Certification is only one part of the Reliability Picture
Steven D. Nicely
In United States v. GonzalezAcosta,  the U.S. Court of Appeals for the Tenth Circuit refused to authorize extensive defense requests for a drug detection dog's training records, veterinary records, alert reports, and other miscellaneous documents. The court allowed the defense to review only limited training records, stating: "We do not believe the documents were relevant because the dog was certified on the day in question and because the dog properly alerted to the presence of contraband.... Indeed, had the dog's records indicated it had false-alerted in the past, the defendant's ability to cross-examine would not have been enhanced because there is no doubt it correctly alerted in this instance." 
Being certified does always not mean the dog will only respond to the intended target odor. Very few certifications test the teams to determine if anything other than the intended target odor will cause the dog to respond. Testing the dog only on the intended target does not assure the dog is only responding to that odor.
Records are very relevant because they paint a picture of the team’s abilities. Field records are of particular importance because they reflect the probability of something being found in actual searches. They also provide information on how training should be conducted for that team’s environment.
Just because drugs were found after a dog’s response does not mean the dog responded to them. The reliability of the dog’s response can only be articulated when certification, testing, and training are geared to identify non productive response caused by such things as associated odors, novel odors, handler’s mental set, or schedule of reinforcement.
When responses caused by other factors have been eliminated through extinction training and consistent ongoing training to prevent or correct the occurrence of spontaneous recovery is documented then and only then can the assertion the dog’s response was induced by a target odor have a valid argument.
A very critical and important testing procedure is rarely. If administrators would arrange for random in field testing for both detection of the target odor, and for non productive responses reliability would drastically improve. As it stands administrators are only left with the information they receive from their trainers. Some trainers are very competent and concerned while others are not. Random in field testing would provide administrators with information about their trainer’s abilities.
Random in field testing would provide impressive and strong supporting evidence if the team performed correctly in every unannounced testing situation. Think of the power a team would have in court when defending the dog’s responses. Why is this tactic not employed very often? Could it be fear of the dog’s failure? Failing to identify failures only leads to more and more failures in critical situations.
Court decisions such as United States v. GonzalezAcosta promote poorly trained dogs. They do not take non productive response seriously, and seems to give the handlers and trainers the idea that the end justifies the means. With these type of ruling more drugs are going to get by because officers are spending more time searching for drugs that are not present. Since administrators don't know the amount of time being wasted and very few courts take action why should handlers, or trainers?
If law enforcement and prosecutors really want to make a dent in the drug trafficking they need to take every non productive response seriously instead of making excuses for the team.