Predictive Relevance for Formative Measures

Frequently Asked Questions about the SmartPLS software. Please refer to this section first, if you have problems with the software.
Post Reply
MeiPeng
PLS User
Posts: 14
Joined: Sun May 24, 2015 3:33 am
Real name and title: LowMeiPeng

Predictive Relevance for Formative Measures

Post by MeiPeng » Mon Jul 06, 2015 3:10 am

Dear All

May I ask, what are the steps and procedures to perform predictive relevance if my endogenous latent variables are formative measures?
I read that Q square through blindfolding is meant for reflective measurement model and endogenous single-item constructs. (pg. 178 in Hair, Hult, Ringle & Sarstedt, 2014, A Primer on PLS-SEM)
My case is that my ultimate targeted endogenous variable is using formative measures and it is not single-item.
Could anyone enlighten me?
My exogenous and endogenous variables are using formative measures.
What should I do for my Step 5 in Assessing PLS-SEM result, Blindfolding & predictive Relevance Q square?

Thanking in advance for all your feedback.

User avatar
cringle
SmartPLS Developer
Posts: 791
Joined: Tue Sep 20, 2005 9:13 am
Real name and title: Prof. Dr. Christian M. Ringle
Location: Hamburg (Germany)
Contact:

Re: Predictive Relevance for Formative Measures

Post by cringle » Tue Jul 14, 2015 8:07 pm

Hi

This procedure does not work for formative measurment models.

Best regards
Christian Ringle

MeiPeng
PLS User
Posts: 14
Joined: Sun May 24, 2015 3:33 am
Real name and title: LowMeiPeng

Re: Predictive Relevance for Formative Measures

Post by MeiPeng » Fri Jul 17, 2015 3:53 am

Dear prof.

Thanks for reply. May I know formative measures need to perform predictive relevance analysis or not? If yes, what would your goodself recommend?
Thank you. Looking forward for your reply.

User avatar
cringle
SmartPLS Developer
Posts: 791
Joined: Tue Sep 20, 2005 9:13 am
Real name and title: Prof. Dr. Christian M. Ringle
Location: Hamburg (Germany)
Contact:

Re: Predictive Relevance for Formative Measures

Post by cringle » Fri Jul 17, 2015 5:42 am

No recommendations exist for that case so far.

Best
CR

MeiPeng
PLS User
Posts: 14
Joined: Sun May 24, 2015 3:33 am
Real name and title: LowMeiPeng

Re: Predictive Relevance for Formative Measures

Post by MeiPeng » Fri Jul 17, 2015 1:06 pm

Dear prof.

Thanks for your swift reply. Therefore, if I'm using formative measures, I will not be penalized for not running predictive relevance analysis?

Thanks.

Man18
PLS User
Posts: 16
Joined: Mon Apr 18, 2016 1:55 pm
Real name and title: Manuela FR

Re: Predictive Relevance for Formative Measures

Post by Man18 » Mon Apr 18, 2016 2:12 pm

cringle wrote:Hi

This procedure does not work for formative measurment models.

Best regards
Christian Ringle

Hi Dr Ringle,
I need your help since I'm little confused you mean Blindfolding procedure doesn't work for formative 1st oder also ? it works when we have type II model : second order reflective-formative endogenous construct or it's not possible? I have 3 LV all type II reflective-formative 2 exogenous and 1 enondogenous I want to test predictive validity of each one is it possible with the Q square?
Thank you in advance for replying

User avatar
cringle
SmartPLS Developer
Posts: 791
Joined: Tue Sep 20, 2005 9:13 am
Real name and title: Prof. Dr. Christian M. Ringle
Location: Hamburg (Germany)
Contact:

Re: Predictive Relevance for Formative Measures

Post by cringle » Mon Apr 18, 2016 2:49 pm

You certainly can compute a Q² value. But does it really mean anything? You explain the indicators of the second order construct with its own indicators of the lower order construct. Maybe this kind of assessment is not a good idea for second order constructs.

Best regards
Christian Ringle

Man18
PLS User
Posts: 16
Joined: Mon Apr 18, 2016 1:55 pm
Real name and title: Manuela FR

Re: Predictive Relevance for Formative Measures

Post by Man18 » Tue Apr 19, 2016 9:31 pm

Thank you Dr for your kind reply. if I understand your idea predictive validity for formative endogenous construct is problematic and q square is unuseful in that case.

Post Reply