Formative vs reflective construct

[eflina]

Hi,

I have an endogenous formative variable which I initially treated as "formative first order and second order" construct. In relation to other exogenous variable, the regression weight was almost zero. But when I inverted the second order of the endogenous variable to reflective (so it's now conceptualized as "first order formative and second order reflective"), I got a satisfying regression weight from exogenous to endogenous (it's increased to .40). Question is: to what degree do I have a freedom to determine whether a construct should be treated as full formative or half formative? Does it make sense if I say that "first order formative and second order reflective" is a formative construct? Thank you.

Debby

[Hengkov]

Hi,
Formative or Reflective based your conceptualization constructs.
Next, used repeated indicators for check significant weight, Why your mention weight is zero?

[eflina]

That's what I was asking about. When I conceptualized the second order relationship with LV (the moderator) as formative, the path coefficient of my IV to MV became zero. However, when I inverted them into reflective, the path coefficient became significant (almost. 40). I had no choice than treating them as first order formative second order reflective.

Furthermore, can SmartPLS test logistic regression? Thanks for your help.

Bests,
Debby

[Hengkov]

Hello,

You have Type III 2nd order. For test this model used combination Repeated Indicators and Two-Stage. Next, PLS cannot ran Logistic Regression with categorial dependent variable.

Regards,
Hengky

[jmbecker]

You might want to read

Becker, Jan-Michael, Kristina Klein and Martin Wetzels (2012): Hierarchical Latent Variable Models in PLS-SEM: Guidelines for Using Reflective-Formative Type Models, Long Range Planning (LRP), Vol. 45, Issue 5-6, 359-394.

It discusses some of your problems.

[ahmeduum]

Hi,

I have an endogenous formative variable which I initially treated as "formative first order and second order" construct. In relation to other exogenous variable, the regression weight was almost zero. But when I inverted the second order of the endogenous variable to reflective (so it's now conceptualized as "first order formative and second order reflective"), I got a satisfying regression weight from exogenous to endogenous (it's increased to .40). Question is: to what degree do I have a freedom to determine whether a construct should be treated as full formative or half formative? Does it make sense if I say that "first order formative and second order reflective" is a formative construct? Thank you.

Debby

You might want to read

Becker, Jan-Michael, Kristina Klein and Martin Wetzels (2012): Hierarchical Latent Variable Models in PLS-SEM: Guidelines for Using Reflective-Formative Type Models, Long Range Planning (LRP), Vol. 45, Issue 5-6, 359-394.

It discusses some of your problems.

Dear Debby,

I have that article as mentioned by Dr. Becker. If you want you may PM me. Thanks Dr. Becker for reffering this article.

[eflina]

Dear Ahmed,

Thank you so much for the offer, I already have the article though. All the best!

[bwilson]

You might want to read

Becker, Jan-Michael, Kristina Klein and Martin Wetzels (2012): Hierarchical Latent Variable Models in PLS-SEM: Guidelines for Using Reflective-Formative Type Models, Long Range Planning (LRP), Vol. 45, Issue 5-6, 359-394.

It discusses some of your problems.

Agreed. Start here.
This is a good suggestion.
For type I reflective reflective contrast see:
Handbook of Partial Least Squares. http://www.springer.com/statistics/comp ... 40-32825-4

Chapter by Wilson.
Using PLS to Investigate Interaction Effects Between Higher Order Branding Constructs.

Best regards

Brad