OPINION

KING, J.

In this case we hold that the psychological technique of detecting child sexual abuse by observing the child's behavior with anatomically correct dolls and analyzing the child's reports of abuse constitutes a new scientific method of proof, and therefore is admissible in court only upon a showing that the technique has been generally accepted as reliable in the scientific community in which it was developed.

Ron B. appeals from an order declaring his daughters Amber and Teela to be dependent children of the juvenile court (Welf. & Inst. Code, § 300). We reverse.

A petition filed by the Solano County Department of Social Services (Department) alleged that Ron had sexually molested three-year-old Amber and that one-year-old Teela was at risk of sexual abuse. At the hearing on the petition the Department presented testimony by a psychologist, Dr. Henry Raming, who had examined Amber on three occasions. Over Ron's objection, the court permitted Dr. Raming to testify that, in his opinion, Amber had been sexually molested and she believed she had been molested by her father.

Dr. Raming's opinion was based on two factors. The first was the nature of Amber's reports of abuse, in which she described instances of abuse in varying ways. According to Dr. Raming, it is "fairly well documented in the literature ... that children who have been molested will talk about being abused, but they will do this by consistently giving the ... same facts or the essence in different words such that they have an event or an experience in their minds and are not merely repeating ... rote by rote, someone else's words...."

The second factor was the nature of Amber's behavior with an anatomically correct female doll in Dr. Raming's office. During two of Dr. Raming's examinations Amber placed her index finger in the vaginal and anal openings of the doll and pushed and twisted her finger vigorously. According to Dr. Raming, Amber's behavior with the doll "is fairly consistent with molested children. This is not the usual type of behavior one would see in children who are in a stage of age appropriate sex exploration.... [W]hen children this age describe or graphically demonstrate anal or vaginal penetration, it's pretty much assumed that the child learned that from experience and not from ... sex exploration with other children."

The only other witnesses presented by the Department, a police detective and a social worker, also testified regarding the nature of Amber's reports of abuse and her behavior with an anatomically correct doll. Amber did not testify. Ron testified in his own behalf and denied the abuse.

The court ruled that Amber had been molested while in the custody of her parents, that Amber believed she had been molested by her father, and that Teela was at risk of sexual abuse because of the sexual abuse of Amber. The court did not sustain the allegation identifying Ron as the perpetrator of the abuse. [1] In its dispositional order the court declared both children to be dependent children of the juvenile court, placed them in the custody of their mother, afforded Ron supervised visitation, and required counseling for Amber and both parents.

(1a) Ron challenges the admission of Dr. Raming's opinion testimony that Amber had been molested and she believed she had been molested by her father. He contends the opinion was based on a new scientific method of proof the analysis of Amber's reports of abuse and behavior with anatomically correct dolls that had not been shown to satisfy the Kelly-Frye test of admissibility.

(2a) Under the Kelly-Frye rule, evidence based on a new scientific method of proof is admissible only upon a showing that the procedure has been generally accepted as reliable in the scientific community in which it was developed. (3) (See fn. 2.) People v. Kelly (1976) 17 Cal.3d 24 [ 130 Cal. Rptr. 144 , 549 P.2d 1240 ]; Frye v. United States (D.C. Cir.1923) 293 Fed. 1013 [ 34 A.L.R. 145 ].) [2] (2b) The test is usually applied to novel devices or processes involving the manipulation of physical evidence, such as lie detectors, experimental systems of blood typing, voiceprints, identification by human bite marks, and microscopic analysis of gunshot residue. (See People v. McDonald (1984) 37 Cal.3d 351, 373 [ 208 Cal. Rptr. 236 , 690 P.2d 709 , 46 A.L.R. 4th 1011 ].)

The Kelly-Frye test is not, however, limited to techniques involving the manipulation of physical evidence. In a recent decision applying the test to hypnosis, the California Supreme Court said, "[W]e do not doubt that if testimony based on a new scientific process operating on purely psychological evidence were to be offered in our courts, it would likewise be subjected to the Frye standard of admissibility." ( People v. Shirley (1982) 31 Cal.3d 18, 53 [ 181 Cal. Rptr. 243 , 641 P.2d 775 ].)

The purpose of the Kelly-Frye test is to prevent factfinders from being misled by the "aura of infallibility" that may surround unproven scientific methods. ( People v. McDonald, supra, 37 Cal.3d at pp. 372-373.) "[L]ike many laypersons, jurors tend to ascribe an inordinately high degree of certainty to proof derived from an apparently `scientific' mechanism, instrument, or procedure." ( Id., at p. 372.)

The test does not apply to mere expert testimony as distinguished from scientific evidence. "When a witness gives his personal opinion on the stand even if he qualifies as an expert the jurors may temper their acceptance of his testimony with a healthy skepticism born of their knowledge that all human beings are fallible." ( Ibid. ) In contrast, factfinders may not view scientific evidence with such skepticism, and indeed will "tend to ascribe an inordinately high degree of certainty" to such evidence. ( Ibid. )

(1b) The fundamental issue presented here is whether the technique employed by Dr. Raming for detecting child sexual abuse is a new scientific method of proof which must satisfy the Kelly-Frye test of admissibility, or whether Dr. Raming simply gave expert testimony to which the test does not apply.

In admitting Dr. Raming's testimony the trial court adhered strictly to In re Cheryl H. (1984) 153 Cal. App.3d 1098 [ 200 Cal. Rptr. 789 ]. In that case a psychiatrist testified that in her opinion three-year-old Cheryl had been sexually abused by her father. This opinion was based upon psychological tests and Cheryl's behavior at personal interviews, including Cheryl's sexual play with anatomically correct dolls and her invention of new names for genitalia. ( Id., at pp. 1109-1110.) The appellate court held it was proper to admit the opinion that Cheryl had been abused, because Cheryl's postinjury behavior "appears to be unique to children subjected to child abuse.... The child played with male and female dolls in a way only children who have been sexually abused ordinarily do. She also used words and demonstrated anxiety symptoms characteristic of those who have been sexually abused." ( Id., at p. 1117.)

The Cheryl H. court held it was error, however, to admit the opinion that Cheryl's father was the person who committed the abuse, because this testimony impermissibly drew inferences about conduct by a third party based primarily on hearsay. ( Id., at pp. 1118-1120.) The court said the psychiatrist should only have been permitted to express an opinion that Cheryl believed her father was the one who abused her. ( Id., at p. 1119.) Accordingly, in the present case the trial court permitted Dr. Raming to testify only that Amber had been abused and she believed her father was the one who abused her.

The Cheryl H. opinion did not discuss the potential applicability of the Kelly-Frye rule. The court simply accepted the psychiatrist's analysis of Cheryl's behavior as demonstrating prior sexual abuse. The court drew an analogy to People v. Jackson (1971) 18 Cal. App.3d 504 [ 95 Cal. Rptr. 919 ], which characterized as admissible expert testimony a diagnosis of "battered child syndrome" based on the presence of certain types of physical injuries. (153 Cal. App.3d at pp. 116-117.)

Shortly after Cheryl H. the California Supreme Court decided People v. Bledsoe (1984) 36 Cal.3d 236 [ 203 Cal. Rptr. 450 , 681 P.2d 291 ], which applied Kelly-Frye to testimony pertaining to the closely analogous concept of "rape trauma syndrome" and found the evidence inadmissible. The decision in Bledsoe calls for careful consideration of whether evidence of the sort admitted in Cheryl H. and the present case should be subject to the Kelly-Frye test.

The concept of rape trauma syndrome is based on the theory that after a rape the victim will follow a set pattern of reaction to the trauma of the rape. (36 Cal.3d at pp. 249-243.) The court in Bledsoe , applying the Kelly-Frye test, disapproved the admission of a counselor's opinion testimony that the victim was suffering from rape trauma syndrome, because the theory was not relied upon in the relevant scientific community for the purpose employed by the prosecutor. The theory was developed as a therapeutic tool, not as a scientifically reliable means of determining whether a rape occurred. (36 Cal.3d at pp. 249-251.) The court also distinguished cases in which recent findings of professional research on the subject of victim reaction to sexual assault (e.g., delay in reporting an assault) are introduced solely to rebut misconceptions about the presumed behavior of rape victims. ( Id., at pp. 247-248.) The court did not discuss why the Kelly-Frye rule applied, but simply quoted the prior statement in Shirley that Kelly-Frye would apply to "testimony based on a new scientific process operating on purely psychological evidence." ( 36 Cal.3d at p. 247, fn. 7 , quoting People v. Shirley, supra, 31 Cal.3d at p. 53 .)

The method of proof relied upon by Dr. Raming is closely analogous to that in Bledsoe , because in both cases the expert witnesses relied upon a psychological analysis of a subject's behavior to determine whether sexual abuse had previously occurred. Thus the decision in Bledsoe strongly suggests evidence of the sort admitted here and in Cheryl H. should be subject to Kelly-Frye.

Two post- Bledsoe decisions have addressed the issue, but neither has resolved it.

In the first case, People v. Roscoe (1985) 168 Cal. App.3d 1093 [ 215 Cal. Rptr. 45 ], the court said that under Bledsoe a psychologist could not testify whether he believed the complainant had been the victim of sexual abuse, but could discuss victims as a class in an attempt to rehabilitate the complainant's credibility. ( Id., at p. 1100.) The court did not explain why Bledsoe and Kelly-Frye were applicable, other than to comment that the reasoning pertaining to rape trauma syndrome in Bledsoe "seems to apply to diagnosis as a victim of child molestation." ( Id., at p. 1098.) The court also commented, "Less strict rules of admissibility apply where child abuse is an issue in noncriminal cases, such as Welfare and Institutions Code section 300 dependency proceedings." ( Id., at p. 1100, fn. 4.)

The second decision, People v. Gray (1986) 187 Cal. App.3d 213 [ 231 Cal. Rptr. 658 ], declined to determine the Kelly-Frye issue. The court held the challenged evidence was admissible for rebuttal as set forth in Roscoe because the psychologist confined his testimony to child molest victims as a class and did not render an opinion whether the victim in that case had been molested. ( Id., at pp. 219-220.)

Thus both post- Bledsoe decisions could avoid determining the applicability of Kelly-Frye. The issue cannot be avoided here, because Dr. Raming gave a direct opinion that Amber had been molested, and the fundamental importance of his testimony precludes a determination that any error was harmless.

Three preliminary questions must be addressed before reaching the heart of the Kelly-Frye issue. (4) The first is whether, as claimed in Roscoe , less strict rules of admissibility apply in dependency proceedings under Welfare and Institutions Code section 300 than in criminal cases. The court in Roscoe cited no authority for this claim, and it does not survive scrutiny with regard to the applicability of Kelly-Frye. Welfare and Institutions Code section 355 states that a finding of dependency must be supported by evidence that is "legally admissible in the trial of civil cases." (See also Cal. Rules of Court, rule 1365(c).) The Kelly-Frye rule has been applied in civil cases in California. ( Huntingdon v. Crowley (1966) 64 Cal.2d 647 [ 51 Cal. Rptr. 254 , 414 P.2d 382 ].) The rule therefore applies in dependency proceedings.

(5) The second preliminary question is whether the objection to Dr. Raming's testimony was adequate to preserve the Kelly-Frye question for appellate review. (See People v. Bledsoe, supra, 36 Cal.3d at pp. 246-247.) This question is easily answered: counsel expressly sought exclusion of the testimony on the ground "there's ... no really established scientific evidence that the type of opinion they're [ sic ] giving meets the Frye standard for admissibility...."

The third preliminary question is whether Dr. Raming's testimony suffered from the same deficiency as in Bledsoe , that of being based on a method of therapy rather than a means of determining whether sexual abuse had occurred. In this respect the present case appears to differ from Bledsoe . Counsel below pointed out that Dr. Raming was appointed by the court not to perform therapy but to conduct a psychological evaluation. Dr. Raming's testimony indicates he had sought to determine whether Amber had been sexually abused. (Cf. Comment, The Admissibility of "Child Sexual Abuse Accommodation Syndrome" in California Criminal Courts (1986) 17 Pacific L.J. 1361, 1381 [stating that theory of "child sexual abuse accommodation syndrome" is "used in the mental health field to determine if sexual abuse has occurred"].) [3]

(1c) This brings us to the heart of the Kelly-Frye issue: is the technique employed by Dr. Raming for detecting child sexual abuse a new scientific method of proof which must satisfy Kelly-Frye, or did Dr. Raming simply provide expert testimony? The answer to this question is elusive. Kelly-Frye is clearly applicable where evidence is produced by a machine such as a lie detector, or by a physical testing procedure such as blood typing. The cases have provided little guidance, however, for determining at what point evidence based upon a psychological analysis of behavior transcends expert testimony and becomes scientific evidence. Indeed, one of the recognized weaknesses of the Kelly-Frye test is the difficulty in defining what types of evidence should be classified as scientific. (Giannelli, The Admissibility of Novel Scientific Evidence: Frye v. United States, a Half-Century Later (1980) 80 Colum.L.Rev. 1197, 1219.)

One treatise states the Kelly-Frye rule encompasses psychological "profiles" which, as in the present case, are based on studies showing a correlation between certain traits and characteristics and certain forms of behavior. (McCormick on Evidence (1984) pp. 634-636 & fn. 99.) [4] The treatise does not, however, explain why this is so.

Rather than attempting to determine the presence of a "new scientific method of proof" based upon factors pertaining to the nature of the challenged procedure a difficult task in cases involving psychological analysis of behavior it makes more sense to decide this issue by reference to the purpose of the Kelly-Frye rule. That purpose is to prevent factfinders from being misled by the "aura of infallibility" that may surround unproven scientific methods. ( People v. McDonald, supra, 37 Cal.3d at pp. 372-373.) Accordingly, a scientific procedure, technique or theory should be characterized as a "new scientific method of proof," subject to Kelly-Frye, if factfinders would "tend to ascribe an inordinately high degree of certainty" to it. (See id., at p. 372.)

The determinative question in the present case thus becomes: would a trier of fact ascribe an inordinately high degree of certainty to the technique employed by Dr. Raming for determining the occurrence of child sexual abuse?

We conclude that the practice of detecting child sexual abuse by (1) observing a child's behavior with anatomically correct dolls, and (2) analyzing the child's reports of abuse, is what Shirley characterizes as "a new scientific process operating on purely psychological evidence" ( 31 Cal.3d at p. 53 ) and is subject to the Kelly-Frye test. The specific causes of age-inappropriate child sexual behavior, and indeed the entire field of child sexuality since the theories of Sigmund Freud, are beyond the scope of critical analysis by the average lay person. Thus a psychologist's examination and analysis employing the technique used by Dr. Raming may be surrounded by an "aura of infallibility," and a trier of fact would tend to ascribe "an inordinately high degree of certainty" to the technique. ( People v. McDonald, supra, 37 Cal.3d at p. 372 .) Unlike with expert testimony where a witness gives a personal opinion, triers of fact are in no position to temper their acceptance of the psychological evidence "with a healthy skepticism born of their knowledge that all human beings are fallible." ( Ibid. )

The trial court therefore erred when it failed to require a showing of general acceptance in the relevant scientific community in accordance with Kelly-Frye.

For this reason the dependency order must be reversed. Because of the court's ruling that Dr. Raming could testify pursuant to Cheryl H., the Department made no attempt to establish general acceptance in the relevant scientific community. In cross-examining Dr. Raming, Ron's counsel briefly explored the question whether there was any disagreement among the authorities upon which Dr. Raming relied, but Dr. Raming's general responses were far from sufficient to satisfy Kelly-Frye. (See People v. Shirley, supra, 31 Cal.3d at pp. 55-56; People v. Kelly, supra, 17 Cal.3d at p. 37 ; People v. John W. (1986) 185 Cal. App.3d 801, 806 [ 229 Cal. Rptr. 783 ].) The error cannot be characterized as harmless, because the Department's entire case was based upon Dr. Raming's analysis of Amber's reports of abuse and her behavior with anatomically correct dolls. Any retrial must proceed in accordance with the requirements of Kelly-Frye.

We recognize that unique problems of proof are presented whenever it is claimed that a young child was the victim of sexual abuse by a parent. The child may be incompetent to testify, the act of testifying could be as traumatic as the abuse itself, and the child may be subjected to strong parental influences. The trial judge, the parties, and society in general would greatly benefit from reliable expert testimony addressing the question whether the child was abused, either in lieu of or in addition to testimony by the child. However, we cannot approve such expert testimony until its reliability is shown pursuant to Kelly-Frye. [5]

The order is reversed.

A petition for a rehearing was denied May 21, 1987, and respondent's petition for review by the Supreme Court was denied July 30, 1987.

[1] In a letter sent to the court before the hearing, Amber's mother reported that her son had described instances of possible sexual abuse of Amber by an aunt.

[2] The proponent of the evidence must also show that the witness furnishing the expert testimony is qualified as an expert to give an opinion on the subject, and must demonstrate that correct scientific procedures were used in the particular case. ( People v. Kelly, supra, 17 Cal.3d at p. 30 .)

[3] "Child sexual abuse accommodation syndrome" describes the purportedly typical behavior of children who have been victims of repeated sexual abuse by a family member or an adult with whom the child has a trusting relationship. It is comprised of one or more of five elements: secrecy, helplessness, entrapment and accommodation, delayed disclosure, and retraction. (Comment, The Admissibility of "Child Sexual Abuse Accommodation Syndrome" in California Criminal Courts, supra, at pp. 1364-1367.) Dr. Raming did not mention child sexual abuse accommodation syndrome or its elements. He confined his testimony to Amber's behavior with anatomically correct dolls and her reports of abuse. This case, therefore, does not present the issue whether the theory of child sexual abuse accommodation syndrome is subject to Kelly-Frye. The theory, however, seems quite similar to that of rape trauma syndrome, suggesting that, as in Bledsoe , the Kelly-Frye test would apply.

[4] The treatise criticizes the Kelly-Frye rule in favor of adherence to traditional standards of relevancy and the need for expertise. (McCormick on Evidence, supra, at pp. 605-609.)

[5] Our ruling is based solely on the need for further proceedings to establish the admissibility of the challenged testimony. Nothing in this opinion is intended to disturb the status quo with regard to the placement of the children pending retrial.