Most users seem to have no problems installing and running WarpPLS. A minority have problems with the MATLAB Compiler Runtime, which can be addressed by following the instructions on this post.

A very small number of users seem to be unable to properly install and run WarpPLS, even after following the instructions above. A common, but not exclusive, error message in this case is: “An application has made an attempt load the C runtime library incorrectly”.

One common characteristic among these users is that they have the software XLSTAT installed on their computers. I have already received a few reports suggesting XLSTAT changes operating system settings in such a way as to prevent WarpPLS from properly running.

When those users removed XLSTAT from their computers, they were able to run WarpPLS without problems.

## Sunday, September 26, 2010

## Wednesday, September 15, 2010

### There is no need for two-way arrows in WarpPLS

In covariance-based structural equation modeling (SEM) software tools, often one has to explicitly model correlations between predictor latent variables (LVs). In WarpPLS, correlations between predictor LVs are automatically taken into consideration in the calculation of path coefficients.

That is, the path coefficients calculated by WarpPLS are true standardized partial regression coefficients, of the same type as those calculated through multiple regression analysis. The difference is, of course, that in WarpPLS the model variables are LVs, which are usually measured through more than one indicator. With multiple regression, only one measure (or indicator) is used for each variable in the model.

That is, the path coefficients calculated by WarpPLS are true standardized partial regression coefficients, of the same type as those calculated through multiple regression analysis. The difference is, of course, that in WarpPLS the model variables are LVs, which are usually measured through more than one indicator. With multiple regression, only one measure (or indicator) is used for each variable in the model.

Subscribe to:
Posts (Atom)