Direction of repeated indicators for a Type II 2nd-order LV

moataz · Post by **moataz** » Wed Apr 11, 2012 9:30 pm

Dear All,

In my model, I have one Type II second‐order factor (that has first-order factors as formative indicators, while those first-order factors themselves have reflective indicators). I only have one endogenous LV in my model.

Now, to specify the model, I need to use repeated indicators. However, the question is whether -in my case- I should have the repeated indicators loading on the 2nd order factor as reflective (Mode A) or formative (Mode B)? Wetzels et al.'s (2009) MISQ paper had all-reflective 2nd order factor (first-order factors as reflective indicators, and those first-order factors having reflective indicators). The authors used Mode A for repeated indicators. On the other hand, also in MISQ, Rai et al. (2006) had an all-formative 2nd order model (first-order factors as formative indicators, and those first-order factors having formative indicators). As expected, the authors used Mode B for repeated indicators. I have came across no study, however, that specifically indicated the direction of repeated indicators for Type II factors. Any advice is greatly appreciated, thank you.

Kind Regards,

Moataz Soliman

Hengkov · Post by **Hengkov** » Fri Apr 13, 2012 6:17 am

Hi Moataz,
For you construct Type II (reflective first-order, formative second-order) => repeated indicators approach used Mode A (second order), but this is difficulties of identification when using Type I and Type IV.
Regards,
Hengky

moataz · Post by **moataz** » Fri Apr 13, 2012 2:10 pm

Hi Hengky,

(If you prefer that I address you with your first name, please tell me!)

Thanks again, this is the second question you answer for me in two days, really appreciated.

You know the strange thing is that if use Mode A, the R2 for the endogenous variables decrease by almost 50%! This is something I may have just to live with, but another puzzling issue is that - if I remove the second-order factor from the model and connect the first-order ones directly to the endogenous factors - the R2 goes back to the values that I've seen when I used Mode B!

One more thing, if there is any reference you can point to, it would be great. Thank you again.

Kind Regards,

Moataz

Hengkov · Post by **Hengkov** » Fri Apr 20, 2012 1:20 am

Hi Moataz,
For reference please some article below:
Ringle et al. 2012. “A Critical Look at the Use of PLS-SEM I MIS Quarterly,” MIS Quarterly (36:1), pp. iii-xiv. => look Appendix B: Hirarchical Component Models.

Polites, G.L., Roberts, N., and Thatcher, J.B. 2012. “Conceptualizing Models Using Multidimensional Constructs: A Review and Guidelines for their Use.” European Journal of Information Systems (21:1), pp. 22-48. => look p. 30 reflective first-order, formative second-order (Type II) and look p.36 dimension sets: a critique of multidimensional constructs.

Peter, S., Straub, D., and Rai, A. 2007. “Specifying Formative Construct in Information Systems Research,” MIS Quarterly (31:4), pp. 623-656.

MacKenzie, S. B., Podsakoff, P.M., and Jarvis, C.B. 2005. “The Problem of Measurement Model Misspecification in Behavioral and Organizational Research and Some Recommended Solutions,” Journal of Applied Psychology (90:4), pp. 710-730.

Jarvis, C.B., Mackenzie, S.B.,and Podsakoff, P.M. 2003. “A critical review of construct indicators and measurement model misspecification in marketing and consumer research,” Journal of Consumer Research, (30:2), pp. 199–218.

Regards,
Hengky

moataz · Post by **moataz** » Fri Apr 20, 2012 1:33 am

Hi Hengky,

Thank you very much, I really appreciate your help and advice.

Have a nice day.

Regards,

Moataz

Hengkov · Post by **Hengkov** » Fri Apr 20, 2012 2:06 am

Ok, Your welcome and good luck.
Regards,
Hengky