¶1 — The trial court admitted the opinion of a physician’s assistant that a child had probably been sexually abused despite the absence of any physical signs of abuse. The physician’s assistant based his opinion on a theory that effectively posits that if the child relates events within a given level of specificity then the child has probably been abused. The theory does not satisfy the Frye test. Frye v. United States, 293 F. 1013, 1014 (D.C. Cir. 1923). The testimony of the physician’s assistant usurped the function of the jury to determine guilt. And we reverse the convictions here for three counts of first degree child rape and one count of first degree child molestation and remand for a new trial.
Sweeney, A.C.J.*585FACTS
¶2 C.M. is Larry Dunn’s step-granddaughter. She said he sexually abused her at his home in Spokane from June 2000 to June 2001. She was then seven and eight years old.
¶3 C.M.’s parents were at times separated. C.M. lived primarily with her mother in Spokane, but reported the abuse to her father shortly after returning to live with him in Pueblo, Colorado. She then gave more details of the abuse to her mother and police in Pueblo.
¶4 C.M. was then referred to physician’s assistant James Kramer for a physical examination. He is experienced in the field of child sex abuse examinations. He found no physical evidence of sexual abuse. But he used a diagnostic method attributed to Dr. Sally Adams. And he concluded that because of C.M.’s “clear, concise, and explicit, and detailed” disclosures, sexual abuse was “probable.” Report of Proceedings (RP) at 164.
Procedural History
¶5 The State charged Mr. Dunn with three counts of rape of a child in the first degree and one count of child molestation in the first degree. His first trial ended in a hung jury. His case went to trial a second time in April 2003.
¶6 Mr. Dunn moved in limine to exclude as cumulative, repetitive, and unduly prejudicial, the State’s proposed hearsay testimonies of both parents, Sergeant Joseph Garcia (including introduction of his videotaped interview with C.M), and Mr. Kramer. The court denied that motion, both pretrial and during the trial.
|7 Mr. Dunn also moved to exclude Mr. Kramer’s proposed opinion testimony that sexual abuse was “probable” in C.M.’s case because it failed the Frye test (based upon novel theory not generally accepted in the scientific community). And the proposed opinion invaded the right of the jury to find guilt. The court allowed the testimony.
*586Trial
¶8 C.M. testified in some detail about the abuse. She said it happened 7 to 12 times. She described the places where it happened and when. Mr. Kramer testified he examined C.M. on a November 2001 law enforcement referral for suspected sexual abuse. He interviewed C.M. before his examination. He wrote down in some detail the events supporting C.M.’s claim including time, place, and exactly what Mr. Dunn did.
¶9 Mr. Kramer thoroughly examined C.M. physically. And he concluded there were no medical indicia of sexual abuse, a “class one normal examination.” RP at 162. Mr. Kramer explained that lack of physical findings is not surprising, as the most common finding is a normal physical exam. He then referred to a report of Dr. Adams positing that 65 to 75 percent or more of physical exams are normal in light of the accusation of sexual assault.
¶10 Mr. Kramer next referred to an overall assessment of sexual abuse divided into four classifications: “Class one ... no evidence of abuse, class two . . . possible abuse, class three . . . probable abuse, and class four . . . definite abuse or sexual contact.” RP at 164. He selected class three, probable abuse. The prosecutor asked why he chose that classification. Mr. Kramer responded: “[Biased upon the criteria of. .. Dr. Adams’ classification system, which stipulates specifically that the history is clear, concise, and explicit, and detailed, it falls into . . . class three, probable.” RP at 164. Mr. Kramer then testified that the only classification more indicative of abuse is definite physical evidence of sexual abuse or sexual contact.
¶11 Mr. Kramer said his interview with C.M. was “a medical history, it’s not a psychological interview,” and did not include family dynamics. RP at 171. He said his conclusion of probable abuse was a subjective finding based on C.M.’s statements—physical evidence of abuse is an objective finding not pertinent here.
¶12 C.M.’s father, Mr. M., testified that during the short time he was united with the family in Spokane, the children *587did spend time alone with Mr. Dunn. C.M. spent a couple of nights at the Dunns’ house. And Mr. Dunn saw her at least once a week. Mr. M. observed that Mr. Dunn wanted to be around C.M., but generally not G.M, her younger brother. Mr. M. said he has always disliked Mr. Dunn and questioned whether he could be trusted around his children.
¶13 The children returned to Pueblo to stay with their father in May 2001. Mr. M. noticed that the usually talkative and articulate C.M. was acting very withdrawn. He asked if something was wrong. She finally disclosed that “Papa Larry” (Mr. Dunn) had been touching her. RP at 194. C.M. pointed to her privates and told her father that Papa Larry had abused her. Mr. M. told C.M. to discuss it further with her mother. Mr. M. telephoned Ms. M. that day.
¶14 Both parents encouraged C.M. to seek counseling or go to the police. But C.M. did not want to talk about it and was afraid she was in trouble. She was finally willing to speak with Sergeant Joseph Garcia in October.
¶15 The trial lasted for four days. The evidence can be characterized as conflicting. Mr. Dunn denied that he molested C.M. Other witnesses, including Mr. Dunn’s wife, said they saw no obvious signs of molestation and then explained how they were in a position to notice if anything was amiss. Everyone who C.M. repeated the abuse to, including her parents, Sergeant Garcia, and Mr. Kramer, related her statements to the jury.
¶16 The jury found Mr. Dunn guilty as charged. The court imposed concurrent sentences of 300 months for each rape and 198 months for the molestation.
DISCUSSION
Cumulative Child Hearsay
¶17 Mr. Dunn first assigns error to the trial judge’s decision to allow all the damning child hearsay—C.M.’s mother and father, Mr. Kramer, and Sergeant Garcia. He complains that the repetitive and cumulative hearsay overemphasized C.M.’s trial testimony and aroused an emo*588tional, rather than rational, response from the jury. And therefore the judge should have limited it.
¶18 Child hearsay is admissible where the child is available and competent to testify. RCW 9A.44.120. The statute alleviates difficult proof problems that often frustrate prosecution of child sex abuse cases. See State v. Jones, 112 Wn.2d 488, 493-94, 772 P.2d 496 (1989); State v. Bedker, 74 Wn. App. 87, 92, 871 P.2d 673 (1994).
¶19 The statements are nonetheless subject to ER 403. It permits exclusion of evidence if the probative value is substantially outweighed by the danger of prejudice from needless presentation of cumulative evidence. Bedker, 74 Wn. App. at 93; see also State v. Rice, 48 Wn. App. 7, 12-13, 737 P.2d 726 (1987); State v. Ammlung, 31 Wn. App. 696, 700, 644 P.2d 717 (1982). Our review is for abuse of discretion. Bedker, 74 Wn. App. at 93.
¶20 Mr. Dunn argues that the prejudice mounted with the testimony of each additional child hearsay witness to the point that it became a vouching for the victim. True, there was considerable overlap in the child hearsay testimony. But the repetitiveness stemmed largely from the logical sequence and timing of events.
¶21 C.M.’s disclosure to her father was a general statement of the allegations and involved little detail. The father deferred the matter to C.M.’s mother. C.M. disclosed considerably more detail to her. C.M. was later willing to speak with Sergeant Garcia. Reporting to law enforcement is part of the logical progression in a child sexual abuse case. The details given by C.M. on the videotaped interview by Sergeant Garcia are largely repetitive of her previous disclosures to her father and mother. But the videotape also provided the jurors with visual and audio information (C.M.’s demeanor and voice inflections) that went beyond repetition of other hearsay statements. Sergeant Garcia also presented drawings on which C.M. identified the private parts she said were touched by Mr. Dunn. Sergeant Garcia’s conclusion from this interview was that C.M. should be referred for a physical medical examination for *589evidence of sexual abuse—again, a logical progression of the case.
¶22 Mr. Kramer interviewed C.M. to take a history-in preparation for the physical examination for evidence of abuse. C.M. knew the purpose for the questions. The statements to Mr. Kramer may repeat and overlap. But the admission of merely cumulative evidence is not necessarily prejudicial error. State v. Todd, 78 Wn.2d 362, 372, 474 P.2d 542 (1970); State v. Acheson, 48 Wn. App. 630, 635, 740 P.2d 346 (1987). Here, the trial court admitted them as part of a medical diagnosis. ER 803(a)(4).
¶23 We find no abuse of discretion in the trial court’s decision to admit the child hearsay statements (including the videotape) or Mr. Kramer’s testimony relating C.M.’s disclosures. The trial court’s determination that the testimony was not unduly prejudicial under ER 403 and did not impinge on the jury’s responsibility to determine credibility is supported by this record.
Opinion of Physician’s Assistant and Frye
¶24 Mr. Dunn next argues that the court erred in admitting Mr. Kramer’s opinion that sexual abuse was probable because the theory supporting the opinion lacked any foundation in recognized scientific principle or theory. Frye v. United States, 293 F. 1013, 1014 (D.C. Cir. 1923). Essentially he posits that the opinion does not represent the view of the scientific community.
¶25 A witness qualified as an expert may testify on the basis of “scientific, technical, or other specialized knowledge” if the testimony “will assist the trier of fact to understand the evidence or to determine a fact in issue.” ER 702. The admission of scientific testimony involves two related inquiries. The first is whether the scientific principle or theory from which the testimony is derived has garnered general acceptance in the relevant scientific community under the Frye standard. The second inquiry is whether the expert testimony is properly admissible under ER 702. See State v. Riker, 123 Wn.2d 351, 359, 869 P.2d 43 *590(1994); State v. Janes, 121 Wn.2d 220, 850 P.2d 495 (1993). The first question is the focus of Mr. Dunn’s argument.
¶26 We look at: “(1) whether the underlying theory is generally accepted in the scientific community and (2) whether there are techniques, experiments, or studies utilizing that theory which are capable of producing reliable results and are generally accepted in the scientific community.” Riker, 123 Wn.2d at 359. Our concern is not to determine if the scientific theory underlying the proposed testimony is correct, but rather, whether it is generally accepted in the appropriate scientific community. Id. at 359-60. “The core concern of Frye is only whether the evidence being offered is based on established scientific methodology. This involves both an accepted theory and a valid technique to implement that theory.” State v. Cauthron, 120 Wn.2d 879, 889, 846 P.2d 502 (1993). And we review the trial court’s decision to admit novel scientific evidence de novo. Id. at 887.
¶27 The question here is whether that part of Dr. Adams’ theory is generally accepted scientific theory in the medical child sex abuse examination community. Said another way, does the conclusion of probable sexual abuse follow solely from statements of the alleged child victim when there is no physical evidence of abuse?
¶28 We begin with a review of State v. Carlson.1 There, the State charged the defendant with molesting a six-year-old child on multiple occasions. A first trial ended in a hung jury. At the second trial, the child testified that the defendant had touched her in inappropriate ways. A medical expert had examined the child. She testified there were no physical findings from which to conclude the child had been sexually abused. She nevertheless testified that the child had been sexually abused based on the child’s statements in the assessment interview; all this based on a study. The defendant denied touching the child. State v. Carlson, 80 Wn. App. 116, 118, 906 P.2d 999 (1995).
*591¶29 The Carlson court noted the obvious—the State offered the expert’s opinion not to prove the child’s credibility, but to prove the child had been sexually abused. Id. at 123. The court concluded the requirements of Frye had not been met and the opinion was inadmissible:
Dr. Feldman’s physical findings were inconclusive, and Washington law has never recognized the ability of a doctor or other expert to diagnose sexual abuse based solely on the statements of an alleged victim. If Dr. Feldman had a scientific basis for her opinion concerning sexual abuse, that basis surely is novel.
Id. at 125 (footnote omitted).
¶30 Here, the prosecutor asked Mr. Kramer at the outset of the Frye hearing if he had an opinion whether Dr. Adams’ methodology meets the Frye standard. Mr. Kramer responded:
Dr. Adams is a nationally recognized expert... in child abuse, and has been published in many journals, as well as text as has her methodology for classifications. And to the best of my knowledge that has been accepted by the sexual examination community as a legitimate methodology for evaluating children.
RP at 140 (emphasis added).
f 31 Mr. Kramer said the Adams’ methodology is used in the Denver, Colorado area in a nationally recognized child abuse center, and that a similar type of methodology is used at the San Diego Children’s Hospital. Mr. Kramer said his own report and finding using the classification system has been admitted in court before, and to his knowledge never excluded when he personally testified. He then referred to particular literature co-authored by Dr. Adams on the topic of findings in child sexual assault examinations. Mr. Kramer also referred to the required training manual at San Diego Children’s Hospital, which breaks down the Adams’ methodology into very detailed subsections.
¶32 Mr. Kramer admitted, however, that other than the hospital training manual, the theme of the literature being discussed was that lack of physical findings is not unusual *592in a child sex abuse case. And he admitted the articles do not address the same criteria as the hospital training manual. Mr. Kramer did not state where, when, and how many times he has testified as to this methodology. Moreover, as in Carlson, the record does not include any of the literature considered by the court at the Frye hearing. Carlson, 80 Wn. App. at 125.
¶33 Here, on appeal, the State reiterates Mr. Kramer’s testimony. But it makes no attempt to explain how Dr. Adams’ methodology has gained general acceptance in the scientific community on the point that probable sexual abuse can be diagnosed solely on the alleged child victim’s statements. The State cites no authority for this proposition. And we find none. We cannot conclude on this record that Dr. Adams’ methodology has resulted in any generally accepted theory or principle. The court therefore erred in allowing Mr. Kramer to testify as to his conclusion of probable abuse. Id. at 125-26.
¶34 Mr. Dunn next argues that Mr. Kramer’s testimony was not harmless given the conflicting versions of what happened. The State responds that the statement of Mr. Kramer is no different than that of any other expert who looks at the sequela and offers an opinion on the cause of physical findings. But there is a difference between accepting the subject matter of the victim’s statements to evaluate for truthfulness and forensically examining those statements for particular characteristics that are hallmarks of abuse and then formulating a medical opinion.
¶35 An expert’s opinion on an ultimate issue of fact that is “based solely on the expert’s perception of the witness’ truthfulness” is unfairly prejudicial and thus inadmissible because it takes an ultimate issue of fact from the jury. State v. Alexander, 64 Wn. App. 147, 154, 822 P.2d 1250 (1992); see also State v. Black, 109 Wn.2d 336, 348-49, 745 P.2d 12 (1987); State v. Florczak, 76 Wn. App. 55, 74, 882 P.2d 199 (1994); State v. Fitzgerald, 39 Wn. App. 652, *593694 P.2d 1117 (1985).2 In Fitzgerald, for example, the court reasoned: “Dr. Griffith’s opinion is based solely on her evaluation of the children’s version of the events. ‘ “An expert may not go so far as to usurp the exclusive function of the jury to weigh the evidence and determine credibility.” ’ ” Fitzgerald, 39 Wn. App. at 657 (quoting 5A Karl B. Tegland, Washington Practice: Evidence § 292, at 39 n.4 (2d ed. 1982) (quoting United States v. Samara, 643 F.2d 701, 705 (10th Cir. 1981))).
¶36 And the admission of such testimony is constitutional error. Florczak, 76 Wn. App. at 74. Any error that infringes on a constitutional right is presumed prejudicial. And the State must show that the error was harmless beyond a reasonable doubt. State v. Miller, 131 Wn.2d 78, 90, 929 P.2d 372 (1997). The error is harmless if there is overwhelming untainted evidence that the jury would have reached the same result without the erroneous introduction of evidence. State v. Guloy, 104 Wn.2d 412, 426, 705 P.2d 1182 (1985).
¶37 In Carlson, the court reversed a conviction based on the admission of the pediatrician’s opinion the child was abused when there was no physical evidence of abuse or independent witness to the charged events. And the case boiled down to the victim’s word against the defendant’s. Carlson, 80 Wn. App. at 129. Similarly, in Fitzgerald, the results of the children’s physical examinations were inconclusive and the pediatrician’s opinion of sexual molestation was based solely upon her evaluation of the children’s version of the events. Fitzgerald, 39 Wn. App. at 656-57.
¶38 Carlson and Fitzgerald are helpful here. The only evidence that C.M. was sexually abused was her own testimony and her hearsay statements to her parents, Sergeant Garcia, and Mr. Kramer. Mr. Dunn denied the abuse. There is no physical evidence or independent witness to the charged events. Mr. Dunn also denied the opportunity to commit the crimes; he testified he was never *594alone with C.M. This was, then, a credibility contest—her word against his, as well as family-member-witness against family-member-witness. Thus, as in Carlson and Fitzgerald, admitting Mr. Kramer’s testimony of probable abuse usurped the exclusive function of the jury to weigh the evidence and determine credibility. It was not harmless error. Carlson, 80 Wn. App. at 129; Fitzgerald, 39 Wn. App. at 657.
f 39 Finally, as in Carlson, the first trial ended in a hung jury. Mr. Kramer did not testify at the first trial, and no one in that trial testified to an opinion of probable abuse.
¶40 Our determination of prejudicial error is further supported by Mr. Dunn’s cited case, United States v. Azure 3 There, the issue was whether the court abused its discretion in ruling Fed. R. Evid. 702 allowed a pediatrician to give an opinion as to the truth of the story of the child abuse victim. United States v. Azure, 801 F.2d 336, 339 (8th Cir. 1986). The court emphasized that by allowing the pediatrician to put his stamp of believability on the child’s story, he essentially told the jury that the defendant was the person who sexually abused her:
No reliable test for truthfulness exists and [the pediatrician] was not qualified to judge the truthfulness of that part of [the child’s] story. The jury may well have relied on his opinion and “surrender [ed] their own common sense in weighing testimony . . . .”
Id. at 341 (quoting United States v. Barnard, 490 F.2d 907, 912 (9th Cir. 1973)). Given that the erroneously admitted opinion likely bolstered the credibility of the key government witness, the court deemed the error reversible even though the evidence was otherwise sufficient to support the conviction. Azure, 801 F.2d at 341.
¶41 The same is true here. The evidence is sufficient to support Mr. Dunn’s convictions. But the evidence is not so overwhelming that it renders Mr. Kramer’s opinion harmless.
*595¶42 As for Mr. Dunn’s additional grounds for review, credibility determinations are for the jury.
143 We reverse the convictions and remand for a new trial.
Schultheis and Brown, JJ., concur.
State v. Carlson, 80 Wn. App. 116, 906 P.2d 999 (1995).
Mr. Dunn’s several federal cases cited at pages 30-31 of his opening brief state the same principles.
United States v. Azure, 801 F.2d 336 (8th Cir. 1986).