There are no statistics on how frequently dogs "alert" and the subsequent search finds no contraband.
This reasoning reminds me of how people allege that the fact that there is no real evidence that NASA tried to cover up that they "never really went to the moon" is somehow indicative of or suggests that they actually *are* covering it up. It's called circular reasoning, and it's a logical fallacy.
Have you considered that the possibilty that reason there aren't any published statistics for it is because it doesn't tend to actually happen that often in the first place?
As I said elsewhere, it is not generally going to be in the interests of even a genuinely maliciously inclined officer to allege that his dog "smelled" drugs when they hadn't actually found any, because that undermines any confidence with which anyone could reasonably claim that dogs have any reliability in this capacity in the first place... unless, as I said above, the officer were intending to plant drugs for the dog to find. A dog's sense of smell can be over a million times more sensitive than a human's, and there is no doubt whatsoever that sniffing out such things even in extremely concealed locations would be well within their sensory capability.
I do not dispute that police use dogs to attempt to find drugs on people where none are found.... that actually *does* happens a lot, and there are unfortunately plenty of published statistics to support it. But it makes almost no sense for the police to actually claim that their dog had alerted them to the presence of drugs when they don't actually find any because that just makes them look stupid, as well as like the dog needs a whole lot more training.