HGN: How reliable is the Horizontal Gaze Nystagmus test?

Robustness of the HGN Test“The Robustness of the Horizontal Gaze Nystagmus (HGN) Test,” an article by the top federal researcher of Standardized Field Sobriety Tests, Marcelline Burns, was supposed to be released in 2004. This study was supposed to explore different variations of the mysterious HGN eye test, where officers move a finger or pen before a person’s face, declaring the person drunk or sober in under a minute. The study was designed to explore the impact of varying speeds of moving the finger, whether holding the finger too high mattered, and if laying down could produce flawed results. But the study was never released in 2004. Or in 2005. Or in 2006…

In 2007, a lot of DUI defense lawyers started asking, “Whatever happened to that ‘robustness’ study those fellas in
Washington were supposed to release?”
According to NHTSA, if a person has four or more clues out of six possible clues on the HGN eye test, they are correctly categorized as over 0.10 BAC. According to these folks, officers are correct 77% of the time, and police love nothing more than to talk about this “Gigmus” test before juries because it sounds so cool, like CSI. When the legal limit dropped to .08, the same researchers “proved” that the tests were accurate at the new lower limit without modification, and they even started to claim greater accuracy. In recent years, NHTSA has been claiming something akin to a nearly flawless ability to discriminate between drunk and sober people based solely upon the eye test.
Well, I kept looking for this robustness study because I’m stubborn, surfing the NHTSA site to see if the study was published. One naysayer claimed it would never be published because something didn’t sit well with the field sobriety gurus.
In March 2008, I came across a study claiming to be the so-called “robustness” study and reported my findings to DUI lawyers across the country. I flipped through it, and it essentially claimed that no matter how an officer does the test or under what conditions, the test can always prove that a person was drunk. I guess that is what “robustness” means. But I tend to disagree with the conclusions of the report since I see first hand how police officers screw the eye test up day after day. That, and I do not see how a study could be named four years before it is published, unless the researchers had a goal in mind and something to prove.
Jim Medley, a great Texas DUI lawyer, actually sat down and worked on the numbers based upon the raw data reported in the study. Apparently the NHTSA people did have something to prove, and they intentionally set out to do it regardless of the evidence.
Mr. Medley found that 67% of people tested by experienced police officers found sober people to be drunk even though the officers properly performed the test. When the test was improperly performed too fast, that number rose to 78%. It was even more alarming when Mr. Medley uncovered the fact that 92% of sober participants were declared intoxicated when the officers held the eyes too high or held the finger or pen too close to the eyes.
Steve Rubenzer, PhD, ABPP, one of the nation’s leading field sobriety test experts in statistical analysis of field sobriety tests and the HGN test pointed out that 77% of sober people failed the HGN test overall, bringing us full circle to the original numbers touted by NHTSA.  This leads one to wonder, “If the test was ever 77% accurate, how can it also be 77% inaccurate too?”
In 1996, Michigan courts acknowledged that the HGN test was scientifically acceptable and reliable without ever holding an evidentiary hearing. The courts jumped on a case where nobody understood the test, with the officer improperly describing how the test was supposed to be performed.  See People v. Berger, 217 Mich. App. 213, 551 N.W.2d 421 (1996). Despite several motions and appeals, Michigan courts refuse to revisit the issue.