f2 is greater than 1

Frequently asked questions about PLS path modeling.
Post Reply
nadja-mazh
PLS Junior User
Posts: 2
Joined: Mon Apr 04, 2011 10:20 am
Real name and title:

f2 is greater than 1

Post by nadja-mazh » Tue Nov 08, 2011 10:31 pm

I need help with my PLS model to.

All quality criteria of it are actually met, but the effect size (f2) of my exogenous variables is greater than 1.

I have 3 IV and 2 DV: A->D, B->D, C->D; A->E, B->E, C->E

What do means an effect size greater than 1? Where I made a mistake?

I'm confused and don't know how to proceed further ...

Thank you very much.

Nadja

christian.nitzl
PLS Expert User
Posts: 248
Joined: Sat Jul 25, 2009 1:34 pm
Real name and title:

Post by christian.nitzl » Wed Nov 09, 2011 9:59 am

Your results should not disturb you. There is no such thing for the effect size that it should be smaller than 1. A higher effect size than 1 means that your exogenous variable has a very big effect on your endogenous variable.

Cheers,

Christian

nadja-mazh
PLS Junior User
Posts: 2
Joined: Mon Apr 04, 2011 10:20 am
Real name and title:

Post by nadja-mazh » Mon Nov 14, 2011 4:03 pm

Hey Christian and all,

thank you very much!

Could you perhaps recommend a pair of references to the topic, which I can mention in my diploma thesis ...

thank you!

Nadja

christian.nitzl
PLS Expert User
Posts: 248
Joined: Sat Jul 25, 2009 1:34 pm
Real name and title:

Post by christian.nitzl » Mon Nov 14, 2011 4:53 pm

Hey Nadja,

The typical references for effect size are:

Chin, W. (1998): Issues and Opinion on Structural Equation Modeling, in: ManagementInformation Systems Quarterly, Vol. 22, No. 1, p. 317.

Cohen, J. (1988): Statistical Power Analysis for the Behavioral Sciences, 2nd edition,
Hillsdale, p. 413.

Some further nice information for the effect size you can also find in my working paper (only German) on pp. 34 f.:

https://www.unibw.de/ifc/team/wimi/nitz ... load/down1

Greetings,

Christian

MBauch
PLS Junior User
Posts: 6
Joined: Mon Jan 02, 2012 11:52 am
Real name and title:

Post by MBauch » Tue Feb 14, 2012 12:17 pm

Dear christian!

I have a brief question regarding the effect sizes:

I have a model with 4 IV and 1 DV, R² is 0.54. It showed that two of the paths are significant, but even these two paths only have effect sizes of 0.11 and 0.12. How can that be explained? Is it dependent on the measurement errors?

Thank you!

Matthias

christian.nitzl
PLS Expert User
Posts: 248
Joined: Sat Jul 25, 2009 1:34 pm
Real name and title:

Post by christian.nitzl » Tue Feb 14, 2012 3:38 pm

Please check following post at the end for a possible answer:

viewtopic.php?t=1453&postdays=0&postorder=asc&start=60

Cheers,

Christian

Theo2711
PLS Junior User
Posts: 4
Joined: Sat Sep 01, 2018 8:30 am
Real name and title: Theo Limt

Re: f2 is greater than 1

Post by Theo2711 » Sat Sep 29, 2018 7:13 pm

In my analysis I have one negative significant moderating effect and a non-significant one. The R² of the model with the moderating effects is 0.850 as opposed to R² of 0.2108 in the model without the interaction effects. This leads to an effect size of over 4. Is this still possible? And according to Cohen 1998 who stated that the definition of a high effect size could be different in different fields?

I would appreciate an answer.

Thanks!

jmbecker
SmartPLS Developer
Posts: 879
Joined: Tue Mar 28, 2006 11:09 am
Real name and title: Dr. Jan-Michael Becker

Re: f2 is greater than 1

Post by jmbecker » Thu Oct 04, 2018 8:01 am

That sounds very strange. Effects sizes of moderating effects are usually quite small. Do you use the f² effect size from the SmartPLS output or a self-calculated one?
Dr. Jan-Michael Becker, University of Cologne, SmartPLS Developer
Researchgate: https://www.researchgate.net/profile/Ja ... v=hdr_xprf
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de

Post Reply