Monday, May 19, 2014

The Nostalgia for Reliability Testing

Professor Ben Trachtenberg of the University of Missouri Law School and I recently had an exchange on Confrontation Clause matters published in the University of Florida Law Review.  Here are links to his original article, to my responsive essay, and to his reply.  I’m not going to ask the Law Review for the opportunity to do a surreply.  But I’ll offer a few comments here.

Prof. Trachtenberg’s pieces reflect nostalgia for the “good-old days” of the pre-Crawford era, when reliability testing under Ohio v. Roberts prevailed.  He certainly is right that some of the members of the Supreme Court have indicated that they would like to return to this era.  I think it’s clear – and the majority opinion in Crawford itself makes it clear – that reliability testing did a pretty rotten job of protecting the confrontation right.  Indeed, the fact that the Washington courts came out the other way in Crawford, in what should have been an absurdly easy case, indicates how weak the right was.

One reason, perhaps the principal one, why reliability testing was so limp is that it is indeterminate to the point of incoherence.  What does reliability mean?  It is notable that, for all the extolling of reliability testing by Prof. Trachtenberg and others, they rarely if ever define reliability; I haven’t found any attempt to define the term by Prof. Trachtenberg in this exchange.  I pointed this out in my essay, but his reply did not take me up on the invitation.  I do think a coherent definition of reliability can be framed – but it demonstrates the impossibility of making reliability a useful test.

In my essay, I wrote: “Evidence is reliable proof of a given proposition if and only if, given the evidence, it is highly improbable that the proposition is false.”  The trouble is that virtually no evidence is reliable under this standard.  The live testimony of a live witness with personal knowledge of the subject of the testimony is the epitome of acceptable testimony that, under any theory of the Confrontation Clause of which I am aware, does not pose any problem under the Clause – and yet eyewitness testimony is notoriously unreliable. It seems to me that candor requires us to recognize that trials are full of items of evidence that, taken individually, may not be reliable indicators of the proposition for which they are offered.  Cumulatively, though, they may create compelling proof.

Prof. Trachtenberg and I appear to agree (and with many others) that conspirator statements are unreliable.   So far as Crawford doctrine and I are concerned, that is irrelevant to any Confrontation Clause decision.  But where does this leave Prof. Trachtenberg?  He contends that necessity is the real justification for admitting such statements.  In my essay, I expressed the belief that this contention exposes a weakness of his approach; it is a remarkably weak right that is overcome by the government's need to gain convictions of a crime that is difficult to prove.  In his reply, Prof. Trachtenberg says that in discussing necessity he is only "describing reality, not announcing . . . approval."  But I don't believe that solves his problem.  Does he believe conspirator statements should be barred by the Confrontation Clause or not?  If he believes they should be, then he is adopting a view of the Clause that would not only cause a radical change to traditional law but that has no historical bearings and that, I believe, would not gain the adherence of a single Supreme Court Justice.  And if he believes these statements should not be barred by the Clause, given how unreliable they are, on what plausible basis?  I am still left wondering.

So I think that when judges and commentators speak of reliable evidence, what they may really have in mind is evidence that probably (let's put aside the question of just how probably) will lead to accurate fact-finding.  After all, accuracy is good, and if we put aside systemic and procedural concerns anything that leads to accurate fact-finding is good and anything that leads away from it is bad.

But a big problem, of course, is that it is pretty much impossible to pick out the evidence that will hurt fact-finding.  This is especially true given that there is no real basis for confidence that jurors are unable to discount for the weaknesses of hearsay; empirical evidence suggests that they're pretty good at it, and may even over-discount.  Hearsay, after all, has real probative value, and excluding it denies the jurors potentially useful evidence.

The bigger problem is the qualification I just stated – that we put aside systemic and procedural concerns.  But those are really what the Confrontation Clause is all about.  That's certainly what the language of the Clause suggests – it gives the accused the right "to be confronted with the witnesses against him," which sure sounds like a procedural right with respect to witnesses and not a right to exclude evidence that lacks certain substantive qualities. 

And this procedural concern is clearly what the confrontation right has been about historically.  Prof.  Trachtenberg vigorously assails the merits of taking an originalist view of the Clause.  But the value of history in this area is not limited to determining what the meaning of the Framers would have been understood to be in 1791, and it's not limited to persuading those who put preeminence on that meaning.  (Note that while Crawford itself is quite an originalist opinion, as one would expect given that it was written by Justice Scalia, the majority that joined it included some Justices of very non-originalist orientation, and that has been true as well of Davis v. Washington, Melendez-Diaz v. Massachusetts, and Bullcoming .v New Mexico.)  What the history shows is that for hundreds of years protecting the conditions of testimony under which witnesses against an accused testify – ensuring thath tey do so face to face with the accused, and subject to cross-examination – has been a central aspect of the common-law system of criminal adjudication.

I have found that this is still a principle that resonates with people.  Does anybody doubt that prosecution witnesses should be required (absent forfeiture of the right) to testify face-to-face with the accused?  Or, looking at the matter from the other direction, does anybody think that a prosecution witness should be able to testify by talking to police officers in a closed room in the station-house, or in the witness's living room?  And isn't it clear that if Crawford and Hammon v. Indiana had come out the other way, then we would have a system in which witnesses would be able to testify in precisely those ways?  It was, after all, entirely clear to all participants in those conversations that the witnesses were creating evidence for possible use in prosecution.

In arguing Hammon, I proposed a per se test that would have yielded a different result in its companion case, Davis v. Washington – that a statement made to a known police officer accusing another person of a crime be deemed testimonial.  That was not meant to be a comprehensive definition of  the term "testimonial," as one might think from reading Prof. Trachtenberg's reply, but rather the delineation of a particular type of statement that should be categorically deemed testimonial, a subset of the broader set of testimonial statements.  I do think it's unfortunate that the Court didn't adopt that test; I think if it had some of the muddle that has since arisen would have been avoided (and more of the muddle would have been avoided if the Court had come out the other way in Giles v. California, for if there it had adopted a robust conception of forfeiture doctrine it almos tcertainly would have avoided the very unfortunate reuslt in Michigan v. Bryant.)  But I don't think that test was essential to sound confrontation doctrine.  I think the ultimate test of whether a statement is testimonial should be whether, assuming admissibility, a reasonable person in the declarant's position at the time of the statement would have anticipated likely litigation use of it.  (It doesn't seem to me that it's really difficult to define "testimonial," as Prof., Trachtenberg says, or at least to come up with a serviceable definition; I just don't think the Supreme Court has clealry signed on to the best definition, though in Melendez-Diaz it seemed on the verge of doing so.)  And under one view, not the one that I would have adopted but nonetheless a coherent one, the first part of the 911 conversation is Davis might be deemed to fall outside this, because one might say that in the heat of the moment the complainant would not have been anticipating likely evidentiary use.  (Pretty quickly, though, it became clear that this was what she was doing.  Note that at the end of the call the 911 operator said that the police would go find Davis and then come talk to the complainant – not that hey would come right away to protect her.)

Prof. Trachtenberg suggests that Davis would have "walk[ed]" if the Supreme Court had come out the other way in his case.  I don't know what would have happened in that case, but I think it's important to emphasize that excluding statements of the type involved in Davis does not mean as a general matter that the accused cannot be convicted.  The prosecution might work harder to secure the live (or deposition) testimony of the complainant; it might demonstrate that her unavailability was the result of wrongdoing causing forfeiture of the confrontation right; and it might, as it ordinarily does in a murder case, prove guilt without relying on a statement by the victim.


Anonymous said...

The underlying issue that drives debate in this area is the reality that prosecutors do not like anything that makes it more difficult--in their view--to secure a conviction. The reliability standard under Ohio vs Roberts is facile? What is exactly is the problem? Come on man, your not a criminal coddler are you?

Of course those who support the tyranny of the prosecutor attack an orginalist understating of the confrontation clause. The Founders did not view the prosecutor as a big fluffy teddy bear but as an agent of the Crown who would be tempted to abuse his authority as any other monarchical figure (King George was on their mind). Sorry, there I go again being a criminal coodler.

Anonymous said...

I disagree with your argument that conspirator statements are unreliable. Some are. Some aren't. And the only way to intelligently make that determination is by evaluating the substance of the particular statement in light of all of the evidence in the case.

This can be done in a pre-trial hearing. The court can exclude the unreliable statements and admit the reliable statements. Trial courts engage in this pre-trial screening process every day -- permitting juries to hear evidence that passes a low bar of relevance and excluding evidence that has too little probative value, and too much potential for distracting from the issue at hand, to allow it to play a part in the truth-finding process.

Simply stated, your reliance upon the proposition that all conspirator statements are unreliable in order to bolster your argument that a reliability test can't function to serve the purpose of confrontation -- which is itself reliability in the fact-finding process -- is misplaced.

And, by the way, there is just as much room for judicial manipulation in an ill-defined "testimonial" approach, as there is in a "reliability" based approach. Nine of the wisest persons in America can't agree on what the term "testimonial" encompasses, indeed on what the words "witnesses against" means.

So what we are left with is the common sense of 12 members of the community to decide how much weight, if any, to afford to a particular hearsay statement. And, although far from perfect, for the past 200 years or so years, most would agree that the jury system is the best filter against tyranny, against government overreaching, against convictions based on patently unreliable evidence, that exists.

Anonymous said...

Anon from 12:33:
If to satisfy the CC an out-of-court statement need only satisfy the low-threshold reliability test applied to any other type of evidence, what would the point of the CC be? The rules of evidence already protect against unreliable evidence. The text of the CC states that the protected right is the right to confront, not that right to reliable evidence. In my view, the problem with applying a reliability test to the CC is not such much the difficulty in measuring reliability, but rather that it measures the wrong thing.

Richard D. Friedman said...

I certainly agree with the last comment that reliability measures the wrong thing. It's not what the Confrontation Clause is all about. But a reliabiity-based Clause still has some impact -- basically it federalizes and constitutionalizes the law of hearsay, so that the Supreme Court, or any federal court on habeas, can determine that a given piece of evidence should not have been admitted.

I didn't mean to say that all conspirator statements are unreliable; if there is a sufficiently lenient definition of reliability to allow some possibility that the proposition for which the evidence is offered is not true notwithstanding the evidence, presumably some conspirator statements will qualify. But under Roberts, the courts did not pick and choose among conspirator statements -- the rule was that if the statement came within the hearsay exemption for conspirator statements, end of story, no Confrontation Clause violation.

I don't think that the failure of the Court so far, in a dozen years of Crawford jurisprudence, to come up with a good definition of testimonial to which it adheres regularly proves very much. Notice I refer to a good definition to which it adheres regularly, because it did use a good definition in Melendez-Diaz. Crawford was a major transformation, and some justices have been uneasy about thinking in a very different way. But it's clear that a pretty sturdy definition is possible. Picking about reliable evidence that the jury cannot hear -- not a function prescribed by the Clause -- is, on the other hand, as Crawford itself pointed out, inherently indeterminate.

Anonymous said...

Isn't there a historically sound argument that the Framers intended the term "witness" to have a very limited meaning that encompassed only persons testifying in court and out of court declarants whose statements were the product of government conduct intended on building a case against a particular suspect?

And isn't it true that statements of persons in the later category are subject to confrontation because the Framers deemed them inherently unreliable?

Richard D. Friedman said...

I think there's no case whatsoever in support of that proposition. Most prosecutions in the centuries before adoption of the Clause were private, not by government prosecutors. If the prosecuting witness did not appear, the case was routinely dismissed. This happened thousands of times; I discussed it at the argument of Hammon v. Indiana and I think also in the briefs. The whole point of the confrontation right was to ensure that witnesses testified by coming to court. And you don't see in the historical sources any attempt to sort out between reliable and unreliable testimony: Testimony had to be given in court.

Anonymous said...

Professor Friedman,

Regarding your "ultimate test": How would a trial judge determine whether a reasonable person in the declarant's position would (or would not) have "anticipated likely litigation use" of their statement?

What factors would guide this pre-trial determination?

Would a reasonable four year old who is interviewed by a police officer dressed up as a clown (to make the child feel more comfortable) "anticipate[] likely litigation use" of their statements accusing their step-father of molesting them? How about a reasonable six, eight and ten year old victim?

Would a reasonable woman who has been violently beaten by her husband and is being interviewed by an investigating officer (in her home) minutes after the crime, while still trembling and bleeding, "anticipate[] likely litigation use" of their accusatory statements? Or is it reasonable to believe that they are not thinking, at that moment, about future litigation? What if the interview took place two seconds, thirty minutes, two hours after the crime? Does that change the result under your "ultimate test."

Isn't the trial judge's determination of what a reasonable person in the declarant's position would have anticipated subject to as much result-oriented manipulation as a determination of the statement's reliablity?

Anonymous said...

Professor Friedman writes, "Picking about reliable evidence that the jury cannot hear -- not a function prescribed by the Clause -- is, on the other hand, as Crawford itself pointed out, inherently indeterminate."

This isn't to say, however, that reliability is inherently neutral. When a standard is inherently indeterminate then some other basis must be found by the judge for making an actual decision on whether the evidence must be admitted. The fact that 50% of all federal judges are ex-prosecutors goes a long way towards explaining how that indeterminacy is resolved in practice and entirely explains why prosecutors are so upset by Crawford and want Roberts back.

Anonymous said...

Professor Friedman's reasonable-anticipation-of-litigation "ultimate test" is as "inherently indeterminate" as the Roberts reliability test.

The clearest, and most determinate, test was set forth by Justice Alito writing for four justices in Williams: Was the primary purpose of the government produced statement to accuse a targeted person of a crime?

This test is also as (if not more) consistent with the history that motivated the Framers to draft the Confrontation Clause (as exemplified by the accusatory statements of Lord Cobham that were produced by the coercive interrogation by the Crown for the purpose of implicating Raleigh in a criminal conspiracy) than any other test that has been proposed.

Richard Friedman said...

I think reasonable anticipation of litigation is a plenty workable test. Obviously, as with all legal tests, there's some gray area around the edges, but the basic idea is clear and sensible and in the vast majority of cases its application is not difficult.

To be blunt, I think the targeted-individual test, which five justices rejected in Williams, is nonsense. Cobham's statement, targeting Raleigh, may be the core type of statement at which the Confrontation Clause is aimed, but that doesn't mean it's the limit. So are as I know, such a limit was never articulated before, and never applied. Under that test, if a witness describes a crime scene in great detail but doesn't identify the perpetrator -- and what, I wonder, constitutes identification? -- it's not testimonial because there's no targeted individual. I regard this test as a none-too-subtle assault on the confrontation right.

Anonymous said...

Maybe the "core" is the whole?

The CC's "faded parchment" resume doesn't necessarily suggest otherwise.

And, ambiguous history (at least beyond the "core") aside, there are powerful structural and policy arguments against expanding its scope (beyond the "core"), including federalism concerns about interfering with a state's right to admit highly relevant, highly reliable and, in some cases, indispensable evidence of serious criminal wrongdoing.

Anonymous said...

Prof. Friedmam,
Do you think Alito and the rest of the Williams plurality would have found a CC violation if Williams had been a targeted individual, or do you think he used the case as an opportunity to try to narrow how the Court's post-Crawford precedent is interpreted?