Covariance or correlation matrix as input

[ghozali]

Hi
Is it possible running Smartpls using covariance or correlation matrix as data input? In covariance based SEM we can input the data using Covariance matrix

[cringle]

Hi!

The basic PLS-algorithm - as proposed by World and implemented in SmartPLS – calcuated scores for the latent variables in the structural model. Thus, scores for manifest variables are required as data input.

However, Lohmoeller (1989, viewtopic.php?t=16) and others make suggestions for correlation/covariance matrix based PLS-model estimations.

If you need such capabilities, please develop such an alternative PLS-algorithm/PLS-methodology. One of the strength of SmartPLS is its extendibility. So it might be possible to add such capabilities to the existing software platform. A sound “cooking recipe” (that describes step by step the calculations you want to perform plus data you use) or a little Visual Basic/Java program would be helpful to create such an extension of the SmartPLS software for you.

Best regards,
Christian