Page 1 of 1

Evaluation of CTA-PLS results

Posted: Thu Mar 31, 2016 2:13 pm
by MRodrigues
Dear all,

I constructed a third-order reflective model of psychological empowerment (see model in attachment), which is composed by:
- 12 first-order latent variables: PC, ACT, CIE, OCE, PTR, PF, SI, CC, CP, OI, SN, and SB
- 3 second-order latent variables: CE (reflected in PTR, PF, and SI), RE (reflected in CC, CP, OI, SN, and SB), and BE (reflected in ACT, CIE, and OCE).

To establish the measurement model of higher-order variables, all the indicators from the lower-order variables were assigned to the higher-order variables in the form of a repeated indicators approach.

Then, CTA-PLS was used to investigate the directionality for indicators associated with the psychological empowerment construct (see CTA-PLS results in attachment, some tetrads were ommited Due to the document's length).

1) Should I use significance test at p = 0.1 or 0.05 level?

2) Since I'm testing for multiple model-implied vanishing at the same time, should I assess the significance of the tetrads based on CI Low adj. and CI Up Adj. values?

3) Should I analyze and report both first and second-order LV results?

4) Taking into account my results, only 3 first-order LV (SI, SB and SN) and 1 second-order LV (RE) are formative ([CI Low Adj., CI Up Adj.] don't include 0). Should I conclude that psychological empowerment model s better measured formatively?


Your help will be sincerely appreciated.

Regards,
Mariana