Indeed, and if they have a good understanding of the expected nature of the failure curve, they can get pretty good estimates of life expectancy from surprising short periods of testing. Were not the case, then many/most IT products (like hard drives) would not complete their life testing and get to market before they were quite seriously obsolete!Normally they'd run a large sample in parallel and apply some statistics to the failures that occur...
Quite so. They obviously should specify the test conditions, but rarely do - and, in any event, those conditions will differ from most in-service conditions in the big wide world. A cynic might suspect that they choose test conditions which give them a 'favourable' answer!... but under what conditions? Ambient temperature, orientation, airflow etc.
Kind Regards, John