¶ 1. Defendant Maureen Wilt appeals a conviction for driving under the influence (DUI) on grounds that the trial court improperly allowed a police trooper to testify about the results of a field-sobriety test he administered to defendant. We affirm defendant’s conviction.
¶ 2. Defendant was arrested on December 26, 2011, while a passenger in her own car, on grounds that she had been seen driving earlier in the night when intoxicated. According to testimony by defendant’s neighbor, he came over to defendant’s house to cook a post-holiday dinner with her in the late afternoon. It took well over an hour to prepare the meal, and by neighbor’s estimation they ate around 6:00 p.m. During his visit, neighbor observed defendant having two or three five-ounce glasses of blush wine. After dinner, neighbor returned to his house, and did not see defendant again until she showed up at his door, injured, approximately an hour and a half later.
¶ 3. In the meantime, an acquaintance of defendant, Mr. Rondeau, testified that he received a call from defendant around 8:00 p.m., asking him if she could come over. He said no, telling her that he had to get up early and that he was going to unplug the phone and lock the doors. Mr. Rondeau estimated that the call lasted twenty minutes. Shortly thereafter, he heard someone loudly knocking on the front door and a window. After looking outside and seeing defendant’s car in his driveway, Mr. Rondeau waited about ten minutes, hoping defendant would leave. When her car remained in the driveway, he checked the garage, heard a noise in the cellar, and then found defendant lying at the bottom *3of the cellar stairs, apparently unconscious and bleeding from her head. Mr. Rondeau instructed his son to call 911, but before the son was able to, defendant got up, walked to, her car and drove away, wearing only one shoe. Mr. Rondeau estimated that defendant left around 8:40 p.m. and testified that her driving “seemed fine” — defendant backed up on the correct side of the road and then drove straight away from the house.
¶ 4. Defendant then arrived on neighbor’s doorstep at close to 9:00 p.m., very upset and with blood running down her face. Neighbor invited defendant in and assessed her injury, locating a wide gash near defendant’s forehead that was bleeding profusely. Neighbor also observed that defendant’s eyes “were very glassy” and she had a “faraway look in her eyes,” which he feared was an indication that she might pass out. Neighbor got his shoes and coat on to take defendant to the hospital, and as he was doing so, he saw defendant drink from a bottle of 100-proof Southern Comfort alcohol. Neighbor estimated that the level in the bottle had “gone down about an inch.” Based on his thirty years of bartending experience, he estimated defendant drank two ounces of alcohol. At that point, neighbor- grabbed defendant, helped her into her car, and started toward the hospital. Not far down the road, police — who had been alerted to look for defendant by Mr. Rondeau — pulled the car over.
¶ 5. Two state troopers began asking questions of both neighbor and defendant. The trooper speaking with defendant observed blood matted in her hair and smelled intoxicants coming from the car and defendant, who admitted she drank wine with dinner. The trooper also noted that defendant had difficulty getting out of the car and walking unassisted, although he also noted she was not wearing shoes. He then asked defendant to complete three standardized field-sobriety exercises: a horizontal-gaze nystagmus (HGN) test, a walk-and-turn test, and a one-leg-stand test. The trooper’s assessment was that defendant failed all three tests, and based on these test 'results, the odor of intoxicants, and the difficulty in walking unassisted, he concluded that defendant was intoxicated. Defendant was eventually transported to the hospital, where her blood was drawn. At the time of the sample, defendant’s blood-alcohol concentration (BAC) was .160.
¶ 6. A jury trial was held in which neighbor, Mr. Rondeau, the two troopers, and blood-alcohol experts for both the State and defendant testified. Both of the experts addressed defendant’s *4blood sample result and, using relation-back reasoning, estimated what defendant’s BAC would have been at the time she was alleged to have driven intoxicated on the night in question. As both experts testified, the accuracy of the relation-back BAC estimate was heavily dependent on how much alcohol defendant consumed at neighbor’s house between the time she drove and the time she was arrested. The State’s expert testified that if defendant had two ounces at neighbor’s house, her BAC at the time of operation would have been .136, but if she had four ounces her BAC would have been closer to .068, which is below the legal limit. Defendant’s expert corroborated these estimates.
¶ 7. On appeal, defendant’s two claims of evidentiary error both stem from the testimony of the police trooper who conducted the field-sobriety exercises, and specifically regard the administration and results of the HGN test. Defendant first argues that the trial court erred in allowing the trooper to offer his assessment of defendant’s BAC based on the results of the HGN test. Similarly, defendant argues that it was also error for the court to allow the trooper to testify about the HGN test after he admitted that he did not strictly follow the standard procedure for a subject with an obvious head wound. Defendant asserts that without the trooper’s HGN testimony, there is no link between defendant’s BAC level and her level of impairment at the time of operation, and therefore the jury would not have convicted her. We disagree.
¶ 8. Before addressing each argument in turn, we note that this Court reviews the question of whether evidence was properly admitted deferentially, reversing only for an abuse of the court’s discretion. State v. Fuller, 168 Vt. 396, 404, 721 A.2d 475, 481 (1998) (“The admissibility of evidence is addressed to the discretion of the trial judge, and this Court will reverse only if the trial judge has abused that discretion.” (quotation omitted)).
¶ 9. Turning to defendant’s first claim, the State concedes in its brief that it was error for the court to allow the trooper to estimate that defendant’s BAC was “over a .10” based on defendant’s performance on the HGN test.2 Although both parties *5acknowledge that the trooper was unqualified to offer such a quantitative assessment, there is a dispute as to whether this error should be analyzed under the plain or harmless error standards. The State argues for a plain error analysis, contending that defendant did not object to the trooper’s estimation during his testimony, and therefore failed to preserve an objection on appeal. See V.R.Cr.P. 52(b) (“Plain errors . . . may be noticed although they were not brought to the attention of the court.”); V.R.E. 103 (stating the objections to rulings on evidence must be timely and specific or constitute plain error); see also State v. Beattie, 157 Vt. 162, 169, 596 A.2d 919, 923 (1991) (noting that where defendant failed to object to testimony, review is limited to plain error). Defendant maintains that the objection was preserved and therefore the harmless error rule applies. Ultimately, it matters little whether the objection was preserved, as we conclude that any error in the trooper’s estimation of defendant’s BAC was harmless, and therefore not grounds for reversal.
¶ 10. Under Vermont’s Rule of Criminal Procedure 52(a), “[a]ny error . . . which does not affect substantial rights shall be disregarded,” and only requires reversal where this Court cannot say beyond a reasonable doubt that the jury would have returned the same verdict. State v. Brooks, 2013 VT 27, ¶ 27, 193 Vt. 461, 70 A.3d 1014. The two most important factors we look at in this determination are (1) the strength of the State’s case without the admitted evidence, and (2) the strength of the admitted evidence itself. Id. Here, the trooper’s BAC estimate carried very little weight on its own. It was clear that it was only an estimation, and the jury subsequently heard expert testimony on defendant’s actual BAC from the blood sample, which was higher than the trooper guessed. The State’s case that defendant drove while she was intoxicated relied on the actual BAC and the expert’s relation-back testimony to show that defendant was intoxicated at the time of operation, not on the trooper’s estimate. Thus, whatever error was carried to the jury in the trooper’s statements about defendant’s potential BAC based on the HGN test was cured by the expert testimony of defendant’s actual BAC. In other *6words, the trooper’s BAC testimony was cumulative evidence of defendant’s intoxication at the time of arrest, and not what ultimately substantiated that defendant drove while intoxicated; rather, her blood sample BAC was what allowed the jury to convict defendant. See State v. Hunt, 150 Vt. 483, 494, 555 A.2d 369, 376 (1988) (holding error harmless where evidence admitted was cumulative). The error was therefore harmless.
¶ 11. We reach the same conclusion on defendant’s second claim of error. Defendant argues that the court should have excluded the trooper’s testimony of the HGN test results after he admitted that he did not alter his administration of the test to try to rule out the possibility that defendant’s head injury was causing the nystagmus he observed. Following this admission, the defense attorney asked the court to preclude any further testimony from the trooper on the HGN test. The court denied the request, stating that it was satisfied that the trooper was qualified to administer the test and that defendant’s objection went to the weight of the evidence, on which defendant was free to cross-examine the trooper in front of the jury.
¶ 12. Defendant’s attorney then cross-examined the trooper about the head injury issue, and the trooper again admitted that he did not follow the specific protocol for administering the test to someone with a head injury, as outlined in the training manual. As a result, the jury was well aware that the HGN test results, such as they were, were potentially unreliable. The fact that the results may not have firmly substantiated defendant’s intoxicated state does not require reversal, however. This Court has “long recognized that judging the credibility of witness testimony is a duty left to the jury.” State v. Couture, 169 Vt. 222, 227, 734 A.2d 524, 528 (1999). Moreover, the trooper’s testimony on the HGN test was by no means the “key” link between defendant’s alcohol consumption and her “alleged impairment at the time of operation,” as defendant contends. Rather, the key link was again the State’s expert testimony relating-back defendant’s .160 BAC to the time she allegedly drove her vehicle while intoxicated.
¶ 13. Defendant further claims that because the testimony by the State’s expert was contradicted by defendant’s expert, the HGN testimony is all the more vital. We cannot agree. The ultimate point of contention in this case is how much alcohol *7defendant drank at neighbor’s house because the relation-back to defendant’s illegal BAC at the time of operation relies entirely on this point. Therefore, the critical information is not the HGN testimony, but the experts’ testimony about relation-back amounts and neighbor’s estimate of how much Southern Comfort defendant drank at his house.
¶ 14. Even assuming arguendo that it was error for the court to allow the trooper to testify to potentially unreliable HGN test results, any error would be harmless because the HGN testimony proved nothing about defendant’s condition that was not established by her BAC at the time of the blood sample. Defendant’s reliance on the trooper’s HGN testimony as the key to her conviction is entirely misplaced; it was the blood sample that confirmed her BAC, and the relation-back testimony based ón neighbor’s estimate that allowed the jury to find she drove while over the legal limit.
¶ 15. Admittedly, the facts presented in this case are unusual for a DUI charge in that police never observed defendant driving, and she was arrested while an intoxicated passenger who had consumed a disputed amount of alcohol between the time of operation and arrest. The State’s expert admitted that the relation-back evidence, reliant as it is on a timeline pieced together by rough estimates from the individuals who testified, was not exact and she could not say for sure what defendant’s BAC was at the time she drove her car. Nonetheless, defendant has not raised a challenge to the sufficiency of the relation-back evidence; rather, her appeal is limited to the two narrow evidentiary claims regarding the trooper’s HGN testimony addressed above. For the reasons stated, we affirm defendant’s conviction.
Affirmed.
According to the field-sobriety testing manual used by the Vermont Criminal Justice Training Council, the HGN test is designed to test for an involuntary jerking of the eyes (nystagmus) as they move to the side that can be caused by alcohol and other drugs. See State v. Blouin, 168 Vt. 119, 120 n.1, 716 A.2d 826, 827 n.1 (1998) (“The HGN test involves moving an object such as a pen across the subject’s field of vision to observe the manner in which the subject’s eyes follow *5the object. . . . [A]n overabundance of eye twitching indicates possible intoxication.”). The trooper testified that he was looking for six clues throughout the HGN exercise, that the presence of four or more clues indicates failure of the test, and that he observed all six in defendant. It was from this result that the trooper estimated defendant’s BAC to be .10.