Defendant appeals from a judgment of conviction for driving while under the influence of intoxicants (DUII), ORS 813.010.1 On appeal, defendant makes eight assignments of error. We reject without discussion defendant’s other assignments of error and write to address only the admission of certain scientific evidence, namely, testimony by a police officer concerning the vertical gaze nystagmus (VGN) test. We review for errors of law. State v. Hernandez, 227 Or App 319, 321, 206 P3d 197 (2009). Because we conclude that the evidence concerning the VGN test should not have been admitted and that the error was not harmless, we reverse and remand.

The central issue at trial was whether defendant “was under the influence of intoxicating liquor.” Most of the testimony at trial came from the arresting officer, Gutierrez.

Gutierrez testified to the following. Using LIDAR, a device for measuring a vehicle’s speed, Gutierrez observed defendant driving 64 miles per hour in a zone where the posted speed limit was 45 miles per hour. As Gutierrez put down his LIDAR and got into his patrol car, defendant slammed on his brakes and turned onto another street. Defendant was not weaving, and his braking was “consistent with a person going from a high rate of speed going to a low rate of speed in order to execute a turn.”

Gutierrez turned down a different side street to intercept defendant. When Gutierrez activated his emergency lights to stop defendant, defendant pulled to a stop on the wrong side of the street — that is, with his car pointing to face oncoming traffic. (One of defendant’s witnesses, who lives near where defendant was stopped, later testified that *536people sometimes park that way on that street.) When asked for his license and proof of insurance, defendant acknowledged that he had no driver’s license and no proof of insurance. Defendant told Gutierrez that he thought that he had been going 35 miles per hour in a 45-mile-per-hour zone.

Gutierrez “noticed that there was a strong odor of alcoholic beverage coming from [defendant] as he spoke” and observed that defendant’s eyes were bloodshot and his speech was slurred. Gutierrez asked if defendant had consumed any alcoholic beverages, and defendant replied that he had had two beers. Gutierrez later found two open beer cans in defendant’s car.

Gutierrez asked defendant to perform field sobriety tests (FSTs), and defendant agreed to do so. Gutierrez, who had completed drug recognition expert (DRE) training, administered “[t]he standardized field sobriety tests, which are the horizontal gaze nystagmus [HGN], walk-and-turn, and the one-leg stand.” In addition, after the HGN test, Gutierrez administered the VGN test, discussed in more detail below. A person fails the HGN test if four clues2 out of six are observed; defendant displayed six clues. Two out of eight clues is the standard for failing the walk-and-turn test; defendant displayed four clues. On the one-leg-stand test, defendant failed to raise his foot as high as instructed (raising it only two inches instead of six inches from the ground), but he maintained his balance perfectly and did not display any clues. Gutierrez testified that those three tests are “scientifically proven” and that, although other tests exist, those are the tests that he had been trained to administer for roadside DUII investigations.

Both the prosecution and the defense examined Gutierrez at some length about the accuracy of the FSTs as indicators of impairment. Those examinations addressed various studies concerning the accuracy of the FSTs singly *537and in combination. Gutierrez initially testified that, with four clues, the HGN test is 80 percent accurate in identifying a blood alcohol content of .08 percent and that the error rate is attributable to human error. Later, he testified that a 1975 study using only the HGN and walk-and-turn tests found an 80 percent accuracy rate, but a more recent study found that all three FSTs, combined, had a 95 percent accuracy rate when used in field conditions.

After completing the standardized FSTs, Gutierrez “advised [defendant] that based on what I had seen, starting with the driving, the odor of an alcoholic beverage, the fact that he told me he drank alcohol, that he was under arrest for driving under the influence of intoxicants.” Defendant, who had been calm and cooperative, became upset after being arrested. After transporting defendant to the police station, Gutierrez read defendant the “rights and consequences” information regarding implied consent. Defendant refused the breath test.

At trial, defendant offered the testimony of two witnesses who testified about their interactions with defendant shortly before and again after his arrest. Each witness had an Oregon Liquor Control Commission (OLCC) license and, as part of the training required to obtain that license, had received training on recognizing signs of impairment. Each testified that defendant was not under the influence of alcohol. Each also testified that Gutierrez had a bad attitude toward defendant.

The testimony at issue concerns the VGN test. That test was first mentioned in the prosecutor’s opening statement. Describing the anticipated evidence to the jury, the prosecutor stated:

“So [Gutierrez] does another test [after the HGN test]. He does what’s called vertical nystagmus. Now, this is something where you look up, and, again, he’s going to tell you exactly how this test is applied, but what it tells you is, it’s not like those other tests. Remember, earlier, the HGN is meant to treat everyone the same, big guy, little lady, heavy drinker, light drinker.

“This test tells you something different, it tells you that this person has consumed — if they have vertical nystagmus *538—a large quantity, a large amount, a large dose of alcohol for that particular person. It’s about that person, so it takes into effect the dose.

“And what happened here? [Defendant], again, had vertical gaze nystagmus.

“So you’re going to hear from the officer this isn’t the kind of thing you see every day. In fact, it’s not all that common for him to pick that type of thing up, this vertical gaze nystagmus. You have to be intoxicated, for that person.

“So that’s the first test. [Defendant] not just has these six clues [from the HGN test], he has an additional clue that just tells you that, whether he had too much alcohol to drink.”

During the state’s case-in-chief, the prosecutor asked Gutierrez about the VGN test. After the trial court overruled defendant’s objection based on a lack of scientific foundation for the test, Gutierrez gave the following testimony.

“[Prosecutor:] Again, did you perform any other eye tests on [defendant]?

“[Gutierrez:] Yes. We did the vertical gaze nystagmus, which instead of going side to side, I went up and down twice, and nystagmus was present in his eyes.

“[Prosecutor:] And what does that test tell you?

“[Gutierrez:] It indicates that that individual has consumed more alcohol than his body usually normally takes.

“[Prosecutor:] Now, you found this on [defendant]. Is there anything else that could cause that other than alcohol?

“[Gutierrez:] A depressant, an over-the-counter depressant. That will do it. The other drug is cannabis, which is marijuana. Inhalants will also cause it, and PCP will also cause the vertical nystagmus.

“[Prosecutor:] But you did see that in [defendant’s] eyes?

“[Gutierrez:] Yes.:

*539In closing, the prosecutor again referred to the VGN test:

“You also have the fact that [defendant] then had what they call — remember the officer was talking about vertical nystagmus? Well, what does that tell you? The officer says it tells you that that’s a high dose of alcohol for this particular individual. For him, this is more alcohol than he could handle.

“You can also look and see all of this right here in the manual if you want to, it’s all in here, but that’s what the officer’s testimony was, and that’s what they found, so he had that.”

The manual to which the prosecutor referred (which was offered by defendant and received into evidence) included a description of VGN and instructions for conducting the VGN test, as well as a statement that, “[although this type of nystagmus was not addressed in the original research, field experience has indicated that the presence of Vertical Gaze Nystagmus has proven to be a reliable indicator of high doses of alcohol for that individual or certain other drugs.” U.S. Department of Transportation, DWI Detection and Standardized Field Sobriety Testing, Student Manual (Feb 2006) at VII-3, VIII-8.

The jury returned a guilty verdict. Although the transcript does not contain the outcome of the jury poll, a colloquy during the sentencing proceedings indicates that two jurors voted not guilty.

We begin with principles concerning admission of scientific evidence. As the court explained in State v. O’Key, 321 Or 285, 291, 899 P2d 663 (1995), “[e]vidence perceived by lay jurors to be scientific in nature possesses an unusually high degree of persuasive power. The function of the court is to ensure that the persuasive appeal is legitimate.” (Footnote omitted.) Hence, the trial court must, “in the absence of a clear case, a case for judicial notice, or a case of prima facie legislative recognition,” assess the scientific validity of proffered scientific evidence. Id. at 293 (footnote omitted). The determination of scientific validity involves considering a number of potential factors. Id. at 299-306; State v. Brown, *540297 Or 404, 417, 687 P2d 751 (1984).3 The proponent of the evidence has the burden to establish its scientific validity. State v. Perry, 347 Or 110, 122, 218 P3d 95 (2009).

Here, the VGN test was offered to show that defendant had, in Gutierrez’s words, “consumed more alcohol than his body usually normally takes.” There is no dispute that the VGN test is scientific in nature. See, e.g., State v. Sampson, 167 Or App 489, 496, 6 P3d 543, rev den, 331 Or 361 (2000) (noting that VGN test “is based on medical science”). Although the manual received in evidence contains a brief statement that “field experience” demonstrates the reliability of the VGN test, at no point in the proceedings was a foundation laid under Brown and O’Key for admission of the test.

Defendant contends that Gutierrez’s testimony regarding the VGN test therefore was inadmissible. The state responds that such a foundation was unnecessary because the VGN test “is based on the same scientific principle that the court in O’Key concluded was valid — namely, that consumption of a certain drug, in this case alcohol— causes nystagmus.” The state also contends that, because the VGN test is a component of the DRE protocol, this court necessarily approved the use of the VGN test in Sampson, in which we concluded that the 12-step DRE protocol was admissible scientific evidence.

We reject the state’s argument that the scientific principle underlying the VGN test has already been accepted by this state’s appellate courts. In O’Key, the court considered the admissibility of the HGN test, not the VGN test. The court explained that the HGN test “purports to draw its convincing force from a principle of science, namely, the asserted *541scientific proposition that there is a causal relationship between consumption of alcohol and the type of nystagmus measured by the HGN test.” 321 Or at 296 (citation omitted). The HGN test is designed to detect jerking movements when the eyes move from side to side. Id. at 294. Here, the scientific proposition underlying the evidence at issue is that there is a causal relationship between the consumption of alcohol in quantities greater than usual for the individual consumer and the type of nystagmus measured by the VGN test — that is, jerking when the eyes move up and down. That proposition may be valid (although the record here does not establish its validity); however, it is not the same proposition that underlies the HGN test, nor is it a logical consequence of the scientific presumption underlying the HGN test.

Sampson does not aid the state either. There, we considered the admissibility of the 12-step DRE protocol, one component of which is the VGN test.4 The defendant there did not challenge the individual components of the DRE *542protocol “but, rather, presented] only the question of whether the DRE protocol is, in general, admissible.” 167 Or App at 493. After considering the various factors concerning admissibility, we concluded that “the underlying proposition of the DRE protocol — that ingestion of controlled substances causes a variety of symptoms detectable by a trained officer — is sufficiently reliable to justify admission of the protocol’s results into evidence.” Id. at 511.

Here, of course, the issue is not the admissibility of the DRE protocol generally, nor did the state offer the VGN test as a component of the DRE protocol for proving one of the three things that the DRE protocol is designed to detect.=5 Rather, the state contends that the scientific validity of the protocol suggests that each of its individual components must be scientifically valid — apparently to prove the degree of alcohol consumption, as well as the consumption of controlled substances. As we have explained before, the admissibility of the complete DRE protocol as scientific evidence does not demonstrate the general admissibility of each component of the protocol. Rather, “in Sampson, we approved the 12-step DRE protocol as scientific evidence because its complete administration by a competent examiner qualified for admission as scientific evidence under O’Key.” State v. Aman, 194 Or App 463, 472, 95 P3d 244 (2004), rev dismissed as improvidently allowed, 339 Or 281 (2005). Further, in Aman, we concluded that an incompletely administered DRE protocol is not admissible as scientific evidence, although we noted that evidence of individual components of the DRE protocol were not necessarily inadmissible as nonscientific evidence of impairment.6 Id. at 473; see also Hernandez, 227 Or App at *543324 (rejecting the state’s argument that individual components of an incomplete DRE protocol were admissible as nonscientific expert evidence where the state failed to identify which elements of the protocol were nonscientific). Nothing in Sampson establishes that the VGN test, merely because it is part of the DRE protocol, is scientifically valid evidence of impairment caused by an unusually high dose of alcohol. Because the scientific validity of the VGN test has not been established, the admission of Gutierrez’s testimony on that point was error.

We next consider whether that error requires reversal. Defendant contends that the error in admitting the VGN evidence was prejudicial. The state argues that any error was harmless, given the emphasis in Gutierrez’s testimony on the HGN test and the test’s accuracy.

We must determine whether there is little likelihood that the error affected the verdict. State v. Davis, 336 Or 19, 32, 77 P3d 1111 (2003). In making that determination, we review the record, not to reweigh the evidence, but to assess the likely effect of the error on the jury’s verdict. Id. Because scientific evidence has a manifest potential to influence the jury, erroneous admission of such evidence weighs against a conclusion that an error was harmless. State v. Johnson, 225 Or App 545, 555, 202 P3d 225 (2009).

Here, the key issue at trial was whether defendant was under the influence of alcohol while he was driving. The jury was instructed that “under the influence of intoxicating liquor” means that defendant’s “physical or mental faculties were adversely affected by the use of intoxicating liquor to a noticeable or perceptible degree.” The jury could have found *544that defendant’s driving was fast but not indicative of impairment. Likewise, the jury could have found that defendant’s performance on the FSTs — failing the HGN and walk-and-turn tests but maintaining his balance on the one-leg-stand test — was inconclusive as to impairment.

The prosecution offered the VGN evidence to show that defendant had consumed more alcohol than usual. Gutierrez testified that defendant acknowledged drinking two beers — an admission that was consistent with the two beer cans found in defendant’s car. Whether defendant had consumed more alcohol than usual — a point that was qualitatively different from the other evidence, not merely cumulative of it — was a fact that may well have affected the jury’s assessment of whether defendant’s faculties were adversely affected by alcohol consumption. Certainly the evidence was sufficient, even without the VGN testimony, to allow the jury to find defendant guilty. We may not, however, reweigh the evidence, but instead must assess the likely effect of the erroneously admitted evidence on the jury’s verdict. Although there was other evidence that defendant was under the influence and Gutierrez’s testimony about the VGN test was a relatively brief part of the evidence offered at trial, the prosecutor referred to the VGN evidence in the opening statement and closing argument, thus emphasizing the evidence for the jury’s consideration. Two of the jurors apparently voted against a guilty verdict based on a record that included the erroneously admitted testimony about the VGN test. Under the circumstances, we cannot say that there is little likelihood that the erroneous admission of the VGN evidence affected the jury’s verdict.

Reversed and remanded.

dissenting.

I agree with the majority’s analysis of the admissibility of the vertical gaze nystagmus (VGN) test. I do not agree, however, with its conclusion that the error in admitting evidence of the test was not harmless. In my view, there is little likelihood that any error in admitting evidence of the VGN test results affected the verdict. I would affirm on that ground and therefore respectfully dissent.

*545First, there is abundant evidence of defendant’s intoxication. The arresting officer testified that he observed defendant driving 64 miles per hour (mph) in a 45 mph zone, quickly slam on the brakes to turn, and park on the wrong side of the street. Defendant misperceived his driving speed and the speed limit, and told the officer that he was going 45 mph in a 35 mph zone. The officer noticed a strong odor of alcohol when defendant spoke. Defendant’s eyes were bloodshot and his speech was slurred. Defendant admitted to having consumed two beers, and the officer found the open cans in the vehicle. Defendant failed the walk-and-turn field sobriety test, and the horizontal gaze nystagmus (HGN) test revealed six out of six “clues” indicating intoxication. Defendant refused to take a breath test, from which the jury could draw an inference of defendant’s own awareness of his intoxication. State v. Ohm, 224 Or App 390,197 P3d 1136 (2008). In short, the evidence of defendant’s intoxication was strong.

Second, the testimony about the inadmissible VGN test results was brief. The arresting officer’s testimony takes up more than 22 pages of transcript. Of that, his description of the VGN test takes up a little more than a dozen lines. Moreover, when asked about the basis for his decision to arrest defendant, the officer replied that he based his decision on “[defendant’s] driving, the odor of an alcoholic beverage, the fact that he told me he drank two beers, bloodshot eyes, slurred speech.” The officer made no mention of either the HGN or the VGN test results.

Third, the prosecutor’s references to the VGN test results were likewise quite brief. In both opening and closing statements, the prosecutor emphasized at some length the HGN test results and gave the VGN test results little more than passing reference. The prosecutor’s closing, for example, takes up more than seven pages of transcript, the majority of which is devoted to a description of the HGN test results, which were properly admitted. After the extended discussion of the HGN test results, the prosecutor added:

“You also have the fact that [defendant] then had what they call — remember what the officer was talking about vertical nystagmus? Well, what does that tell you. The officer says it tells you that that’s a high dose of alcohol for this particular *546individual. For him, this is more alcohol than he could handle.”

That is the extent of the state’s emphasis of the inadmissible evidence pertaining to the VGN.

The majority makes much of the fact that VGN evidence is “scientific” and thus has more potential to influence a jury. The statement is true, as far as it goes. But the fact remains that, in this case, the record shows that the scientific evidence on which the state overwhelmingly relied was not the VGN test, but the HGN test, which was admissible and is regarded as a “fairly reliable” indicator of alcohol impairment. State v. O’Key, 321 Or 285, 313, 899 P2d 663 (1995).

In my view, in light of the brevity of the references to the inadmissible VGN test results, and in light of the abundance of other, stronger evidence of intoxication and the state’s clear emphasis on that other evidence at trial, I conclude that there was little likelihood that admission of the evidence affected the verdict. In other words, any error in admitting the VGN evidence was harmless, and the judgment of conviction should be affirmed in that basis.