Behavior of Abused and Non-Abused Children in Interviews With Anatomically Correct Dolls

William McIver II, Hollida Wakefield, and Ralph Underwager*

ABSTRACT: Fifty non-abused and ten abused children were given anatomically-correct dolls and their behaviors were observed.  None of the abused children had been previously interviewed or treated for sexual abuse.  For seven of the children, following the initial portion of the interview, the interviewer deliberately used leading questions, cues, modeling and reinforcement in an attempt to elicit sexual behaviors with the dolls.

There were no differences in the behavior and responses to the dolls between the abused and the non-abused children.  In general, the children did not identify the gender of the dolls on the basis of primary sexual characteristics.  However, two-fifths of the children spontaneously talked about and/or touched the dolls' genitals and three-fifths placed the dolls in clear sexual positions and/or played with the dolls in an overtly aggressive manner.  Six of the seven children who were given leading questions, cues, modeling and reinforcement responded by performing the behaviors that were cued, modeled and reinforced.

The authors conclude that information obtained by the use of dolls in interviews to assess whether a child has been sexually abused will probably be misleading.  The dolls are likely to increase the error and decrease the reliability of the information gathered.


The so-called anatomically correct dolls are widely used as diagnostic tools with children in cases of alleged sexual abuse. They are used by social workers, police, prosecutors, and sometimes by parents.  They are generally used with children ages two through six, although some workers have used them with children from 18 months to 11 years of age.  We have a videotape of an interrogation in which a social worker used the dolls with a 17-year-old female with a history of three abortions.

These dolls are made of plastic or cotton and are usually about 20 to 25 inches in length.  Pubic hair is simulated with dark embroidery or synthetic fur.  The breasts of the mature females protrude and the boy and mature male dolls have penises.  These penises are often disproportionately large, although this is less true of the more recent dolls.  There are representations of oral and anal openings and the female dolls have crude representations of the pubic area and vaginal openings.  The penis is able to fit into any of these openings.  The dolls are dressed in easily removable clothing.  There is generally a mature male and female doll and a boy and girl doll which lack the pubic hair and large breasts.  The dolls may be purchased from manufacturers or handmade by someone.  There is no standardization for their design.

These dolls were originally used in therapy as toys and as aids in helping sexually abused children deal with the experience.  But now they have come to be used as diagnostic tools in the investigation of suspected cases of sexual abuse.  They are routinely used by some mental health professionals in the assessment of a child and claims are made that the type of doll play can prove sexual abuse.  Yet their reliability and validity have not been established.

Their use has been based on the assumption that children who have been sexually abused will demonstrate sexual behaviors with the dolls which children who have not been sexually abused will not demonstrate.  Underlining this assumption are several other assumptions:

1. Children will identify gender (or the "sex" of the dolls) on the basis of primary sexual characteristics. That is, children will "see" the dolls the way adults do.

2. The things children do with these dolls will be a reflection of underlying psychological dynamics, and this will indicate the same things to a child as it may to the adult interviewer.

3. Children will not demonstrate, either spontaneously or in response to the interviewer's verbal and/or nonverbal cues, sexual acts that did not occur.

4. Children will, somehow, naturally identify the dolls as "Mommy ... Daddy ... Uncle ... Sister ... Brother ... ," etc.  That is, the dolls will symbolize persons in their lives.

However, there is no research on these assumptions.  What is the antecedent probability of a child's response to the dolls?  What is the response to a doll with genitals when children accustomed to department store dolls without genitals are exposed to them?  Questions such as the effect of naming the dolls, who undresses the dolls and what is said when they are undressed, what leading questions about explicit sexual behavior do, and what is done when the child loses interest or moves to different play need to be researched.  Such baseline information is necessary in order to interpret what responses to such dolls mean.

To date there are no solid empirical data that support a differential behavior of abused and nonabused children when the dolls are used as diagnostic or assessment devices.  The data that are available suggest that they cannot be used to distinguish abused from non-abused children.

There is evidence that the dolls can suggest behaviors that may be interpreted as abuse.  King and Yuille (1987) point out that "... the dolls serve the function of a suggestive question with young children.  The genitals and orifices of the dolls suggest a play pattern to children, and that play may be misinterpreted as evidence for abuse" (p. 31).  Gabriel (1985) states about the use of dolls in assessing sexual abuse that "Many persons working in the child protection field are untrained in play therapy and do not know about the projection-evoking properties of toys.  The result has been that material produced by children in this manner can appear to confirm suspicions of sexual abuse when it may actually be no more than a normal reaction to the dolls and the situation" (p.42).

Gabriel observed 19 non-abused children who were exposed to the dolls as well as other toys.  Many of the children did things with the dolls which could have been interpreted by other interviewers as indicating likely sexual abuse.  Roughly half of the children showed interest in the genitals and manipulated them.  Gabriel stated that "On the evidence of the dolls alone, when used as part of a 'fishing expedition' exercise, the suspect will almost always be found 'guilty,' especially if the examiner is already biased in that direction" (p.49).  He concludes "Thus it can be seen how extremely important it is not to allow suspect 'evidence' to be introduced by unevaluated diagnostic tools" (p.50).

There have been some beginning efforts at developing standardized procedures for the dolls.  White, Strom and Santilli (1985) developed a protocol for interviewing preschoolers.  They suggest not using the dolls with children who have reached the age of social awareness (five to seven years).  Their protocol emphasizes maintaining objectivity by not receiving information prior to the interview, avoiding leading questions, taking care not to cue responses, and interviewing the child away from the parents.  But many of their suggested questions in the protocol are leading.  Nevertheless, their protocol as described represents an improvement over the videotaped doll interviews we have observed.

White, Strom, Santilli, and Halpin (1986) report that with this protocol, their sample of non-sexually abused children interacted differently with the dolls than did the abused sample.  However, there is no information on other differences which may exist between the two groups (interviews about sexual abuse, previous therapy for sexual abuse, prior experiences with the dolls, etc.).  Without this information, no conclusions can be drawn from their results.  In a comment accompanying this article, the editors of Child Abuse & Neglect state: "We wish to emphasize, however, that to date no data have been published which clearly delineate the responses to these interviews by children who have not been sexually abused.  This study is necessary before the results of interviews of children who have been sexually exploited can be accurately interpreted" (p. 519).

In a study from which preliminary results were reported at the American Academy of Child Psychiatry, October, 1986, the protocol suggested by White, et al. was used (Jensen, Realmuto, & Wescoe, 1986).  Two control groups, one, a normal non-abused group, and the other, a psychiatric group but non-abused, and a sample of abused children were interviewed by a single therapist who followed the protocol faithfully.  All interviews were videotaped.  A panel of raters then viewed the videotapes and rated the behaviors along White's scale from not at all suspicious to very suspicious.  No differences between groups were found.  Some of the non-abused children got the highest rating of very suspicious and some of the abused children got ratings of no suspicion of abuse.

Jampole and Weber (1987) studied 16 girls and 14 boys ages three to eight.  The children were divided into two groups (sexually abused and non-sexually abused) and observed with the anatomical dolls.  Assignment to the abused group was based upon a decision by police and child protection workers that the child had been abused.  This decision was made in advance of any final determination by the justice system.  Neither group had been previously interviewed with the dolls.  The behavior of the children was rated on a scale which was developed on the basis of observing four children.  The researchers report a significant difference between the two groups in that the children who had been abused (as determined by caseworkers and/or police) demonstrated more sexual behavior than the non-abused group.  On this basis they conclude that the anatomically correct doll is a reliable, valid instrument as an investigative and treatment tool.

The major difficulty with this study is that, although none of the children had been previously interviewed with the dolls, there is no information concerning the content of any interviews about sexual abuse, any therapy given to the sexually abused children, or what discussions were held with these children about sexual abuse by foster parents or social workers.  The criterion measure means that the contrast is not between abused and non-abused children but rather children interrogated in the process of substantiating an accusation and children who have not been interrogated.

Aman and Goodman (1987) report a study that they interpret as indicating that the dolls do not lead to false reports of sexual abuse.  However, this is an over interpretation of this study.

Their sample was 30 three-year-old and 30 five-year-old non-sexually abused children.  The real life event, an interaction with an adult male, was ten minutes long and included five play activities, permitting about two minutes for each activity.  (Such brief exposures are not analogous to the real life events likely to be the subject of an investigative interview using the dolls.)  A week later, the children were questioned about the event.  There were three conditions (which resulted in ten children per cell).  In the first condition, children were questioned with anatomically-correct dolls; in the second condition they were questioned with ordinary dolls; in the third condition no dolls were used.  All children were asked a set of objective and subjective questions.  The number of correct verbal and behavioral responses by each child in response to the free recall portion of the interview was analyzed.

The most important finding of this study is a clear age effect throughout.  The younger children were consistently more suggestive across all variables.  False alarm errors, that is, errors that could lead to false reports of abuse, were significantly greater for three-year-olds than five-year-olds.  One in five three-year-old children made such errors.

Although the authors reported that there was no overall doll effect, there was a significant age X doll interaction.  In responding to questions described as objective, the three year olds in the doll condition were significantly less accurate in answering the objective questions than were the three year olds in the no doll condition.

Sivan and Schor (1987) report a study of 144 children, ages three to eight, on labeling of body parts using allegedly anatomically correct dolls.  Younger children gave more responses but older children gave clearer responses.  Younger children were reported to be willing to provide answers to questions when they have no meaningful response.  The attitude and approach of the interviewer was a crucial factor in eliciting responses.

Sivan, Schor, Koeppl, and Nobel (1988) report that girls play with dolls more than boys; all children play with the dolls more when the interviewer is female, suggesting that females subtly suggest doll play.  Of most significance is the finding that the dolls are of little interest to children.  They do not spontaneously choose them to play with them.

In a study of non-abused children and their behavior with the dolls, Boat and Everson (undated) found that manual exploration of the sexual parts of unclothed dolls is not uncommon.  They report that a small number of non-abused older preschoolers will demonstrate explicit genital intercourse.

Friedemann and Morgan (1985) have written a guidebook for interviewing with the anatomically correct dolls.  They caution against many of the abuses that have become common in the use of these dolls but present no standardized format.

There has been no research to establish standardized procedures for using the dolls in an investigation.  Although White, Strom, and Santilli (1985) describe a protocol, there is no normative data to standardize it.  Standardized procedures that can be repeated by others are an absolute requirement before anything sensible can be said or any conclusions drawn beyond the immediate situation.  Without standardized procedures there is no way of determining what effect the procedures have on the responses elicited.

Underwager, Wakefield, Legrand, Bartz and Erickson (1986) and Wakefield and Underwager (1988) point out that the use of the dolls can provide a modeling effect.  All of the social learning literature shows that one of the most powerful ways of teaching children is by modeling.  The learning is enhanced when the adults involved in the modeling are reinforcing to the child.  The research evidence establishes that verbal reinforcement such as, "good," "that's right," or attention from an adult sharply increase the imitation of the modeled behavior.  Therefore the use of the dolls in interviews must not be viewed as a pursuit of truth but rather as a learning experience.  The child is being taught to produce the responses favored and reinforced by the interviewer.

The present study was designed to examine the general assumption that children who have been sexually abused will demonstrate sexual behaviors with the dolls which children who have not been sexually abused will not demonstrate.  Specifically, the underlying assumptions discussed above were examined for a group of abused and a group on non-abused children.  The assumptions tested were:

1. Children will identify gender (or the "sex" of the dolls) on the basis of primary sexual characteristics rather than on the basis of their clothing.

2. Children will not demonstrate, either spontaneously or in response to the interviewer's verbal and/or nonverbal cues, sexual acts that did not occur.

3. Children will, somehow, naturally identify the dolls as "Mommy ... Daddy ... Uncle ... Sister ... Brother ... ", etc. That is, the dolls will symbolize persons in their lives.

4. The things non-abused children do with these dolls will be different from the things abused children do with them.



The subjects were 50 non-abused and 10 abused children.  Of the 60 children, 36 were females and 24 were males.  The non-abused children (28 females and 22 males) ranged in age from 2 years 6 months to 7 years; the mean age was 4 years, 7.8 months.  The abused children (8 females and 2 males) ranged from 3 years 4 months to 5 years 8 months; the mean age was 4 years, 5.4 months.  The children were recruited in various ways from friends, patients and their relatives and friends.  The first author obtained sufficient information about the children to feel confident that the abused children actually had been abused and that the non-abused children had had no history of abuse.

The 10 abused children had experienced a variety of abusive situations including fondling, being rubbed with the penis, oral sex and attempted penetration.  For 4 of the children these were single instances and for 6 of them the abuse occurred repeatedly the range was from 1 to 15 times.  The children were interviewed from 3 days to 70 days following the abuse, the median length of time was 14 days.  The abuse was verified apart from the children's statements by eyewitnesses or by admission from the perpetrators.  None of the abused children had been interviewed by counselors, police or anyone outside of the immediate family.


The children were interviewed individually by the first author.  Twenty-two of the children were interviewed in an office and 38 were interviewed in their homes.  The office was informally furnished with a couch where the dolls were placed.  When the children were interviewed in their homes, the interview was conducted in a spot removed from normal family activity.  In a few instances the interview was briefly interrupted by other children.

Four anatomically-correct dolls from Migima Designs, Inc. (P.O. Box 70064, Eugene, OR 97401) were used for all of the children.  There was a adult female, an adult male and a boy and a girl doll.  The adult dolls were approximately 20" and the child dolls 16."  All of the dolls had oral and anal openings and both the adult and child female dolls had vaginal openings.  The male dolls had testicles and penises which fit easily into any of these openings.  The adult female doll had breasts.  The penis, testicles, breasts and vagina were fairly proportionate to the rest of the body.  There were a variety of extra dresses, blouses and slacks for the female dolls and pants and shirts for the male dolls.  The clothing was all easily removable with Velcro fasteners.

All of the interviews were video or audiotaped.  In the non-abused group, 38 were videotaped and 12 were audiotaped; in the abused group, 8 were videotaped and 2 were audiotaped.  It was originally planned to videotape all of the interviews, but this proved to be unworkable due to equipment failure.  In these instances, the interviews were audiotaped.

The interviews ranged in time from 8 minutes (a hyperactive 2 year, 6-month-old non-abused child) to 46 minutes.  The mean time was 27.13 minutes (non-abused 27.52, abused 25.20).

In the interviews the children were greeted and about 5 to 10 minutes was spent building rapport.  The children were then shown the dolls.  The dolls were always presented dressed with the extra clothing placed alongside.  Cross-dressed dolls (male dolls in dresses and female dolls in pants and shirts) were presented to 21 non-abused and to 5 abused children.  It was randomly determined whether the child received the sex-appropriate or the cross dressed dolls.

The children were encouraged by open, non-leading questions to say whatever they wanted and to do whatever they wanted with the dolls.  The goal was to see if and how children identified the dolls' genders and differentiated between the child and the adult dolls, and to observe what the children would spontaneously say and do with the dolls.  The children were allowed to freely play with the dolls if they wished.

All of the children were asked general questions or given standard statements about the dolls:

"What are these?" (dolls)
"Are they girls or boys?"
"You can change the clothes if you like."
"What do these look like?"
(Pointing to various body parts) "What's this?"
"What do you call this?"
"Are you a boy or a girl?" "How do you know?"

The subsequent questions were geared to the child's responses.  When, in response to "Are they boys or girls?" the child gave an answer, he would then be asked "How do you know?"  If the doll was identified as a parent or a boy or girl, the interviewer asked in a deliberately low-keyed fashion, "Uh huh, How do you know?"  When body parts were identified (hands, rear end, lips, belly button, etc.) the experimenter asked "What does he (or "she," depending upon the identification given by the child) do with it?"  Stringent efforts were made to keep from attributing gender or any personality characteristics to the dolls.

For 7 of the children (6 non-abused and 1 abused), following the initial portion of the interview, the interviewer deliberately used leading questions, cues, modeling and reinforcement in an attempt to elicit sexual behaviors with the dolls.  The behaviors engaged in by the interviewer in this portion of the interview were similar to those observed in typical videotaped interviewers of allegedly sexually abused children.

Following the interviews, the behaviors were classified from the videotaped and audiotaped interviews independently by the interviewer and one of two other raters.  The interviewer was aware of whether the children were in the abused or the non-abused groups; the two other raters were not.  In cases where there was disagreement as to the classification of a behavior, the tape was viewed again by the two raters and agreement was reached.

The children's responses were rated according to the following categories:

  1. Gender identification
  1. On basis of clothes, eye or lip color, size and other non-sexual characteristics.
  2. Took notice of breasts.
  3. Took notice of penis.
  4. Took notice of vagina-pubic area.
  5. Children with cross-dressed dolls who identified and did not change gender.
  6. Children with cross-dressed dolls who identified gender and did not change on the basis of one or more primary sexual characteristics.
  7. Identification primarily on the basis of primary sexual characteristics with either one or more dolls.
  1. Spontaneous behavior or comments about things the dolls did, interest in genitals, sexual and/or aggressive behavior with dolls.
  1. Showed interest in genitals (talking about, touching).
  2. Made spontaneous comments about what the dolls do ("Daddy pooped on his head").
  3. Spontaneously placed the dolls in clear "sexual positions" (intercourse-vaginal and anal, oral sex, kissing), hitting, kicking, etc.
  1. Spontaneously used dolls to symbolize people (family, etc.) in their lives.
  2. Knew physiological reasons for their own gender.
  3. Led to demonstrate with minimal cues, leading questions, and reinforcement (7 children who received this at the end of the interview).

The results were compared for abused and non-abused children and t tests were used to test for significant differences. Comparisons were also made between the 36 younger (below age 5) and 24 older (age 5 and older) children.


Gender Identification

In general, the children did not identify the dolls as males or females on the basis of the primary sexual characteristics.  The children were able to say that the dolls were "boys" or "girls" (or "Mommy" or "Daddy"), but when asked how they knew, they generally responded in terms of size, dress, hair, or color of eyes (the male dolls had brown eyes and the female dolls blue eyes) or color of lips (the lips of the female dolls were red and the male dolls were brown).  Some of the children gave very vague reasons: "I don't know," "Mommy told me," "Because."  This happened even when the dolls were undressed and the genitals were recognized.  For the children as a whole, 87% identified the dolls using non-primary sexual characteristics and only 20% identified them using primary sexual characteristics (Table 1).

All of the children correctly identified themselves as boys or girls but only 16% could give a physiological reason for this.  The most common responses were vague answers such as "Because," or "I know," "Cause my name's 'Mary'," "Cause my Mommy told me," "I just know."  The abused and non-abused children reacted similarly there were no significant differences between them.

Table 1
Gender Identification

Identify dolls using non-primary sexual characteristics Identify dolls using primary sexual characteristics Know physiological reasons for own gender
N % N % N %
Non-abused 43 86% 9 18% 8 16%
Abused 9 90% 3 30% 0 0%
Total* *52 87% *12 20% 8 13%
*Total does not equal 60 because of unclear classification for 4 children

As one would expect, there are significant differences depending upon the age of the child.  Only 2 of the 36 children under age 5 (5.6%) identified the dolls using primary sexual characteristics whereas 10 of the 24 children 5 and over (42%) identified them in this way (p=.002).  Also, all 8 of the children who were able to state a physiological basis for their own gender were over 5 (p=.001).

Most of the children (78%) recognized and identified the breasts on the adult female doll.  A somewhat smaller number (62%) recognized the penis on the adult male and the boy doll but fewer (27%) recognized the vagina and/or pubic area on the adult female and girl dolls (Table 2).  There were no significant differences between the non-abused and abused children.

Table 2
Recognition, Notice or Identification of
Primary Sexual Characteristics

Breasts Penis Vagina/pubic area
N % N % N %
Non-abused 40 80% 32 64% 13 26%
Abused 7 70% 5 50% 3 30%
Total* 47 78% 37 62% 16 27%

Again, this was related to the age of the children.  For the 36 children under 5 years of age, 24 (67%) recognized the breasts, 18 (50%) recognized the penis and 5 (14%) recognized the vagina and/or pubic area.  For the 24 children over 5, these numbers were 23(96%), 19(79%), and 11(46%).  These differences were statistically significant (breasts, p=.007; penis, p=.023; vagina/pubic area, p=.006).

What happened when the child was presented with cross-dressed dolls?  Two thirds (69%) of the children identified the doll according to the clothing and didn't change the gender when the dolls were undressed.  One third (35%) changed the gender on the basis of the primary sexual characteristics after the dolls were undressed.  Of these 9 children, 8 were over 5 years old (Table 3).  There were no significant differences between abused and non-abused children, but the age difference was significant (p=.001).

Table 3
Gender Identification in Children with
Cross-dressed dolls

Identified and did not change gender Identified and changed gender on basis of primary sexual characteristics
N % N %
Non-abused 14 57% 8 38%
Abused 4 80% 1 20%
Total* *18 69% *9 35%
*Total does not equal 26 because of unclear classification for 1 child

Spontaneous Behavior and/or Comments About the Dolls

Table 4 summarizes the spontaneous behaviors and comments made by the children while interacting with the dolls.  There were no significant differences between abused and non-abused children on any of these measures.  Two-fifths of the children (44% non-abused and 30% abused) spontaneously talked about and/or touched the doll's genitals.

Table 4
Spontaneous Behavior or Comments about Dolls
Talk about or touch doll's genitals Comments about what dolls do Place dolls in clear sexual positions and aggressive behavior Use dolls to symbolize people in child's life
N % N % N % N %
Non-abused 22 44% 29 58% 31 62% 7 14%
Abused 3 30% 4 40% 5 50% 2 20%
Total 25 42% 33 55% 36 60% 9 15%

Three-fifths of the children (62% non-abused and 50% abused) placed the dolls in clear sexual positions and/or played with the dolls in an overtly aggressive manner.  This included placing one doll on top of another doll in the missionary position, sticking a male doll's penis in one of the openings (anal, vaginal, oral), wrestling, spanking, throwing or hitting the dolls or having one doll do these to another doll.  Many of these spontaneous behaviors could have elicited suspicion of a sexual abuse in an interviewer who accepted the assumptions that the doll play reflects actual experiences in a child's life.

Around half (58% non-abused and 40% abused) of the children made spontaneous comments about what the dolls did.  Examples of these comments are "He did something naughty," "(He) jumped on his bed," "Daddy went poopy on my head," "He peed" (while holding the doll on the interviewer's head), "I jump on top of him" (using the doll to illustrate), "These guys throw each other around" (demonstrated by throwing dolls against the wall), "A rhinoceros stands on the Dad's head. ... A car falls on my head. ..."

Only 9 out of 60 children, or 15% (14% non-abused and 20% abused), of the children used the dolls to symbolize persons in their lives.  These identifications generally were "This is a Mommy," This is a Dad, he's bigger than the little kid," etc.

When children under 5 years old and children over 5 years old were compared as to these behaviors, no significant differences were found.

Led to Demonstrate

For 7 of the children (6 non-abused and 1 abused), following the initial portion of the interview, the interviewer deliberately used leading questions, cues, modeling and reinforcement in an attempt to elicit aggressive and sexual behaviors with the dolls.  The interviewer asked questions such as "Can you show me?," "How else could they go together?" and then reinforced the responses with "Uh Huh," "Anything else?"

In response to this, all but one of the children responded by performing the behaviors that were cued, modeled and reinforced.  Behaviors elicited in these 6 children included hitting the Daddy doll, hitting, punching and kicking the boy doll, cunnilingus, fellatio, anal and vaginal intercourse and the Mommy doll sitting on the boy doll's face.  As this portion of the interview progressed, the children continued to demonstrate and produce more and more behaviors.


The assumption, Children will identify gender (or the "sex" of the dolls) on the basis of primary sexual characteristics rather than on the basis of their clothing is not supported by this study.  All of the children gave the dolls a male or female identity.  When asked "How do you know?," most of the children responded in terms of clothing, color of lips, color of eyes, etc.  Only about a fifth of the children gave the primary sexual characteristics (breasts, penis, vagina and/or pubic area) as a reason for the gender identification of the dolls.  This was true even though most of children noticed or attended to at least some of the genitals on the dolls.

When presented with the cross-dressed dolls, roughly one third of the 26 children identified and changed the gender of the doll on the basis of the breasts, penis, or vagina/pubic area.  All but one of these children were over 5 years old.

Similarly, even though all of the children were very clear as to whether they were boys or girls, only 8 out of the 60 children appeared to know the physiological basis for their gender.  These 8 children were all over 5 years old.

The assumption, Children will not demonstrate, either spontaneously or in response to the interviewer's verbal and/or nonverbal cues, sexual acts that did not occur was not supported.  Over half of the children played with the dolls in ways that suggest overt sexual behavior and/or aggressive behavior.  Children placed the dolls in missionary positions and put the penises of the boy dolls into the anal, vaginal and oral openings of other dolls.

This behavior was particularly dramatic in all but one of the 7 children who, at the end of the interview, were cued and reinforced for producing such behavior.  One child put the male doll's penis into every possible opening of the other dolls.  When he was then asked "Could she ever sit on him?...," "Could you show me?," he proceeded to get up and sit on the face of the doll.  This demonstrates the concrete manner by which a child often interprets the questions and statement of adults.  Another child stated that "Daddy pee peed on my head," when asked "What else can he do?"  It is noteworthy that this behavior was obtained with comparatively subtle questions and cues.

In that the typical interview with the dolls contains a large proportion of leading questions, modeling, selective reinforcement and even coercion (Wakefield & Underwager, 1988), this finding casts doubt on the validity of the information obtained from children in such interviews.  The dolls, in the way they are typically used, are likely to increase the error and decrease the reliability of information obtained in interviews of children concerning alleged sexual abuse.

This hypothesis is strengthened by the fact that the assumption The things non-abused children do with these dolls will be different from the things abused children do with them was not supported.  There were no significant differences between abused and non-abused children on any of the measures.  A major difference between this study and other ones is that the abused children in this study had not been talked to by anyone outside of their immediate families.  It is probable that interviews about abuse with social workers, police, investigators and/or therapists could account for any differences found between abused and non-abused children in other studies.

Only 9 of the 60 children (15%) used the dolls in their playing with them to symbolize people in their lives.  Therefore, the assumption Children will, somehow, naturally identify the dolls as "Mommy ... Daddy ... Uncle ... Sister ... Brother...," etc.  That is, dolls will symbolize persons in their lives is not supported.  It is likely that an interviewer could label the dolls as "Daddy," "Mary," "Uncle Fred," etc., model and reinforce this identification to the child, with the result that the child would then play with the dolls using these identifications.  But there is no evidence that such a learning experience for the child means that subsequent play reflects actual incidents in the child's life.  There is no evidence that if children identify a doll as a "Daddy" or "Mommy" this indicates that they are referring to their Daddy or Mommy.

A recent study by Herbert, Grams, and Goranson (1987) produced results similar to ours.  These investigators used a standardized investigative interview approach in studying 14 children, ages three to five, with no suspicion of sexual abuse.  The investigative method used was the allegedly anatomically correct dolls.  Questions asked of the child when the dolls were in use were typical of the first level of questions asked in such interviews and then were followed by the use of leading questions.

Herbert, et al. found that the children did not approach the dolls spontaneously but required the interrogator to direct them to approach the dolls.  None initiated undressing the dolls but all accepted the direction of the interrogator to undress them.  The children showed inability to identify gender on the basis of the symbolic genitalia.  Although none of the children placed the dolls in postures representing sexual behavior between the dolls, all the children showed behavior that is often interpreted as demonstrating sexual abuse.

The children demonstrated admixture of fantasy into their responses which was not detected by the interviewer.  The authors regard this as a troubling finding demonstrating how easily an interview can be misinterpreted.  Almost all children demonstrated suggestibility in responding to the interrogation.  Evaluation of the interviews "as if" a sexual abuse allegation had been made resulted in half the sample being identified as probable victims of abuse.  This suggests a high level of false positives when such techniques are used.  Exploratory play with the symbolic genitals or anus of dolls is in itself not an indicator of of child sexual abuse and the authors conclude that a search for proof of child sexual abuse is best done by other methods.

The Aman and Goodman (1987) study, despite the conclusions stated by its authors, cannot be said to show that the dolls have no effect on increasing false reports.  Younger children showed both suggestibility and effect of dolls.  In real-life sexual abuse interviews, it is the younger children with whom the dolls are more likely to be used.  The study demonstrates the possibility of false reports from younger children.  If one in five can be led to produce false alarms by questions asked once, what will happen when the dolls are used with repeated questions and pressure to interact with the dolls until the desired behavior is elicited?  When children are pressured and coerced with repeated questions and demands for demonstrating with the dolls, there will likely be a much greater proportion of false reports.

The efforts to develop protocols (White, et al., 1985) represent attempts to improve the ways the dolls have typically been used in interviews of children who are alleged to be sexually abused.  But if the underlying assumptions are unsupported, the use of the dolls will not increase the accuracy of the information obtained.  At best, use of the protocols may prevent some of the most blatant errors that have been observed with the dolls in such interviews (Wakefield & Underwager, 1988).  However, they are likely to lull interviewers into falsely believing they have a valid and reliable method of obtaining accurate information.  At worst, they are likely to remain teaching devices which contaminate whatever information is obtained, teach non-abused children explicit sexual behavior, and train emotionally healthy children to believe and act as though they had been sexually abused.

The American Psychological Association's Committee on Psychological Testing and Assessment determined in its March, 1988 meeting that the dolls "are considered to be a psychological test and are subject to the standards when used to assess individuals and make inferences about their behavior" (Landers, 1988).  The significance of this determination is that any psychologist who uses the dolls without making a prior statement as to their lack of reliability and validity to the adult caretaker of the child is vulnerable to an accusation of negligent and unethical practice.  Any psychologist who testifies in court without making a clear statement about their limitations and the necessary qualifications in interpreting what behavior with them means is vulnerable to an accusation of unethical practice.  Any social worker or law enforcement official who uses them is vulnerable to criticism for exceeding his level of competence.

The situation is roughly similar to the use of phrenology, that is, the belief that the bumps on a person's head could be used to understand the personality and predict the behavior of an individual.  Phrenologists developed very precise maps of the skull and the differential location of bumps, the size, and the height of the bumps, are carefully and painstakingly detailed with attendant meanings that are supposed to be diagnostic.  The difficulty was that the fundamental assumptions of phrenology are false.  No matter how much precision is claimed in the implementation of the theory, the whole thing is nonsense because the beginning is nonsense.


This study indicates that assumptions underlying the use of the anatomically correct dolls are not supported.  Non-abused and abused children did not differ in their comments about the dolls or their behaviors and play with them.  Most children did not identify gender (of themselves or of the dolls) in terms of primary sexual characteristics.  Children frequently demonstrated aggressive and sexual acts that had not occurred; this was particularly marked when the interviewer asked leading questions and cued and encouraged the child.  The behaviors obtained were of a nature that would lead many interviewers to believe that the child had, in fact, been sexually abused.

Our results suggest that the information obtained by the use of these dolls in interviews is misleading.  The dolls are likely to increase the error and decrease the reliability of the information gathered.  Any information obtained from such interviews should be discarded.  There is nothing to support their use as diagnostic or assessment tools.

The dolls are not generally accepted in the scientific community and nothing obtained from their use should ever be admitted as evidence in any legal setting.  A California Appeals Court ruled in 1987 that the use of the dolls was not supported by the scientific evidence and their use did not meet the Frye test for admissibility.  Testimony based upon the use of the dolls was therefore ruled inadmissible (Law Week, 1987).

It is also the case, in our opinion, that with the present status of the research evidence related to the use of the dolls, any psychologist who uses them as assessment and diagnostic tools and bases any expression of opinion about an allegation of child sexual abuse primarily on use of the dolls is guilty of medical malpractice and unethical conduct under the code of ethics of the American Psychological Association.  The dolls simply do not have the necessary demonstrated validity and reliability to be used as diagnostic or assessment devices.


Aman, C. & Goodman, G. S. (1987, September). Children's use of anatomically detailed dolls: An experimental study. Symposium on Interviewing Children, Washington, D.C.

Boat, B. W. & Everson, M.D. (undated). The Anatomical Doll Project: An Overview. Department of Psychiatry, University of North Carolina, Chapel Hill.

Friedemann, V. M. & Morgan, M. K. (1985). Interviewing Sexual Abuse Victims Using Anatomical Dolls: The Professionals' Guidebook (Paperback). Eugene, OR: Migima Designs, Inc.

Gabriel, R. M. (1985). Anatomically correct dolls in the diagnosis of sexual abuse of children. The Journal of the Melanie Klein Society, 3(2), 40-51.

Herbert, C. P., Grams, G. D., & Goranson, S. E. (1987). The Use of Anatomically Detailed Dolls in an Investigative Interview: A Preliminary Study of "Non-Abused" Children. Department of Family Practice, University of British Columbia. Vancouver, B. C.

Jampole, L. & Weber, M. K. (1987). An assessment of the behavior of sexually abused and non-abused children with anatomically correct dolls. Child Abuse & Neglect, 11, 187-194.

Jensen, J. B., Realmuto, G. & Wescoe, S. (1986, October). Paper presented at the American Academy of Child Psychiatry, Washington, DC

King. M.A., & Yuille, J. C. (1987). Suggestibility and the child witness. In S. J. Ceci, M. P. Toglia, & D. F. Ross (Eds.) Children's Eyewitness Memory (Out of Print) (pp.25-35). New York: Springer-Verlag.

Landers, S. (1988) Use of "detailed dolls" questioned. APA Monitor, 19, 24-25.

Law Week (1987, May 26). Techniques used to diagnose child sexual abuse held subject to Frye Test.

Sivan, A. B., Schor, D. P., (1987 August). Children's labels for sexually related body parts, Poster presented at the American Psychological Association convention, New York, New York.

Sivan, A. B., Schor, D. P., Koeppl, G. L., & Noble, L. D. (1988). Interaction of normal children with anatomically-correct dolls. Child Abuse & Neglect, 12, 295-304.

Wakefield, H., & Underwager, R. (1988). Accusations of Child Sexual Abuse (Hardcover)(Paperback). Springfield, IL: C.C. Thomas.

White, S., Strom, G. S. & Santilli, G. (1985, October). Interviewing young sexual abuse victims with anatomically correct dolls. Paper presented at the 32nd Annual Meeting of the American Academy of Child Psychiatry, San Antonio, Texas.

White, S., Strom, G. S., & Santilli, G. & Halpin, B. M. (1986). Interviewing young sexual abuse victims with anatomically correct dolls. Child Abuse & Neglect, 10, 519-529.

Underwager, R., Wakefield, H., Legrand, R., Bartz, C. & Erickson, J. (1986, August). The role of the psychologist in the assessment of cases of alleged sexual abuse of children. Paper given at the 94th Annual Convention of the American Psychological Association, Washington, D.C.

* William Mclver is a psychologist in private practice. His address is 1910 Norwood, Eugene Oregon, 97401.

Hollida Wakefield is a licensed psychologist and Ralph Underwager is a licensed consulting psychologist at the Institute for Psychological Therapies, 2344 Nicollet Avenue South, Suite 170, Minneapolis, Minnesota 55404.  [Back]

[Back to Volume 1, Number 1]  [Other Articles by these Authors]

Copyright 1989-2014 by the Institute for Psychological Therapies.
This website last revised on April 15, 2014.
Found a non-working link?  Please notify the Webmaster.