This thread is a background on the numbers behind NHTSA’s alleged “validation” of the “standardized” field sobriety tests in DWI arrests and those statistics they claim for alcohol-caused deaths. Click here to read more about the actual standardized field sobriety testing procedures in Missouri DWI cases.
Background of NHTSA’s Validation Studies
Years ago, the methods the police used to determine if someone was “intoxicated” while driving were only limited by their imagination. The tests utilized to determine if someone was DWI were wildly varied in jurisdictions across the United States, but the goals of all of these tests was obviously the same– to convict as many drunk drivers as possible.
The old-fashioned tests included: ABC’s, counting, drawing, picking up coins from the hood of the officer’s car, blowing up balloons, and anything else that could be conceived of.
Beginning in 1975, studies began which were sponsored by NHTSA to determine which of the field sobriety tests being utilized around the country were the most accurate. These studies were conducted by the Southern California Research Institute. The goal of the studies was not for SCRI to develop new field sobriety tests, but to determine which of the field sobriety tests in use by law enforcement at that time were the most accurate. SCRI was directed by Marcelline Burns, who created a technical and cost proposal and submitted it to be awarded NHTSA’s contract.
1977 Study
The first NHTSA study was reported in 1977. In first attempting to determine which tests then in use were the most accurate 15 to 16 tests were considered, and 238 drinking subjects and 10 police officers were utilized. The study first reduced down to the six most “reliable,” and then reduced down to three. The 1977 felt that the HGN, horizontal gaze nystagmus eye test, the walk-and-turn test, and the one-leg stand test were the most reliable. The finger-to-nose test, the finger-count test, and the tracing test were disregarded as unreliable, although this will, of course, not stop Missouri police officers from STILL utilizing them in making DWI arrests.
Of course, one of the biggest mistakes DWI attorneys make is not actually reading the original studies, and instead relying on the “summary” of the study where NHTSA gives the information that it wants the public to focus on when it publishes it training “manuals” for the police. When anyone bothers to read the actual underpinnings of NHTSA’s conclusions in their manuals about how accurate these tests are, it is an eye-opening experience.
There is much information in the original 1977 study, which you will, of course, not find in the NHTSA training manuals, which is astounding:
The error rate in the 1977 NHTSA study was an astounding 46.5%. The officers made the decision to “arrest” a total of 101 people. 47 of these 101 people had a BAC (blood alcohol content) below 0.10% (the legal limit at the time)! Even the study authors admitted that the results were “totally unacceptable.” They also were completely subjective in the subjects they chose for the 1977 study to get to the end result they were seeking: 80% of the subjects were in their twenties, and about 2/3 of them were male.
General principles of statistical analysis and good science were also largely ignored:
No neurologists or ophthalmologists were brought in for consultation for the eye testing; no true reliability testing was administered in the true scientific meaning of the word; Important testing variables such as swaying while performing the test, the person’s eyes or the person’s attitudes, or odor of alcohol were not given “controls” for the test; elderly or obese people were not subjects in the test at all, and females in general only made up about 1/3 of the subjects; there was no contact lens testing for the eye tests; and the prevailing scientific literature in the field was only generally referenced, therefore making the study largely exist in a vacuum outside of reality.
The 1981 Study
Despite the shortcomings of the 1977 study, and they were numerous, NHTSA gave SCRI a second attempt to retest and “standardize” field sobriety testing. In the 1981, only the new three-test battery the 1977 indicated was the most reliable was utilized– The HGN, horizontal-gaze-nystagmus eye test, the walk-and-turn test, and the one-leg-stand test.
The 1981 test utilized 296 subjects, and attempts were made to add “divided attention” aspects to the three tests, thereby making the administration portion of these three tests allegedly standardized. (example: the verbatim walk-and-turn instructions phase).
It is highly touted that the 1981 study bettered the unacceptable error rate of the 1977 study. Again, it is important to read the statistics in the actual 1981 study, as opposed to the canned summary of what NHTSA says about the 1981 study in the NHTSA manuals.
In the 1981 study:
32& of the 118 arrest decisions of the officers for DWI were WRONG!
18% of the subjects who had NO alcohol in their system whatsoever were misjudged by the officers to be impaired.
The officers in the 1981 study believed 31% of the people who were actually at a 0.05% blood alcohol content to be impaired above the legal limit (which was then 0.10%– meaning that they were guessing 31% of the people actually at 0.05% to BE TWICE THE LIMIT THEY ACTUALLY WERE)!
Marcelline Burns and SCRI’s “explanation” for these problems? She believed that the results were poor because the study was done “next to the drug capital of the world.”
Still, how did the 1981 study improve the error rate over the 1977 study? By skewing their numbers with improper dosing differential, of course. (You won’t find THAT in the NHTSA training manuals, but you can read it for yourself in the study).
In other words, they loaded the deck by making sure many of the test subjects were dosed at levels OTHER than the 0.10% level that they were trying to say these three tests were accurate in gauging people to be at.
In the 1981 Study:
33% of the subject’s actual BAC was 0.0%, no alcohol;
34% of the subject’s actual BAC was 0.05% , half the legal limit they were testing for (0.10%);
11% of the subject’s actual BAC was 0.15% of higher.
In other words, these subjects represent a disproportionately high number of “gimmie’s,” that my grade school daughter could accurately assess. It seems without question that the reduction in the “false arrest” rate from 46.5 percent in 1977 study down to 32 percent in 1981 study is due in large part to this “dosing differential.”
Further skewing the numbers of the 1981 study in favor of a new found “accuracy” is the fact that the number of subjects dosed in the mid-range of 0.05% BAC and 0.15% BAC (the most important group as it is the group that they are allegedly trying to find) went down from 27% of the total subjects in the 1977 study to 22% in the 1981 study.
Despite these nonsensical testing discrepancies (why weren’t the same number of subject’s used at the same BAC levels in each study for scientific accuracy and reliability accept to get to the result you are seeking?) NHTSA claimed an overall accuracy rate of 80% from the 1981 study when using the three-test battery of the HGN eye test, the walk and turn test, and the one leg stand test.
As an aside, it is also interesting to note that NHTSA did NOT EVER RELEASE the results of the subjects dosed between 0.05% and 0.15% BAC (those most important to determining whether or not these tests actually are accurate in finding impairment at the mean level of 0.10%, the legal limit at the time). I wonder why that is that they chose not to release the numbers for the most relevant subjects closest to the limit they are looking for? Assuredly, these individuals would undoubtedly have SIGNIFICANTLY lowered the accuracy rate.
The “reliability” portion of the 1981 study consisted of asking 100 of the subjects “back” for retesting two weeks after the original study. The “reliability” factor was 0.77. The inter-rater reliability coefficient dropped to 0.57.
Similar to the 1977 Study, the age and gender factors of the 1981 Study were also unfairly skewed:
80% of the subjects were between the ages of 21 and 34, and like the 1977 Study, about 2/3 of them were male.
Even assuming NHTSA’s 1981 study numbers are not completely cooked and self-serving, and that the three tests were administered, demonstrated, and scored correctly, you are still talking about 2 out of 10 people being wrongly accused by these tests.
There were other errors in the 1981 study similar to those in the 1977 study, in that general principles of statistical analysis and good science were largely ignored:
No neurologists or ophthalmologists were brought in for consultation for the eye testing; no true reliability testing was administered in the true scientific meaning of the word; Important testing variables such as swaying while performing the test, the person’s eyes or the person’s attitudes, or odor of alcohol were not given “controls” for the test; there was no contact lens testing for the eye tests; and the prevailing scientific literature in the field was only generally referenced, therefore making the study largely exist in a vacuum outside of reality.
The Good-Ausberger Study
One of the earliest non-NHTSA studies of SFSTs was published in in the American Journal of Optometry & Physiological Optics by two optometrists at Ohio State University, Gregory W. Good and Carol R. Augsberger in 1986.
In particular, this study focused on the HGN horizontal gaze nystagmus test. The study noted that 92% of subjects scoring four ”points” or higher on the HGN registered BAC’s above 0.10, and NHTSA, of course, seized on this research to claim to everyone that would listen that HGN is “92% accurate in identifying intoxicated people.”
While it may be factually accurate that 92% of the subjects in the study who scored four “points” on the HGN test were above 0.10%, this overlooked the study’s FALSE POSITIVES, where the study’s own charts indicate that 81.5% of those with blood alcohol content BELOW 0.10% ALSO SHOWED FOUR OR MORE CLUES.
NHTSA made a concerted effort to ignore the fact that the study’s data indicates the HGN test is 82% inaccurate as applied to INNOCENT PEOPLE, and screamed from the ramparts instead that the HGN was “92% accurate in identifying intoxicated people.”
From the 1981 Study and the Good-Ausberger Study, the NHTSA publicity machine swung into high gear, and everyone has been improperly convinced that these three tests are valid every since, based on bogus numbers.
NHTSA then proceeded on with three standardized field sobriety validation studies between 1995 and 1998: The Colorado Validation Study in 1995, The Florida Validation Study in 1997, and the San Diego Study in 1998.
In general all three of these studies are fairly consistent in terms of low false negative rates, and believe me, that is all you will hear NHTSA talk about in their SFST training manuals for the police– how wonderful and reliable everything is.
What you don’t get in the NHSTA student manual or the NHTSA instructor manual is the high numbers of false positives for these three field sobriety tests in all three of these studies– you won’t see that unless you read the hard data of the original NHTSA studies themselves.
The false positives rate is what is completely alarming– where the wrongly accused are being arrested because of flawed “science,” and the term is used loosely. NHTSA just glosses over the false positives and hopes no one will notice.
The reason why so many people over the limit of 0.08% BAC and 0.10% BAC show four or more clues on the HGN horizontal gaze nystagmus test, is because so many people ALSO HAVE FOUR OR MORE CLUES ON THE HGN TEST at BAC LEVELS OF 0.04%, 0.05%, and 0.06%, within the legal limit to drive!
NHTSA says that the Colorado study was the first field study that utilized police fully trained in the use of the three-test SFST battery. They also claim that the correct arrest decisions were made 93% of the time based on the officer’s implementation of the three-test battery in the Colorado study, which is obviously considerably better than the miserable results of the 1977 Study and the 1981 Study. I wonder how they got there in so short a time?
NHTSA claims the Florida validation study was to answer the question of whether or not the newly-developed three-test battery is valid and reliable as indices of the presence of alcohol when used under “present day” traffic and law enforcement conditions. In the Florida Study, NHTSA alleges that the arrest decisions of the officers were correct 95% of the time! Isn’t it amazing how much better they are allegedly getting?
NHTSA claims the San Diego study was undertaken because NHTSA wanted these three tests to be recognized as capable of discriminating between BACs above and below 0.08% BAC (remember the original studies were only “standardized,” if you can call it that, for the old legal limit of 0.10% BAC), as NHTSA and MADD had campaigned to reduce the per se limits to 0.08% BAC across the nation. Not surprisingly, NHTSA claims the San Diego study had the “correct arrest decision” was made 91% of the time at the 0.08% BAC level or above.
THE REST OF THE STORY AND THE COOKED NUMBERS OF THE FALSE POSITIVES IN THE COLORADO, FLORIDA AND SAN DIEGO STUDIES THAT NHTSA GLOSSES OVER AND DOESN’T WANT YOU TO KNOW ABOUT
Believe me, the attorneys for the State of Missouri in any DWI case, or the attorneys for the Missouri Director of Revenue will use the validation numbers from these three tests for the new 0.08% standard, as opposed to the ridiculously low numbers of the original NHTSA validation research in 1977 and 1981.
When you cut through the propaganda of the NHTSA training manuals and get to the raw data of these three actual studies, the numbers from the studies are shocking, and an affront to real science.
In the Florida validation study, 16% of all people below 0.08% BAC HAD SHOWED ALL 6 CLUES FOR THE HGN TEST!
The Florida validation study also suggests that OVER 50% OF THE SUBJECTS HAD AT LEAST 4 OF 6 CLUES FOR THE HGN TEST, but of course, it does not come out and say it……Instead, NHTSA attempts to conceal these numbers by saying that 1/2 of the correctly released drivers had 0 of 6 clues for the HGN test, or 2 of 6 clues for the HGN test, which suggests that HALF of the correctly released drivers (under 0.08% BAC) had MORE than 2 of 6 HGN clues.
The Colorado validation study found that 1 in 8 people UNDER 0.05% BAC SHOWED 4 OF 6 OR MORE HGN CLUES!
Of course, the police will never readily admit that 13% of people under 0.05% by NHTSA’s own touted numbers would be improperly deemed “impaired” by an HGN test with four or more clues………….
The Florida validation study also states that 67% OF ALL INCORRECT ARREST DECISIONS (meaning that they wrongly thought they were above 0.08% when they were actually below 0.08) IN THE FLORIDA STUDY HAD ALL 6 CLUES FOR THE HGN TEST!!!!
In the San Diego study, the one most frequently cited, and the one prosecutors will most frequently blindly regurgitate for the notion that “79% reliability for walk and turn! 83% reliability for the one leg stand! 88% reliability for the HGN!,” the AVERAGE BAC OF THOSE ARRESTED IN THE SAN DIEGO STUDY was 0.15%!!
This means the average was almost two times the legal limit of 0.08% BAC! How loaded can they make the deck? My grade-schooler can see with her own eyes if someone is at 0.15% BAC.
The major literature in the scientific community indicates that someone at 0.15% BAC is completely ripped…. Kurt Dubowski, (one of the world leaders in breath testing) indicates that at 0.13 — 0.15 BAC, a subject will show: “gross motor impairment and lack of physical control- Blurred vision and major loss of balance. Euphoria is reduced and dysphoria is beginning to appear. Judgment and perception are severely impaired. Dysphoria is an emotional state of anxiety, depression, or unease.” A subject at this level is not exactly who a real scientist would go looking for if they were attempting to get valid results for a test.
False positives in the San Diego study (those arrested which had BAC levels below 0.08%) were six times as common as false negatives (persons NOT arrested who actually had BAC levels of 0.08% or greater. For all people stopped and investigated 29% of those with a BAC under 0.08% were arrested!
In the 0.07-0.09% BAC range (the most scientifically relevant, of course for testing a 0.08% BAC standard) the accuracy was unbelievably low.
36% of those arrested were over the legal limit of 0.08% BAC, while 64% OF THE PEOPLE ARRESTED WERE UNDER THE 0.08% BAC LIMIT! Think about THAT when you have had two beers and get stopped driving home before you take these nonsensical tests…….
Perhaps most shockingly of all, in my humble opinion, is the fact that ALL of the police officers in the San Diego study were equipped with portable breath testing devices to skew their already pathetic and scientifically unreliable guesses in their favor!! What else could they possibly do to get the results they were looking for?
Most everyone tested in the Florida validation and the Colorado were RIPPED, as the average BAC of those arrested in the Florida study was 0.15% BAC, and in the Colorado study 0.152% BAC…… like shooting ducks in a bathtub… but that is what they are looking for to begin with…
In the Colorado and San Diego studies, we have no idea what instructions were given for the walk and turn and one leg stand test (even though the instructions are now deemed “critical” to “standardization”) because they did not bother to record it during testing. Further, the instructions for these two tests in the Florida study differ substantially from the NHTSA training manuals, and are therefore basically different tests!
The Colorado study reported that only 13 errors of administration occurred during testing and that 6 errors in instructions were observed in 305 administered field sobriety tests. What an amazingly high number… Funny how SCRI only bothered to actually observe the officers administer the tests 41% of the time in the study, despite getting immense amounts of money from the taxpayers through their NHTSA contract to get it right.
Even better, “no errors” were allegedly observed in the 313 field sobriety testing batteries given in the Florida study, even though 1/3 of the tests which were administered weren’t even monitored for accuracy!
The NHTSA “validation” numbers for standardized field sobriety testing accuracy are the biggest joke I can think of that has been foisted on the public as truth in my lifetime. In general, “garbage in and garbage out.” You can make numbers anything you want when such scientifically invalid methods are used, and that is exactly what NHTSA got to justify it’s DWI witch hunt of the last thirty years.
REAL STUDIES FROM THE SCIENTIFIC COMMUNITY ABOUT THE RELIABILITY OF “STANDARDIZED” FIELD SOBRIETY TESTING
There are, of course, actual scientific studies on the reliability of these tests, and I am sure you would be surprised to find that the numbers in the studies in the scientific community (peer-reviewed by others in the field with real statistical methods utilized) are nothing like the numbers NHTSA trots out.
In the study, “End Position Nystagmus as an Indicator of Ethanol Intoxication,” 40(2) Science and Justice 113 – 116 (2001), by Booker, the reliability of the HGN test is discussed at length, and specifically the comparison of the jerking of the eyes of the driver during the HGN test from alcohol impairment v. general fatigue in non-drinking drivers. The study found that there was a 55% false positive rate due to fatigue as opposed to alcohol consumption. Most significantly in that study, MORE THAN 50% OF THE SUBJECTS TESTED POSITIVE FOR EYE JERKING (NYSTAGMUS) WITH AN ACTUAL BAC LEVEL OF 0.0% (no alcohol) !!
In the study, “Field Sobriety Tests: Are they Designed for Failure?” 79 Perceptual and Motor Skills, 1994, by Cole and Nowaczyk, two groups of seven law enforcement officers each viewed videotapes of 21 sober individuals performing a variety of field sobriety tests or normal-abilities tests, e.g., reciting one’s address and phone number or walking in a normal manner. Officers judged a significantly larger number of the individuals as impaired when they performed the field sobriety tests than when they performed the normal-abilities tests.
The work of Dr. Spurgeon Cole (one of the authors) is also discussed at length in the case, United States v. Horn, 185 F. Supp. 2d 530 (D. Md. 2002). The Horn Court noted Dr. Cole’s critique of NHTSA’s original validation research not being peer-reviewed by the scientific community and it’s overall unreliability:
47% of the subjects tested in the 1977 NHTSA laboratory study who would have been arrested by the testing officers for driving while intoxicated (BAC of 0.10 or greater) actually had BACs below 0.10;
in the 1981 Final Report, 32% of the participants in the lab study were incorrectly judged by the testing officers as having BACs of 0.10 or greater; and
the accepted reliability coefficient for standardized clinical tests is .85 or higher, yet the reliability coefficients for the SFSTs, as reported in the NHTSA studies, ranged from .61 to .72 for the individual tests and .77 for individuals that were tested on two different occasions while dosed to the exact same BAC. More alarmingly, inter-rater reliability rates (where different officers score each subject) ranged from .34 to .60, with an over-all rate of .57.
Dr. Cole theorized that the SFSTs, particularly the walk and turn test and one leg stand test, required subjects to perform unfamiliar, unpracticed motions and noted that a very few miscues result in a conclusion that the subject failed and had a BAC in excess of 0.10.
His hypothesis was that individuals could be classified as intoxicated / impaired as a result of unfamiliarity with the test, rather than actual BAC.
He tested this hypothesis by videotaping twenty-one completely sober individuals performing either “normal-abilities tests” (such as reciting their addresses or phone numbers or walking in a normal manner) or the WAT and OLS tests. The results of the study were that 46% of the officers that viewed the videotape of the sober individuals performing the SFSTs rated the subjects as having had too much to drink, as compared to only 15% reaching this decision after seeing the videotape of the subjects performing the normal-abilities tests.
The study, “Statistical Evaluation of Standardized Field Sobriety Tests,” 50(3) J Forensic Sci (2005), by Hlastla, Polissar, Oberman, is a critique of NHTSA’s report on standardized field sobriety testing which alleges the SFSTs as being 91% accurate in predicting Blood Alcohol Concentration (BAC) as lying at or above 0.08%.
The study notes that NHTSA’s conclusions regarding accuracy are heavily weighted by the large number of subjects with very high BAC levels, and re-analyzes the original NHTSA data with a more complete statistical evaluation.
The study’s evaluation indicates that the accuracy of the SFSTs depends on the BAC level and is much poorer than that indicated by the NHTSA research using the same numbers. The study found that while the SFSTs may be usable for evaluating suspects for BAC, the means of evaluation must be significantly modified to represent the large degree of variability of BAC in relation to SFST test scores.
OTHER NUMBERS COOKED BY NHTSA IN RESULTS-BASED, BIASED STATISTICS REGARDING DWI ACCIDENTS
For at least the last twenty years NHSTA and MADD have justified the erosion of Constitutional rights and the move toward what is for all practical purposes prohibition by claiming there is ““carnage on our highways caused by drunk drivers.”
Year after year NHTSA claims that there are approximately 18,000 alcohol related deaths on our highways. “Alcohol caused” deaths are not the same as “alcohol related.” NHTSA’s sleight of hand is that if anyone has consumed alcohol, they deem it alcohol related.
When the General Accounting Office (GAO) reviewed these figures form the National Highway Traffic Safety Administration, the GAO reported that they “raised the methodological concerns calling their conclusions into question.” NHTSA’s numbers “fall short of providing conclusive evidence that 0.08% BAC laws were, by themselves, responsible for reductions in alcohol related fatalities”.
This means that the statistics were not valid when examining alcohol related deaths, much less alcohol caused deaths.
Some independent investigations have come up with numbers much different from those published by NHTSA.
The Los Angeles Times found that only about 5,000 deaths per year involved a drunk driver causing the death of a sober driver, passenger, or pedestrian. Responsibility in DUI Laws, Inc. put the number at closer to 3,000.
Not that any of these deaths is a good thing, I personally had a good friend from childhood recently killed in a DWI head-on collision, but you see the point. NHTSA cooks it’s numbers however is convenient to stand for whatever proposition it is attempting to jam down some State’s throat at that moment. Just you wait, the 0.05% per se limit, is on it’s way. If States won’t comply, they will lose their federal highway money just like when we dropped to 0.08%. Never mind the fact that all we are doing is increasing the net to arrest people who are not impaired because they have crossed a threshold that is scientifically unreliable, and are being persecuted for having a couple of drinks and driving home, despite it being allegedly legal, rather than focus on the true problem in drunk driving, which is repeat offenders who continually blow off of the chart and are actually impaired and a menace to everyone driving. If we as a society really want prohibition all over again, why all the smoke and mirrors?
NHTSA continues to mislead the public to justify the passage of increasingly punitive DWI laws throughout the country.
Posts
0 comments:
Post a Comment