“Good Enough” for Probable Cause

Sunday, July 1st, 2012

“Bono” is a state police drug dog in North Carolina. Of the 85 times Bono has alerted his handler to the presence of drugs, a subsequent search turned up actual drugs just 23 times.

No matter. Federal District Court Judge Glen Conrad ruled late week that Bono’s record is still good enough to establish probable cause.

Bono “may not be a model of canine accuracy,” Conrad wrote in an opinion filed Thursday in U.S. District Court in Roanoke.

However, the judge ruled that other factors, including the dog’s training and flawless performance during re-certification sessions, were enough to overcome a challenge raised by Green’s attorney, public defender Randy Cargill.

Conrad’s justification for allowing the search illustrates just how clueless federal judges can be about these things—and why they can be such poor custodians of the Fourth Amendment. Judges have far too little skepticism for law enforcement officials.

I wrote about this in a column for Reason a couple years ago. The reason Bono performs so well in certification exams is because those exams test his ability to detect drugs. Dogs are great at that. But when Bono is with his handler alongside the highway, he isn’t detecting drugs. He’s pleasing his handler. Dogs are great at that, too. And that’s what we’ve bred them to do. On the road, Bono reading his handler’s body language, and alerting to confirm his handler’s suspicions. Of course, this assumes Bono’s handler is on the up and up, and isn’t deliberately cuing alerts. Which is also a problem.

It gets worse.

At a hearing earlier this month, Assistant U.S. Attorney Ashley Neese defended the performance of the German shepherd.

In some cases where nothing was found after an alert by Bono, police later determined that drugs had been in the vehicle earlier, likely leaving an odor the dog was trained to detect, Neese said.

Taking those cases into account, Conrad found that Bono’s accuracy rate was at least 50 percent.

I wonder if the judge asked the U.S. Attorney to provide some documentation for his claims, or if these were cases of an officer claiming to have found “shake” or residue—both of which, conveniently, are never tested.

But let’s take the U.S. Attorney’s claims at face value. Let’s say the dog was alerting to odors from drugs that had been in the car days or weeks or months earlier. (Which may or may not have had anything to do with the person driving when the car gets pulled over.) As one dog trainer told me in the piece I wrote for Huffington Post earlier this year, drugs dogs can be trained to ignore residue, shake, and lingering odors. That is, they can be trained to alert only when there’s a measurable quantity of illicit drugs.  The cops don’t want those dogs. They want dogs that will give them probable cause to search as often as possible. And because the courts have said a dog’s sniff is, in itself, enough for probable cause, there’s a strong incentive for police departments to want dogs that will alert to just about anything.

What’s incredible is that even if everything the U.S. Attorney says is true, the dog and his handler still have a 50 percent rate of error. Which means they’re no better than a coin flip. A coin flip is good enough for Judge Conrad.

Digg it |  reddit |  del.icio.us |  Fark

36 Responses to ““Good Enough” for Probable Cause”

  1. #1 |  Bob | 

    “And that’s what we’ve bread them to do.”

    Hungry this morning?

    So basically, Bono has a flawless record in certification testing, but only a 23% rate while in the field.

    Clearly, the handler is the problem, and is using the dog as a probable cause factory. And this is perfectly OK with a U.S. Attorney.

  2. #2 |  kino | 

    about 60% of alerts are bogus

  3. #3 |  C. S. P. Schofield | 

    Has anyone tried to deal with this through legislation rather than relying on judges? Anybody know? I imagine that in such an event the cries of police anguish would be audible on Mars, so I thought it likely that it hasn’t come up, but I’m curious.

  4. #4 |  perlhaqr | 

    That’s because the judge doesn’t give a shit.

    He’s part of the system.

  5. #5 |  Burgers Allday | 

    “He then worked as a U.S. probation officer for the Western District of Virginia from 1975 until 1976.”

  6. #6 |  Yizmo Gizmo | 

    How come the FEDS didn’t utilize “beer dogs” during Prohibition?
    Or Moonshine Hounds?
    Gov’t tyranny in this Drug War endeavor must be far worse, and more desperate….

  7. #7 |  Burgers Allday | 

    Here is a bad thing written by professor Kerr relating to the legal issues here:

    http://www.volokh.com/2011/03/28/new-draft-paper-why-courts-should-not-quantify-probable-cause/

  8. #8 |  Bramblyspam | 

    Just a nitpicky observation from a math geek… the “coin flip” analogy is a bit suspect.

    Let’s suppose, for sake of argument, that 10% of all search targets actually carry drugs. Using coin flips, you’d have false positives in 90% * 0.5 = 45% of stops, and accurate positives in 10% * 0.5 = 5% of stops, for an accuracy of 10%. (That’s 5/(5+45)). An accuracy rate of 23/85 is clearly better than a coin flip.

    Assuming that the “coin” would correctly find drugs on all the 10% of targets actually carrying them, that changes to 45% false positives and 10% accurate positives, for an accuracy of 10/55, which is still worse than 23/85.

    Of course, your analysis of the problem is still spot-on. I just take issue with the coin flip analogy :-)

  9. #9 |  supercat | 

    #8 | Bramblyspam | “Let’s suppose, for sake of argument, that 10% of all search targets actually carry drugs.”

    Bearing in mind that search targets are not selected at random, I’m not sure that’s a reasonable assumption.

    Still, I would posit that rather than trying to define by statute what the success rate of a drug dog needs to be to constitute “probable cause”, it might be good to make the handler’s record a legitimate subject of defense questioning, and inform jurors that before they may construe the results of a search in any manner against the defendant, they must determine *based upon their assessment of the cops’ credibility* whether bona fide “probable cause” existed. If there’s simply a rule that dogs have to find drugs 51% of the time, cops will simply make sure to bring their dogs along on enough “searches” where drugs are known to exist to ensure they meet the required success rate. But allowing a defense attorney to question the dog’s handler more intensely could help make jurors realize that the cops spent more time harassing innocent people than catching crooks.

  10. #10 |  Burgers Allday | 

    I haven’t read the case (guessing it is suppression in a criminal case, rather than a civil suit for an alleged 4a violation against a regcit). Does the case say whether the certification tests are double blind and have legal scents (eg, beggin strips) deployed to attempt to fool the k9?

    Bono may be as good as Mr. Balko says, but I am not so sure he would pass a fairly administered test.

  11. #11 |  Becky | 

    I don’t understand why having dark tinted windows and an obscured license plate call for bringing a drug sniffing dog to the site at all.

  12. #12 |  Stormy Dragon | 

    While I agree with your general point that police dogs are basically being used to generate excuses to ignore the fourth ammendment, the engineer in me has to object to the faulty use of statistics on display in your conluding statement:

    What’s incredible is that even if everything the U.S. Attorney says is true, the dog and his handler still have a 50 percent rate of error. Which means they’re no better than a coin flip. A coin flip is good enough for Judge Conrad.

    This statement is only true if we start from an assumption that half of all cars have drugs in them. Given that only a small percentage of cars have drugs in them, this dog is still significantly better than a coin flip.

    The problem here is not that the dog is doing a bad job, but that the vast majority of cars have no drugs in them. It’s a known problem with tests for unusual events that even if you have an accurate detector, you’re going to get a lot of false positives.

  13. #13 |  Bob | 

    Bramblyspam:

    Let’s suppose, for sake of argument, that 10% of all search targets actually carry drugs…

    Actually, we don’t have to suppose. We know. In the sample selected, 27% of the 85 cars contained drugs, while 100% of them were searched. The fact that the dog was in fact able to detect drugs was irrelevant, the same results would have resulted from the use of a dousing rod or … a coin toss.

    Even then, the coin toss would be irrelevant, as it’s made after the search to serve only the purpose of verifying the randomness of coin tosses.

    Police Departments should just go to the “Build a Bear” workshop and get stuffed bears dressed like K9 units. It would be cheaper and result in just as many searches.

    And just think of the comedy value.

  14. #14 |  Radley Balko | 

    Given that only a small percentage of cars have drugs in them, this dog is still significantly better than a coin flip.

    Wasn’t the best way of phrasing it, but I don’t think you’re quite right, either. The pool of cars we’re talking about are those where the police officer suspects drug activity. Presumably, a significantly higher percentage of those cars will have contraband. I guess my point is that the dog is wrong as often the dog is right. So maybe the better way to put it would have been, “The dog is no more accurate than someone trying to predict the outcome of a coin flip.”

  15. #15 |  Jonathan Foreman | 

    “I don’t understand why having dark tinted windows and an obscured license plate call for bringing a drug sniffing dog to the site at all.”

    The courts have already solved that problem on behalf of the police. Bringing a drug dog to walk around the car is not a “search”. The Supreme Court just defined the pesky Fourth Amendment issue away, and the police don’t need any reason at all to direct a dog around your car. Illinois v. Caballes, 543 U.S. 405 (2005).

    Please take a seat in the back of my patrol car while Savage here walks around your vehicle, peasant. /s

  16. #16 |  Stormy Dragon | 

    Radley, please see the Wikipedia article on Baye’s Theorem, specifically this section:

    http://en.wikipedia.org/wiki/Bayes%27_theorem#Drug_testing

  17. #17 |  C. S. P. Schofield | 

    When I think of the ways that the asinine War on Drugs has eroded the traditional protections of the Law, I want to look up an anti-drug crusader and spit in his face.

  18. #18 |  sailshonan | 

    Bramblyspam:
    But isn’t that assuming that 10% of the people being sniffed by drug dogs are carrying drugs? Also, we will never know how many times the dog didn’t alert when the people stopped actually were carring drugs (false negatives).

  19. #19 |  Highway | 

    Stormy, I think that loses out with the fact that it’s not anything close to a random sample. Presumably, *all* the cars they call out the dog to search are already thought to be carrying drugs, so the sample group is pre-screened by the cops. If they were actually pulling the dog sniff shenanigans on every car they pulled over, then the Bayes Theorem issue might apply, but they don’t. The only call out the dog when they still suspect drug activity after first interacting with the driver, passengers, and vehicle, and second being stymied in all previous efforts to search the car, including seeing things through the window, ‘smelling’ things, asking for a search. So presumably the cops think that *everyone* they call out the dog on is worth searching. whether for drugs or for something else.

    Given that that’s the case, the large group of ‘negatives’ that should lead to a large number of false positives becomes far less significant. So the real ‘problem’, as has been argued, is that they’re using the dog as a way to fake probable cause for a search in a group that they think they *should* be getting a much higher percentage of positives from. That they still have such a crappy rate of true to false positives means there is something else confounding the system.

  20. #20 |  Andrew Roth | 

    If this ruling is appealed to the Supreme Court, my fear is that it will be upheld, possibly due to the justices’ disregard for the Fourth Amendment but more likely because most of the justices are too sheltered and impractical to accurately assess hard data. For a panel whose members are expected to be impeccably dispassionate, the current court is very easily swayed by emotion and prejudice. This can work to the advantage of justice and equity in some cases, e.g., the California prison overcrowding case, but I’d say more often it results in bad rulings that make an ass of the law and abet government brutality. Antonin Scalia’s unhinged comments about “ripped prisoners” come to mind. I don’t think most of the justices have a clue about what goes on within prisons unless they’re shown graphic pictures, as they were in the California case.

  21. #21 |  Andrew Roth | 

    I’m afraid that the only answer to false alerts by drug dogs is to educate the public about how unreliable they are and how eager they are to please their masters, i.e., the cops. Skeptical, independent-minded citizens need to be gotten into jury pools so that it is impossible for judges and prosecutors to stack juries with nitwit authoritarians. This is feasible, but it’s a fucking tall order.

  22. #22 |  Andrew Roth | 

    Yet another thought (since I’m too scatterbrained today to organize everything into a single coherent comment): I have a hunch that juries will become more skeptical of drug dogs if police procedural dramas start featuring more good-for-nothing drug dogs. The same thing applies to faulty forensics, police perjury, incompetent detectives, and probably just about everything else that is wrong with policing. The public doesn’t usually pay much attention to police misconduct when it’s exposed in the newspapers, but it pays a lot more attention when Captain Cragen dresses down his subordinates. Of course, we’ll still have to deal with the shallowness of the forensics shows like “CSI,” which distort the hell out of real forensics and make Grissom and his buddies out to be demigods, and with the sad fact that these shitty shows are more popular than well-researched shows like the “Law and Order” franchises. Still, I think part of the solution is a more frequent dose of “Elliot. My office.”

    If drama is what it takes to educate the public, I’m all for it. It’s a lot better than nothing.

  23. #23 |  MingoV | 

    “… Conrad’s justification for allowing the search illustrates just how clueless federal judges can be…”

    That statement assumes that the judge is ignorant on this issue. I believe the judge knows that drug sniffing dogs are misused by the cops and supports that misuse.

  24. #24 |  Matthew F | 

    I think Bramblyspam is off a bit as well. Take a look at http://en.wikipedia.org/wiki/Sensitivity_and_specificity to get an idea of how statistics for classification tests work.

    A dog makes a decision, either Positive or Negative, about every vehicle it is presented. Generally, we care more about how accurate those positives are *relative to the entire universe of tests*. In this case, the article tells us the number of Positives, and the number of those later determined to be True Positives (22/85 = 26%), and hence false positives (63/85 = 74%). But to understand how accurate the dog is, we need to know how many Negative indications the dog made – which the article does not tell us.

    A large number of negatives would make a huge difference on the dog’s specificity (“to the ability of the test to identify negative results”), or false positive rate (the percentage of negative vehicles identified as positive), since those all include the number of true negatives in the denominator of the calculation.

    Assume every Negative result is true – which seems reasonable given how well dogs can detect drugs. If the dog did not alert on 1000 vehicles over the same period, the specificity would be 1000 / (1000 + 63) = 95%, and the false positive rate would be 63 / (63 + 1000) = 6% — and that would be doggone remarkable, especially compared to the medical diagnostic tests on which we make life-and-death decisions, and in my mind definitely sufficient for probable cause.

    But if there were only, say, 50 Negative results, the specificity would be 44%, the false positive rate would be 55%, and that would be atrocious.

    So, I say, does anyone know where to find the number of negatives? Without it, any discussion is statistically deficient.

  25. #25 |  Bob | 

    #24 : Matthew F

    I think Bramblyspam is off a bit as well.

    Missing the big picture you are, explain I will.

    This has nothing to do with the dog. You could replace the dog with a divining rod, a hat with two magic stones, or a stuffed animal dressed like a drug sniffing dog.

    It’s all about the handler. And the handler would get the same results with a lighter. “The lighter lit! That’s a sure sign that drugs are in the car!” Or a stick. “The stick has detected drugs!” Or hell, a Corn Dog. “The Holy Corn Dog has vibrated in the presence of illegal drugs!”

    This has absolutely nothing to do with statistics or mathematics… it’s all about the perceived “accuracy” of drug sniffing dogs, along with the preconceived notion that the handlers of these dogs are honest.

    In addition, there is a prescribed notion among the “Law and order” types that egregious abuses can’t happen to them, thus… it’s ok that dirtbags and lowlifes are subject to searches. After all, they’re dirtbags and lowlifes. They SHOULD be searched. These people will be happy to wink approval to otherwise illegal searches.

    Stop using statistics or math in defense of the dog. It’s reasonable to assume that a perfectly trained and handled dog will be accurate 100% of the time. If the actual statistic is LESS, than that’s on the handler. Always focus on the handler.

    The public needs to be trained to expect 100% accuracy from drug dogs. Not this sorry ass “Well! It was shake!” or “Uh… Drugs WERE there.” or “Derp! 50% is fine!” crap.

  26. #26 |  KPRyan | 

    At this point The State should just do away with silly, historic ideas like ‘probable cause’. Just let the peace (oops… another archaic idea) I mean, police enforcers search whomever they wish, whenever they wish.

    It would at least be more honest.

  27. #27 |  Red | 

    The judge isn’t clueless. Every judge in the country knows how corrupt and dishonest both the cops and the DAs are. The judge is evil. He knows he must be corrupt to part of the system and at the same time pretend that he’s there for justice to make sure people don’t give up on the system. This makes him worse than the corrupt cops and the justaceless DAs. He’s the one who gives the system legitimacy.

  28. #28 |  Gordon Clason | 

    Y’all are going under the assumption that when the dog alerts, there is a real genuine alert there and the DOG made the determination. In most cases I’ve seen a dog’s “alert” is more likely to be a handler CLAIMING the dog alerted than a real alert.

    Anyone who would believe anything a cop says would buy a bridge in Brooklyn.

  29. #29 |  Burgers Allday | 

    The judge isn’t clueless. Every judge in the country knows how corrupt and dishonest both the cops and the DAs are. The judge is evil. He knows he must be corrupt to part of the system and at the same time pretend that he’s there for justice to make sure people don’t give up on the system. This makes him worse than the corrupt cops and the justaceless DAs. He’s the one who gives the system legitimacy.

    I am going to try to be an apologist for the judge here, at least in a limited sense. Here is his mitigation:

    He doesn’t want to let a guilty person go free. That is understandable. I looked it up. This was a suppression case. If he suppresses, then the guilty person goes free. If Green didn’t have cocaine, and he sued police for tossing his SUV then police might well have gone the other way with respect to Bono’s ability to give probable cause.

    Now, I understand that the judge isn’t supposed to treat innocent litigants different from the guilty ones. But it is hard not to (it might be easy if one doesn’t think cocaine should be a criminal offense — but imagine that it was evidence of some crime that personally makes you angry that was sitting hidden in Green’s SUV). I think the judge gave into the understandable temptation not to let the guilty man walk. I also think that this temptation is too much for judges. Not all of them, but many of them — their intelligence and integrity notwithstanding.

    This is why I often try to convince Mr. Balko (and now Ken White) that we should get rid of the exclusionary rule. It is the exclusionary rule that is making a j/k out of 4a jurisprudence and making the legal system a mockery. Now there obviously need to be other remedies to lower the incidence of unreasonable searches by police. It is fun to imagine what those remedies could be, but I will save that discussion for other thds.

    For now: GET RID OF THE EXCLUSIONARY RULE! This here case under discussion shows how it leds to bad results and alwys will.

  30. #30 |  Yizmo Gizmo | 

    “Just let the peace (oops… another archaic idea) I mean, police enforcers search whomever they wish, whenever they wish. ”

    Good idea, but then how would we convince the kids we are the
    most wonderful and most free country in the world, stuff we drill
    into their heads starting at age 3? With the drug dog “alert” system
    we can have both the phony 4th amendment for show/indoctrination
    AND be able to rifle through anyone’s stuff, anywhere…

  31. #31 |  Bramblyspam | 

    I actually agree with getting rid of the exclusionary rule. People who are guilty of heinous crimes (i.e. not necessarily drug smuggling) shouldn’t walk free because of some technicality. However, there needs to be some mandatory penalty for the cop/prosecutor if evidence that’s used in a trial was obtained illegally.

    Which is more “just”: penalizing two guilty people for their crimes, or allowing two guilty people to escape without penalty because the second crime was obtaining evidence in an illegal manner?

  32. #32 |  Personanongrata | 

    What’s incredible is that even if everything the U.S. Attorney says is true, the dog and his handler still have a 50 percent rate of error. Which means they’re no better than a coin flip. A coin flip is good enough for Judge Conrad.

    As any fool knows when flipping coins with the state the flip always hinges on the states: heads I win, tails you lose doctrine.

  33. #33 |  Stevenson v. Kwiecinski | 

    This is the same judge who ruled in favor of two police officers who arrested a man for violating a law that didn’t actually exist. Judge Conrad found that the officers were entitled to qualified immunity because of their “sincere belief” that they weren’t breaking the law.

    https://docs.google.com/file/d/0B61j4W5ufFbcNmJjN2IxODItMjNiOC00OTgwLWFmZjItNjMyYjkwYzhiZTFi/edit?authkey=CLrg3LkK

  34. #34 |  supercat | 

    The Exclusionary Rule is necessary because a judge who determines that evidence was obtained in an unreasonable and illegitimate fashion will generally not be in a position to punish those who obtained it in any fashion other than by denying them the ability to use such evidence, and those who would be in a position to punish them in other ways seldom have much motivation to do so, especially if the illegitimately-gained evidence will get them the results they want.

    That having been said, one major problem with the Exclusionary Rule, as implemented, is that it is generally only effective in cases where evidence was obtained in a fashion which no reasonable person could judge to be reasonable. I would suggest that it needs to be bolstered by affirming the right of defendants to have jurors examine the factual matters surrounding the legitimacy of evidence, and have the jurors instructed to disregard any evidence they find was not gathered in legitimate fashion. To be sure, this check by itself would not eliminate the need to have judges enforce the Exclusionary Rule (since jurors may in some cases be loath to ignore evidence of a heinous crime, no matter how patently illegitimate the means by which it was obtained), but it would also provide some protections the exclusionary rule does not (e.g. if based on the cross examination of a cop who sought and received a search warrant, a jury concludes that the cop knew or should have known the his basis for “probable cause” was bogus, the jury could disregard the evidence gained thereby; a judge examining the issue, by comparison, would probably only strike the evidence if there was no possible way the cop could have believed his basis for “probable cause” was legitimate).

  35. #35 |  Power Corrupts – Just Ask John Roberts (and other news…) » Scott Lazarowitz's Blog | 

    [...] Radley Balko: “Good Enough” for Probable Cause [...]

  36. #36 |  Prosecution and police roundup - Overlawyered | 

    [...] with police dogs as evidence [Balko, [...]

Leave a Reply