Validation of Statistical Software

Has anyone had any experience with validation of statistical software such as SAS JMP and Minitab.

Could their be an argument that if the software supplier has performed validation activities (validating that the statiscal functions compute correctly), that the end user would just confirm the installation is correct.

To validate all the functions in JMP / Mintab would be a huge task.

Another approach would be just to validate for the few tasks you intend to use the software for. This would create a difficulty in ensuring that no one used the functions you have not validated.

Any opinions?

Your spot on there. Do a standard IQ and in the OQ and PQ, validate purely the functions you use.

We recently used this method to validated an SPSS stats system that was used to predict product quality. A tricky subject, but all went well and the regulators accepted what we did.

Dont forget to write the scope of validated use into your SOP’s.

Alex Kennedy