forum.smartpls.com

Hi,

I am working on a construct whose model is currently formative formative, but I suspect the relationship between indicators and sub-dimensions is reflective, while the relationship between the dimensions and the focal variable remains formative.

Hence I am conducting a CTA test on the sub-dimensions, but I am not sure I am analyzing data in the right way. 0 is within the Confidence Interval, so I should conclude that my model should be reflective. But the P-value is quite high, so I suspect the results are not significative ...

What should I do ?

Here is the result of CTA for one construct :

Orig.S (O) Mean (M) STDEV T Stats P Val. Bias CI Low CI Up Alpha adj. z(1-alpha) CI Low.adj CI Up.

0.184 0.171 0.370 0.494 0.622 -0.013 -0.534 0.927 0.025 2.242 -0.638 1.032

0.097 0.093 0.330 0.293 0.769 -0.004 -0.546 0.746 0.025 2.242 -0.639 0.839

Thanks a lot for your help !

A high p-value indicates the same as a confidence interval including zero that is a non-significant (not different from zero) parameter.

You are estimating tetrads that are covariance components. If they are all non-significant you likely have a reflective model, if you have at least one significant tetrad you likely have a formative model.

Also interesting to read:
viewtopic.php?f=5&t=3887
viewtopic.php?f=12&t=16234&p=27280&hilit=CTA#p27280

Hi, thanks a lot for this reply

Kind regards