forum.smartpls.com

Dear Sir

Can we model Multi-dimensional construct in smartPLS. Now if we can model it then do we need to check for cronbach alpha and other reliability and validity measures for each dimension or we need to check only for a construct.

Thanks
Ruchi

Hi,

Yes, see

WETZELS, M.; ODEKERKEN-SCHRÖDER, G.; VAN OPPEN, C. Using PLS path modeling for assessing hierarchical construct models: guidelines and empirical illustration. MIS Quarterly, v. 33, n. 1, p. 177–195, 2009. [p.181]

Yes, we should assess the AVE, CR for all LV (1st and higher order LV).

See viewtopic.php?t=640

Best regards,

Bido

Dear Sir

I read the paper of wetzels, Some of the doubts are -

1) should only check for first order or we should also check for second order also??

2) If my first order LV is having AVE above 0.5 , composite reliability above 0.7 but cronbach alpha is less than 0.6 (say 0.5) (also checked in spss and smartpls-same cronbach alpha comes), then should i delete that dimension and make it unidimensinal ??

3) Now if i see my second order LV cronbach alpha with all items of my first order latent variables , it goes below 0.6 and also AVE is o.4.

Please let me know about it.

Thanks
Ruchi

1) should only check for first order or we should also check for second order also??
Yes, see viewtopic.php?t=640

2) If my first order LV is having AVE above 0.5 , composite reliability above 0.7 but cronbach alpha is less than 0.6 (say 0.5) (also checked in spss and smartpls-same cronbach alpha comes), then should i delete that dimension and make it unidimensinal ??
Consider just C.R. (alpha is useful in the summated rating scale context). Do not delete the dimension.


3) Now if i see my second order LV cronbach alpha with all items of my first order latent variables , it goes below 0.6 and also AVE is o.4.
You should compute the AVE and C.R. by hand, just with the paths from 2nd order to the 1st order LV,
See viewtopic.php?t=640

Best regards,

Bido

Dear Sir

Thanks a lot sir for such a quick reply and solving all my queries.

Warm Regards
Ruchi

It means if my AVE or CR anyone of them is less than their critical values, then only i should remove those dimensions , should not look cronbach alpha.

Regards
Ruchi

Dear Sir


I have a independent variable (construct) - "Usefulness" . when i did factor analysis of this construct 3 components were extracted -
1)economic ( 3 items)
2) efficiency (4 ietms)
3) emergency (2 items)

It means it is not uni dimensional. It has 3 dimensions.
Now i tried to model it in smartPLS as

economic dimension measured by 3 items in reflective mode (E1, E2, E3)

efficiency dimension measured by 4 items in reflective mode (EF 1, EF 2, EF 3, EF 4)

emergency dimension measured by 2 items in reflective mode (EM 1, EM 2)

These 3 dimensions i can term as first order contructs

Then I connected all these 3 dimension with my construct usefulness (PU)
PU ( all the items of all dimensions included in reflective form(E1, E2, E3, EF 1, EF 2, EF 3, EF 4,EM 1, EM 2). This becomes my second order construct

and then i run PLS.

I found that composite reliability and AVE are fine (above cut off value) for 3 dimensions (first order construct) but AVE for usefulness (second order construct) is low as 0.35.

Please guide me for this issue. I dont know why its happening and how should i model this.

Thanks
Ruchi

Dear,

do you have a reference that I can use if my cronbach's alpha of a lower-order component is less than 0.6 (i.e. 0.525)? My AVE and CR are good.

Thank you very much.

Kind regards,

Yes. Please review this latest open-access article published in the Journal of Retailing where we have developed a measurement model (scale) for the multi-dimensional construct named OCX (omnichannel customer experience): https://doi.org/10.1016/j.jretai.2022.03.003. Table 1 in the article offers an extensive step by step guideline for scale development and model validation. Also, review the web appendix for details. This article has been included in the SmartPLS recommended reading list https://www.smartpls.com/documentation/ ... -readings/

Article reference: Rahman, S. M., Carlson, J., Gudergan, S., Wetzels, M., Grewal, D. (2022) Perceived Omnichannel Customer Experience (OCX): Concept, Measurement, and Impact. Journal of Retailing, https://doi.org/10.1016/j.jretai.2022.03.003