GILBERT, P.J.
*880Sports fans often use "statistical odds" to predict the outcome of a sporting event. Statistical odds, however, are not a substitute for admissible evidence to decide the guilt or innocence of the defendant.
Cody Adam Julian appeals a judgment following his conviction after a jury trial of four counts of lewd acts upon a child ( Pen. Code, § 288, subd. (a) ) and one count of sexual penetration with a child under 10 years old (id. , § 288.7, subd. (b)). We conclude, among other things, that 1) the People's expert witness introduced inadmissible statistical evidence that went beyond the permissible scope of child sexual abuse accommodation syndrome (CSAAS) evidence and deprived Julian of a fair trial; 2) Julian's counsel provided ineffective assistance by not objecting to this evidence; and 3) Julian's defense was prejudicially undermined by his counsel's question that *881invited a police detective to give the opinion that the People's witness against Julian was credible. We reverse and remand for a new trial.
FACTS
Julian, 28 years old, moved into a home where four minor sisters resided - child 1, child 2, child 3, and child 4.
Child 2's Testimony
Child 2, 10 years old, testified Julian played games with her and her sisters, including tag and hide and seek. She played a tickle game with Julian and her father. The prosecutor asked her whether Julian "ever put his hands down [her] pants when [she] played the tickle game." She responded, "I don't think so. I don't remember." When he asked the question again, she said, "No."
The prosecutor asked child 2 how many times Julian touched her "private area" when she was seven years old. She answered, "I think maybe, like, once or twice." He asked, "Can you tell me about the very first time you remember it happening[?]" Child 2 responded, "I don't really remember it." She remembered playing hide and seek near a truck. There were "pokey leaves" on the ground. Julian asked her to sit on his lap. She sat there because she did not want to sit on those leaves. Julian put his hand "inside [her] underwear" and put his finger in her "private area" - vagina. On a second occasion, they played hide and seek near a truck. Julian asked her "to sit on his lap." She "went on his lap and then he did the same thing that he did [the first time]."
The prosecutor asked child 2, "[W]hen you hid down by [a] tree one time, did [Julian] do this to you?" She said, "Yes." They were playing hide and seek; she sat *520on his lap. Julian also touched her private area on another occasion when they were on a bed in a recreational vehicle (RV) playing hide and seek. He put his finger in her vagina and her anus. She said it hurt. The prosecutor asked, "How many times did something like this happen in the RV?" She said, "I think, like - like, twice or something." She did not immediately report these incidents. She eventually told child 4 about them.
Julian's counsel asked child 2 about a Child Abuse Interdisciplinary Team (CAIT) interview where child 2 told the interviewer that Julian first sexually assaulted her when she "was still eight." Counsel asked child 2, "[I]f you had said that, would that have been true or would that have been a lie?" She said, "I think it would have been a lie." Counsel asked her about another conflict between her sexual penetration testimony and her CAIT interview. The CAIT interviewer asked her whether Julian had "ever gone inside that part? [Her]
*882private area?" She responded, "Um no, not really." Child 2 testified her answer to the interviewer's question was not "the truth." During the CAIT interview, she said Julian had her "sit in his lap every time." She testified, "That would have been mostly true and a little bit of a lie." She said, "[H]e didn't have me sit on his lap when we were in the RV." Her statement to the CAIT interviewer that Julian did not touch inside her "private area" while in the RV "was a lie." After talking with her mother, she remembered more details than she mentioned in her initial CAIT interview.
The Testimony of Child 2's Sisters
Child 1, eight years old, testified she and her sisters played hide and seek with Julian. She and child 4 "would look," and Julian, child 2, and child 3 would hide. She did not remember Julian doing anything that made her "feel uncomfortable." He did not do anything that made her sisters uncomfortable. Julian did not touch her "in her private area" and she did not "ever see" Julian touch child 2 in that area. She did not remember child 2 telling her that Julian did something to make her feel uncomfortable.
Child 3, seven years old, testified that when they played hide and seek she, Julian, and child 2 would hide in an RV, and child 1 and child 4 would try to find them. She was a "look out" for child 1 and child 4 in the top bed of the RV. Julian and child 2 were hiding in a bed in the back of the RV. Child 3 saw Julian "play a tickling game" with child 2 on her armpits and neck. She did not see Julian touch child 2 in her private parts. She did not remember child 2 ever telling her that Julian did something to her or that she was afraid of Julian. She did not see Julian touch child 1 or child 4 inappropriately.
Child 4, 12 years old, testified Julian had a 20- or 30-minute time limit for hide and seek. He and her two sisters would hide in the RV, her father's shop, his truck, and the forest. One time while playing hide and seek, she saw Julian and child 2 in a bed in the RV. Child 2's face was a little "paler than" normal. Child 4 did not see Julian touch child 1, child 2, or child 3 in an inappropriate way. Sometime in 2016, child 2 said Julian did something to her. Child 4 testified, "I thought she was lying. ... I didn't think he could do something like that." Child 2 did not want child 4 to tell her mother. The next time child 2 mentioned this, child 4 "talked [child 2] into telling [her] mom."
Urquiza's Expert Testimony
Anthony Joseph Urquiza, a clinical psychologist, testified about the CSAAS theory. CSAAS dispels myths people have about the reactions children have to sexual *521abuse, including the myths that: 1) children are sexually abused by strangers, 2) they can escape the abuse "environment," *8833) they would disclose abuse "right away," and 4) they "will be significantly distressed." Most sexually abused children are "sexually abused by somebody" they know. They do not report abuse immediately. Fear motivates them to keep the abuse secret. They learn "the ability to cope" with the abuse. "They submit to the experience."
Statistical Data on False Allegations of Child Sexual Abuse
After presenting CSAAS evidence, the People introduced a new issue - the statistical percentage of false allegations by child sexual abuse victims. Urquiza testified false allegations by children "don't happen very often." "The range of false allegations that are known to law enforcement or [Child Protective Services] ... is about as low as one percent of cases to a high of maybe 6, 7, 8 percent of cases that appear to be false allegations ." (Italics added.) Julian's trial counsel did not object to this testimony.
Urquiza testified one study showed that of the 4 percent of cases where there are false allegations, the "largest subgroup" involved "some type [of] custodial dispute." He also said that research bears out that false allegations are "very infrequent, or rare. " (Italics added.)
On cross-examination, Julian's counsel asked Urquiza to name the studies that supported his claim that "false reports of sexual abuse of children are rather rare." Urquiza said a Canadian study supported the claim and it was "a good methodological study." Julian's counsel asked about the data used in the Canadian study to reach the conclusion that "false allegations provided to law enforcement and other authorities range from one to seven to eight percent ." (Italics added.) Urquiza said there may be data limitations in that study, but "the body of research supports" the claim that false allegations are rare and "very" infrequent. He stated that there are 12 studies that show a "[false statement] rate that is as low as one percent or as high as about six or seven or eight percent ." (Italics added.)
Julian's counsel asked about a "Peters article." Urquiza said this involved emergency room reports by "medical personnel" of rape or sexual assault victims. In "six percent of those cases ," there "was some determination of a false allegation." (Italics added.) He said a Denver Department of Social Services study found that in 551 cases, only "2.5 percent ... have been false allegations. " (Italics added.)
When asked by Julian's counsel about an article regarding false abuse allegations when parents separate, Urquiza reiterated that he did not rely on one study. He looked at a "dozen studies" that supported a "one to six *884percent" false allegation rate. He responded to further questioning by repeating the claim that "the best research that we have is that false allegations do occur but they happen infrequently or rarely ...." "[A]ll [of the studies] fall relatively close to each other" in concluding false allegations "happen very infrequently or rarely ." (Italics added.) When Julian's counsel asked about the Denver study, Urquiza said, "[D]o kids make false allegations of sexual abuse? Certainly. Do they make it very often? No ." Counsel responded, "Thank you for stating that once again ."
Julian's counsel asked Urquiza about a book suggesting there are multiple methods to determine false allegations. Urquiza said that "you can have error in just about everything," but there are "multiple studies" reaching the same conclusion on false allegations. When asked about a "Trocme *522& Bala" report, Urquiza said he initially thought it showed a 4 percent rate for false allegations. But it actually showed a "five percent " rate. (Italics added.)
On redirect, the prosecutor asked about the Trocme & Bala report. Urquiza said, "[T]here was not a single instance of a false reported case in that article where the child was the one saying I was abused." (Italics added.) When Julian's counsel also asked about that report's "zero percent" rate of false allegations, Urquiza responded there were 798 cases; 43 "were determined to have been false," but in none of those 43 cases was a child the source of the false information.
Detective Menghrajani's Testimony
Police Detective Devashish Menghrajani testified the detectives decided not to conduct a Sexual Assault Response Team (SART) examination of child 2. Child 2's disclosure was beyond a "72 to 96" hour "guideline" for such exams. Julian's counsel asked, "[D]o you believe [child 2] was honest with you?" Menghrajani responded, "Yes, Sir."
The Defense Case
Linai Poland testified she was a nanny for the family. She observed the "interaction" between the girls and Julian. She did not "ever see [Julian] touch any of the girls inappropriately." None of the girls ever complained to her that he acted inappropriately. Child 2 would tell her about things "she wanted to complain about."
Matt Aanerud, a District Attorney investigator, monitored "[i]n excess of a hundred" phone calls Julian made in jail. Julian's counsel asked regarding those calls, "[D]id my client ever tell anyone to destroy any evidence?" Aanerud said, "I do not recall ever hearing that."
*885Julian testified that child 1, child 2, child 3, and child 4 "would come and get [him]" to see if he "had time before work to ... play games with them, be it tag, hide-and-seek, anything they would come up with." Between June 2015 and August 23, 2016, he played games with them "probably ... 200 times." Child 2's claim that he penetrated "her private areas" is not true. Her claim that he "penetrated her private areas in an RV" is not true. He did not molest her. He cooperated with the police in their investigation because he did not "have anything to hide."
DISCUSSION
Julian contends Urquiza's statistical probability testimony went beyond the permissible scope of CSAAS evidence, was highly prejudicial, and deprived him of his right to a fair trial. We agree.
Urquiza testified in the People's case regarding CSAAS. During that testimony, he said false allegations of sexual abuse by children "don't happen very often." "The range of false allegations that are known to law enforcement or [Child Protective Services] ... is about as low as one percent of cases to a high of maybe 6, 7, 8 percent of cases that appear to be false allegations." (Italics added.)
Expert testimony on "the common reactions of child molestation victims," known as CSAAS theory evidence, "is admissible to rehabilitate such witness's credibility when the defendant suggests that the child's conduct after the incident - e.g., a delay in reporting - is inconsistent with his or her testimony claiming molestation." ( People v. McAlpin (1991) 53 Cal.3d 1289, 1300, 283 Cal.Rptr. 382, 812 P.2d 563.) " 'Such expert testimony is needed to disabuse jurors of commonly held misconceptions about child sexual abuse, and to explain the emotional *523antecedents of abused children's seemingly self-impeaching behavior.' " ( Id. at p. 1301, 283 Cal.Rptr. 382, 812 P.2d 563.)
But such evidence "is not admissible to prove that the complaining witness has in fact been sexually abused." ( People v. McAlpin , supra , 53 Cal.3d at p. 1300, 283 Cal.Rptr. 382, 812 P.2d 563.) "The expert is not allowed to give an opinion on whether a witness is telling the truth ...." ( People v. Long (2005) 126 Cal.App.4th 865, 871, 24 Cal.Rptr.3d 654.)
The expert providing CSAAS testimony may not give " 'general' testimony describing the components of the syndrome in such a way as to allow the jury to apply the syndrome to the facts of the case and conclude the child was sexually abused." ( *886People v. Bowker (1988) 203 Cal.App.3d 385, 393, 249 Cal.Rptr. 886.) Nor is it proper for an expert to present "predictive conclusions" ( ibid. ), such as alleged child abuse victims "should be believed" or "abused children give inconsistent accounts and are credible nonetheless." ( Id. at p. 394, 249 Cal.Rptr. 886.) Such predictive conclusions go beyond the scope of CSAAS evidence and may confuse the jury. "[T]he jurors' education and training may not have sensitized them to the dangers of drawing predictive conclusions." ( Id . at p. 393, 249 Cal.Rptr. 886.) Where expert opinions on the statistical probability of guilt are admitted, the jury may be "distracted" from its "requisite function of weighing the evidence on the issue of guilt," and may rely instead on this "irrelevant" evidence. ( People v. Collins (1968) 68 Cal.2d 319, 327, 66 Cal.Rptr. 497, 438 P.2d 33.)
Here the jury had to decide between the credibility of child 2's testimony and Julian's. But Urquiza's 92 to 99 percent probability evidence invited jurors to presume Julian was guilty based on statistical probabilities, and not decide the evidence properly introduced in the case. ( People v. Collins , supra , 68 Cal.2d at p. 327, 66 Cal.Rptr. 497, 438 P.2d 33.) Urquiza's statistics were not admissible as CSAAS evidence. ( People v. Bowker , supra , 203 Cal.App.3d at pp. 393-394, 249 Cal.Rptr. 886.)
This statistical probability evidence deprived Julian of his right to a fair trial. ( Snowden v. Singletary (11th Cir. 1998) 135 F.3d 732, 739.) In Snowden , the defendant was convicted of five counts of child abuse. The Eleventh Circuit vacated that conviction. At trial an expert witness testified that "child witnesses in sexual abuse cases tell the truth" 99.5 percent of the time. ( Id. at p. 738.) The court said, "That such evidence is improper, in both state and federal trials, can hardly be disputed." ( Ibid . ) "The jury's opinion on the truthfulness of the children's stories went to the heart of the case." ( Ibid . ) "Witness credibility is the sole province of the jury." ( Id . at p. 739.) Allowing this expert testimony to "boost the credibility of the main witness against [the defendant]" resulted in a "fundamentally unfair" trial. ( Ibid. )
In Powell v. State (Del. 1987) 527 A.2d 276, an expert in a child sexual abuse case testified that 99 percent "of the alleged victims involved in sexual abuse treatment programs in which she was ... involved 'have told the truth.' " ( Id. at p. 278.) The appellate court said, "The admission of Cantor's percentage testimony deprived Powell of his right to have his fate determined by a jury making the credibility determinations, so clearly crucial in these cases, without guidance from an expert, in stark mathematical terms, bolstering the credibility of the complainant and thereby impugning his credibility." ( Id. at pp. 279-280.)
In State v. Myers (Iowa 1986) 382 N.W.2d 91, 92, an expert testified that *524statistics show that "it is very rare for a child to lie" about sexual abuse. In reversing the conviction, the appellate court said, "We believe the effect of *887the opinion testimony was to improperly suggest the complainant was telling the truth and, consequently, the defendant was guilty. We conclude the opinion testimony crossed that 'fine but essential' line between an 'opinion which would be truly helpful to the jury and that which merely conveys a conclusion concerning defendant's legal guilt.' " ( Id. at pp. 97-98.)
In Wilson v. State (Tex. Ct.App. 2002) 90 S.W.3d 391, 393, the court held it was error to admit expert testimony about false allegations being only 2 to 8 percent in a child sexual assault case. It said this statistical evidence "did not aid, but supplanted, the jury in its decision on whether the child complainant's testimony was credible." ( Ibid . )
Courts from other jurisdictions have reached the same conclusion. ( State v. Lindsey (1986) 149 Ariz. 472, 720 P.2d 73, 77 [expert's "[q]uantification of the percentage of witnesses who tell the truth" in incest cases "usurps the function of the jury"]; State v. W.B. (2011) 205 N.J. 588, 17 A.3d 187, 202 ["Statistical information quantifying the number or percentage of abuse victims who lie deprives the jury of its right and duty to decide the question of credibility of the victim"]; State v. MacRae (1996) 141 N.H. 106, 677 A.2d 698, 702 [expert testimony was inadmissible "because it improperly provided statistical evidence that the victim more probably than not had been abused"]; U.S. v. Brooks (C.A.A.F. 2007) 64 M.J. 325, 329 [expert's testimony that 5 percent of child sex victims lie about abuse was inadmissible]; State v. Williams (Mo. Ct.App. 1993) 858 S.W.2d 796, 801 [doctor's testimony that incidents of children lying about sexual abuse is "less than three percent" was inadmissible as an "improper quantification of the probability of the complaining witness'[s] credibility"]; Lawrence v. State (Okla. Ct.App. 1990) 796 P.2d 1176, 1177 [social worker's testimony that 10-year-old children tell the truth was inadmissible and its admission constituted reversible error]; State v. Vidrine (La. Ct.App. 2009) 9 So.3d 1095, 1111 [expert's "testimony regarding the statistical probability of false reporting ... of rape cases was irrelevant to the charges at hand and was clearly offered for the sole purpose of bolstering the credibility of [the minor]"].)
Collins , Bowker , Snowden, and the other authorities cited show that Urquiza's statistical evidence on false allegations should not have been introduced.
Ineffective Assistance of Counsel
The People note that Julian's trial counsel did not object to this evidence. Julian responds that his counsel was ineffective in "failing to object." We agree.
*888In deciding ineffective assistance, we determine whether counsel's failure to object fell below the standard required for reasonably competent attorneys and whether counsel's performance was prejudicial to the defendant's case. ( Strickland v. Washington (1984) 466 U.S. 668, 686-692, 104 S.Ct. 2052, 2064-2067, 80 L.Ed.2d 674, 692-696.) Here there is no justification for counsel's failure to object to Urquiza's statistical evidence on false allegations. It was inadmissible and it improperly suggested Julian was guilty based on statistical probabilities that were irrelevant to this case.
The evidence also was highly prejudicial because this case was a credibility dispute between child 2's testimony and Julian's. It was a heavily contested case with strong defense evidence. Julian denied child 2's claims. He said he cooperated with police *525because he had nothing to hide. Menghrajani confirmed that he had cooperated. Poland never saw Julian touch any of the girls "inappropriately" and she said none of the girls ever complained that he engaged in such conduct. Aanerud said child molesters frequently possess child pornography. The police did not find evidence that Julian possessed such material. Aanerud monitored more than 100 of Julian's jail phone calls. He could not recall one where Julian ever requested anyone to destroy evidence.
Child 1 testified Julian did not touch her inappropriately and she did not see him touch child 2 in that manner. She did not recall child 2 ever saying Julian did something to make her "feel uncomfortable." Child 3 and child 4 did not see Julian inappropriately touch child 2 or their other sisters.
When child 2 first told child 4 about her claims, child 4 believed child 2 "was lying." Child 4 felt Julian could not have done what child 2 claimed. There was no SART report to corroborate child 2's claims and no eyewitnesses. Child 2 had difficulty remembering certain facts, gave some tentative responses, and some of her testimony was introduced with leading questions. The conflicts between child 2's trial testimony and a CAIT interview raised credibility issues. During closing argument, the People conceded that child 2's CAIT interviews "were very different from her testimony" and there were "some serious inconsistencies ." (Italics added.)
But Urquiza's statistical evidence tipped the scales in favor of the People based on statistical studies that were irrelevant to the issue of Julian's guilt or innocence. It distracted the jury from its duty to decide the properly admitted evidence. ( People v. Collins , supra , 68 Cal.2d at p. 327, 66 Cal.Rptr. 497, 438 P.2d 33.) Such evidence may not be prejudicial where it occurs in a slight passing reference by the expert. But here the jury was bombarded with it.
Julian's counsel cross-examined Urquiza. But Urquiza used that opportunity to repeatedly reassert his claim that statistics show children do not lie *889about being abused. His counsel's questions about multiple studies only opened the door to a mountain of prejudicial statistical data that fortified the prosecutor's claim about a statistical certainty that defendants are guilty. ( In re Jones (1996) 13 Cal.4th 552, 571, 54 Cal.Rptr.2d 52, 917 P.2d 1175.)
Moreover, in closing argument, the prosecutor asked the jury to rely on Urquiza's statistical evidence that "children rarely falsify allegations of sexual abuse." He reminded jurors that Urquiza "quoted a Canadian study for over 700 cases, not a single one where there was a false allegation ." (Italics added.) The claim that there is a zero percent chance children will fabricate abuse claims replaced the presumption of innocence with a presumption of guilt.
In his closing argument, Julian's counsel discussed his position regarding Urquiza's testimony about the "12 studies," the Canadian study, the Trocme & Bala study, a social worker study showing "four percent or five percent" as false allegations, and the prosecutor's claim that "false allegations are very rare." When he discussed the statistical percentage of false allegations in a study called "false allegations of sexual abuse of children and adolescents," the prosecutor objected. The court stopped the argument for a 15-minute recess. When the jury returned, the court instructed jurors that there was "a disagreement" by counsel about "a certain study." The jury should decide the issue based on the evidence introduced about the study, not what the lawyers remember about it. Consequently, the jurors' attention was directed, once again, to the statistical study *526evidence right before they began their deliberations.
But the jury's duty to decide the facts does not include considering inadmissible statistical information ( People v. Collins , supra , 68 Cal.2d at p. 327, 66 Cal.Rptr. 497, 438 P.2d 33 ) or using studies of statistical odds to determine guilt. ( Ibid. ; Snowden v. Singletary , supra , 135 F.3d at p. 739.) Julian did not receive a fair trial. ( Strickland v. Washington , supra , 466 U.S. at pp. 686-687, 104 S.Ct. 2052.)
Soliciting the Detective's Opinion on Child 2's Credibility
Julian claims he was prejudiced by his counsel's questioning of Menghrajani. Counsel asked Menghrajani if child 2 was "honest with you" in making her claims. The detective answered, "Yes, Sir." This solicited inadmissible opinion evidence, which bolstered the People's case and undermined Julian's defense. ( In re Jones , supra , 13 Cal.4th at p. 571, 54 Cal.Rptr.2d 52, 917 P.2d 1175 ; People v. Long , supra , 126 Cal.App.4th at p. 871, 24 Cal.Rptr.3d 654 ; People v. Sergill (1982) 138 Cal.App.3d 34, 41, 187 Cal.Rptr. 497.)
In Sergill , the defendant was charged with a sexual offense on a child. Police officers testified they believed the child was truthful. The Court of *890Appeal held "the officers' opinions on the child's truthfulness during their limited contacts with her did not have a reasonable tendency to prove or disprove her credibility and were therefore not relevant." ( People v. Sergill , supra, 138 Cal.App.3d at p. 40, 187 Cal.Rptr. 497, italics omitted.) It reversed the judgment of conviction. The court said the trial court's ruling that the officers were qualified to render the opinion and the officers' opinions that they believed the child "may well have caused the jury to place undue emphasis on the officers' testimony," resulting in "the usurpation of the jury's function as fact finder." ( Id. at p. 41, 187 Cal.Rptr. 497.)
It is beyond question that the errors here were prejudicial by any standard. ( Chapman v. California (1967) 386 U.S. 18, 87 S.Ct. 824, 17 L.Ed.2d 705 ; People v. Watson (1956) 46 Cal.2d 818, 299 P.2d 243.) The myriad reasons this happened include the detective's improper opinion testimony, Urquiza's inadmissible statistical evidence, the prosecutor's invitation that the jury consider the Canadian statistical study in deliberations, and defense counsel's failure to object. As we pointed out in People v. Cowan (2017) 8 Cal.App.5th 1152, 214 Cal.Rptr.3d 576, flagrant errors of this type do not serve the public interest or the cause of justice.
DISPOSITION
The judgment is reversed and the case is remanded to the trial court for a new trial.
We concur:
PERREN, J.
TANGEMAN, J.