Mean and SD of latent variables

[harmindersingh]

Hi -

Does anyone know how to obtain the mean and std deviation values for latent variables? I looked at the section called "Latent Variable Scores" in the SmartPLS output file and, as I expected, the mean of those values is 0 (zero) and the SD is 1. As far as I know, that's a requirement for PLS to run.

However, I've seen papers that have published the mean and SD values for the latent variables. Do they calculate them manually (e.g. multiplying the raw data with the factor loadings)? Or are these values somewhere in the SmartPLS output file?

Thanks

[harmindersingh]

I did the calculations manually and figured out that the "Index Values for Latent Variables" in the output are the mean values for the latent variables. I don't think the SDs are available in the output.

- harminder

[allenjia]

We always need to report the mean/std/whether squrared root of AVE are above corrleation. Thus, I searched the forum and found the following answer by Dr. Bido. Hope it helps! viewtopic.php?t=754&highlight=standard+ ... t+variable

[harmindersingh]

Thanks- I'll try that..

[allenjia]

However, the problem comes. I do not know how to figure it. Maybe I understand the method in a wrong way. SD and Mean of LV are always reported in publications, but it looks like that there is not many person interested in how to do that here.


Please see:
viewtopic.php?t=2028&highlight=latent+variable+score

[Diogenes]

Hi,

1) The simple way:
Run the PLS algorithm
Report / Index Values / Results / Latent Variable Scores (unstandardized)
Copy these results to Excel
Compute the mean and s.d.
Done.

2) The complex way:
The procedure above will give you the values with some difference in the second decimal place (bug), and if you want the exact unstandardized scores, you should compute by hand as it was explained in the: viewtopic.php?t=812&highlight=unstandardized++score

Bido