Judge Boggs and Probable Cause:
A number of law bloggers have pointed out the paragraph in a recent concurrence by Judge Danny Boggs that attempts to quantify probable cause. It's only a paragraph, but it's a pretty interesting paragraph and by a prominent judge. Boggs suggests that the threshold of probable cause is in the neighborhood of a 5 to 10 percent likelihood of evidence being found:
While courts have resisted mightily putting a number on probable cause, see Maryland v. Pringle, 540 U.S. 366, 371 (2003), at bottom a review of cases indicates that there must be some, albeit inchoate, feeling as to what kind of probability constitutes probable cause. My reading is that it does not require a belief that there is more than a 50% probability of evidence being found in a particular location. See, e.g., United States v. Gourde, 382 F.3d 1003, 1015 (9th Cir. 2004) (Gould, J., concurring) (collecting cases). If that were the case, one could never get a search warrant to search all three cars of a person for whom there was overwhelming evidence of general drug dealing, and specific evidence of a drug transaction the proceeds of which were now certainly in one of three cars in his garage, and certainly not in any of the others. However, to be more than a hunch or a supposition, in my own mind, requires a legitimate belief that there is more than a 5 or 10 percent chance that a crime is being committed or that evidence is in a particular location.
  I think Judge Boggs is wrong, and that his hypothetical is based on a simple analytical error. Judge Boggs assumes that the "place to be searched" for the purposes of a warrant to search for evidence in a car must be an individual vehicle. In his hypothetical, the drug dealer has three cars in a garage, and there is evidence in only one of the three cars. Boggs concludes that probable cause must be a low probability because the chances that the evidence is in any one car are only 1 in 3, and yet the warrant in that case obviously would be issued.

  The problem with Boggs' hypothetical is that the "place to be searched" in this hypothetical wouldn't be an individual car, or even the garage itself; the "place to be searched" normally would be the entire property that contained the garage and all of the cars inside it. The chances that evidence would be found in the place to be searched are 100%, not 1 in 3, as we know that the evidence will be found in one of the cars.

  Boggs' hypothetical provides a good reminder of how probable cause to search property is always contingent on the size and scope of the space searched. The larger the space to be searched, the higher the probability that evidence of crime exists there. If you make the place to be searched big enough, probable cause will always exist; for example, there is a 100% certainty that crack cocaine will be found somewhere in New York City. Conversely, the more you narrow the place to be searched, the less likely it is that probable cause exists; if you make the place to be searched infinitesmally small, you'll never have probable cause.

  With exceptions that need not be covered here, the key determinant of the scope of the place to be searched is the particularity requirement. Warrants must "particularly describe the place to be searched," meaning that cops can't ask for permisssion to search New York City, or the Upper West Side, or even a city block. Instead, they normally need to limit the place to be searched to a particular house and premises, or an office or group of offices. The only question is whether probable cause exists that somewhere on the property some evidence of the crime, contraband, instrumentality, or a fruit of the crime will be discovered.

  Finally, I think Boggs' estimate that probable cause means a 5% to 10% chance is pretty far off. In my experience, probable cause exists when the police have some pretty strong signals that there is really something unusual going on in the place to be searched. It's impossible to translate that into a numerical likelihood, as the police never know whether the selection of signals they have observed is representative of the whole. It's like the old poem about The Blind Men and the Elephant; the police never know what part of the elephant they're touching, so it's really hard to quantify the certainty of the evidence they have.

  At the same time, the studies I have seen indicate that about 90% of search warrants end up leading to the discovery of the evidence of crime sought in the search. This doesn't mean that probable cause requires 90% certainty, obviously, but I think it does indicate that probable cause in practice is more certain that Judge Boggs suggests.
You could certainly construct a hypothetical where the three locations to be searched are distinct places that each require a separate search warrant. There is a more basic problem with any attempt to quantify probable cause, which is that "something more" is required than mere probability. For example, in certain neighborhoods, it may well be that over 5-10% of the houses contain illegal drugs, but I doubt the police could obtain a warrant for every single house based upon that probability.

There is an analogy to the oft-cited claim that a "preponderance of the evidence" means greater than a 50% probability of being correct. This prompted a hypothetical by my civil procedure professor: a man is struck by a bus, but he is unable to identify the bus, and there are no witnesses. However, conclusive evidence establishes that 3 of the 5 buses to drive through town that day were owned by Greyhound. Can he sue Greyhound and win, on a theory that it was 60% likely to be a Greyhound bus? I suspect most of us would like to see something more than just statistical evidence. Indeed, even if 90% of the buses that day belonged to Greyhound, I still would find it hard to say Greyhound should be liable.
12.5.2005 12:11pm
If the requirement for the likelihood of finding evidence were less than 50%, wouldn't we call it 'possible cause'?
12.5.2005 12:30pm
Steve, I think you're on to something, but wouldn't a little more investigation/discovery lead to the right bus line owner?

IANAL, so I might not be playing the hypo game correctly. Is it ever that the only evidence in a case would be "hit by a bus" and "x out of y (the majority) were defendant's buses"? That seems a little out there to me.
12.5.2005 12:31pm
Sam Heldman:
In a related vein, I have always wondered -- yet never gotten a good answer -- whether there would be "probable cause" in the relevant sense to search each and every car in the parking lot of a Greatful Dead show (yes, I am dating myself), if one takes as true for present purposes the proposition that more than 50% of the cars going in to any Dead show have marijuana in them.
12.5.2005 12:50pm
ChrisW (mail):
IANAL either, and just speaking from what seems "right" to me. But I should think that the criteria for granting a search warrant ought to be informed by much more than a simple probability of a discovery.

Specifically, if I were to formalize the factors, they would certainly include

- Expected probability of a discovery (as discussed)

- Import of the expected discovery (the expectation to find evidence of shoplifting is less pressing than murder; maybe we should lower the bar when the crime is particularly important)

- Intrusiveness of the search (we can set a lower bar for "checking the car trunk" than for a body cavity search).
12.5.2005 12:51pm
There is a whole literature about Steve's hypothetical problem. The upshot, as Steve suggests, is that the nature of the evidence seems to matter (in addition to the numerical probability that evidence establishes). For just one argument in favor of that result, consider what would happen in a series of accidents fitting that hypothetical description. Greyhound would be liable in every case, and the smaller bus companies in none of the cases, even though Greyhound was only responsible for 60% of the accidents.

Anyway, I also suspect that Steve is right about these evidentiary issues playing a role in the "probable cause" context. As his other hypothetical suggests, certain kinds of evidence could lead to the police getting warrants for every house in a high crime neighborhood--which, in fact, is much like the hypothetical series of accidents in which the biggest bus company is always liable for accidents.
12.5.2005 12:53pm
NaG (mail):
To be fair, Judge Boggs thinks probable cause means "more than a 5 or 10 percent chance." Maybe he thinks it's a 20 percent chance.

The very term "probable" indicates to me a more-likely-than-not scenario.
12.5.2005 1:10pm
Hans Bader (mail):
Restraining orders -- under which one spouse is kicked out of the marital residence at the behest of the other spouse, whom he has purportedly annoyed and/or threatened, typically in divorce cases -- are granted by family courts based on mere "probable cause."

If Boggs is right about the meaning of probable cause, does that mean that a 5 to 10 percent chance that the complainant is telling the truth is justification enough for restraining orders to issue?

If so, then restraining orders, which are difficult for an innocent spouse to defend against in practice (no judge wants to be on the Six O'Clock News in the one-in-a-million chance that a spouse against whom he has refused to grant a restraining order subsequently goes bonkers and kills the other spouse), will now become difficult to challenge even in theory, giving free reign to false allegations.

(Note: I'm not divorced, and have never been the subject of a restraining order or application for a restraining order).
12.5.2005 1:35pm
Houston Lawyer:
Someone should do a statistical analysis of the percentages of search warrants that produce the expected evidence (normalized for other criminal evidence produced that was not the evidence being searched for). It would be very interesting to know what types of searches are most likely to turn up the suspected evidence.

DA: Judge, I'll give you three to one odds that there is contraband in that house.

Judge: I'll take that bet.
12.5.2005 1:55pm
I think ChrisW raises some interesting criteria above:

- Expected probability of a discovery (as discussed)

- Import of the expected discovery (the expectation to find evidence of shoplifting is less pressing than murder; maybe we should lower the bar when the crime is particularly important)

- Intrusiveness of the search (we can set a lower bar for "checking the car trunk" than for a body cavity search).

Basically, this amounts to the expected positive utility of a search (first two criteria) minus the known negative utility due to the intrusiveness of the search (third criterion). If the net result is positive, do the search; otherwise, don't.

It seems that probability alone really isn't enough. We might not want every car at a Grateful Dead concert to be searched, even if many (most?) contained small amounts of marijuana. However, we would probably decide differently if we knew that even one car had, say, 5 bodies in the trunk.

12.5.2005 3:18pm
Blar (mail) (www):
Note to self: buy 21 cars.
12.5.2005 3:41pm
Hattio (mail):
21 cars (assuming you can't afford really nice ones, and a garage to hold them all) leads to a whole 'nother set of assumptions, which may not lead to police surveillance, but will definitely get the neighbors talking about rednecks.
12.5.2005 3:59pm
snead16 (mail):
Blackstone used the the "probable cause" standard as the test for issuance of a writ of habeas corpus.

In Chapter 3, p. 132 of his Commentaries, Blackstone said:

“In the court of king's bench it was, and is still, necessary to apply for it by motion to the court, as in the case of all other prerogative writs (certiorari, prohibition, mandamus, &c) which do not issue as of mere course, without shewing some probable cause why the extraordinary power of the crown is called in to the party's assistance. For, as was argued by lord chief justice Vaughan, ‘it is granted on motion, because it cannot be had of course; and there is therefore no necessity to grant it: for the court ought to be satisfied that the party hath a probable cause to be delivered.’”

So if "probable cause" is the common-law standard for issuance of the Great Writ, and Boggs is right that PC requires only a showing of something not much more than a 10% likelihood of success, and given that the Constitution's key phrases -- like "probable cause" -- require reference to the common law for their content [see D'Oench, Duhme &Co. v. FDIC, 315 U.S. 447, 468 n. 9 (1942)(Jackson, J., concurring)("In the Bill of Rights Amendments, the necessity for resort to the common law for constitutional interpretation is even more obvious.")], then a preponderance standard for habeas petitions would be contrary to the Framers' -- and the original -- understanding and, thus, unconstitutional. No?

In light of Blackstone, it seems more likely that probable cause is indeed more akin to a preponderance standard.

And Judge Boggs would seem to be off on quite a wild, wild fox hunt. And it's a hunt that doesn't bode well for those of us who think that each of the Bill of Rights means something.
12.5.2005 6:51pm
NickM (mail) (www):
For those people who assert that probable cause requires a "more likely than not" standard, how would you deal with the following situation:
Police car comes around a corner and sees what appears to be a drug deal taking place. Dealer's face is seen. Customer has his back to the cops, and is wearing a gray hooded sweatshirt with long sleeves and dark blue jeans. Both take off running. Dealer trips and gets caught by one cop. He has cash on him but no drugs. Customer runs into a nearby bar. Other cop enters that bar a half minute later and finds 6 people who meet that description. Back door (emergency exit) is alarmed, and the alarm has not gone off.
Now, one of those 6 people has on him the evidence of a crime, but the cop does not know which one. Does the cop have probable cause for a search of any or all of those 6?

12.5.2005 10:59pm
Visitor Again:
Boggs' error is very dangerous for freedom.

In the Greyhound civil liability hypo above, the probability is entirely statistical (not an evidentiary likelihood) and imposing liability on Greyhound would be arbitrary and a violation of due process. There is a discussion, as I remember, of such improper use of statistical evidence to prove criminal liability in a California Supreme Court case from the 1960s. I believe it was a murder case in which the prosecution introduced an expert who testified to the statistical likelihood of more than one yellow Cadillac being driven by a black man with a white woman as passenger in the Long Beach area on a certain day. The expert came up with huge odds against a duplicate such occurrence, and the man was convicted. Conviction reversed because of arbitrariness and a false indicia of guilt. This is all from memory of almost 40 years ago, so please don't criticize me for getting the facts jumbled if you find the case. I cannot remember its name, although it may come to me. It may be a Justice Traynor opinion.

Whether probable cause to search--as opposed to civil or criminal liability--may be based on purely statistical likelihood may be a different question, of course. The hypo of the six guys in the bar poses that one. I would hold that probable cause to search a person must exist as to THAT individual person. Therefore, no search allowed. Let the police question these guys. Let them come up with other evidence. But no search based on mere statistical likelihood. That is an arbitary search and unreasonable on its face.
12.6.2005 5:54am
Visitor Again:
An addendum to my message on the yellow cadillac case. I don't know whether it's been overruled or how it compares with DNA cases in which witnesses testify to infinitesimally small odds of duplication. I haven't thought about it and I'd have to find and read the case again. I'd like to think the DNA cases are distinguishable because there there is an evidentiary basis for the statistical likelihood testimony, i.e., some foundation other than mere chance for the statistical likelihood. Or some other distinction. But I am not up on the area, and hence this caveat.
12.6.2005 6:06am
Visitor Again:
I found the case although I haven't read it yet. It's People v. Collins, 68 Cal.2d 319 (1068) and you can find it here.

It's an opinion by Justice Sullivan, not Traynor, and its headnote says:

As we explain in detail, infra, the testimony as to mathematical probability infected the case with fatal error and distorted the jury's traditional role of determining guilt or innocence according to long- settled rules. Mathematics, a veritable sorcerer in our computerized society, while assisting the trier of fact in the search for truth, must not cast a spell over him. We conclude that on the record before us defendant should not have had his guilt determined by the odds and that he is entitled to a new trial.
12.6.2005 6:20am
Visitor Again:
Have read it now. There is a lot more in it than I remembered, of course, and it is not as clear on what I thought it stood for as I would have hoped. But it does somewhat support what I said above about guilt based solely on statistical likelihood being arbitrary in this snippet from the opinion:

Urging that the Collinses be convicted on the basis of evidence which logically establishes no more than this seems as indefensible as arguing for the conviction of X on the ground that a witness saw either X or X's twin commit the crime.

The DNA cases, I presume, meet the Collins case objections to the misuse of statistical probability.
12.6.2005 6:41am
Hans Bader (mail):
Probable cause doesn't just give cops the discretion to search or arrest people -- it also sometimes requires them to do so.

For example, many domestic violence statutes require police to arrest when probable cause exists.

Yet Boggs says probable cause is satisfied by anything over ten percent.

Do we really want a society where a police officer who believes that there is a 90 percent chance that someone is innocent (for example, because the complainant has a history of making false allegations), feels compelled to arrest the accused because there is a mere 10 percent chance he is guilty?

(I have seen cases in Massachusetts and other states where police in fact did arrest people who had previously been held to be falsely accused by the very individual currently making the complaint, despite the police officer's deep misgivings, pursuant to mandatory domestic violence arrest statutes).

The fact that probable cause can be construed so broadly shows the wisdom of the Supreme Court's Castle Rock decision refusing to let people sue in federal court for cops' refusal to make arrests under state laws mandating arrests for mere probable cause.
12.6.2005 1:18pm

Does a warrant require not only a specific place, but also a specific crime? I would guess that a warrant to search all houses on a block on the grounds that X percent of them probably have something illegal would be problematically vague not only in defining the area but in defining the specific evidence or crime being investigated. In the drug dealer example, there may be some ambiguity in the place but the crime alleged is very specific.

If an inmate (known to be dangerous) escapes from a prison, and the police are reasonably certain (based on time since the escape and other hard evidence) that the inmate is hiding somewhere in a nearby neighborhood, are the police allowed to go door to door?
12.6.2005 3:08pm
jack simms (mail):
I know Michigan is only one state, but, it seems that at least here, to me, probable cause is mathematically less than 50%. In Michigan, “Probable cause” requires “only the probability, and not a prima facie showing of criminal activity”. Illinois v Gates, 462 US 213, 235, 103 S Ct 2317, 76 L Ed 2d 527 (1983). This is also the rule in the sixth circuit. "Probable cause exists 'when there is a 'fair probability,' given the totality of the circumstances, that contraband or evidence of a crime will be found in a particular place." US v Johnson, 351 F 2d 254, 258 (CA 6, 2003). "Probable cause is defined as reasonable grounds for belief, supported by less than prima facie proof but more than mere suspicion."
US v. Bennett, 905 F2d 931, 934 (CA 6, 1990). Michigan
law is the same: a proper finding of "probable cause" simply does not require as its basis a conclusion that it is overwhelmingly likely that a search will uncover contraband or other evidence--or that it is more likely than not that a search will turn up the type of item suspected. People v Garvin, 235 Mich App 90, 104-105
(1999), see, generally, People v Russo, 439 Mich 584, 607 (1992). The probable cause standard from People v Russo, 439 Mich 584, 609 (1992) is: “when the issue is a probability determination with respect to certain specific items being in a particular place, LaFave does not disagree with the Gates majority that probable cause does not require a more probable than not definition of probable cause. LaFave observes that it is a fair statement of the law that the "more probable than not" standard has not been actually applied and that
[t]o the extent such rulings permit searches to be made upon something less than a 50% probability as to any one particular place, they do not appear objectionable. The fact remains that it is unlikely that the privacy of an innocent person will be disturbed under such circumstances. [1 LaFave, supra, § 3.2(e), pp 598-599.]”
12.6.2005 4:28pm