Although it is widely assumed that it is possible to accurately predict a dog's adult temperament from early testing, this has never actually been satisfactorily demonstrated in any scientific study. An interested breeder or trainer can endeavor to answer this question
Early testing only predicts how the puppies might develop if left to their own devices and if all treated equally. But pups and adolescents are never treated equally. Some grow up in great homes, others in good homes and yet others in bad homes — some poor dogs grow up with owners who should not be allowed to keep a rock, let alone a dog. The owner-variable far overshadows potential puppy predispositions. And without a doubt, a poor puppy prospect in a good home almost always becomes a better canine companion than a good puppy prospect in a poor home.
Human intervention is certainly the most important factor determining predictability: whether or not the new owners capitalized on the dog's potential good characteristics and/or resolved expected or incipient bad characteristics that were revealed in earlier testing. It would be naive to expect a dog to cure his own faults. Behavior and temperament are in a state of constant flux, and without human guidance, faults generally tend to get worse rather than better. It would be tantamount to stupidity to test a pup and discover that he is fearful, rambunctious, or aggressive, yet leave him to develop in this expected fashion. Surely a major reason for early testing is to locate potential or incipient problems and solve them before they become full-blown. Similarly, it would be utter folly to assume that a dog's naturally good temperament will necessarily remain that way indefinitely. As soon as owners become presumptively audacious about their pet paragon with the perfect personality, the dog's demeanor will predictably begin to deteriorate.