one with 113 observations (and not two subgroups both having 130 observations as I reffered at the previous post).
Thank you again,
Panagiotis