WERDEGAR, J.

Law Offices of Chad R. Maddox and Chad R. Maddox, Santa Ana, for Plaintiff and appellant.

Kamala D. Harris, Attorney General, Alicia M.B. Fowler, Kenneth C. Jones and Kevin K. Hosn, Deputy Attorneys General, for Defendant and Respondent.

Opinion WERDEGAR, J.

A California Highway Patrol officer stopped a car driven by plaintiff Ashley Jourdan Coffey after he observed her driving erratically. Four subsequent chemical tests revealed her blood-alcohol concentration (BAC) ranged from 0.08 to 0.096 percent. The officer then confiscated plaintiff's driver's license and served her with a notice that her license would be suspended pursuant to Vehicle Code section 13382. In an administrative hearing to review the suspension, plaintiff's expert witness opined that her BAC was rising at the time of the chemical tests, suggesting her BAC was below the 0.08 percent threshold at the time plaintiff was driving. Both the Department of Motor Vehicles (DMV) hearing officer and the trial court discounted the expert's testimony in part by relying on arrest reports, which described the physical manifestations of plaintiff's intoxication, such as her general appearance, erratic driving, poor performance on field sobriety tests, and the strong odor of alcohol she projected.

We decide in this case whether the trial court erred by considering, in addition to the results of breath and blood tests, other circumstantial evidence of intoxication to conclude by a preponderance of the evidence that plaintiff drove with a BAC at or above 0.08 percent. As we explain, we conclude the trial court did not err. Facts

On November 13, 2011, at 1:32 in the morning, Sergeant Martin of the California Highway Patrol was traveling southbound on State Route 55 in Orange County when he saw a car traveling 60 miles per hour, swerving erratically from side to side. From the number four, or right-hand, lane, the car swerved one foot to the left into the number three lane before correcting. It then twice swerved one to two feet to the right, onto the highway's shoulder. Sergeant Martin positioned his patrol vehicle behind the car and activated his emergency lights, whereupon the car slowly moved left across the highway into the number one lane. When Martin activated his siren, the car veered even further left, into the carpool lane. Only when Sergeant Martin used his public address system and directed the driver to pull to the right did the car eventually comply.

Upon making contact with the driver of the vehicle, plaintiff Ashley Coffey, Sergeant Martin noticed her eyes were red and a strong odor of alcohol emanated from her car. Officer White arrived to provide backup and confirmed these observations. To both officers she denied having consumed any alcoholic beverages, offering the rather implausible story that she had just turned 21 years old, had been in a bar, but had not herself consumed any alcoholic beverages. The officers then had plaintiff perform various field sobriety tests. Plaintiff failed the horizontal gaze nystagmus test, “display[ing] a lack of smooth pursuit in both eyes.” Asked to complete the “walk-and-turn test,” in which she was asked to walk heel to toe for nine steps, turn counterclockwise, and then walk back heel to toe, “she missed heel to toe on five of those nine steps by 2–4 inches on each step. When she reached step nine, ... she turned clockwise instead of counter clockwise as instructed.... [She] used both feet to make the turn instead of keeping her front foot in place” and on the return similarly “missed heel to toe three of the steps by 2–4 inches.” Plaintiff did somewhat better on the “one-leg stand” test, standing on one foot and counting out loud beginning with 1,001; the test was terminated when she reached 1,022 after 30 seconds. On the Romberg test, “[s]he swayed slightly in all directions from center by 1–2 inches” and “estimated 30 seconds at 37 actual seconds.” Plaintiff refused to perform a preliminary alcohol screening, or PAS. Based on the officers' observations of plaintiff and her poor performance on the field sobriety tests, they placed her under arrest at 2:00 a.m. Officer White advised her of the implied consent law and she chose to perform a breath test, although she failed several times to provide an adequate breath sample and had to be retested multiple times. At 2:28 a.m., 56 minutes after she was stopped by Sergeant Martin, her breath test registered a BAC of 0.08 percent. Three minutes later, at 2:31 a.m., her second breath test measured a BAC of 0.09 percent. Police then transported plaintiff to the Orange County jail, where she elected to have her blood drawn. The blood draw occurred at 2:55 a.m., one hour 23 minutes after plaintiff was pulled over by Sergeant Martin. The first test of the blood sample showed a BAC of 0.095 percent; the second measured 0.096 percent. As a result of these chemical test results, Officer White confiscated plaintiff's driver's license and issued her an “administrative per se suspension/revocation order” and temporary driver's license. (§ 13382 ; see Lake v. Reed (1997) 16 Cal.4th 448, 454–455, 65 Cal.Rptr.2d 860, 940 P.2d 311 (Lake ).)

Plaintiff, charged with drunk driving (§ 23152 ), was allowed to plead to a “wet reckless” (§§ 23103 [misdemeanor reckless driving], 23103.5 [prosecutorial statement that alcohol was involved]; see People v. Claire (1991) 229 Cal.App.3d 647, 650 & fn. 2, 280 Cal.Rptr. 269 ), but requested a hearing before the DMV to challenge her license suspension (§ 13558).

At the ensuing administrative hearing, the DMV hearing officer had before her the “Officer's Sworn Statement” form, Officer White's arrest report and the supplemental reports of Sergeant Martin and Officer White. In addition to considering these documents, the hearing officer heard telephonic testimony from Jay Williams, a forensic toxicologist with extensive experience, who testified for plaintiff. Williams noted the result of plaintiff's first breath test was 0.08 percent, the second test three minutes later was 0.09 percent, and her blood sample taken about 20 minutes later tested at 0.095 and 0.096 percent. According to Williams, these results suggested the alcohol level in plaintiff's body was rising at the time of the tests and, given the totality of the circumstances, were consistent with plaintiff's BAC being below 0.08 percent at 1:32 a.m. when she was first pulled over by Sergeant Martin.The DMV hearing officer rejected Williams's testimony regarding a rising BAC, explaining in her ruling that the witness's two conclusions—first, that plaintiff's BAC was rising at the time she was pulled over, and second, that it may accordingly be deduced that her BAC was below 0.08 percent when she was driving—were not supported by reliable evidence, were “too speculative to support the contention,” and were “based on a subjective interpretation of the evidence.” In addition, Williams's conclusions were “insufficient to rebut the official duty presumption,” which in this context we take to be a reference to the presumption the chemical test results were valid. The hearing officer reached this conclusion, she explained, because Williams had not himself examined the breath-analyzing device used in the case, offered no opinion whether it was in working order, conducted no scientific tests himself, and “did not show that any other experts in the scientific community had reached similar conclusions.” Finally, the hearing officer specifically found credible Officer White's recordation of the “events as they occur[red],” which we assume meant White's observations of plaintiff's appearance and her performance on field sobriety tests. Accordingly, the hearing officer concluded plaintiff's license suspension was proper because the state had shown by a preponderance of the evidence that she had been driving with a BAC of 0.08 percent or higher.

Plaintiff filed a petition for a writ of mandate with the trial court to challenge the DMV hearing officer's decision. After first noting that section 23152, subdivision (b) makes it a rebuttable presumption that a person was driving with a BAC of 0.08 percent or higher if so tested at that level or higher within three hours of driving (see discussion, post ), the trial court denied the writ, explaining that “[e]ven assuming that petitioner Coffey rebutted [this] presumption ..., there was sufficient evidence based on the blood-alcohol tests and the other circumstantial evidence based on the assessment, observations and tests by the arresting officers at the scene to support the DMV hearing officer's decision under the weight of the evidence.” (Italics added.)

The Court of Appeal affirmed. In determining whether the trial court's decision was supported by substantial evidence, the appellate court opined that “[t]he issue boils down to whether non-chemical-test circumstantial evidence can prove that Coffey's BAC at the time of driving was consistent with her BAC at the time of her chemical tests.” Relying on this court's opinion in Burg v. Municipal Court (1983) 35 Cal.3d 257, 198 Cal.Rptr. 145, 673 P.2d 732, the appellate court held in the affirmative. We granted review.

Discussion

A. Background

The DMV suspended plaintiff's license to drive pursuant to the “administrative per se” law, “under which a person arrested for driving under the influence of alcohol, and who is determined to have a prohibited amount of alcohol in his or her blood, must have driving privileges suspended prior to an actual conviction for a criminal offense.” (Lake, supra, 16 Cal.4th at p. 454, 65 Cal.Rptr.2d 860, 940 P.2d 311.) As we explained in that case, “ ‘[t]he express legislative purposes of the administrative suspension procedure are: (1) to provide safety to persons using the highways by quickly suspending the driving privilege of persons who drive with excessive blood-alcohol levels; (2) to guard against erroneous deprivation by providing a prompt administrative review of the suspension; and (3) to place no restriction on the ability of a prosecutor to pursue related criminal actions.’ ” (Ibid. ) “[T]he administrative per se laws were deemed necessary due to the time lag that often occurs between an arrest and a conviction for driving while intoxicated or with a prohibited BAC. During this interim period, arrestees who could eventually be convicted of an intoxication-related driving offense were permitted to continue driving and, possibly, endangering the public thereby. Moreover, without administrative per se laws, persons with extremely high BAC levels at the time of arrest could escape license suspension or revocation by plea bargaining to lesser crimes or entering pretrial diversion. Thus, by providing for an administrative license suspension prior to the criminal proceeding, the law affords the public added protection.” (Id. at pp. 454–455, 65 Cal.Rptr.2d 860, 940 P.2d 311.)

Pursuant to the administrative per se law, “[a]fter either the arresting officer or the DMV serves a person with a ‘notice of an order of suspension or revocation of the person's [driver's license],’ the DMV automatically reviews the merits of the suspension or revocation. [Citation.] The standard of review is preponderance of the evidence [citation], and the department bears the burden of proof.” (Lake, supra, 16 Cal.4th at p. 455, 65 Cal.Rptr.2d 860, 940 P.2d 311.) A driver served with such a suspension notice is entitled to a hearing on request (§ 13558, subd. (a)), at which the only issues to be decided for drivers such as plaintiff are whether the arresting officer had reasonable cause to believe she was driving, whether she was arrested for an enumerated offense, and whether she was driving with 0.08 percent BAC or higher. (§ 13557, subd. (b)(3)(A), (B), & (C)(i).) If the DMV hearing officer finds these three statutory prerequisites proved by a preponderance of the evidence, the accused's driver's license will be suspended for four months if the driver has had a clean driving record (§ 13353.3, subd. (b)(1)). Higher penalties apply to those with prior drunk driving convictions. (§ 13353.3, subd. (b)(2).)

B. The Rebuttable Presumption in Section 23152

We first address whether the presumption created by section 23152, subdivision (b) controls this case. That provision states in part: “In any prosecution under this subdivision, it is a rebuttable presumption that the person had 0.08 percent or more, by weight, of alcohol in his or her blood at the time of driving the vehicle if the person had 0.08 percent or more, by weight, of alcohol in his or her blood at the time of the performance of a chemical test within three hours after the driving.” Although the statutory language speaks in terms of a “prosecution,” several Courts of Appeal have held this presumption is not limited to criminal prosecutions but also applies in administrative license suspension proceedings. (Corrigan v. Zolin (1996) 47 Cal.App.4th 230, 236, 54 Cal.Rptr.2d 634, citing Jackson v. Department of Motor Vehicles (1994) 22 Cal.App.4th 730, 740, fn. 9, 27 Cal.Rptr.2d 712, and Bell v. Department of Motor Vehicles (1992) 11 Cal.App.4th 304, 310–313, 13 Cal.Rptr.2d 830.)

Extending the reach of Vehicle Code section 23152, subdivision (b)'s evidentiary presumption to administrative per se proceedings would be consistent with the legislative history of that provision. The need for the presumption “arose from the absence in ‘[e]xisting law’ of any ‘provision for the delay involved between the time a person is arrested for [driving under the influence] and when the chemical test for BAC is actually administered,’ of any ‘means to determine a person's BAC at the time the person is actually driving the car,’ or of any ‘mention of time parameters for the administering of chemical tests and for their admission as [admissible] evidence into a court of law.’ (Health & Welf. Agency, Dept. of Alcohol & Drug Programs, Enrolled Bill Rep. for Sen. Bill No. 745 (1981–1982 Reg. Sess.) Sept. 1982, original italics.) Thus, in enacting the presumption, the Legislature intended (1) to ‘diminish the arguments that ha[d] arisen when extrapolating the [BAC] at the time of the test back to the time of the driving’ (Bus. & Transportation Agency, DMV, Enrolled Bill Rep. for Sen. Bill No. 745 (1981–1982 Reg. Sess.) Sept. 1982), (2) ‘to close a potential loophole in the current law, whereby a person ... could claim that he or she had consumed ... alcohol which had not yet been absorbed into the bloodstream while the person was operating the vehicle, but which later raised the blood alcohol level’ (Governor's Office, Dept. of Legal Affairs, Enrolled Bill Rep. for Sen. Bill No. 745 (1981–1982 Reg. Sess.) Sept. 1982), and (3) ‘to recognize that alcohol concentrations dissipate over time, so that a person whose blood alcohol levels exceed the permissible concentrations at the time of the test, was likely to have had unlawfully high blood alcohol levels when driving....’ ” (Bell v. Department of Motor Vehicles, supra, 11 Cal.App.4th at p. 311, 13 Cal.Rptr.2d 830.) These three statements of legislative intent would arguably apply to administrative per se proceedings as well.

Consistent with the previously cited Court of Appeal cases, both parties assume section 23152's presumption applies in administrative per se hearings. We need not resolve that question, however, because even were the presumption applicable, it was rebutted in this case. An explanation of how the presumption operates can be found in the Evidence Code. Section 601 of that code provides: “A presumption is either conclusive or rebuttable. Every rebuttable presumption is either (a) a presumption affecting the burden of producing evidence or (b) a presumption affecting the burden of proof.” Vehicle Code section 23152, subdivision (b), by its terms, creates a rebuttable presumption, and we agree with the parties that it establishes a presumption affecting the burden of producing evidence, not the burden of proof. A statute transferring the burden of proof to a driver facing a criminal charge of drunk driving would raise serious constitutional questions (see Ulster County Court v. Allen (1979) 442 U.S. 140, 157, 99 S.Ct. 2213, 60 L.Ed.2d 777 [a permissive presumption is constitutional if, among other things, it “does not shift the burden of proof”]; see People v. Gamache (2010) 48 Cal.4th 347, 376, 106 Cal.Rptr.3d 771, 227 P.3d 342 [rejecting claim that jury instruction “impermissibly alters the burden of proof,” explaining the “instruction does not establish an unconstitutional mandatory presumption in favor of guilt [citation] or otherwise shift or lower the prosecution's burden of establishing guilt beyond a reasonable doubt”] ), and although those concerns are inapplicable here because it is an administrative per se proceeding, neither party argues for a different construction. (See People v. Dubon (2001) 90 Cal.App.4th 944, 953, 108 Cal.Rptr.2d 914 [“When a presumption is established to facilitate the determination of the particular action in which the presumption is applied, rather than to implement public policy, it is a presumption affecting the burden of producing evidence.”].)

A rebuttable presumption requires the trier of fact, given a showing of the preliminary fact (here, that a chemical test result showed plaintiff had a BAC of 0.08 percent or more within three hours of driving), to assume the existence of the presumed fact (here, that plaintiff had been driving with a prohibited BAC) “unless and until evidence is introduced which would support a finding of its nonexistence, in which case the trier of fact shall determine the existence or nonexistence of the presumed fact from the evidence and without regard to the presumption.” (Evid.Code, § 604.) In other words, if evidence sufficient to negate the presumed fact is presented, the “presumption disappears” (Craig v. Brown & Root, Inc. (2000) 84 Cal.App.4th 416, 421, 100 Cal.Rptr.2d 818 ) and “has no further effect” (In re Heather B. (1992) 9 Cal.App.4th 535, 561, 11 Cal.Rptr.2d 891 ), although “inferences may nevertheless be drawn from the same circumstances that gave rise to the presumption in the first place” (Craig v. Brown & Root, Inc., supra, at p. 421, 100 Cal.Rptr.2d 818 ; see Evid.Code, § 604 [“Nothing in this section shall be construed to prevent the drawing of any inference that may be appropriate”] ).

Assuming the results of her breath and blood tests gave rise to a presumption she was driving with a BAC of 0.08 percent or more, plaintiff argues the testimony of her expert witness, Jay Williams, supplied the necessary contrary evidence sufficient to rebut the presumption. By contrast, the DMV argues Williams's testimony was insufficient, arguing that evidence necessary to rebut the presumption must be substantial, i.e., “ ‘reasonable in nature, credible, and of solid value; it must actually be “substantial” proof of the essentials which the law requires in a particular case.’ ” Asserting the DMV hearing officer's refusal to credit Williams's views shows his testimony was insubstantial, the DMV argues his testimony was accordingly insufficient to rebut the statutory presumption.

The DMV misapprehends Evidence Code section 604. That section provides that evidence is sufficient to rebut a presumption if it “would support a finding of [the] nonexistence of” the presumed fact. (Italics added.) The most reasonable meaning of this phrase is that if the predicate facts are found, Vehicle Code section 23152's presumption will apply unless the driver presents evidence which, if believed, “would support a finding of [the] nonexistence of” the presumed fact. This plain meaning of the statutory language is supported by the Assembly Committee on Judiciary's comment on section 604, which states: “Such a presumption is merely a preliminary assumption in the absence of contrary evidence, i.e., evidence sufficient to sustain a finding of the nonexistence of the presumed fact. If contrary evidence is introduced, the trier of fact must weigh the inferences arising from the facts that gave rise to the presumption against the contrary evidence and resolve the conflict. For example, if a party proves that a letter was mailed, the trier of fact is required to find that the letter was received in the absence of any believable contrary evidence. However, if the adverse party denies receipt, the presumption is gone from the case. The trier of fact must then weigh the denial of receipt against the inference of receipt arising from proof of mailing and decide whether or not the letter was received.” (Assem. Com. on Judiciary com., reprinted at 29B pt. 2, West's Ann. Evid.Code (1995 ed.) foll. § 604, p. 59.)

Viewing the presumption in section 23152, subdivision (b) in this way, and assuming without deciding that it applies in administrative per se proceedings, we find Williams's testimony was sufficient to rebut the presumption that plaintiff's BAC was at least 0.08 percent at the time she was driving. Williams was qualified as an expert in the field and his testimony was clear and direct. If believed, his evidence would have justified a conclusion that plaintiff's BAC was rising at the time of her chemical tests and was thus quite possibly below the 0.08 percent threshold at the time she had been driving. As a consequence, the DMV was required to prove plaintiff's BAC at the time she was driving without resort to the statutory presumption.

C. The Circumstantial Evidence of Plaintiff's Intoxication Was Relevant and Thus Admissible

Having found the statutory presumption in section 23152, subdivision (b), even if applicable, does not control this case, we turn to the main issue presented here: Did the DMV hearing officer properly admit and consider non-chemical-test evidence to reach her conclusion that plaintiff was driving with at least a 0.08 percent BAC?

The crime of drunk driving is set forth in section 23152 and can be established in two ways: Subdivision (a) states that “[i]t is unlawful for a person who is under the influence of any alcoholic beverage to drive a vehicle.” To prove a violation under subdivision (a), the People must present evidence the driver's alcohol consumption impaired his or her ability to drive. (People v. McNeal (2009) 46 Cal.4th 1183, 1197, 96 Cal.Rptr.3d 261, 210 P.3d 420.) Subdivision (b) offers an alternative method of proof; it states that “[i]t is unlawful for a person who has 0.08 percent or more, by weight, of alcohol in his or her blood to drive a vehicle.” To prove a criminal violation under subdivision (b), the People need not prove the accused's driving ability was impaired or diminished, but only that the driver's BAC reached or exceeded the prohibited level at the time the accused was driving.The administrative per se law's license suspension provision, although not a criminal matter, is linked to the second prong of section 23152. Thus, license suspension under the administrative per se law does not require proof of a person's impairment to safely operate a motor vehicle due to alcohol consumption, but only that the person's BAC level was 0.08 percent or more. Section 13382, subdivision (a) provides in pertinent part: “If the chemical test results for a person who has been arrested for a violation of Section 23152 or 23153 show that the person has 0.08 percent or more, by weight, of alcohol in the person's blood, ... the peace officer, acting on behalf of the [DMV], shall serve a notice of order of suspension or revocation of the person's privilege to operate a motor vehicle personally on the arrested person.” (Italics added.) In this case, the DMV presented the results of four chemical tests showing plaintiff's BAC was at or above 0.08 percent. Given the presentation of these facially qualifying chemical test results, did the hearing officer abuse her discretion in concluding that non-chemical-test evidence was relevant and thus admissible to bolster or corroborate the chemical test results?

Plaintiff argues reliance on the non-chemical-test circumstantial evidence was improper because such evidence cannot by itself establish whether her BAC was 0.08 percent, higher than 0.08 percent, or lower than that level. She cites to scientific evidence showing that physical manifestations of alcohol intoxication can occur at levels much lower than a BAC of 0.08 percent and that observable physical symptoms correlate poorly to actual BAC levels. She also argues that one study has shown that poor performance on field sobriety tests has a low correlation to whether a driver's BAC is over 0.08 percent. (Hlastala, Polissar & Oberman, Statistical Evaluation of Standardized Filed Sobriety Tests (May 2005) 50 J. Forensic Science, No. 3, p. 662.)

Unmentioned is that some studies have reached a contrary conclusion. For example, the National Highway Traffic Safety Administration (NHTSA) released the results of a study in 1998 that evaluated the accuracy of the standardized field sobriety test (SFST) battery at BACs below 0.10 percent. (Stuster and Burns, Final Rep. to NHTSA, Validation of the Standardized Field Sobriety Test Battery at BACs Below 0.10 Percent (1998).) The NHTSA's study found that the battery of SFSTs, which includes three of the tests administered to plaintiff (the horizontal gaze nystagmus test, the “walk-and-turn test,” and the “one-leg stand test”), when administered by a trained officer, are “extremely accurate in discriminating between BACs above and below 0.08 percent. ” (Id., at p. v, italics added.) The NHTSA's report expressly dispelled a common misapprehension “that field sobriety tests are designed to measure driving impairment.” (Id., at p. 28.) According to the NHTSA, the SFST battery is instead designed specifically to “provide statistically valid and reliable indications of a driver's BAC, rather than indications of driving impairment.”

We are not here attempting to resolve the scientific debate over the use of SFSTs to predict BAC. As plaintiff acknowledges, the test for admissibility of evidence is not a strict one: As a general matter, evidence may be admitted if relevant (Evid.Code, § 350 ), and “ ‘[r]elevant evidence’ means evidence ... having any tendency in reason to prove or disprove any disputed fact that is of consequence to the determination of the action” (id., § 210 ). “ ‘ “The test of relevance is whether the evidence tends, ‘logically, naturally, and by reasonable inference’ to establish material facts....” ' ” (People v. Wilson (2006) 38 Cal.4th 1237, 1245, 45 Cal.Rptr.3d 73, 136 P.3d 864.) “The trial court has broad discretion to determine the relevance of evidence [citation], and we will not disturb the court's exercise of that discretion unless it acted in an arbitrary, capricious or patently absurd manner.” (People v. Jones (2013) 57 Cal.4th 899, 947, 161 Cal.Rptr.3d 295, 306 P.3d 1136.)

Past cases applying this standard have found circumstantial evidence of intoxication may be admissible when later-administered chemical tests show a BAC exceeding the legal limit. In Burg v. Municipal Court, supra, 35 Cal.3d 257, 198 Cal.Rptr. 145, 673 P.2d 732 (Burg ), decided at a time when the BAC threshold for drunk driving was 0.10 percent, we explained that the crime of drunk driving set forth in section 23152, subdivision (b) “prohibits driving a vehicle with a blood-alcohol level of 0.10 percent or higher; it does not prohibit driving a vehicle when a subsequent test shows a level of 0.10 percent or more. Circumstantial evidence will generally be necessary to establish the requisite blood-alcohol level called for by the statute. A test for the proportion of alcohol in the blood will, obviously, be the usual type of circumstantial evidence, but of course the test is not conclusive: the defendant remains free to challenge the accuracy of the test result, the manner in which it was administered, and by whom. (People v. Lewis (1983) 148 Cal.App.3d 614, 620, 196 Cal.Rptr. 161 ; accord, Fuenning v. Superior Court (Ariz.1983) 680 P.2d 121, 127... [rejecting argument that analogous statute represents ‘substitution of a machine test result for a jury verdict’ because defendant is given an opportunity to challenge accuracy of test result, and state must prove beyond a reasonable doubt that defendant's blood-alcohol level was 0.10 percent at the time he was driving]; Cooley v. Municipality [of ] Anchorage (Alaska App.1982) 649 P.2d 251, 254–255.) Of course, both parties may also adduce other circumstantial evidence tending to establish that the defendant did or did not have a 0.10 percent blood-alcohol level while driving. (See, e.g., Fuenning, supra, at p. 130.)” (Burg, supra, at p. 266, fn. 10, 198 Cal.Rptr. 145, 673 P.2d 732, italics added.)

Plaintiff would distinguish Burg, supra, 35 Cal.3d 257, 198 Cal.Rptr. 145, 673 P.2d 732, as a case involving a criminal drunk driving prosecution, not an administrative per se matter, but if non-chemical-test circumstantial evidence of intoxication may be admissible in a criminal case where the People's burden of proof is beyond a reasonable doubt, we see no reason why such evidence would be categorically inadmissible in an administrative proceeding where the People's burden of proof is only a preponderance of the evidence. Decisions in the Courts of Appeal support this conclusion. Thus, in McKinney v. Department of Motor Vehicles (1992) 5 Cal.App.4th 519, 7 Cal.Rptr.2d 18, the Court of Appeal considered whether sufficient evidence supported the suspension of a driver's license under the administrative per se law. Citing Burg, the McKinney court noted that administration of a chemical test (blood, breath or urine) “is not the only means of establishing that a driver's [BAC] was .08 or more. [¶] ... [B]oth parties are free to introduce circumstantial evidence bearing on whether the driver's [BAC] exceeded the permissible level. [Citation.] ‘Evidence regarding the manner in which a defendant drove, performed field sobriety tests, and behaved is admissible and relevant as tending to establish that he did or did not have a 0.10 [now 0.08] [BAC] while driving.’ ” (McKinney, at p. 526, fn. 6, 7 Cal.Rptr.2d 18 ; see Jackson v. Department of Motor Vehicles, supra, 22 Cal.App.4th at p. 741, 27 Cal.Rptr.2d 712 [“circumstantial evidence other than chemical test results may properly be admitted to establish a driver had the proscribed level of blood-alcohol at the time of the offense”]; People v. Randolph (1989) 213 Cal.App.3d Supp. 1, 7, 262 Cal.Rptr. 378 [same].)Plaintiff argues Burg, supra, 35 Cal.3d 257, 198 Cal.Rptr. 145, 673 P.2d 732, is unpersuasive because, apart from the single sentence referencing non-chemical-test circumstantial evidence, the opinion does not elaborate on the use of such evidence to prove a driver's BAC. But Burg's citation to Fuenning v. Superior Court, supra, 139 Ariz. 590, 680 P.2d 121 (Fuenning ), following that sentence is significant. In that case, a person charged with drunk driving argued that circumstantial evidence “regarding the manner in which he was driving, [and] the manner in which he performed the field sobriety tests,” while relevant to the issue of whether he drove under the influence of alcohol, was “irrelevant to the question of whether he ... [was] driving with a .10% or greater BAC.” (Fuenning, supra, at p. 599, 680 P.2d 121.) The Supreme Court of Arizona rejected the argument, explaining the evidence was relevant and thus admissible: “We agree with defendant that the only ultimate issue is whether defendant had a BAC of .10% or greater. In each case in which a violation of [the .10 percent BAC law] is charged, the state will present evidence of the test and the issue will be whether the test results were an accurate measurement of the defendant's BAC at the time of arrest. Typically, defendants will attack the margin of error, the conversion rate, the calibration of the test instrument, the technique used by the operator, the absorption and detoxification factors, etc. Evidence of defendant's conduct and behavior—good or bad—will be relevant to the jury's determination of whether the test results are an accurate measurement of alcohol concentration at the time of the conduct charged. For instance, the test in the case at bench was given several hours after the arrest and showed a .11% BAC. Defendant attacked the results, presenting evidence regarding margin of error, time lapse and other factors. Such evidence might raise considerable doubt whether the test result of .11% indicated .10% or greater BAC at the time defendant was arrested. Evidence that at that time the person charged smelled strongly of alcohol, was unable to stand without help, suffered from nausea, dizziness or any of the other ‘symptoms' of intoxication would justify an inference that a test administered some time after arrest probably produced lower readings than that which would have been produced had the test been administered at the moment of arrest. The converse is also true. Evidence that at the time of arrest defendant was in perfect control, displayed none of the symptoms of intoxication and had not driven in an erratic manner, is relevant to show that a reading of .11% from a test given some time later does not prove beyond a reasonable doubt that the defendant was driving with a .10% or greater BAC at the time of his arrest. Such evidence has been held admissible. [Citations.] Again, evidence is admissible when it is relevant.” (Fuenning, supra, at p. 599, 680 P.2d 121, italics added.)

Plaintiff argues Fuenning is neither controlling nor even on point because it did not address the issue before us in the instant case. Although Fuenning is an Arizona case and thus admittedly not controlling here (see Farmers Ins. Group v. County of Santa Clara (1995) 11 Cal.4th 992, 1018, 47 Cal.Rptr.2d 478, 906 P.2d 440 ), it persuasively explains why circumstantial evidence of intoxication, while not dispositive, may be relevant and thus admissible to help interpret the results of a chemical test for a driver's BAC. Moreover, because the Legislature has prohibited driving with a BAC of 0.08 percent or higher, that being the threshold at which a person cannot safely operate a motor vehicle due to alcohol consumption, circumstantial evidence that plaintiff was weaving erratically all over the roadway, smelled strongly of alcohol, and failed a battery of field sobriety tests may bolster chemical test results showing that she had attained or exceeded that BAC level.The administrative per se scheme in section 13382 is triggered by a chemical test result showing a BAC of 0.08 percent or more, and we do not here confront a case in which the DMV failed to present such test results; indeed, the DMV produced the results of four such tests. In other words, neither the DMV hearing officer nor the trial court considered circumstantial evidence of intoxication in the absence of any chemical test results. Although we find non-chemical-test evidence of plaintiff's intoxication may be relevant and thus admissible in the typical administrative per se proceeding triggered by a BAC of 0.08 percent or more to help connect those test results to a driver's BAC at the time she was driving (subject, of course, to the hearing officer's routine exercise of discretion), we would in any event affirm the hearing officer's decision in this case because the non-chemical-test evidence was admissible to rebut plaintiff's proffered defense that her BAC was low at the time she was driving and only later rose to exceed the legal limit. Her expert, Jay Williams, testified that, in his opinion, the four chemical test results indicated plaintiff's BAC was rising at the time of the tests; from that supposition, he further deduced that plaintiff's BAC was below the 0.08 percent threshold at the time she was driving. Even assuming that non-chemical-test evidence cannot by itself prove a driver's exact BAC at the moment the driver is stopped by a police officer, in this case plaintiff's erratic driving, outward appearance of substantial intoxication, implausible story of having just turned 21 years old, and was coming from a bar without having imbibed alcohol at all, and her failure on multiple field sobriety tests, together tend to rebut Williams's theory of a rising BAC and corroborate the BAC test results. For example, plaintiff's extremely erratic driving, observed by Sergeant Martin from before the moment he first made contact with her, suggests she was quite intoxicated from that early point in the timeline and tends to refute the expert's speculation that her BAC was low at the time she was driving, but rose to 0.08 percent and above only after she was stopped. Whether the circumstantial evidence of plaintiff's intoxication was admitted to bolster the results of the chemical tests or merely to rebut plaintiff's defense of a rising BAC, the hearing officer did not act in “an arbitrary, capricious or patently absurd manner” and thus did not abuse her broad discretion. (People v. Jones, supra, 57 Cal.4th at p. 947, 161 Cal.Rptr.3d 295, 306 P.3d 1136.) Brenner v. Department of Motor Vehicles (2010) 189 Cal.App.4th 365, 116 Cal.Rptr.3d 716, cited by plaintiff, does not warrant a contrary result. In that case, an expert examined the maintenance records for the device used to measure the driver's breath and opined that it produced slightly elevated results. As the driver's test results showed a BAC of 0.08 percent exactly, the expert opined the driver's actual BAC was slightly below the legal limit. Because the trial court in Brenner had granted the driver's petition for a writ, the Court of Appeal was required to uphold that decision if supported by substantial evidence. By contrast, the instant case comes to us in the opposite procedural posture: the trial court here denied relief to plaintiff and we are bound to uphold that decision if supported by substantial evidence. To the extent the appellate court in Brenner asserted that “the impressions of the officer may have a bearing on plaintiff's level of impairment, [but] they have no bearing on the precise level of his BAC” (id. at p. 373, 116 Cal.Rptr.3d 716 ), the statement must be read in the context of the case: Because the chemical tests in Brenner were done on a miscalibrated device used to measure the BAC in a breath sample, no accurate chemical test results were presented. No such problem exists in the instant case; the DMV presented the results of four chemical tests and, although plaintiff's expert would have interpreted those results to conclude her BAC was rising, he did not testify they inaccurately measured her BAC at the time the tests were run. To the extent the court's statement in Brenner v. Department of Motor Vehicles, supra, 189 Cal.App.4th 365, 116 Cal.Rptr.3d 716, may be taken to preclude consideration of non-chemical-test evidence to bolster or corroborate chemical test results, we disapprove it.Having concluded the DMV hearing officer properly admitted the circumstantial, nontest evidence of plaintiff's intoxication, we also conclude substantial evidence supported the trial court's decision to deny writ relief, thereby sustaining the DMV hearing officer's decision to suspend plaintiff's license to drive. A driver whose license has been suspended under the administrative per se law can seek review of the DMV's decision by seeking a writ of mandate in the trial court. “In ruling on an application for a writ of mandate following an order of suspension or revocation, a trial court is required to determine, based on its independent judgment, ‘ “whether the weight of the evidence supported the administrative decision.” ’ ” (Lake, supra, 16 Cal.4th at p. 456, 65 Cal.Rptr.2d 860, 940 P.2d 311.) Following the trial court's denial of the writ, the scope of our review on appeal is limited: “[W]e ‘need only review the record to determine whether the trial court's findings are supported by substantial evidence.’ [Citation.] ‘ “We must resolve all evidentiary conflicts and draw all legitimate and reasonable inferences in favor of the trial court's decision. [Citations.] Where the evidence supports more than one inference, we may not substitute our deductions for the trial court's. [Citation.] We may overturn the trial court's factual findings only if the evidence before the trial court is insufficient as a matter of law to sustain those findings.” ’ ” (Lake, supra, at p. 457, 65 Cal.Rptr.2d 860, 940 P.2d 311.)

Applying this standard, we have no trouble concluding substantial evidence supported the trial court's ruling, for it acted well within its discretion in rejecting the expert's testimony and placing primary emphasis on the four chemical tests that showed plaintiff's BAC met or exceeded the statutory threshold at the time plaintiff was driving. Neither the DMV hearing officer nor the trial court was required to accept Williams's testimony at face value. (People v. Prince (1988) 203 Cal.App.3d 848, 858, 250 Cal.Rptr. 154 [trier of fact “is permitted to consider the credibility of the expert witnesses, the reasons given for their opinions, and the facts and other matters upon which their opinions are based”]; cf. CALCRIM No. 332 [instructing the jury it is not required to accept a proffered expert opinion as “true or correct”].) Both reasonably relied on circumstantial evidence of plaintiff's intoxication—her general appearance, the odor of alcohol about her person, her erratic driving, and her failed field sobriety tests—to support the accuracy of the chemical test results and to reject Williams's view of the evidence, including his opinion of a rising BAC, as unduly speculative.

Conclusion

The judgment of the Court of Appeal is affirmed.

We Concur: CANTIL–SAKAUYE, C.J., CHIN, CORRIGAN, CUÉLLAR, and KRUGER, JJ.

Concurring Opinion by LIU, J.

I write separately to clarify the limited way in which evidence of behavioral impairment was relevant in this case to determining whether the driver's blood-alcohol concentration (BAC) was 0.08 percent or higher at the time of arrest.

“ ‘Relevant evidence’ means evidence ... having any tendency in reason to prove or disprove any disputed fact that is of consequence to the determination of the action.” (Evid.Code, § 210.) Evidence that a driver's behavior is not impaired tends to prove that her BAC was below 0.08 percent because we can rationally surmise, given the Legislature's choice of the 0.08 percent BAC threshold, that a BAC of 0.08 percent is associated with an unsafe degree of impairment. But the converse is not true. Absent foundational evidence, a driver's impairment does not generally tend to show that her BAC was 0.08 percent or higher because we have no way of correlating a specific type or degree of impairment with a particular BAC in a close case. The fact that 0.08 percent BAC is a threshold associated with an unsafe degree of impairment does not imply that no impairment occurs below that threshold. Without evidence that correlates particular behavioral impairments with particular BAC levels, signs of impairment do not generally establish that it is more likely a driver's BAC is 0.08 percent as opposed to 0.06 or 0.07 percent. In other words, signs of impairment do not generally have a tendency in reason to prove a BAC of 0.08 percent or greater.I say “generally” because sometimes evidence of impairment can be so extreme—for example, if the driver passed out at the scene of arrest—that the trier of fact may infer on the basis of common experience that the driver likely was driving with a BAC of 0.08 percent or higher. But there is no evidence of such extreme impairment in this case.

Today's opinion properly refrains from suggesting that there is a correlation between evidence of impairment, including field sobriety test results, and a BAC of 0.08 percent or greater. The court cites a study commissioned by the National Highway Traffic Safety Administration (NHTSA) reporting that at least some field sobriety tests, properly administered, may accurately measure a BAC of 0.08 percent or greater. (Stuster and Burns, Final Rep. to NHTSA, Validation of the Standardized Field Sobriety Test Battery at BACs Below 0.10 Percent (1998) p. v.) But today's opinion is careful to express no view about the merits of these claims (maj. opn., ante, 185 Cal.Rptr.3d at p. 549, 345 P.3d at p. 905), and other courts have questioned them. In United States v. Horn (D.Md.2002) 185 F.Supp.2d 530, for example, the court undertook an extensive review of expert testimony and academic literature critiquing an earlier NHTSA study and concluded that “presently there is insufficient data to support these claims of accuracy” of field sobriety tests in predicting BAC. (Id. at p. 556.)

This lack of correlation between evidence of impairment and BAC does not mean that such evidence is always irrelevant. As the New Mexico Supreme Court has explained, “behavioral evidence by itself cannot be sufficient to show the required nexus between a BAC test and an earlier BAC. It may, however, have limited relevance when the factors that underlie the shape of the concentration time curve [showing the level of alcohol absorption over time] are subject to conflicting testimony.” (State v. Day (2008) 143 N.M. 359, 176 P.3d 1091, 1100.)

That is precisely how such evidence is relevant in this case. The results of the four tests of Ashley Coffey's BAC indicated that it was rising, and the parties offer competing explanations. Coffey argues that her BAC was in fact rising and that her BAC must have been below 0.08 percent at the time of her arrest. By contrast, the Attorney General attributes the rising BAC test results to the test's margin of error (which would explain the 0.01 percent rise within three minutes between the first and second breath tests) and to the fact that the last two tests were blood tests rather than breath tests and thus produced slightly different results. The Attorney General asserts that the testimony of Coffey's expert, Jay Williams, lacked key findings to support a rising BAC theory, such as when Coffey had her last drink before getting behind the wheel, her food intake, her weight, and other factors. As the Department of Motor Vehicles (DMV) hearing officer concluded, Williams did not show that other experts in the scientific community had reached similar conclusions based on facts similar to those in this record.Evidence of Coffey's impairment was relevant to deciding which theory of rising BAC was more likely correct. Such evidence was relevant in the following limited sense: The fact that Coffey did show signs of impairment made her theory of rising BAC weaker than if Coffey had not shown signs of impairment at the time of her arrest. (See maj. opn., ante, 185 Cal.Rptr.3d at p. 552, 345 P.3d at p. 907 [“[T]he non-chemical-test evidence was admissible to rebut plaintiff's proffered defense that her BAC was low at the time she was driving and only later rose to exceed the legal limit.”].) This point, on which all members of the court agree, renders unnecessary any broader suggestion that in a typical administrative per se proceeding, evidence of a driver's impairment tends to prove a BAC level of 0.08 percent or higher at the time of arrest. Here the evidence of Coffey's impairment was, like the other weaknesses in the expert's testimony discussed above, relevant to the DMV hearing officer's conclusion that the rising BAC theory was “too speculative.”Because the evidence of Coffey's impairment was relevant in the limited way just described, and because the evidence is sufficient to support the determinations of the DMV hearing officer and the trial court that Coffey drove with a BAC at or above 0.08 percent, I concur in the judgment.