As the police DECP.org explains, a traffic-police Drug Influence Evaluation has lots of parts—measurements, "indicators", interpretations, a final opinion. Where do these parts come from?

1971 - 1984: No science

As recounted in NHTSA's DRE Training Manual, DIEs were made up in the early 1970s, by two Los Angeles policemen dissatisfied that juries were not convicting defendants the officers wanted convicted. The officers tried taking drivers to medical doctors, but doctors disagreed with their assessments and refused to diagnose impairment. So the two policemen "undertook independent research by consulting with physicians, enrolling in relevant courses, studying text books and technical articles, etc.," and began telling juries they had a scientific test for identifying drug impairment.

The scheme worked. Juries convicted.

"Proof of the effectiveness of the DEC program began to be accumulated from the very outset of the program. LAPD personnel demonstrated that they could conduct examinations that led directly to the conviction of drug impaired drivers and other drug law violators."

Drug Evaluation And Classification Training Program, The Drug Recognition Expert School, HS 172A R1/10, US DOT, NHTSA, January 2010; Chapter 3, page 4

Read that again. Notice that the program was effective because it led directly to convictions. According to NHTSA, the purpose of the DIE is not accurate scientific determination of impairment; the purpose is convictions.

For more than a decade Los Angeles police got the convictions they wanted without bothering to do the scientific studies necessary to see whether their testing actually worked. The traffic-police DIE test had no basis in science.

1984 - 2000: Ready, fire, aim

In the early 1980s the federal traffic safety agency, NHTSA, learned of the LAPD program, saw that it was good, and moved to take it nationwide. But the "scientific" test NHTSA was adopting had never been scientifically tested —it had no basis in science.

Bigelow, Compton and Adler each use methodologies that render their conclusions about DIE accuracy scientifically wrong.

1. NHTSA paid for and self-published a lab project, Bigelow 1984 (aka Johns Hopkins).
2. NHTSA had an employee write up an in-house field project, self-published as a NHTSA Technical Report, Compton 1985 (aka LAPD 173).
3. Through the Arizona DOT, NHTSA funded a third project, reported in several stages, finally as Adler 1995 (aka the Arizona DRE Validation Study).

These three NHTSA studies each shared a failure that is important for DIE practice and for DRE officers' courtroom testimony. Each study failed to say exactly how traffic-police DIE tests were done. In the patois, they failed to reproducibly describe how the DIE-like testing they studied was done.

Because the studies failed to report what indicators and interpretations officers relied on to form their unstandardized opinions, NTHSA could not identify which individual indicators and interpretations made up the validated DIE. Turn that around and no DRE officer could then, or can today, correctly testify that any specific indicator or interpretation his opinion relied on is scientifically validated by any of these studies. Key elements of the DIEs currently done by US law enforcement have no basis is science.

2000 - now: Bureaucrats make it up

NHTSA called a meeting to make up the missing science.

In 2000 NHTSA, the National Safety Council, and the State of Washington Traffic Safety Commission put on the International Consultative Panel on Drugs and Driving Impairment. Delegates in "the fields of psychopharmacology, behavioral psychology, drug chemistry, forensic toxicology, medicine, and law enforcement experts trained in the recognition of drug effects on drivers in the field" guessed what DIE validation science would probably show. If there were any. Which there wasn't, or NHTSA wouldn't have needed the Consultative Panel.

The ICPDDI appears not to have published its minutes, or proceedings, or any official report. It was nonetheless used to justify NHTSA's Drugs and Human Performance Fact Sheets, a compilation purported to reflect committee guesses about what validation science would probably show are indicators of drug intoxication. If there were any validation science. Which there isn't, or NHTSA wouldn't have needed the Consultative Panel. The guesses memorialized in the Fact Sheets make their way to NHTSA's DIE Officer training manual and from there into officer testimony.

NHTSA's second strategic move was to redefine DIE evidence in a way that makes it in principal impossible for defendants to refute.

NHTSA brought in the International Association of Chiefs of Police to oversee and standardize the US law enforcement Drug Evaluation and Classification Program. The IACP publishes The International Standards Of the Drug Evaluation and Classification Program and revises and OKs the official NHTSA Drug Recognition Expert Training Manual. Notice that what the IACP standardizes is the DECP program—the DIE bureaucracy. The IACP does not standardize the DIE test itself. You'll see the importance of this in a second.

Watch carefully. As NHTSA hands DIE oversight to the IACP, the agencies fix their DIEs-have-no-basis-in-science problem. They switch the focus of the DIE—the definition of what a DIE is—from the DIE test to the DRE officers. Their claim is not that the DIE test is accurate, but that properly trained DRE officers are accurate.

This maneuver changes how DIE evidence works in court, where the dispositive principal becomes not science but officer training. To prove that a defendant's DIE test was valid and reliable, the prosecutor doesn't prove the test results were interpreted according to validated scientific standards, he just proves the DRE officer is trained and certified.

DRE officers are trained to do valid DIEs. A valid DIE is whatever a trained officer does. Because the IACP says so. A defendant can never prove the DIE officer formed his opinion in the wrong way, because the correct way is defined as whatever a trained DRE officer does. Identifiable science is not part of the analysis because identifiable science is not part of the definition.