COMMON METHOD BIAS

Frequently asked questions about PLS path modeling.
Post Reply
belinda
PLS Junior User
Posts: 3
Joined: Thu Jun 12, 2008 11:45 am
Real name and title:

COMMON METHOD BIAS

Post by belinda » Sun Nov 14, 2010 9:55 am

With regard to checking for common method bias, I have just used SmartPLS according to the following advised method:

LIANG, H.; SARAF, N.; HU, Q.; XUE, Y. Assimilation of enterprise systems: the effect of institutional pressures and the mediating role of top management. MIS Quarterly. v.31, n.1, p.59-87, mar/2007.


These authors advise that for the indicators created as single indicator constructs their "measurement error and loading have to be constrained to zero and one, respectively".


MY QUESTION: How does one accomplish this using Smartpls?
Belinda Dewsnap

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

Post by christian.nitzl » Tue Nov 16, 2010 12:58 pm

Hey Belinda,

Please check following post for an answer:

viewtopic.php?t=639&highlight=cmv

Best regards,

Christian

belinda
PLS Junior User
Posts: 3
Joined: Thu Jun 12, 2008 11:45 am
Real name and title:

CMV

Post by belinda » Tue Nov 16, 2010 2:11 pm

Thanks, but it is something quite specific from the procedure advised by Liang et al that I don't know to do, vis:

These authors advise that for the indicators created as single indicator constructs their "measurement error and loading have to be constrained to zero and one, respectively".

Hopeful thanks,
Belinda
Belinda Dewsnap

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

Post by christian.nitzl » Tue Nov 16, 2010 3:12 pm

Sorry!

Isn't that condition automatically fulfilled for a single indicator measurement?

If we use only one indicator for measurement the loading has to be 1 and therefore the measurement error is 0. I am not fully understand the comment of Liang et al. ;-)

Please correct me, if I am wrong!

Best regards,

Christian

belinda
PLS Junior User
Posts: 3
Joined: Thu Jun 12, 2008 11:45 am
Real name and title:

CMV

Post by belinda » Tue Nov 16, 2010 3:16 pm

Yes, that's right. So, their instruction is somewhat redundant!

Thank you very much for resolving this; it is much appreciated.

Best wishes,
Belinda
Belinda Dewsnap

mronkko
PLS User
Posts: 14
Joined: Tue Jan 06, 2009 6:45 pm
Real name and title:

Post by mronkko » Tue Dec 07, 2010 1:35 pm

You should note that the Liang et al. does actually nothing to control for method variance. The reason for this is explained in detail in our paper that is cited in this thread. We also provide an alternative with simulation evidence supporting the method.

viewtopic.php?t=1573

Post Reply