Monday, September 26, 2011

Model policy under development for eyewitness ID procedures

On Friday I was in Huntsville to attend a working group meeting at Sam Houston State University, where academics at the Law Enforcement Management Institute of Texas (LEMIT) are preparing a state-mandated model policy on eyewitness identification procedures. Perhaps 40 people were there, including representatives from law enforcement, prosecutors and a handful of advocates including yours truly (attending on behalf of the Innocence Project of Texas) rounding out the mix.

LEMIT was charged in legislation passed this spring with creating a model policy for dissemination to law enforcement agencies that conduct eyewitness identification procedures. Agencies are not required to adopt what LEMIT comes up with, but in the past, when this same structure was used for creation of racial profiling policies, most agencies adopted the model policy or something pretty close to it.

I won't post their draft since the document hasn't been finalized, but I was pleased to see that the initial version produced by the folks at LEMIT - led by Dr. William Wells - relied closely on best practices and the latest science in the field regarding blind administration, admonitions to witnesses, sequential presentation, and documenting witness reactions. The science underlying the suggested policy is quite well developed at this point, Dr. Wells told the group, and is remarkable for the level of consensus demonstrated on best practices.

Perhaps the most disputed element of the policy was reliance on sequential presentation, but arguments against that method were undercut significantly by a just-released study (the most comprehensive of its kind, described here) which found witnesses choose "fillers" (non-suspects) 12% of the time in sequential lineups and 18% of the time when photos are presented in a group. A robbery detective from Houston opined that the peer-review process on that study hasn't been completed and these were only preliminary findings. But peer review won't change the raw data, and a 50% higher error rate for the group presentations - to this writer, anyway - seems like too big a difference to ignore.

The issue is whether the witness is comparing the photo they're looking at to their own memory or to others in the array. Group presentations, the theory goes, encourage witnesses to pick the subject that looks most like the perpetrator among the available options, instead of comparing each one directly to the image they recall. Human memory is less reliable than courts and juries have often supposed. Research developed over the last three decades or so has demonstrated eyewitness testimony is essentially "trace evidence," and like other trace evidence it may be easily contaminated using shoddy collection techniques.

Some law enforcement folks at the event seemed to be in a state of denial. The police chiefs association argued that the draft LEMIT policy was too detailed, including elements that shouldn't be considered "policy" but fall more in the realm of "procedures." This struck me as downright bizarre, not only because LEMIT was charged with creating a "detailed written policy," but because the statute in at least four different places directed them specifically to write "procedures" for various aspects of live lineups and photo arrays. The distinction being suggested by law enforcement interests would contradict the plain language of the statute (in case anybody cares about such things).

Indeed, hanging their hat in part on this spurious distinction between policies and procedures, a breakout group made up of law enforcement folks declined to come up with specific recommendations during the working group meeting, apparently hoping they could slow-walk the matter and delay the process. But SHSU must soon publish their draft in the Texas Register so they can receive formal public comment and finalize it by a statutory deadline in December. Participants in that subgroup were encouraged to stay after the event to complete their recommendations or to provide them via email in the next couple of weeks.

Court of Criminal Appeals Judge Barbara Hervey spoke up to debunk complaints that blind administration would be too difficult for small agencies to implement, demonstrating the simplicity of the "folder method" of blind administration, where photos are placed in file folders, shuffled, and presented sequentially so that the officer administering the procedure doesn't know which one is the suspect. That method was suggested by LEMIT as an alternative for smaller agencies to a truly blind administrator who's not one of the investigators in the case. In serious cases, though, if an agency is so small it can't provide a blind lineup administrator, they probably should be calling in the Texas Rangers or seeking help from larger agencies.

At the end of the event, a police chief approached your correspondent to provide his own department's newly developed policy and offered a rather odd warning. He opined that, just like with the racial profiling law, some agencies will refuse to create policies and just do whatever they want, predicting that such recalcitrance would cause the statute to be ineffective. By that logic, though, laws against murder, theft and drug taking are ineffective because, despite the statutes, some people still murder, steal, or take drugs. He's probably right some agencies will flout the law or adopt policies that fail to accept the state of modern science on eyewitness IDs. But in an environment where most agencies currently have no policies at all, my sense is that most departments will adopt acceptable policies and most officers will follow them.

Bottom line: This bill wasn't even controversial at the Legislature, so if too many agencies balk at implementing best practices, they risk the Lege returning to the subject and imposing more prescriptive rules in the future, without the discretion afforded in the current statute.

 See related Grits posts:


Anonymous said...

"That method was suggested by LEMIT as an alternative for smaller agencies to a truly blind administrator who's not one of the investigators in the case. In serious cases, though, if an agency is so small it can't provide a blind lineup administrator, they probably should be calling in the Texas Rangers or seeking help from larger agencies."

It isn't clear to me why the seriousness of a case should be a factor here. If a method isn't good enough for a serious case, it ought not to be used for a less serious case.

Anonymous said...

The size of the department shouldn't be an excuse. Even very small police departments can access other resources. They can ask for assistance from the sheriff's department. There are DPS officers in every county. The DA's office has investigators that can help. I think there are game wardens who are certified peace officers in every county. They could ask for someone from another department nearby to assist. Surely, among all these (and more that aren't included) they can find one person who isn't involved in the investigation to do a lineup.

I think the real reason for the resistance to this is that they know it may make some cases harder to prove. It will make it harder, when a witness is unsure, for them to prod them in the right direction. But, if it makes them work harder to make sure they actually get the right person, its a good thing.