Defendant, who was convicted after a bench trial of one count of first-degree sexual abuse, appeals, asserting that the trial court erred in admitting an emergency room doctor’s diagnosis that the complaining witness had suffered “abusive contact.” He further asserts that his mandatory minimum sentence of 75 months pursuant to ORS 137.700 is unconstitutional as applied. We reverse defendant’s conviction, and therefore do not address the sentencing issue.
We state the necessary facts, including all reasonable inferences, in the light most favorable to the state. State v. Cervantes, 319 Or 121, 873 P2d 316 (1994). In December 2008, the 10-year-old complainant was spending the night at her friend H’s house when defendant, the boyfriend of H’s mother, came into the room where the two girls were sleeping on the floor, knelt down beside the complainant, and started rubbing her thigh over her clothes, eventually moving up to her pubic area. The complainant was awake but pretended to be asleep. Defendant got on top of the complainant, placed his pelvis between her legs, and made a motion simulating sexual intercourse. The complainant could feel defendant’s penis against her. Defendant tried unsuccessfully to remove the complainant’s shorts. The complainant rolled over onto her stomach, and defendant left the room.
The complainant recounted the incident that night to her mother, who took her to the emergency room. Medical personnel could find no physical evidence of abuse. However, based on the complainant’s reports, the emergency room doctor, Van Eaton, diagnosed “abusive contact of an adult with a patient, no penetration or genital contact.”
In a police interview, defendant denied any sexual contact with the complainant. He said that he went into the room where the girls were sleeping to turn off the television and the light and then left.
Defendant was charged with several sex offenses, and his case came to trial on December 21, 2009. Among other witnesses, the defense called Van Eaton, who described his examination of the complainant in the emergency room. He testified that the examination lasted “no more than ten *34to 15 minutes,” that the victim was not crying during the examination, and did not appear traumatized. He testified that he did not find any physical signs of trauma, and that a fluorescent black light did not reveal any semen on the complainant or her shorts. On the state’s cross-examination, the prosecutor asked Van Eaton whether he had made a diagnosis, and he testified without objection that he had diagnosed “abusive contact of an adult with a patient, no penetration or genital contact.” The prosecutor then asked, “So you saw her — [you did the] black light [to check for semen], talkfed] to her, talkfed] to her mom, reviewed the situation and that’s the diagnosis you came up with?” Van Eaton responded, “Yeah. To be clear, it was a diagnosis based on history, not on any physical findings.”
The theory of the defense was that the victim made up the allegations, either to gain her mother’s attention or because she felt animosity toward her friend, H, and H’s family. Defense counsel offered many reasons why the evidence gave rise to reasonable doubt: The room was too small; there was no physical sign of abuse; the complainant’s reports were inconsistent; the complainant did not appear traumatized shortly after the alleged event; defendant had previously appeared uninterested in the girls and had behaved normally with them.
The court convicted defendant of one count of first-degree sexual abuse. Defendant contends on appeal that, under both State v. Southard, 347 Or 127, 218 P3d 104 (2009), and State v. Lupoli, 348 Or 346, 234 P3d 117 (2010), the court plainly erred in admitting the doctor’s diagnosis in the absence of physical evidence of abuse. To consider correcting a plain error, this court must first determine whether the purported error meets three requirements: (1) The error must be one of law; (2) it must be “apparent,” i.e., the point must be obvious, not reasonably in dispute; and (3) it must appear on the face of the record, i.e., the court must not need to go outside the record to identify the error or choose between competing inferences, and the facts constituting the error must be irrefutable. Ailes v. Portland Meadows, Inc., 312 Or 376, 381-82, 823 P2d 956 (1991).
*35We describe briefly the underlying bases of defendant’s plain error argument. Under OEC 702, expert testimony is admissible if “specialized knowledge will assist the trier of fact to understand the evidence or to determine a fact in issue.” In Southard, decided on October 1, 2009, two and one-half months before defendant’s trial in December 2009, the Supreme Court considered the admissibility of a medical diagnosis of “sexual abuse” and whether it was “scientific evidence” that met the minimum standards for scientific validity under OEC 702, State v. Brown, 297 Or 404, 687 P2d 751 (1984), and State v. O’Key, 321 Or 285, 899 P2d 663 (1995). The court held that the diagnosis of sexual abuse, as made by a medical expert associated with KIDS Center, a nationally accredited medical facility, met the criteria for scientific evidence under OEC 702 and, further, that the evidence was relevant in a prosecution for sexual abuse. Southard, 347 Or at 138-39. The court explained, however, that the testimony showed that the methodology used to diagnose sexual abuse in the absence of physical evidence of abuse was largely dependent on an implicit credibility assessment. The court concluded that such a diagnosis is therefore inadmissible under OEC 403, because the probative value of the diagnosis is outweighed by the risk that a jury would determine the defendant’s guilt, based not on the its own credibility determination, but on the expert witness’s implicit conclusion that the victim’s reports of abuse are credible. Id. at 141-43. The court said:
“Our holding today is narrow. The only question on review is whether a diagnosis of‘sexual abuse’ — i.e., a statement from an expert that, in the expert’s opinion, the child was sexually abused — is admissible in the absence of any physical evidence of abuse. We hold that where, as here, that diagnosis does not tell the jury anything that it could not have determined on its own, the diagnosis is not admissible under OEC 403.”
347 Or at 142. The court was careful to point out that its decision “does not resolve [ ] whether any subsidiary principles that inform that diagnosis are themselves admissible.” Id.
In Lupoli, decided after defendant’s trial, the court took on that question. The expert witnesses in that case had *36explained that, in making a diagnosis of sexual abuse in the absence of physical evidence, they had evaluated the manner in which the child victim had reported the incident, and whether the victim’s disclosures were “clear,” “appropriate,” “detailed,” “consistent,” made in such a way that “you think that a kid is telling it like it is,” “developmentally appropriate,” and did not reflect risk factors such as “inappropriate sexual knowledge.” 348 Or at 353-56. In Lupoli, the defendant made no objection to the diagnosis of sexual abuse in itself. He did, however, object to the explanatory testimony concerning that diagnosis on the ground that it amounted to improper vouching by the experts for the victim’s credibility. Citing the well-accepted law that one witness may not give an opinion on whether he or she believes another witness is telling the truth, State v. Middleton, 294 Or 427, 438, 657 P2d 1215 (1983), the Supreme Court agreed with the defendant. The court held that such testimony — which might be admissible in' other contexts — is impermissible if it is “inextricably bound up with” portions of testimony that constitute clear “vouching” for the witness’s reports. The court said that, absent physical evidence of abuse, a sexual abuse diagnosis is necessarily based on an assessment of the child’s believability, and the testimony explaining the bases for the diagnosis could not be meaningfully separated from its context and was inadmissible. Lupoli, 348 Or at 362; see also State v. Bainbridge, 238 Or App 56, 59, 241 P3d 1186 (2010).
Under that case law, the state contends that the trial court’s admission of Van Eaton’s testimony is not plain error for two reasons: (1) It is not “apparent” that Van Eaton made a diagnosis of “sexual abuse,” or that he made a diagnosis that includes an implicit credibility determination that would be inadmissible under either Southard or Lupoli. (2) The court may infer that, because Southard had been decided more than two months before defendant’s trial, trial counsel had a strategic reason for not objecting to the testimony.
The state asserts that Van Eaton’s diagnosis does not necessarily impart that he believed the complainant had been sexually abused and or that he believed she had experienced the type of contact necessary to prove *37first-degree sexual abuse.1 The state makes a creditable argument that a diagnosis of “abusive contact” could cover a broader range of contact than sexual abuse and that it is possible that Van Eaton’s diagnosis does not mean that he had concluded that the complainant had been sexually abused. For example, the state suggests that an unwanted strike to the arm might constitute “abusive contact.” The state speculates that Van Eaton might have concluded that defendant had made “abusive contact” but disbelieved that any “sexual contact” had occurred.
We might agree with the state’s view if Van Eaton’s diagnosis had only been “abusive contact.” But, from the doctor’s further description of that contact — “of an adult with a patient, no penetration or genital contact” — the only reasonable inference is that the “abusive contact” was sexual. If Van Eaton had not been referring to sexually abusive contact, there was no reason to include in his diagnosis “no penetration or genital contact.” In the absence of physical evidence of abuse, we conclude that the diagnosis was inadmissible under Southard.2
The state’s further contention is that the error is not plain because, in light of the fact that Southard had been decided before defendant’s trial, defendant’s counsel must be presumed to have known of it and must be assumed to have made a tactical decision not to object to Van Eaton’s diagnosis. It is true, as the state points out, that defense counsel is presumed to have known about the Supreme Court’s opinion in Southard at the time of trial. However, the record does not reflect any plausible tactical reason why *38counsel would have chosen not to object to Van Eaton’s diagnosis testimony under Southard, and we can conceive of no strategic purpose in not objecting to that testimony.3 It was plain error for the trial court to admit it.
The remaining question is whether we should exercise our discretion under Ailes to correct the error. One factor that we consider in determining whether to exercise our discretion to correct an unpreserved error is “the gravity of the error ” Ailes, 312 Or at 382 n 6. The state posits that, in a case such as this, where the trial was to the court, any error in admitting the evidence was not particularly grave, because it is unlikely that the error led the trial court to make a decision based on improper evidence. Citing State v. Cafarelli, 254 Or 73, 76, 456 P2d 999 (1969) (“In a trial to the court we can assume that any questionable evidence would be disregarded.”); but see State v. Marrington, 335 Or 555, 565-66, 73 P3d 911 (2003) (calling Cafarelli into question and rejecting contention that failure of trial court to mention witness’s testimony in announcing its decision means that evidence played no role in trial court’s assessment of the state’s evidence), the state contends that this court may presume that the trial court did not consider Van Eaton’s diagnosis “in an impermissible way.” The state points out that the trial court said, “I agree and do not dispute that one witness cannot vouch for the credibility or comment on the credibility [of] another witness,” allowing the presumption that the court understood that Van Eaton’s diagnosis should not be considered. In context, however, it appears that that statement addressed the testimony of a different witness and not Van Eaton’s diagnosis. The court described the evidence that it found most persuasive and did not separately discuss Van Eaton’s diagnosis. The trial court placed particular emphasis on the fact that it could not conclude that the complainant was motivated to make the abuse report by some improper purpose. All in all, however, in the absence of a statement by the trial court that it did not consider Van Eaton’s diagnosis, we must conclude that the error was *39not harmless. See State v. Davilia, 239 Or App 468, 478, 244 P3d 855 (2010) (citing Marrington). As we held in State v. Bahmatov, 244 Or App 50, 55, 260 P3d 592 (2011), in a case such as this, which is largely dependent on the witnesses’ credibility, we cannot say how Van Eaton’s testimony might have affected the trial court’s determination. We therefore exercise our discretion to correct the error, for the reasons set forth in State v. Gonzales, 241 Or App 353, 360-61, 250 P3d 418, rev den, 350 Or 571 (2011).
Reversed and remanded.
ORS 163.305(6) defines the “sexual contact” necessary to prove first-degree sexual abuse under ORS 163.427 as “any touching of the sexual or other intimate parts of a person ***[.]”
In further support of its first contention, the state makes the somewhat circular argument that Van Eaton’s diagnosis was not a diagnosis of sexual abuse, because it is not apparent that Van Eaton followed the methodology outlined in Southard for the diagnosis of sexual abuse and it is not apparent that Van Eaton followed any methodology that is dependent on an implicit evaluation of the complainant’s credibility. The state’s argument is off point. Whether the diagnosis was a diagnosis of sexual abuse depends on the diagnosis itself, not on the method that was used to reach it. But the fact that the evidence does not show what methodology Van Eaton used does not mean that his diagnosis was not a diagnosis of sexual abuse and was not dependent on an assessment of the complainant’s credibility.
It is possible, given the differences between this case and Southard in terms of the evidentiary record concerning the making of the diagnosis, that defense counsel, like the state here, believed that Southard was inapplicable and that an objection under Southard would have been futile.