Missing values

johann · Post by **johann** » Tue Jun 20, 2006 8:29 am

How does smartpls treat missing values if you only define missing values in the data, but select not to treat the missing values with either mean replacement or case wise deletion.

Best regards
Johann

stefanbehrens · Post by **stefanbehrens** » Tue Jun 20, 2006 9:27 am

It doesn't.

I.e. if you had your missings coded with say -999, SmartPLS would take these values for real and your results would obviously be very far off.

Good luck,
Stefan

johann · Post by **johann** » Tue Jun 20, 2006 10:18 am

Thank you for your answer.

It would be good if Smartpls could do pair wise deletion. If you use mean replacement you will reduce varians in the data set. Which is not so good when using PLS. Case wise deletion will reduce your data to much if you don't have to many cases.

Best regards,
Johann

stefanbehrens · Post by **stefanbehrens** » Tue Jun 20, 2006 12:59 pm

I don't think smartPLS' algorithm would allow pair-wise deletion.
The only way this could work is to use pair-wise deletion in SPSS/ SAS to calculate a correlation/ covariance matrix. Then feed this matrix into Lohmoeller's LVPLS, which is also freely available on the web, to calculate your model.

A much better approach, however, has been suggested by C. Ringle in this forum. Simply do all your missing value treatments (listwise deletion, mean replacement, imputation, EM, etc.) in the statistical package of your choice BEFORE you load the data into SmartPLS. This way you have complete control over the process.

Cheers,
Stefan