Anatomically-Detailed Dolls

Although the anatomically-detailed dolls are widely used by many different types of professionals (Boat and Everson, 1988; Conte et al., 1991; Kendall-Tackett and Watson, 1992), they are extremely controversial and there is disagreement in the professional community as to whether they should be used (e.g., Koocher et al., in press; Yates and Terr, 1988). The mental health professional is especially vulnerable if the child's interaction with the dolls forms the basis for an opinion or conclusion about sexual abuse. Even professionals who believe it is all right to use the dolls as interview aids are sharply critical of this use. For example, doll supporters Everson and Boat (1994) state:

Although there seem to be widespread perceptions in both lay and professional circles that young children's behavior with the dolls is commonly used to make definitive diagnoses of sexual abuse (Diagnostic Test Use), such a use of the dolls was not endorsed by any of the guidelines reviewed and is open to significant criticism. (p. 113)

The American Psychological Association anatomical doll task force (Koocher et al., in press) concludes it is all right to use the dolls with certain caveats:

First, AD dolls are not a psychological test with predictive (or post-dictive) validity per se.

Second, diagnostic statements about child sexual abuse cannot be made on the basis of spontaneous or guided "doll play." A clinical interview by a skilled clinician is not play.

Third, particular caution is called for when interpreting the reports of children ages 4 and under, at least so far as reports of "being touched" are concerned and when repeated misleading questioning has been employed.

Fourth, in light of current knowledge, we recommend that APA reconsider whether valid "doll-centered assessment" techniques exist and whether they still "may be the best available practical solution" for the pressing and frequent problem of investigation of child sexual abuse.

Finally, special recognition of normative differences between children of different racial groups and socioeconomic strata should be a part of training professionals who use AD dolls in clinical inquiry.

These have been seldom, if ever, observed in the hundreds of cases we have reviewed. The attorney may therefore be able to use these in questioning a mental health professional who has used the dolls.

We, as well as many other scientists, believe that the dolls should never be used, even with these caveats. There are no commonly accepted standards for the use of the dolls nor normative data on them (APA Council of Representatives, 1991). The dolls may become a modeling and learning experience for a child (Wakefield and Underwager, 1988a; Underwager and Wakefield, 1990). Interviewers model handling the dolls, suggest that they be undressed (or undress them for the child) and label them for the child. They ask the child to show with the dolls what the accused perpetrator did and they may even place the dolls in sexually explicit positions for the child. Although some researchers claim the dolls are not suggestive (e.g., Everson and Boat, 1994), studies show that some nonabused children engage the dolls in sexual play (Dawson and Geddie, 1991; Dawson et al., 1992; Everson and Boat, 1990; Gabriel, 1985; Glaser and Collins, 1989; McIver et al., 1989).

The studies that claim to show differences between the doll interactions of sexually abused and nonabused children have major methodological shortcomings which limit any conclusions that can be drawn from them (Ceci and Bruck, 1993a; Skinner and Berry, 1993; Underwager and Wakefield, 1990; Wakefield and Underwager, 1988a, 1989, 1994c; Wolfner et al., 1993). DeLoache (1995), whose research is on the developmental aspects of symbolic representation, notes that the basic reason for using anatomical dolls is the belief that the dolls will elicit information from children that they are unable or unwilling to give verbally. But she observes that, not only is there no good evidence that dolls help in interviews with very young children (age 3 and below), but that the presence of the dolls might result in the youngest children providing less information. Younger children cannot understand the basic self-doll relation assumed by interviewers who use the dolls. They cannot use dolls as symbols or representations for themselves and therefore cannot use the dolls to enact their own experiences. She concludes:

To my mind, the most important research finding about the use of dolls with very young children is that there is no good evidence that the dolls help. . . .My study ... suggested that the presence of the doll might even interfere with the memory reports of the youngest children . . .(p. 178)

Levy (1989) argues that any statement by a child that is the product of a doll-aided evaluation should be inadmissible as evidence:

There is literally neither theoretical nor any empirical basis for drawing any conclusion about what a given child's play with the dolls means. In addition, there is at least a possibility that some children, evaluated by professionals who want them to acknowledge sexual abuse, may come to use the dolls in a fashion that leads fact finders to easily to believe incorrectly that the children have been abused. Mental health professionals who testify have made, and if the testimony is admissible, are likely to continue to make extravagant and baseless claims about the significance of children's play with dolls. And because the dolls purport to be a scientific demonstration that establishes an "aura of infallibility," the implicit message of doll-play testimony is likely to be much more influential with fact finders than any other uncorroborated clinical conclusion by an expert. (p. 407)

Skinner and Berry (1993) observe that distinct patterns of play of abused versus nonabused children have not been identified and that the lack of norms calls into question the forensic use of the dolls and conclude that:

The lack of sufficient evidence supporting the psychometric properties of AD dolls calls into question the use of those dolls in the validation of child sexual abuse allegations. . . .Moreover, given that validity is the principal issue underlying the admissibility of psychological evidence in the courtroom. . .and the inadequate evidence for the construct and criterion-related validity of AD dolls, evidence collected using AD dolls should not be admitted in court in child sexual abuse cases at this time. (p. 418)

Wolfner et al. (1993) point out that the necessary research to determine whether using the dolls has any incremental validity in establishing abuse would involve a group of children who were all suspected of being abused who, based on subsequent evidence, could be definitely divided into those who have and have not been abused. The doll interviews would have to take place prior to the children undergoing the standard procedures for investigating sexual abuse, since the process of being questioned about abuse could affect their reactions to the dolls. Such research has not been done-the studies that are claimed to support the use of the dolls only compare children suspected of abuse to those who are not suspected. Wolfner et al. conclude:

We are left with the conclusion that there is simply no scientific evidence available that would justify clinical or forensic diagnosis of abuse on the basis of doll play. The common counter is that such play is "just one component" in reaching such a diagnosis based on a "full clinical picture" (or portrait). ADD play cannot be validly used as a component, however, unless it provides incremental validity, and there is virtually no evidence that it does. . . . we urge that the lack of evidence for the incremental validity of ADD use in diagnosing such abuse, and the interpretive and research problems highlighted in this review, be taken very seriously. We believe that the ethical principle that application should follow knowledge gained from research results, rather than precede it, is self-evident, particularly in an area where an incorrect diagnostic conclusion can have horrific effects on people's lives. Using a diagnostic technique that may simply exacerbate error benefits no one. (p. 9)

In summary, there is no evidence that doll interviews are a valid and reliable method for getting accurate information about sexual abuse, including the claim that they can be used as demonstration aids. The use of the dolls as an assessment or investigatory technique is not generally accepted within the scientific community and would not meet the Frye test. Rather, their use remains highly controversial and the scientists quoted above who have carefully reviewed the body of research on the dolls recommend that they not be used. For the reasons discussed by Skinner and Berry (1993) and Wolfner et al. (1993) they also fail to meet Daubert.