How would you analyze data from automated tests to determine whether a failure is caused by software or hardware?
