In attachment, you can find the model.
I didn't include any global item to assess PE construct in the questionnaire (Hair et al. 2014). What can I do to assess the convergent validity of this model?
I will appreciate your help.
Since 2nd order constructs usually are multidimensional, this kind of analysis does not work for 2nd order constructs. A second-order construct itself already is very difficult to establish and to justify content-wise; here it's multidimensional content may consist of the lower order constructs (it is what you define) and no additional tests may be needed.
As per the words of Christian Ringle
Can be accessed at http://forum.smartpls.com/viewtopic.php?t=15579
Redundancy analysis: You should calculate the variance inflation factor (VIF) to assess multicolinearity among the independent variables. Values higher than 10 indicate high multicolinearity (Henseler et al., 2009).
Convergent validity: The formative construct must be correlated with a global measure (when I did that, I used the first item of each formative dimension to create this global measure). Results above .8 are considere adequate.
Thank you Larissa! Do you have any reference to justify the way how you explored the convergent validity (using the first item of each formative construct)?
I believe there may be something in this book, but not about the way I created the global measure. I believe that, if you have a global measure, it would be more apropriate to use it.
HAIR, J. F.; HULT, G. T. M., RINGLE, C. M.; SARSTEDT, M. A primer on partial least squares structural equations modeling (PLS-SEM). Los Angeles, SAGE: 2014.
Thanks Mariana Rodrigues for asking a such important question, if you have already found any solution, please share
Using my previous PE model:
To evaluate the convergent validity, the PE construct was included in a wider nomological network, which encompasses quality of participation experiences construct, and their relationship was measured (Bagozzi, 1994). It was hypothesized that quality of participation experiences is a consequence and/or antecedent of PE (Rodrigues, Menezes, & Ferreira, 2015; Veiga, 2008). If the previous hypothesis is verified, the convergent validity of the model is supported (Diamantopoulos & Winklhofer, 2001).
There isn't any information in Hair et al(2014) "A Primer on PLS SEM" on how to create global measure. Larissa could you please share in detail how you have created the global measure?
Since 2nd order constructs usually are multidimensional, this kind of analysis does not work for 2nd order constructs. A second-order construct itself already is very difficult to establish and to justify content-wise; here it's multidimensional content may consist of the lower order constructs (it is what you define) and no additional tests may be needed.
As per the words of Christian Ringle
Can be accessed at http://forum.smartpls.com/viewtopic.php?t=15579
as far as i concern, you need to create the global item which will measured the overall of the measurement. I don't see any other way except using the global item. What has been done by the previous author above is not suggested by the experts in this area. Personally, i did not agree with replacing the global item with the 1st item for the dimension, because it will not work as global item as it should be.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Structural Equation Modeling