2nd-order construct & AVE: smartPLS vs. manual calculation

Questions about the implementation and application of the PLS-SEM method, that are not related to the usage of the SmartPLS software.
Post Reply
SHU
PLS Junior User
Posts: 10
Joined: Tue Jan 03, 2017 5:58 pm
Real name and title: Susanne Hügel
Location: Wiesbaden (Germany)

2nd-order construct & AVE: smartPLS vs. manual calculation

Post by SHU »

Hello community,

I have a first-order reflective second-order formative construct (type II) being an exogenous variable in the structural model. Its subconstructs have an unequal number of items, thus, I use the two-stage approach. In order to assess validity and reliability of the reflective subconstructs, I have run PLS algorithm in the first stage to obtain the latent variable scores and to check the AVE and CR values in the quality criteria section - everything looks fine (AVE > 0.50, CR > 0.70). But, when I calculate AVE and CR manually with Excel (factor loadings have been generated in SPSS using principal component analysis with varimax rotation), the values are different (i.e., lower) in comparison to the values shown in smartPLS.

Why is that? Shouldn't it be the same? Is it incorrect to assess AVE and CR of the reflective subconstructs in the first stage of the two-stage approach?

p.s. The cronbach values calculated in both programmes are identical.

Many thanks for you reply!
jmbecker
SmartPLS Developer
Posts: 1282
Joined: Tue Mar 28, 2006 11:09 am
Real name and title: Dr. Jan-Michael Becker

Re: 2nd-order construct & AVE: smartPLS vs. manual calculati

Post by jmbecker »

AVE and CR are based on loadings and thus model dependent. PLS uses a different modelling approach than PCA. Thus, the results are different.
In particular, PLS incorporates the information from other related constructs via the structural model in the estimation of weights and loadings. PCA does not incorporate structural model information.
If you use the first-order scores from a PLS model, you should report those AVE and CR values. If you think that the first-order model should not take the structural model into account (and you need a good reason for that) you can use the PCA scores and the PCA estimates of AVE and CR. But you should not mix the two.

Cronbachs Alpha is based on correlations and thus model independent.
Dr. Jan-Michael Becker, BI Norwegian Business School, SmartPLS Developer
Researchgate: https://www.researchgate.net/profile/Jan_Michael_Becker
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de
SHU
PLS Junior User
Posts: 10
Joined: Tue Jan 03, 2017 5:58 pm
Real name and title: Susanne Hügel
Location: Wiesbaden (Germany)

Re: 2nd-order construct & AVE: smartPLS vs. manual calculati

Post by SHU »

Thank you very much for the detailed answer! Now, it sounds clear to me.
Post Reply