Hi,
Is it sufficient to confirm (1) configural invariance and (2) compositional invariance, to conduct PLS-MGA (Multigroup analysis)?
or is it also necessary to confirm the equality of composite mean values and variances?
Thank you,
Measurement Invariance
-
MrcJnk
- PLS User
- Posts: 16
- Joined: Thu Oct 17, 2013 10:04 am
- Real name and title: Marc Janka
- Location: Germany
Re: Measurement Invariance
Dear Ten
Yes, it is sufficient to confirm both configural and compositional invariance in order to conduct PLS-MGA. When both are established, PLS-MGA can be performed with standardized variable scores each per group. Since PLS commonly works with standardized scores there are no further steps needed before conducting PLS-MGA (Henseler et al., 2015). If the two groups are homogeneous as a result of PLS-MGA you can pool your data to enhance the performance of PLS by estimating an aggregated dataset model. Nevertheless, if there are significant differences in the means and standard deviations of your variables between the two groups, I would advice to standardize your variable scores for each group seperately before pooling your dataset and then use the standardized scores as input for your model.
Best
MJ
Henseler, J., Ringle, C. M., & Sarstedt, M. (2015). Testing measurement invariance of composites using partial least squares. International Marketing Review, Forthcoming.
Yes, it is sufficient to confirm both configural and compositional invariance in order to conduct PLS-MGA. When both are established, PLS-MGA can be performed with standardized variable scores each per group. Since PLS commonly works with standardized scores there are no further steps needed before conducting PLS-MGA (Henseler et al., 2015). If the two groups are homogeneous as a result of PLS-MGA you can pool your data to enhance the performance of PLS by estimating an aggregated dataset model. Nevertheless, if there are significant differences in the means and standard deviations of your variables between the two groups, I would advice to standardize your variable scores for each group seperately before pooling your dataset and then use the standardized scores as input for your model.
Best
MJ
Henseler, J., Ringle, C. M., & Sarstedt, M. (2015). Testing measurement invariance of composites using partial least squares. International Marketing Review, Forthcoming.