Formative second-order construct Type IV

Frequently asked questions about PLS path modeling.
Post Reply
PLS Junior User
Posts: 3
Joined: Wed Jan 16, 2013 11:00 am
Real name and title:

Formative second-order construct Type IV

Post by Troasi » Sat Jun 01, 2013 3:45 pm

Albers and Götz (2006) suggest to handle higher order constructs by first running a preliminary model without the second-order construct and then running the main model with computed weighed mean values of the first-order constructs. Does anybody have experience with that approach?

Moreover, Ringle et. al. (2012) have pointed out that the indicator reuse approach is problematic with an unbalanced number of indicators. Is the same true for the alternative ways to handle Type IV constructs (e.g., first conducting a confirmatory factor analysis followed by an examination of structural relationships)?

Albers, Sönke and Götz, Oliver 2006. "Messmodelle Mit Konstrukten Zweiter Ordnung in Der Betriebswirtschaftlichen Forschung. (German)." Die Betriebswirtschaft 66(6): 669-77.

Ringle, Christian M., Sarstedt, Marko and Straub, Detmar W. 2012. "Editor's Comments: A Critical Look at the Use of Pls-Sem in Mis Quarterly." MIS Quarterly 36(1): iii-xiv.

User avatar
PLS Super-Expert
Posts: 1618
Joined: Sun Apr 24, 2011 10:13 am
Real name and title: Hengky Latan
Location: AMQ, Indonesia

Post by Hengkov » Sun Jun 02, 2013 12:57 pm


Yes, I know this approach. In WarpPLS this approach applied. First, you ran all 1st construct dimension for result LVs. Next, used this LVs for indicators 2nd OK.


Post Reply