Friday, May 25, 2007

Tennessee High Court Overturns Death Penalty Conviction For Failure to Admit ID Expert

I'm on the road so this is brief and nearly old news by now, but with that out of the way:

This great opinion (PDF) came down from the Tennessee Supreme Court this week in State v. Copeland, overturning its long-standing precedent excluding eyewitness testimony as common sense, citing (among other things) a poll (PDF) conducted by the Public Defender Service for DC, showing that jurors tend to overestimate the reliability of cross-racial IDs. The defendant in the case had been sentenced to death, but the Tennessee Supreme Court overturned the conviction for the trial court's failure to admit the expert testimony of Jack Brigham on the cross-race effect.

One point that was raised that never receives enough attention is the disparity in admissibility requirements for government vs. defense expert witnesses. So-called "narcotics experts" (i.e., cops) are routinely allowed to testify as experts on the subtleties of drug transactions, and yet eyewitness ID experts (i.e, PhD psychologists) are routinely precluded from testifying:

Another author has observed that while experts are often not permitted to testify regarding eyewitness testimony, police officers and other law enforcement officials are regularly permitted to testify “concerning the general way criminal
schemes and enterprises operate and the usual meaning of criminal slang and code words.” D. Michael Risinger, Navigating Expert Reliability: Are Criminal Standards of Certainty Being Left on the Dock?, 64 Alb. L. Rev. 99, 132 (2000). The author contrasts the “technical” knowledge of the law enforcement officials with the educational and scientific credentials of experts on eyewitness identification. Id. at 131-35.

It is the educational training of the experts and empirical science behind the reliability of eyewitness testimony that persuades us to depart from the Coley rule. Times have changed. Today, many scholarly articles detail the extensive amount of behavioral science research in this area. See generally Gary L. Wells et al., Eyewitness Evidence: Improving Its Probative Value, 7 Psychol. Sci. Pub. Int. 45, 47-49 (2006) (for a brief history of psychological research of eyewitness testimony).
There are literally hundreds of articles in scholarly, legal, and scientific journals on the subject of eyewitness testimony.

And on how cross-examination is rarely enough to undermine faulty eyewitness testimony in the minds of the jury:
Further, the research also indicates that neither cross-examination nor jury instructions on the issue are sufficient to educate the jury on the problems with eyewitness identification, contrary to the conclusion reached by the majority in Coley. See, e.g., id. (“[E]ven when presented with an eyewitness who was quite thoroughly discredited by counsel, a full 68% still voted to convict.”) (citing Elizabeth Loftus, Reconstructing Memory: The Incredible Eyewitness, 15 Jurimetrics J. 188, 189-90 (1975)). “Considered as a whole, the studies of juror knowledge and decision making indicate that expert psychological testimony can serve as a safeguard against mistaken identification.” Steven D. Penrod & Brian L. Cutler, Preventing Mistaken Identification in Eyewitness Identification Trials, Psychology & Law: The State of the Discipline 89, 114 (1999).

More in the opinion (PDF), which is well worth a read.

Read more :: Permalink

Tuesday, May 22, 2007

Cross-Ethnic IDs Get No Instruction in NJ (For Now); Standard ID Instruction Gets an Upgrade

The New Jersey Supreme Court held in State v. Cromedy in 1999 that uncorroborated cross-racial identifications required a jury charge to offset the impairment that has been observed by social scientists when members of one race attempt to identify members of another race.

Yesterday, the same court identified a limit on the application of that holding, finding the cross-racial jury charge unnecessary in cases of cross-ethnic, yet same-race, identifications. In State v. Romero (PDF), --- A.2d ----, 2007 WL 1461022 (N.J. 2007), the court observed the following:

Social science research does not tie identification unreliability directly to ethnic differences in the same way that racial differences can affect identification reliability. That said, identification testimony is an area that warrants vigilant supervision. An eyewitness's identification carries significant impact in criminal cases. This appeal highlights the importance of the model charge that guides jurors in the assessment of the reliability of that powerful evidence. We use this opportunity to refine the charge so that it will alert jurors in all eyewitness identification cases that such testimony requires close scrutiny.

The case involved the identification of a "Hispanic Caucasian" by a "non-Hispanic Caucasian, which the court found to fall outside of the reach of Cromedy. The court based its distinction in part on the practice of the US Census Bureau, which "considers race and Hispanic origin to be two separate and distinct concepts," in recognition that "Hispanics may be of any race." The court held that applying the same reasoning used in Cromedy would require an expansion of that rule, and declined to do so at this time, citing insufficient research to date on the specific issue of cross-ethnic identifications. It stopped short, however, of holding that such an expansion of the rule might not be appropriate if future research bears out a similar impairment in cross-ethnic scenarios.

Highlighting the demonstrated fallibility of eyewitness identifications generally, the court went on to require a modification to the general charge on eyewitness testimony:
In light of the social science research noting the fallibility of eyewitness identifications, we direct that the charge should underscore, for jurors in all eyewitness identification cases, that eyewitness identification testimony requires close scrutiny and should not be accepted uncritically. Accordingly, we shall require that the following additional language be included in the out-of-court identification charge immediately before the enumeration of factors to be considered by the jury when it gauges the reliability and believability of an eyewitness's identification:

Although nothing may appear more convincing than a witness's categorical identification of a perpetrator, you must critically analyze such testimony. Such identifications, although made in good faith, may be mistaken. Therefore, when analyzing such testimony, be advised that a witness's level of confidence, standing alone, may not be an indication of the reliability of the identification.

New Jersey continues to be a source of hope for progress, as its high court has been consistent in expressing an openness to the social science research that informs the issue. Romero's lawyer Richard Berg, while disappointed with the court's decision, expressed confidence (subscription required) that the court would eventually come to require the Cromedy charge in cross-ethnic cases.

Read more :: Permalink

Monday, May 21, 2007

Texas ID Reform Bill Dies in Committee

As reported by Scott Henson at Grits for Breakfast:

Despite the rash of exonerations of innocent people serving long sentences in Texas prisons, the 80th Texas Legislature is about to close seven days from now having passed virtually no legislation to address the problem, and indeed possibly having passed several bills that make it more likely that innocent people will be convicted.

That's why I was sad to see SB 799 by Ellis died in the Calendars committee last night - the bill would have created a study group to develop best practices for eyewitness identification. That said, the folks on the study group may not have been the best people to develop these best practices, anyway (in particular the TCLEOSE director and Police Chiefs Association President's participation would have made the group pretty politicized).

An unfortunate turn, but as Henson points out, legislation is not the only solution:
Texas doesn't need to wait for legislation to study this issue and develop recommendations. Police could rely on the work of other states like California which developed reform recommendations for lineup procedures last year. And at the capitol, one of the Texas legislative committees concerned with criminal justice should make the issue the subject of an "interim study," or a topic the standing committees research in-depth in between regular legislative sessions.

Wherever or however it happens, the failure of SB 799 and SB 263 (Texas' Innocence Commission legislation) to pass should not end the discussion of how to prevent more wrongful convictions in Texas. It's my hope, in fact, that these bills' death serves to launch that discussion at the Texas Legislature in a serious way.

Read more :: Permalink

Friday, May 18, 2007

Face Shapes, First Names, and the Malleability of Memory

This is might be off-topic, but it's Friday afternoon and I thought this was pretty interesting. Some research came out this week from psychologists at Miami University in Ohio, suggesting that people are wired to expect certain correlations to exist between certain names and certain facial characteristics.

If you want to test out the theory, check out these photos and take a stab at which face you think is a "Tim," and which one is a "Bob."

The researchers only looked at white males, as a method of limiting the study "subtly perceived differences among mostly homogeneous faces." The results suggest that people store "prototype faces" for certain names -- like a round face for "Bob," and a thin face for "Tim" -- and that people have more difficulty remembering names when the characteristics of a person's face conflict with the prototype.

The researchers didn't speculate on the applicability of the research findings outside of the narrow range of white male test subjects and a handful of common first names, but interesting to see what seems intuitively true borne out by some actual research.

Also interesting, at at least conceivably relevant to the eyewitness ID scenario, is that people were observed to actually distort the characteristics of people's faces in their memories, when learning that a person has a particular name for which a "prototype" suggests that certain characteristics should be present. When a test subject learned that a face belonged to a "Bob," she would be more likely to add "roundness" to that person's face in her memory, even when the trait was not present. In other words, the memory was seen to morph to fit the preconception associated with the name, when the name was known.

Read more :: Permalink

Monday, May 14, 2007

Coke Machines, Willie Nelson Lookalikes, and the Failure of Manson v. Brathwaite

The following, from a real court decision:

Barnett testified that Mrs. Hayden told him the intruder was short and stocky, approximately 5'4 to 5'6 in height, that he had long hair, and that he was in his middle 20's. At trial, Mrs. Hayden testified that she told Barnett that the intruder was between 5' and 5'4 tall, that he had red hair, and that he had a red bandana tied around his head. Later in the day, after defendant had been apprehended, Deputy Barnett called Mrs. Hayden, told her he thought they had caught the person who had broken into her home, and asked her to come to the courthouse to see if she could identify the suspect. She went to the courthouse where she viewed a lineup. The lineup, which was reminiscent of a scene from the television show Saturday Night Live, consisted of the defendant, who bore a resemblance to a diminutive look-alike of country singer Willie Nelson, red bandana and all, five other men, the shortest of whom was 5'11 tall, and a red Coca-Cola machine. Defendant was the only person in the lineup who had red hair and who was wearing a red bandana as a headband. At the motion to suppress hearing, Mrs. Hayden was asked if defendant was the only person in the lineup with red hair. Her reply was, "I didn't notice the other ones because when I saw him, I knew it was him."

In case you missed it, or tend to skip over block-quoted text, the police in State v. Kiplinger (591 S.W.2d 207 (Mo.App. S.D. 1979)) used a Coke machine as a filler in a lineup, presumably to minimize the extent to which a red-headed suspect -- wearing a red bandana -- stood out from other members of the lineup. Further to that end, the police included five other men 5'11" or taller, to draw attention away from the 5'4" miniature Willie Nelson lookalike.

But the most astonishing thing about this case is not the lineup itself, nor is it the egregious police negligence, incompetence, and general disregard for the rights of the accused. The most astonishing thing about this case is the fact that the appellate court held that the lineup did not run afoul of due process.

This case is a stark illustration of the failure of the Manson v. Brathwaite due process standard for the admissibility of eyewitness identification evidence. The Supreme Court in Manson told us that "reliability is the linchpin" in the admissibility analysis for eyewitness evidence, which on its face doesn't sound all bad. When there is a suggestive procedure, courts should look to other "indicia of reliability" -- the classic factors, including the witness's opportunity to view the perpetrator, the accuracy of the prior description, and (science notwithstanding) the confidence of the witness.

But the Manson court didn't stop there. The Supreme Court held that against those "indicia of reliability" should be weighed "the corrupting effect of the suggestive identification itself." Which is to say, the presence of suggestive procedures was not intended to be relegated only to the first prong of the analysis, as a threshold requirement to be met prior to engaging in a reliability analysis. Rather, the corrupting effect of the suggestive procedure itself was intended to be carried over into the second step of the analysis, for a determination as to whether or not the "indicia of reliability" outweigh that effect. In a case like the one quoted above, at best it should be impossible to extract any indication of independent reliability from a procedure so corrupting, such that the corrupting effect should easily be found to win the day.

But instead, that court -- and nearly all courts since -- managed to omit a fundamental step of the analysis set forth in Manson, namely that the corrupting effect of the procedure itself should be considered in determining the overall reliability of the identification evidence. The unfortunate reality is that Manson has been interpreted to mean, in direct contradiction to the clear language of the decision, that if the circumstances surrounding the original viewing of the perpetrator are found to be sufficiently reliable, even the most egregiously suggestive procedure will be excused and overlooked.

The result is that courts allow eyewitness evidence in front of juries from procedures that they themselves liken to a Saturday Night Live skit. I'm not sure that the Comedy is the right literary vehicle for the story of wrongful conviction.

(Thanks to occasional fellow-blogger Zeke for digging up this case.)

Read more :: Permalink

Thursday, May 10, 2007

Failure to Resolve ID Issues = New Trial

Just came across an interesting decision from the great state of New Jersey from a couple weeks back, in which the New Jersey Superior Court overturned a conviction because the trial court failed to resolve factual issues relating to ID procedures. (State v. Greene, Not Reported in A.2d, 2007 WL 1223906 (N.J.Super.A.D. 2007))

The case involved an armed robbery of a Domino's Pizza restaurant. Shortly after the robbery, police showed up, and apparently in the presence and earshot of the witness, the officer on the scene received a radio call stating that the "guy who committed the robbery" had been apprehended. The officer on the scene then took the suspect to a field location to conduct a show-up.

There were several factual disputes that were left unresolved at the Wade hearing (determining the reliability of the ID), including a complete lack of information about clothing in original description by the witness that "evolved at the Wade hearing to match precisely the items revealed to Heiman by the prosecutor," unresolved issues regarding whether the suspect was on the ground or in a police car during the show-up and how far away the witness was, and an additional unresolved problem with the fact that the defendant was paraded in front of the witness in an orange jumpsuit and shackles immediately prior to the hearing itself.

Citing this article (PDF) on everything that's wrong with Manson v. Brathwaite, along with other works chronicling the problems with eyewitness evidence and the due process standard that should be serving to protect against the admission of unreliable procedures, the NJ Superior Court overturned the conviction, ordering a new Wade hearing and a new trial.

Read more :: Permalink

Monday, May 7, 2007

Lineup Reforms in Real Life

There are a number of law enforcement agencies out there with real lineup reforms in action -- like Hennepin County, Minnesota, and Northampton, Massachusetts (PDF), among others -- and in every case that we're aware of where police departments have enacted these reforms, the police themselves have become advocates for them. The reality is, as they will be the first to tell you, that implementing basic reforms like double-blind procedures and cautionary instructions not only make for more reliable identifications, but also make lineup procedures less vulnerable to challenges in court.

We just became aware of reforms currently in place (PDF) in Santa Clara County, California -- a county spanning 15 cities and towns presumably with widely varying law enforcement resources, for those who would claim that the reforms are only feasible on a big city budget.

A great quote from the president of the Police Chiefs' Association of Santa Clara County, on the importance of using sequential lineups (showing one photo at a time) over simultaneous (six or more photos at once) ones:

By studying cases where DNA has proven that innocent people have been convicted, we have learned that there are many causes of false convictions. One cause - cases of good witnesses honestly but falsely making an identification. Exhaustive studies have found that witnesses are much more likely to identify the guilty suspect if the lineup is sequential. Under traditional simultaneous lineups, some witnesses will inadvertently begin to compare the photos to one another instead of comparing the photo to their
memory. Consequently, the identifications are not as reliable as those conducted sequentially.

And on the rationale for using an administrator who does not know the identity of the suspect:
The DOJ study found that even the most experienced officer can inadvertently give subconscious hints to the witness to identify the suspect. This can result in false identification. This change brings us in line with other professionals. For example, doctors who are conducting medical research never know
whether their patient is receiving the new experimental drug or a placebo. This way they can never be accused of influencing the results. By using these new protocols we will bring our practice in line with other professionals.

Seems like a level-headed way to look at it to me.

Read more :: Permalink

Eyewitness ID Reform on NPR Today

Just got wind that NPR's show "To the Point" will be devoting today's show (live at noon on the west coast) to DNA and wrongful convictions, with a discussion on the reliability of eyewitness evidence and proposed reforms to police procedures to minimize future errors. Might be worth checking out:

DNA and Wrongful Convictions
MON MAY 7, 2007

After years in prison for crimes they didn't commit, hundreds of innocent people have been exonerated by DNA testing. Monday, on To the Point, can line-ups and other testimony from eye-witnesses be trusted? How can other mistakes be corrected? Is a flawed criminal justice system being repaired?

Read more :: Permalink

Wednesday, May 2, 2007

What about Frye? The Admissibility of IDs from Faulty Police Lineups

It is well-settled among social scientists who have studied the mechanisms of eyewitness memory that certain police practices are likely to enhance the reliability of identifications resulting from police lineups, whereas certain faulty practices have been demonstrated to diminish that reliability. Yet under federal due process law (i.e. Manson v. Brathwaite), identifications from egregiously flawed police lineups are routinely admitted because the result is found to have an independent source, notwithstanding the suggestivity of the lineup procedure.

Courts have widely interpreted Manson -- however wrongly -- to mean that it really doesn't matter how suggestive a procedure is, as long as the witness had, for example, a sufficient opportunity to view the perpetrator such that she would theoretically have been able to ID the person if the police hadn't gone out of their way to suggest that their suspect was the guy. Under that unfortunate interpretation of the governing rule, due process has been rendered a fundamentally ineffective legal vehicle for challenging the admissibility of suggestive and otherwise unreliable police lineups.

But what about Frye? Here in DC, in order to introduce evidence that flows from a scientific procedure, the proponent of the evidence must show that "the thing from which the deduction is made [is] sufficiently established to have gained general acceptance in the particular field in which it belongs." A police lineup is a scientific experiment designed to test an eyewitness's ability to accurately recall the identity of the perpetrator. The community of experts in the field of human memory are in a state of broad consensus with respect to the lineup procedures that make for a reliable test of human memory, and those that make for a poor one. When there is anything short of a consensus among that community -- that is, the community of social scientists who have actually researched the effectiveness of the procedures employed by police to test witness memory -- with respect to the reliability of a particular procedure, identifications flowing from those unreliable procedures should be excluded from use in criminal trials.

The question before the court in Frye itself was similar, when it was faced with a challenge to the admissibility of the results of a polygraph test. The polygraph is designed to test the accuracy of a person's account of some event. A police lineup is also designed to test the accuracy of a person's account of some event, and specifically, the identity of the perpetrator. When a lineup is administered in accordance with settled best practices -- i.e. with proper instructions, conducted by someone who does not know the identity of the suspect, with an adequate number of fillers, where the fillers match the characteristics of the original description, with only one suspect per lineup, etc. -- then the relevant community of scientists are in consensus that the procedure is more likely to yield an accurate identification. When those procedures are not followed, there is broad consensus for the exact opposite proposition -- namely, that the procedure and any identification that flows from it are demonstrably unreliable.

Evidence flowing from an unreliable test of memory should be excluded for the same reasons that the DC Court of Appeals excluded evidence from the polygraph test in 1923: its proponents cannot establish that it reliably produced the evidence it was purported to produce.

So why not use Frye (and, equally, Daubert) to keep out unreliable IDs? It just struck me that you shouldn't even need due process to keep out the results of a procedure whose unreliability is corroborated by a broad consensus among scientists.

Read more :: Permalink