How effec­tive are these field sobri­ety tests?

Con­sider the research funded by the National High­way Traf­fic Safety Admin­is­tra­tion (NHTSA), which resulted in the later adop­tion of the so-called “stan­dard­ized” field sobri­ety tests.

In a study, researchers deter­mined that the three most effec­tive field sobri­ety tests (FSTs) were walk-and-turn, one-leg stand, and hor­i­zon­tal gaze nys­tag­mus. Yet, even using just these sup­pos­edly more accu­rate tests, the researchers found that 47 per­cent of the sub­jects who would have been arrested based upon test per­for­mance actu­ally had blood-alcohol con­cen­tra­tions of less than the legal limit of. 10 per­cent. In other words, almost half of all per­sons ‘fail­ing” the tests were not legally under the influ­ence of alcohol!

Later, these same researchers con­ducted fur­ther tests in an attempt to improve the cred­i­bil­ity of the pro­posed “stan­dard­ized” bat­tery of FSTs. The error rate improved some­what: The false results dropped to 32 percent-i.e., “only” a third of all per­sons judged to be guilty by these tests were, in fact, innocent.

These tests are unfa­mil­iar to most peo­ple and not well prac­ticed mak­ing it more dif­fi­cult for peo­ple to per­form them. As few as two mis­cues in per­for­mance can result in an indi­vid­ual being clas­si­fied as impaired because of alco­hol con­sump­tion when the prob­lem may actu­ally be the result of the unfa­mil­iar­ity with the test.

It appears the NHTSA-funded researchers used meth­ods that ensured improved reli­a­bil­ity fig­ures. Appar­ently, to reduce the num­ber of bor­der­line sub­jects (those with blood-alcohol lev­els of, say, .09 or .11 per­cent), most of the sub­jects received either exces­sive amounts of alco­hol so that their BACs were ele­vated to .15 per­cent, or very small amounts so that they were below .05 percent.

NHTSA con­ducted a study to deter­mine the effi­cacy of the var­i­ous field sobri­ety tests.  The study found that “at present, the tests and pro­ce­dures used vary between local agen­cies and offi­cers” and that “for many of these tests, the rela­tion­ship between per­for­mance and spe­cific BAC lev­els has not been well doc­u­mented.” As a result of this and other stud­ies, NHTSA now rec­om­mends only three field sobri­ety tests: walk-and-turn (walk-the-line), one-leg-stand, and nystagmus.

Another group of researchers tested the effi­cacy of the pro­posed stan­dard­ized FSTs. The study involved test­ing the abil­ity to deter­mine whether a suspect’s blood-alcohol level was above or below. 10 per­cent — that is, whether he was “under the influ­ence” in most states. The test, con­ducted under lab­o­ra­tory con­di­tions, indi­cated that the walk-and-turn tests resulted in a cor­rect assess­ment 75.1 per­cent of the time, the one-leg-stand 75.5 per­cent, and nys­tag­mus 81.8 per­cent; when all three were given, a cor­rect deter­mi­na­tion was arrived at in 83.4 per­cent of the cases. Put another way, these “improved” FSTs still iden­tify roughly one-fourth of inno­cent DUI sus­pects as guilty — and this pre­sumes hon­est and accu­rate admin­is­tra­tion of the tests by an expe­ri­enced offi­cer under ideal lab­o­ra­tory conditions.

Later, many of the orig­i­nal researchers at the South­ern Cal­i­for­nia Research Insti­tute who had been fed­er­ally funded to come up with a stan­dard­ized bat­tery pub­lished find­ings of their research. The study con­cluded that FSTs do not accu­rately mea­sure dri­ving impair­ment. Researchers rec­og­nized that such tests are designed to deter­mine bal­ance, steadi­ness, and reac­tion time but con­cluded that a con­nec­tion between these fac­tors and dri­ving abil­ity “is not appar­ent since nei­ther a steady stance nor sim­ple move­ment time is essen­tial to the safe oper­a­tion of a motor vehi­cle.” While con­ced­ing that field sobri­ety tests may indi­cate the pres­ence of alco­hol, the researchers found that they do not nec­es­sar­ily mea­sure dri­ving ability.

Dr. Spur­geon Cole of Clem­son Uni­ver­sity con­ducted a study on the accu­racy of FSTs. His staff video­taped 21 indi­vid­u­als per­form­ing six com­mon field sobri­ety tests, then showed the tapes to 14 police offi­cers and asked them to decide whether the sus­pects had “had too much to drink to drive.” Unknown to the offi­cers, the blood-alcohol con­cen­tra­tion of each of the 21 sub­jects was .00 per­cent. The results: 46 per­cent of the time the offi­cers gave their opin­ion that the sub­ject was too ine­bri­ated to drive. In other words, the FSTs were hardly more accu­rate at pre­dict­ing intox­i­ca­tion than flip­ping a coin.