Measurement model for longitudinal data

Frequently asked questions about PLS path modeling.
Post Reply
PLS Junior User
Posts: 1
Joined: Thu Oct 27, 2011 9:16 am
Real name and title:

Measurement model for longitudinal data

Post by garbers » Tue Feb 05, 2013 11:10 am


I have some questions concerning the assessment of the measurement model with longitudinal data.
The situation: We used a cross-lagged full-panel model with three waves. Within each wave we measured our constructs in the same way and each wave is a predictor for the next wave. When we analyzed the full model with the three waves, our measurement model (internal consistency, convergent and discriminant validity) was not acceptable. More in detail, the CRs and AVEs were small (CR < .6; AVE < .5), and the square root of the AVE of each construct was not greater than their interconstruct correlations.

1) Has anyone similar problems with the measurement model of longitudinal data and a solution?

2) Are there special recommendations/literature for the measurement model when modeling longitudinal data with Smart PLS?

Thanks a lot and best regards,

User avatar
PLS Super-Expert
Posts: 1619
Joined: Sun Apr 24, 2011 10:13 am
Real name and title: Hengky Latan
Location: AMQ, Indonesia

Post by Hengkov » Mon Feb 11, 2013 10:59 pm


If you used secondary data for longitudinal analysis, not evaluation outer model.

Check my paper below:

Free online access in: ... s.pdf.html

PLS Junior User
Posts: 2
Joined: Tue Mar 21, 2017 7:53 am
Real name and title: Chris

Re: Measurement model for longitudinal data

Post by Lupus » Thu Feb 08, 2018 7:53 am


I have a similar issue with longitudinal data. Unfortunately, the link in the previous pose is not accessible anymore.

Is there any solution for that issue?

Many thanks in advance.

Post Reply