The defendant, Wayne Cressey, was convicted of three counts of aggravated felonious sexual assault, RSA 632-A:2 (1986 & Supp. 1989) (amended 1992), and one count of felonious sexual assault, RSA 632-A:3 (1986), after a jury trial in the Superior Court (Mohl, J.). On appeal, the defendant argues that the trial court erred in admitting the testimony of the State’s expert psychologist to prove that the child victims had been sexually abused. We reverse the convictions on this ground and, therefore, address the defendant’s other claims on appeal only to the extent that they are likely to arise again in a new trial.
The defendant and his wife, Betsy Cressey, were given custody of their nieces, Lisa, Julie, and Lorie, while living in Ossipee in 1983. The children’s father had been institutionalized since an automobile accident in 1981. The children’s mother was killed in another automobile accident in 1982. In 1985, Betsy Cressey also died from complications of injuries sustained in an automobile accident. Lisa, Julie, and Lorie remained in the defendant’s care, and he eventually remarried in 1987.
*404At trial, Lisa testified that the defendant began to sexually abuse her when she was approximately eight years old. The pattern of abuse that continued over the next six years included acts of sexual touching, fellatio, and intercourse. Lisa’s sister, Julie, also testified that when she was eleven years old she was involved in two sexual encounters with the defendant, which included sexual touching and digital penetration.
Lisa had not mentioned the defendant’s abusive acts to anyone until October 1989, when she disclosed them to a woman for whom she babysat. She was fourteen years old at the time. Lisa testified that the defendant had warned her not to tell anyone about the incidents because she would be separated from her sisters and he would go to jail. Julie had not disclosed any incidents of sexual abuse until February 1991, shortly before a trial was scheduled to begin on the charges involving Lisa. In her interviews "with police and counselors prior to that time, Julie had denied having sexual contact with the defendant.
The defendant’s first claim on appeal is that the trial court erred in admitting the testimony of the State’s expert witness, Dr. Kathleen Bollerud. Dr. Bollerud is an expert in the areas of psychology and child sexual abuse. She interviewed Lisa and Julie prior to trial, and her testimony was a substantial part of the State’s case-in-chief. She testified in general about the effects that sexual abuse has on children and about the symptoms and behaviors commonly exhibited by sexually abused children. She testified in particular about the interviewing techniques she used with Lisa and Julie and about her evaluation of each child. Ultimately Dr. Bollerud stated that the symptoms exhibited by each child were consistent with those of a sexually abused child.
To evaluate the admissibility of Dr. Bollerud’s expert testimony, we must return to the basic evidentiary rules regarding experts. New Hampshire Rule of Evidence 702 states:
“If scientific, technical, or other specialized knowledge will assist the trier of fact to understand the evidence or to determine a fact in issue, a witness qualified as an expert by knowledge, skill, experience, training, or education, may testify thereto in the form of an opinion or otherwise.”
Subsumed in the requirements of Rule 702 is the premise that expert testimony must be reliable to be admissible. For only if the expert’s testimony is reliable can it assist the jury in understanding the evidence or determining a fact in issue. The requirement that an ex*405pert’s testimony be reliable is reflected in the evidentiary practices of properly establishing an expert’s qualifications, see, e.g., State v. Coleman, 133 N.H. 713, 715-16, 584 A.2d 755, 757 (1990), and subjecting technical evidence to the scrutiny of the test set forth in Frye v. United States, 293 F. 1013 (D.C. Cir. 1923), see State v. Vandebogart (DNA), 136 N.H. 365, 373, 616 A.2d 483, 489 (1992).
We therefore recognize that an expert’s testimony must rise to a threshold level of reliability to be admissible under New Hampshire Rule of Evidence 702. Cf. Daubert v. Merrell Dow Pharmaceuticals, Inc., — U.S. —, —, 113 S. Ct. 2786, 2795-96 (1993). (We do not reach the issue addressed in Daubert, however, of whether the Frye test has been superseded in New Hampshire by our adoption of New Hampshire Rule of Evidence 702.) The reliability of evidence is of a special concern when offered through expert testimony because such testimony involves the potential risks that a jury may disproportionately defer to the statements of an expert if the subject area is beyond the common knowledge of the average person, and that a jury may attach extra importance to an expert’s opinion simply because it is given with the air of authority that commonly accompanies an expert’s testimony. The determination of whether particular expert testimony is reliable and admissible rests, in the first instance, within the sound discretion of the trial court. Cf. Johnston v. Lynch, 133 N.H. 79, 88, 574 A.2d 934, 939 (1990).
The importance of Dr. Bollerud’s testimony, coupled with the air of authority it gains from being presented by a highly qualified expert, requires us to scrutinize her methods and conclusions carefully in assessing the reliability of her statements. In reaching her conclusions, Dr. Bollerud relied on several different types of information. She interviewed each child, asking open-ended questions about their lives, backgrounds, family situations, and the alleged abuse. Dr. Bollerud then used a disassociative events scale to identify and assess any of the'children’s behaviors that could be interpreted as manifestations of disassociative behavior resulting from post-traumatic stress. The disassociative events scale used by Dr. Bollerud consists of a checklist of twenty-eight possible indicators of disassociative behavior, such as depersonalization, hypervigilance, and numbing.
Finally, Dr. Bollerud evaluated the children through a technique of art therapy. She testified that this technique involves the children being asked to do a standard series of drawings, which serve to “reveal the unconscious psychological world of the child.” Dr. Bollerud first asked each child to draw a person, then a person of the opposite sex of the person in the first drawing, and then a picture of the child’s *406family. Finally she asked the children to draw themselves before, during, and after the abuse. She then interpreted the drawings, analyzing the characters and scenes that were drawn, as well as the types of lines and drawing styles in each drawing. Dr. Bollerud testified that a child drawing a person of the opposite sex in the first drawing, drawings of people with no secondary sex characteristics, drawings of people with no hands, feet, or arms, and the presence of genitals in a drawing are some of the many potential indicators of abuse.
As a result of her information gathering, Dr. Bollerud found Lisa to be exhibiting the specific symptoms of “feeling that her body was not her own; forgetting time; having amnesias; feeling as if she was in another world; describing nightmares and dreams about sexuality; flashbacks; being intensely preoccupied with sexual abuse”; and having a fragmented sense of self. Dr. Bollerud also found indicators of sexual abuse in Lisa’s drawings, including her drawing a person of the opposite sex in the first drawing; a lack of secondary sex characteristics; missing hands and arms, with more missing in the drawings of herself and the defendant; the drawing of genitalia and sexual acts with the defendant when asked to draw the abuse;-the drawing of a transparent figure; and an “adoration in her ability to draw as she was drawing the more emotionally charged pictures.”
With respect to her evaluation of Julie, Dr. Bollerud found her not to be suffering from a disassociative disorder, although she did manifest some symptoms of post-traumatic stress such as hypervigilance, efforts to avoid thinking about what had happened to her in the past, a fear of males, and compulsive overeating. Dr. Bollerud also testified that some of the pictures that Julie drew had no indicators of sexual abuse. Other drawings, however, contained a figure with empty, uncolored eyes; figures with heavy belts around their waists; some figures lacking in secondary sex characteristics or having misshapen feet; and black marks representing her “gross feelings” after the abuse. Dr. Bollerud noted these characteristics as indicators of sexual abuse or emotional distress.
From all of the information gathered and observations made during her evaluations of Lisa and Julie, Dr. Bollerud ultimately testified that the children exhibited symptoms consistent with those of children who have been sexually abused. She emphasized in her testimony that her approach in evaluating children suspected of being sexually abused is to consider a wide variety of factors and gather as much information as possible, using a number of different means. She acknowledged that she did not, nor could not, rely solely on one *407behavior or response to conclude that a child has been sexually abused. Instead, she viewed all of the results, interpreted them in light of her expertise in the field, and based her conclusions on all the information before her.
We hold that Dr. Bollerud’s expert testimony is not sufficiently reliable to be admitted in a criminal trial as evidence that Lisa and Julie were sexually abused. By this opinion we do not seek to disparage the work being done in psychology and the behavioral sciences, for we can surely see its value; however, we are bound to recognize that the separate fields of behavioral science and criminal justice are different enough in their foundations and goals that what may be considered helpful information in one may not be so valued in the other. Generally speaking, the psychological evaluation of a child suspected of being sexually abused is, at best, an inexact science. Dr. Bollerud, herself, acknowledged at trial that the evaluation of such a child is partly a science and partly an art form. As this opinion explains below, a psychological evaluation of a potentially abused child does not present the verifiable results and logical conclusions that work to ensure the reliability required in the solemn matter of a criminal trial.
As a preliminary matter, we reject the State’s assertions that the scope of Dr. Bollerud’s testimony was somehow limited by her statements in conclusion that the children exhibited symptoms consistent with those of sexually abused children. We see no appreciable difference between this type of statement and a statement that, in her opinion, the children were sexually abused. Even if some difference could be identified, our review of the record below reveals that the manner in which Dr. Bollerud’s testimony was presented in its entirety reflects no such limitation and clearly would lead a jury to conclude that Dr. Bollerud believed the children were sexually abused.
The following analysis of Dr. Bollerud’s testimony properly concentrates on the parts of the psychological evaluations extending beyond the children’s recountings of the abusive incidents. Expert psychological evidence can only be admissible in a case such as this if it is at least partly based on factors in addition to and independent of the victim’s accounts. Otherwise, the expert’s conclusions are of no value to the jury because they present no new evidence and are merely vouching for the credibility of the child victim witness. We understand that Dr. Bollerud reached her conclusions by considering many factors in addition to the victims’ statements about the abuse. Her reliance on the children’s accounts was substantial, however, for *408even she conceded that she could not reach a conclusion that a child had been sexually abused, based only on a consideration of secondary factors without a verbal account from the child. Our purpose in reviewing Dr. Bollerud’s testimony is to determine whether these secondary factors, when viewed in conjunction with the children’s recountings, form a reliable ground on which Dr. Bollerud can offer an opinion that the children were sexually abused.
Reviewing Dr. Bollerud’s methods and conclusions in the context of their presentation as evidence in a criminal trial raises a number of specific concerns with respect to the reliability of her testimony. Our first concern is that the evaluations of the children deal almost exclusively in vague psychological profiles and symptoms, and unquantifiable evaluation results. There is much criticism attacking the attempts to compile a list of symptoms and behaviors to serve as an accurate indicator of whether a child has been sexually abused. See Commonwealth v. Dunkle, 602 A.2d 830, 832-36 (Pa. 1992) (citing articles); State v. J.Q., 252 N.J. Super. 11, 33-35, 599 A.2d 172, 184-85 (1991) (citing articles), aff’d, 130 N.J. 554, 617 A.2d 1196 (1993); State v. Rimmasch, 775 P.2d 388, 401-02 (Utah 1989) (citing articles); see also Myers et al., Expert Testimony in Child Sexual Abuse Litigation, 68 Neb. L. Rev. 1, 67-68 (1989). “The consensus among scholars is that there are as yet no scientifically reliable indicators of child sexual abuse.” State v. J.Q., 252 N.J. Super. at 33, 599 A.2d at 184. There are no symptoms or behaviors that occur in every case of child abuse, nor are there symptoms or behaviors that are found exclusively in child abuse cases. See Myers, supra at 62. The symptoms cited by Dr. Bollerud in this case are far from establishing a clear profile by which an abused child can be accurately identified. Many of the symptoms considered to be indicators of sexual abuse, such as nightmares, forgetfulness, and overeating, could just as easily be the result of some other problem, or simply may be appearing in the natural course of the children’s development.
The inherent uncertainty in the use of this vague symptomology is increased by the absence of standardized tests and readily quantifiable results as parts of the psychological evaluations. No standardized tests were administered to Lisa and Julie, as Dr. Bollerud testified that they have not been found to be helpful in determining whether a child has been sexually abused. The responses given by the children also were not quantified through any normative scoring procedure that would enable them to be compared to a standard set of results reflecting the responses of the general population of children, or even of a large grouping of known sexually abused children. The dis-*409associative events scale, for instance, which when used as a research tool yields a quantifiable result along a scale, was used in this case only as a clinical evaluative tool, producing no quantifiable results that could then be compared to a standardized norm. Without a more accurate means of assessing the implications of those symptoms exhibited by the children, and without the ability to somehow compare the results of Dr. Bollerud’s analysis of the children with an objective standard, the reliability of the psychological evaluations overall comes into question.
Our second area of concern regarding the reliability of Dr. Bollerud’s testimony compounds the problems noted above that are inherent in the evaluator’s need to rely on a vague symptomology and unquantifiable results. Although symptoms may be vague and inconsistent from case to case, this does not necessarily prevent them from logically leading to a certain conclusion. The problem present in this case, however, is the absence of any recognizable, logical nexus between many of the identified symptoms and the conclusion that the children have been sexually abused. Many of the factors considered by Dr. Bollerud, while they may accurately indicate that the children’s mental health may be suffering to some degree, do not necessitate a finding that the children have been sexually abused. To the extent that disassociative behavior may result from post-traumatic stress, the manifestations of disassociative behavior in Lisa may have been caused by incidents of sexual abuse, or could have been caused by the traumatic deaths of her mother and aunt, and the institutionalization of her father. The existence of the disassociative behavior, by itself, proves little. In addition, many of the indicators that were noted during the evaluations as signs of abuse, when viewed independently, do not even suggest that anything is wrong with the children, much less lead to a logical conclusion that they have been sexually abused. The fact that the children drew pictures of males when asked to draw a person does not put in motion any logical series of deductions that leads to a conclusion that the children were sexually abused. Without being able to identify this logical nexus, we have no way of accurately knowing whether the drawing of a male is an indicator of child abuse or the result of a fifty-fifty chance. Ultimately, we have no way of independently verifying Dr. Bollerud’s conclusions.
Our third concern regarding the reliability of Dr. Bollerud’s testimony stems from the problems we have previously mentioned. Much of the information relied upon by Dr. Bollerud begins to take on some sort of meaning only when it is interpreted and evaluated by *410someone in her field of expertise. Dr. Bollerud testified that her psychological evaluations are partly an art form, and that her interpretation and evaluation of the available information is somewhat subjective. We realize that Dr. Bollerud is not evaluating the children as a lay person would, and that her training in methodologies and adherence to professional guidelines and practices tend to improve the consistency and accuracy of her evaluations. Nevertheless, the fact that this type of interpretive step must take place adds yet another variable to the whole evaluative process, which can only make the results less certain. Moreover, because this final interpretive step occurs in Dr. Bollerud’s mind, drawing on her experiences and personal knowledge, it is all the more unverifiable.
Finally, we are not convinced that a thorough cross-examination can effectively expose any unreliable elements or assumptions in Dr. Bollerud’s testimony. The methodology used in the psychological evaluations makes her presentation of evidence effectively beyond reproach. Dr. Bollerud’s conclusions do not rest on one particular indicator or symptom, but rather on her interpretation of all the factors and information before her. So even though the defendant may be able to discredit several of the indicators, symptoms, or test results, the expert’s overall opinion is likely to emerge unscathed. An expert using this methodology may candidly acknowledge any inconsistencies or potential shortcomings in the individual pieces of evidence she presents, but can easily dismiss the critique by saying that her evaluation relies on no one symptom or indicator and that her conclusions still hold true in light of all the other available factors and her expertise in the field. In such a case, the expert’s conclusions are as impenetrable as they are unverifiable.
It would be disingenuous for this court to imply through its opinion that Dr. Bollerud’s testimony is entirely bereft of probative or useful information, or founded solely on illogical considerations, for such is not the case. Aside from the children’s recountings of the incidents of sexual abuse, Dr. Bollerud did consider several factors, such as age-inappropriate sexual behavior and knowledge, and obsessions with sexual abuse, that may suggest that sexual abuse did occur. Factors such as these have a close logical association with sexual abuse. See Myers, supra at 59. Absent an alternative explanation for their appearance, such as the victim’s involvement in other sexually oriented activities, these factors may be quite probative of whether a child has been sexually abused. If otherwise admissible, such factors could be offered independently at trial to prove that sexual abuse did occur. We do not believe, however, that the presence *411of such factors is an adequate ground on which an expert may affirmatively state that a child has been sexually abused. The relevance of this type of information is not beyond the ken of the average juror. Therefore, a jury should be able to draw its own conclusions from the evidence without an expert offering an opinion on the ultimate issue. See Vincent v. Public Serv. Co. of N.H., 129 N.H. 621, 625, 529 A.2d 397, 399 (1987).
In sum, we cannot allow an expert to present conclusions on such important issues in a criminal trial without greater assurances of the testimony’s reliability. We cannot consider the admission of Dr. Bollerud’s testimony in this case to be harmless error. Her testimony was lengthy, comprehensive, and directly linked to a determination of the guilt or innocence of the defendant. We cannot say beyond a reasonable doubt that her testimony did not affect the verdict, see State v. Elwell, 132 N.H. 599, 607, 567 A.2d 1002, 1007 (1989), and therefore reverse the defendant’s convictions.
Our holding in this case does not render expert psychological testimony useless in all child sexual abuse cases. There are cases in which an expert may play a valuable role as an educator, supplying the jury with necessary information about child sexual abuse in general, without offering an opinion as to whether a certain child has been sexually abused. Dr. Bollerud testified partially for this purpose when she detailed and explained the elements of the child sexual abuse accommodation syndrome. See generally Summit, The Child Sexual Abuse Accommodation Syndrome, 7 Child Abuse and Neglect 177 (1983). As Dr. Bollerud testified, this syndrome consists of five characteristics commonly found in sexually abused children: secrecy; helplessness; entrapment and accommodation; delayed, inconsistent, and unconvincing disclosure of incidents of sexual abuse; and retraction of the initial disclosure. The child sexual abuse accommodation syndrome was not intended to be a diagnostic device capable of detecting whether a child has been sexually abused. State v. J.Q., 252 N.J. Super. at 28, 599 A.2d at 181. Rather, it proceeds from the premise that a child has been sexually abused and seeks to explain the resulting behaviors and actions of the child. See People v. Bowker, 249 Cal. Rptr. 886, 892 (Cal. Ct. App. 1988).
Several of the common behaviors mentioned by Dr. Bollerud, such as a child’s delayed disclosure of abuse, inconsistent statements about abuse, and recantation of statements about abuse, may be puzzling or appear counterintuitive to lay observers when they consider the suffering endured by a child who is continually being abused. *412These behaviors also present an obvious opportunity for a defendant to superficially attack the testimony of a child victim witness during cross-examination or to argue against the child’s credibility in closing statements before the jury. Therefore, expert testimony explaining the peculiar behaviors commonly found in sexually abused children may aid a jury in accurately evaluating the credibility of a child victim witness. In addressing this issue, the Connecticut Supreme Court recognized that “the overwhelming majority of courts have held that, where the defendant has sought to impeach the testimony of the minor victim based on inconsistencies, partial disclosures, or recantations relating to the alleged incidents, the state may present expert opinion evidence that such behavior by minor sexual abuse victims is common.” State v. Spigarolo, 210 Conn. 359, 377-78, 556 A.2d 112, 122 (citing cases), cert. denied, 493 U.S. 933 (1989); see State v. J.Q., 252 N.J. Super. at 28-33, 599 A.2d at 181-84 (citing cases and articles). For these reasons, we hold that the State may offer expert testimony explaining the behavioral characteristics commonly found in child abuse victims to preempt or rebut any inferences that a child victim witness is lying. This expert testimony may not be offered to prove that a particular child has been sexually abused, and a defendant is entitled to a limiting instruction that so states.
We now turn to the defendant’s other claims on appeal. Although we reverse the convictions on his first claim of error, we will address his other claims in the interest of judicial economy to the extent they are likely to arise again in a second trial.
The defendant argues that his due process rights and right to confrontation were violated when he was prohibited from presenting testimony at trial regarding Lisa’s prior consensual sexual activity. The defendant sought to introduce this evidence in response to the State’s medical testimony that the condition of Lisa’s hymen and vagina was consistent with that of a child who has had sexual intercourse. The defense had some evidence through a former friend of Lisa’s that Lisa had had sexual experiences with three males other than the defendant prior to her medical examination. The trial court concluded that the State’s medical testimony did not “open the door” for any inquiry by the defense into Lisa’s past sexual activity.
New Hampshire’s rape shield law, RSA 632-A:6 (1986) (amended 1992), prohibits the admission of evidence regarding a victim’s prior consensual sexual activity with persons other than the defendant in a prosecution under RSA chapter 632-A, except when a defendant’s right to due process so requires. State v. Howard, 121 *413N.H. 53, 58-59, 426 A.2d 457, 460-61 (1981). The right to due process clearly ensures a defendant’s opportunity to present relevant evidence offering an alternative explanation to medical or physical evidence relied on by the State to prove that sexual intercourse has indeed taken place. See State v. LaClair, 121 N.H. 743, 746, 433 A.2d 1326, 1329 (1981). In this case, the trial judge, at the very least, should have made a further inquiry under State v. Howard to evaluate the evidence proffered by the defendant and determine whether due process considerations required the admission of evidence concerning Lisa’s prior consensual sexual conduct, despite the potential prejudicial effect on Lisa.
The defendant next argues that the trial court erred in failing to conduct an in camera review of a psychologist’s notes from her counseling sessions with one of the children to determine whether they contained evidence that would aid the defendant’s case. Instead of conducting an in camera review, the trial judge relied upon the psychologist’s representation that her files did not contain information exculpatory to the defendant. The trial court explained the departure from its previous practice of conducting an in camera review of confidential records from the New Hampshire Division for Child and Youth Services (DCYS) on the ground that DCYS records are under the control of a State agency, whereas the psychologist, whose notes were in question, was employed by a private mental health facility. As a reading of State v. Gagne, 136 N.H. 101, 612 A.2d 899 (1992), makes clear, this is a distinction without a difference.
In Gagne, we held that a defendant’s due process rights require a court to conduct an in camera review of confidential or privileged records where the defendant establishes a reasonable probability that the records contain information relevant and material to his defense. Id. at 105, 612 A.2d at 901. Gagne did not distinguish between the privileged records of a State agency and the privileged records of a private organization. The rationale in Gagne, balancing the rights of a criminal defendant against the interests and benefits of confidentiality, applies equally in both cases. A record is no less privileged simply because it belongs to a State agency. Likewise, a defendant’s rights are no less worthy of protection simply because he seeks information maintained by a non-public entity. We therefore hold that in the event of a retrial, the trial court must conduct an in camera review of the psychologist’s privileged records should the defendant establish a reasonable probability that the records contain information relevant and material to his defense.
*414Finally, the defendant argues that the trial court erred in not allowing defense counsel to participate in the in camera review of the victims’ DCYS records. In Gagne, we instructed the trial court on remand to conduct an in camera review of a confidential DCYS file. Relying on Pennsylvania v. Ritchie, 480 U.S. 39, 60-61 (1987), we stated that “the in camera review shall be made without the presence of counsel, since counsel for the defendant need not be present to assist the trial court in recognizing exculpatory evidence, and there is a danger that the names of persons who have spoken to the DCYS in confidence will be disclosed.” Gagne, 136 N.H. at 106, 612 A.2d at 902 (quotation and brackets omitted). On appeal, the defendant attempts to argue that the reasons behind excluding counsel from the in camera review in Gagne do not apply in this case because the names of everyone who spoke with the DCYS eventually will be disclosed when they testify at trial. The defendant cannot know if everyone who has spoken with the DCYS will testify at trial, nor can he be sure that the scope of their testimony will be as broad as the information contained in the DCYS records. A substantial interest in protecting confidentiality still exists in this case, and we affirm our presumption in Gagne that the trial court is capable of accurately evaluating the contents of the records without the assistance of counsel. Id.
Reversed and remanded.
All concurred.