Journal of Applied Psychology

2000 by the American Psychological Association

February 2000 Vol. 85, No. 1, 38-49

Allegations of Wrongdoing
The Effects of Reinforcement on Children's Mundane and Fantastic
Claims
Sena Garven
Department of Psychology University of Texas at El Paso
James M. Wood
Department of Psychology University of Texas at El Paso
Roy S. Malpass
Department of Psychology University of Texas at El Paso
ABSTRACT

S. Garven, J. M. Wood, R. S. Malpass, and J. S. Shaw (1998) found that the interviewing
techniques used in the McMartin Preschool case can induce preschool children to make
false allegations of wrongdoing against a classroom visitor. In this study, 2 specific
components of the McMartin interviews, reinforcement and cowitness information, were
examined more closely in interviews of 120 children, ages 5 to 7 years. Children who
received reinforcement made 35% false allegations against a classroom visitor, compared
with 12% made by controls. When questioned about "fantastic" events (e.g., being taken
from school in a helicopter), children receiving reinforcement made 52% false
allegations, compared with 5% made by controls. In a second interview, children
repeated the allegations even when reinforcement had been discontinued. The findings
indicate that reinforcement can swiftly induce children to make persistent false
allegations of wrongdoing.

During the 1980s, a series of highly publicized "daycare ritual abuse cases" erupted across the United
States and Europe ( Nathan & Snedeker, 1995 ). In widely separated locales, groups of preschool children
made similar bizarre allegations, claiming that they had been sexually abused by their teachers and forced
to participate in fantastic ceremonies, often with Satanic overtones. The first such case to receive national
attention was in Manhattan Beach, California. Seven teachers at the McMartin Preschool were accused of
kidnapping children and flying them in a helicopter to an isolated farm, where animals were tortured and
the children were forced to engage in group sex. All charges were eventually dropped against five of the
teachers. The remaining two were tried in criminal court but not convicted.
The prosecution in the McMartin case relied heavily on videotaped interviews of children. However, these
interviews eventually undermined the prosecution's case. After the trial, jurors publicly criticized them as
leading and suggestive ( Reinhold, 1990 ; Timnick & McGraw, 1990 ; Wilkerson & Rainey, 1990 ).
Popular-press books and articles ( Eberle & Eberle, 1993 ; Hicks, 1991 ; Nathan & Snedeker, 1995 ; Tavris,
1997 ) and academic writings ( Ceci & Bruck, 1993 , 1995 ; Green, 1992 ; Mason, 1991 ; but see Faller,
1996 ; Summit, 1994 ) have also criticized the McMartin interviews. In addition, archival and experimental
studies by Wood, Garven, and their colleagues have explored these interviews more systematically, as
discussed below.
Wood et al. (1998) compared transcripts of the McMartin interviews with transcripts of sexual abuse
interviews in an ordinary child protection service (CPS). Quantitative analyses confirmed the impressions
of previous authors: The McMartin interviews were found to be substantially more suggestive than CPS

interviews. In addition, the McMartin interviews were characterized by repeatedly telling children what
other witnesses in the case had already said and providing children with frequent positive reinforcement for
making allegations.
Garven, Wood, Malpass, and Shaw (1998) followed up on the archival research of Wood et al. (1998) using
experimental methods. Garven et al. (1998) questioned preschool children about a classroom visitor, using
interviewing techniques from the McMartin Preschool case. The children who were questioned using the
McMartin techniques made 58% false allegations against the classroom visitor, as compared with 17%
false allegations made by children in a control group. Garven et al. concluded that the McMartin interviews
contained social incentives (i.e., reinforcement and cowitness information) that could induce high error
rates among preschool children in a matter of minutes.
The present experiment set out to extend these findings using a similar methodology. Again, interviewing
techniques from the McMartin Preschool case were used to question children about a stranger's visit to their
classroom. However, the present study addressed three new issues that were not resolved by the original
experiment.

Central Issues of This Study

Reinforcement Versus Cowitness Information
Garven et al. (1998) used several interviewing techniques from the McMartin case simultaneously to
question children. By using a "package" of techniques, the original study could closely mimic the
McMartin interviews and show a clear and robust effect on children's statements. However, some important
theoretical questions could not be addressed: Which of the various McMartin techniques caused children's
error rates to increase? Did some techniques have a stronger effect than others?
To address these theoretical questions, two components of the McMartin package were singled out for
closer examination in the present study: reinforcement and cowitness information. The experimental
strategy was to study these two components individually and in combination, and thus "decompose" their
effects. We hypothesized that if interviewers used reinforcement or cowitness information when
questioning children about a classroom visitor, then each component would independently increase
children's false allegations. In addition, we hypothesized that if the two components were combined, the
effect might be synergistic. The two components, reinforcement and cowitness information, are described
more fully in the paragraphs below.
Reinforcement.
The first component examined in the present study was reinforcement, as operationalized in the
interviewing techniques of Positive and Negative Consequences ( Wood et al., 1998 ). The technique of
Positive Consequences consists of giving, promising, or implying praise, approval, agreement, or other
rewards to a child for saying or doing something, or indicating that the child will demonstrate desirable
qualities (e.g., helpfulness, intelligence) by saying or doing something. A simple "yes" by an interviewer,
indicating that the interviewer has understood the child, would not be considered Positive Consequences.
Similarly, a compliment to the child at the beginning of an interview ("What pretty eyes you have!") would
not usually be considered Positive Consequences, because such a compliment is not conditional on the
child's saying or doing anything.
Many examples of Positive Consequences can be found in the McMartin interviews. For example, in one
interview a child begins to agree with suggestive questions and the interviewer says, "You have got one of
the best memories of any kid that's been here. You are really doing a good job." (Interview Number 100, p.
39). 1

The technique of Negative Consequences consists of criticizing or disagreeing with a child's statement, or
otherwise indicating that the statement is incomplete, inadequate, or disappointing. Simple repetition of a
question would not usually be considered Negative Consequences unless surrounding parts of the interview
indicate that the interviewer is being argumentative. Striking examples of Negative Consequences appear in
the McMartin interviews. One example is the following:
I = Interviewer. C = Child.
I: Heck, everybody was playing naked games in their school. Then they were just little
kids, and they played, too, and some of the naked games were fun. The kids had a good
time. And they were kind of silly. Do you remember that Bear [a puppet], some of those
fun silly games?
C: [Shakes Bear puppet's head, "no".]
I: Oh, Bear, maybe you don't have a very good memory ...
C: [Laughs]
I: ... and your memory must not be as good as Patsy's friend's memories ...
(Interview Number 111, pp. 1920)
Within the framework of learning theory, the interviewing technique of Positive Consequences may be
viewed as a way of delivering positive reinforcement, whereas the technique of Negative Consequences
may be viewed as a way of delivering punishment. A positive reinforcer increases, and a punishment
decreases, the probability that a behavior will be repeated ( Ettinger, Crooks, & Stein, 1994 ). A reinforcer
or punishment that comes from another person is by definition a social reinforcer or social punishment.
Zigler and Kanzer (1962) reported that middle-class children were more apt to change their behavior for a
verbal (social) reinforcer that emphasized correctness ("correct," "right") rather than general praise (e.g.,
"good," "fine"), an effect referred to as Zigler's valence theory of social reinforcement ( Spence, 1973 ).
Gilboa and Greenbaum (1978) found that a "warm" adult was more influential in affecting learning than a
"cold" one, especially when verbal reinforcers emphasized correctness ("correct") over praise ("nice").
The effects of positive reinforcement and punishment on children's accuracy during interviews has received
only limited attention from researchers. Three decades ago, in Behavior Modification in the Natural
Environment, Tharp and Wetzel (1969 , p. 76) reported giving children candy to reward answers during
interviews. Tharp and Wetzel noted anecdotally that some children began to respond at random simply to
obtain the candy. These researchers recommended that interviewers could avoid this problem by rewarding
only "appropriate responses." Of course, such a recommendation is highly problematic in forensic settings.
Goodman, Bottoms, Schwartz-Kenney, and Rudy (1991) considered reinforcement to be "a form of social
support" (p. 72) and operationally defined it as "beginning the interview with cookies and juice" and "the
interviewer being warm and friendly, smiling a lot, and giving the child considerable praise such as 'You're
doing a great job' or 'You've got a great memory' " (p. 78). Goodman et al. predicted that social support of
this type, given randomly, would reduce children's stress and increase their accuracy during an interview
about a stressful event. The prediction was confirmed by some but not all of the findings in their study,
suggesting that the effect of social support was not very strong.
In a related study, Carter, Bottoms, and Levine (1996) examined the effect of social support on children's
accuracy. These researchers intentionally used an operational definition of social support that was similar to
the definition of reinforcement used by Goodman et al. (1991) . Specifically, in the study by Carter et al.,
the interviewer in the social-support condition introduced himself at the beginning of the interview, smiled

frequently, used a warm tone of voice, maintained eye contact, and sat in a relaxed manner with open
posture. However, the interviewer did not praise children's answers during the interview, as in the study by
Goodman et al. Carter et al. found that children interviewed with social support were generally more
accurate than children in a control group.
Cowitness information.
The second component of the McMartin package examined in the present study was cowitness information,
as operationalized in the interviewing technique of Other People ( Wood et al., 1998 ). The technique of
Other People consists of telling the child that the interviewer has already received information from another
person regarding the topics of the interview. For example, in one McMartin interview, the interviewer tells
the child that "you know that all the kids told us that they got touched in yucky places" (Interview Number
100, p. 46). Another interview (Interview Number 101, p. 66) unfolds as follows:
I: ... Mr. Floppy Family person [a puppet]. Have you ever seen anything so floppy as
that? That's ridiculous, isn't it?
C: I'm going to choke him.
I: You're going to choke him? You know, some of the kids said they got choked.
By telling a child about the statements of other people, an interviewer may create pressures toward
conformity, "the tendency to change or modify our own behaviors so that they are consistent with those of
other people" ( Ettinger et al., 1994 , p. 685). Binet (cited in Siegler, 1992 ) discovered that children's
statements regarding factual matters can be influenced by conformity. Binet showed a group of children
one card with a single line, and a second card with several lines, then asked the children to choose the line
on the second card that matched the line on the first card. In the first few trials there was an obvious correct
answer. Later trials had no matching line. Nevertheless, children often agreed with a child who had
emerged as an unofficial leader, even when the leader was obviously wrong.
Pynoos and Nader (1989) , conducting interviews at a school that had been attacked by a sniper, found that
some children absent from school during the attack gave fabricated stories of having been present.
Apparently the children had heard accounts of the attack from their parents, other children, or news reports,
and created stories to match. An unpublished study by Pettit, Fegan, and Howie (as described in Ceci &
Bruck, 1995 , pp. 9091) reported similar findings. After a staged classroom event, several children who
had not been present made reports as if they had. Apparently they concocted stories on the basis of what
they had heard from other children, although it is also possible that they were responding to leading
questions by interviewers who had preconceptions about what had happened.
The influence of cowitness information on children is probably best understood in light of the much more
extensive literature on cowitness information in adults. In a classic study that has been replicated many
times, Asch (1956 ; see also Larsen, 1991 ) demonstrated that adults' reports often conform to a group
norm. Shaw, Garven, and Wood (1997) found that the immediate memory reports of an adult eyewitness
could be substantially affected by the statements of another witness, an effect that remained stable after a 2day delay. Luus and Wells (1994) found that eyewitnesses who made false identifications from a
photospread became more confident in their choice if they were told that a cowitness had made the same
identification.
Effects on Children's Subsequent Statements
Because children were interviewed only once in the original study by Garven et al. (1998) , important
forensic questions were left unanswered: Can interviewing techniques like those from the McMartin case
exert a lasting effect on children's statements? If a child is questioned once with these techniques, but

reinterviewed later without them, will the second interview be "tainted"? Such questions can be of great
importance in real sexual abuse cases.
To address these issues, children were interviewed twice in the present study. Approximately half of the
children who had been questioned using reinforcement (Positive and Negative Consequences) and
cowitness information (Other People) during the first interview were reinterviewed without these
techniques. We hypothesized that these children would continue to have a high error rate in the second
interview, even though the reinforcement and cowitness information were no longer present.
There are several reasons to expect that reinforcement and cowitness information can exert a lasting effect
on children's statements. First, behavior that has been learned and reinforced (e.g., making false allegations)
is likely to persist even after reinforcement has stopped. Second, social psychological research on
consistency and commitment ( Cialdini, 1993 ) suggests that children who make initial false reports might
be likely to stick with their story. Third, research indicates that postevent misinformation can have a lasting
effect on reports by both children and adults ( Cassel, Roebers, & Bjorklund, 1996 ; Ceci & Bruck, 1995 ;
Loftus & Davies, 1984 ; Poole & Lindsay, 1995 ) and that misinformation from other witnesses can have a
similar enduring effect ( Luus and Wells, 1994 ; Shaw et al., 1997 ). Therefore, the effect of cowitness
information (Other People) would be expected to exert a lasting influence on children's reports. In general,
reinforcement and conformity are among the most powerful effects on social behavior of children and
adults that have been found in more than 50 years of research on these topics.
Fantastic and Mundane Allegations
In the original study by Garven et al. (1998) , 64% of children assented to false allegations of wrongdoing
(e.g., "Did he steal a pen from the teacher's desk?"), and 44% assented to false allegations of touching
("Did he put a sticker on your knee?"). However, these allegations were rather mundane when compared
with some of the fantastic and bizarre accusations made by children in the McMartin Preschool and other
similar cases ( Nathan & Snedeker, 1995 ). For example, some McMartin children alleged that they had
been flown away in a helicopter or plane or that they had been taken to a farm full of animals where they
had witnessed a horse being killed with a baseball bat.
Although Garven et al. (1998) found that the McMartin interviewing techniques could induce children to
make false allegations of mundane wrongdoing, the question remains whether the same techniques could
induce fantastic allegations like those reported in the McMartin case. To explore this question, the present
study attempted to elicit both mundane and fantastic allegations from children. For example, children in the
present study were asked whether the visitor to their classroom had taken them on a helicopter ride and to a
farm. We hypothesized that reinforcement and cowitness information would cause some children to assent
to these fantastic allegations. However, we expected that fewer children would assent to the fantastic
allegations than to the mundane ones. Specifically, we expected that most children in the study would view
the fantastic allegations as absurd and reject them.
Overview of This Experiment
In the present experiment a young man visited five grade schools, read a story, and distributed treats. One
week afterward, children were interviewed about his visit. Interviewing techniques involving reinforcement
and cowitness information, taken from the McMartin Preschool case, were used to question some children.
Two to 3 weeks after the initial interview, children at four of the schools were interviewed a second time.
Half the children who had been interviewed with reinforcement and cowitness information during the first
interview were interviewed without these techniques during the second interview. The study tested three
main hypotheses, as outlined above. Specifically, we hypothesized that (a) reinforcement and cowitness
information would each independently cause children to make false allegations against the classroom
visitor, (b) children interviewed with reinforcement or cowitness information during the first interview
would continue to make false allegations during a second interview in which there was no reinforcement or
cowitness information, and (c) reinforcement and cowitness information would cause some children to

make bizarre or fantastic allegations, although such fantastic allegations would be less frequent than
mundane allegations.

Method
Participants
Participants were children attending kindergarten and first grade at five different elementary schools.
Children were excluded from the study if they were less than 5 years old or had to be interviewed in
Spanish. The children in the remaining sample ( N = 120) ranged in age from 5 to 7 years, including thirtynine 5-year-olds (mean age = 65.75 months), sixty-seven 6-year-olds (mean age = 77.08 months), and
fourteen 7-year-olds (mean age = 85.39 months). There were 51 boys and 69 girls in the sample. Informed
consent was obtained from a parent or guardian of each child before the interview session.
Procedure
Children at each elementary school attended a special story time led by a male undergraduate student
introduced as Paco Perez. Paco, wearing an enormous colored hat, was presented by a teacher who
mentioned his name several times. Paco then said:
Hi. I'm Paco and I'm here to tell you a story. The name of the story is The Hunchback of
Notre Dame and I want you all to sit quietly and listen. How many of you saw the movie?
Did you like it? After the story I brought some special treats to share with you. But first I
have to take off my hat. Isn't it a silly hat?
Paco also put on a pair of goofy glasses with a large plastic nose and mustache and proceeded to tell the
story in an engaging way. He then placed a sticker on the back of each child's hand and handed out treats to
each child. Finally Paco said goodbye to the class and left. The entire visit took approximately 20 min and
was videotaped.
Children were interviewed two times regarding Paco's visit. The first interview was 1 week after Paco's
story time. The second interview was 1421 days after the first interview. Children were interviewed
individually on videotape and audiotape, away from their regular classrooms but in the same building. A
different person interviewed each child at the first and second interview. The McMartin interviews began
with extended rapport building. To create a similar warm atmosphere, the interviewers in the present study
adopted the social support techniques of Carter et al. (1996) , as described above.
Manipulation and Design
Type of interview was the independent variable, and the number of times the child answered "yes" to
misleading questions (mundane or fantastic) was the dependent variable. For the first interview, the
experimental design was a 2 _ 2 between-subjects design. Each child was interviewed in one of four ways:
(a) reinforcement only, (b) cowitness information only, (c) both reinforcement and cowitness information,
and (d) a suggestive control condition ( Garven et al., 1998 ). In the reinforcement-only condition, the
interviewer praised children when they answered "yes" to a question (Positive Consequences) and
expressed mild disappointment if the child answered "no" (Negative Consequences). In the cowitnessinformation-only condition, the interviewer told the child what other kids had supposedly said about a
question (Other People). In the reinforcement-and-cowitness-information condition, children were exposed
to both praise and mild disappointment from the interviewer, as well as the mention of the other children
(Positive and Negative Consequences and Other People). Finally, in the suggestive-control condition,
children were asked questions that were solely suggestive in nature, without praise or disappointment from
the interviewer or any mention of the other children.

The questioning format of the reinforcement-and-cowitness-information condition is illustrated by the

following excerpt (interview techniques are identified in brackets):
I: ... The other kids say that Paco took them somewhere on a helicopter [Cowitness
Information]. Did Paco take you somewhere on a helicopter?
C: No.
I: You're not doing good. Um-uh (negative) [Reinforcement]. The other kids say that
Paco took them to a farm [Cowitness Information]. Did Paco take you to a farm?
C: Yes.
I: Great. You're doing excellent now [Reinforcement]. The other kids say that Paco
showed them the animals on the farm [Cowitness Information]. Did Paco show you the
animals on the farm?
C: Yes.
I: Great. You're doing excellent [Reinforcement]. One last question. The other kids say
that Paco took them on a horse ride [Cowitness Information]. Did Paco take you on a
horse ride?
C: Yes.
The questioning format of the suggestive-control condition is illustrated by the following excerpt:
I: Did Paco break a toy while he was visiting?
C: No.
I: Did Paco tickle your tummy?
C: No.
As may be seen, this child in the suggestive-control condition was asked simple suggestive questions,
without any feedback from the interviewer or mention of other children.
Children were asked 8 misleading questions concerning mundane items, 4 leading correct questions, and 4
misleading questions concerning fantastic items. Thus each child was asked 16 questions. Specifically, the
8 misleading mundane items were whether Paco (a) tore a book, (b) stole a pen from the teacher's desk, (c)
broke a toy, (d) tickled the child on the tummy, (e) told the child a secret, (f) threw a crayon at a child who
was talking, (g) said a bad word, and (h) kissed the child on the nose.
Interspersed among these mundane leading incorrect items were four leading correct items (questions about
things Paco did do). Specifically, children were asked if Paco (a) told the children to sit quietly and listen,
(b) took off his hat, (c) put on goofy glasses, and (d) put a sticker on the child's hand. The four final
questions of each interview were what we deemed the fantastic items. The four misleading fantastic items
were whether (a) Paco took the child on a helicopter ride, (b) Paco took the child to a farm, (c) the child
saw animals on a farm, and (d) Paco took the child on a horse ride.

The main purpose of the second interview was to investigate the extent to which being reinforced or
hearing about the other children in the first interview would carry over to the second interview. Therefore,
during the second interview, half of the children were questioned using the same format that had been used
during the first interview. That is, if they had been questioned using reinforcement during the first
interview, the second interview was a repeat of the same format. The other half of the children were
interviewed with what was dubbed the new format. This consisted of asking the children about the 16 items
without either reinforcement or cowitness information. Thus, some children who had been interviewed with
reinforcement in the previous interview were interviewed the second time with questions that were solely
suggestive, (i.e., essentially the same format that was used with the control group in the first interview).
Half of the children in the suggestive-control condition were interviewed with the new format during the
second interview, whereas the other half were interviewed using the same format as the first interview.
Thus, no children in the suggestive-control condition were exposed to reinforcement or cowitness
information during either interview.
All children were randomly assigned to the repeat or the new format for the second interview. For the
children assigned to the new format in the second interview, the topic of Paco's visit was introduced with
this paragraph:
Remember the day Paco Perez came and read you the Hunchback of Notre Dame? He
had on a silly hat didn't he? Well, I know another lady came already and asked you some
questions but some of the things she said might not have really happened. I wasn't there
that day and I'd like you to answer some questions about what happened when he visited,
okay?
Another purpose of the second interview was to see if children who answered "yes" to the interviewer's
questions would stick with their story if they were mildly challenged and asked to provide more
elaboration. Thus the interviewer kept track of "yes" answers during the initial portion of the second
interview. After asking the child all 16 questions, the interviewer returned to the questions to which the
children had answered "yes" and challenged the child with a follow-up question. For example, "You said
that Paco stole a pen from the teacher's desk. Did you see that happen, or just hear about it?" For the
fantastic elements only, if the child affirmed that he or she had personally observed the event, he or she was
asked to elaborate by telling the interviewer details. For example:
I: You said that Paco took you on a helicopter ride. Did you see that or just hear about it?
C: I saw it.
I: Can you tell me more about that?

Results
First Interview Manipulation check.
To confirm that reinforcement and cowitness information had actually been used in the appropriate
interviews, audiotapes were scored using a system developed by Wood et al. (1998) . 2 Raters were blind to
the experimental hypotheses and design of the experiment. The scoring categories were Positive
Consequences (PC), Negative Consequences (NC), and Other People (OP). One rater scored interview
transcripts for 32 (26%) of the 120 interviews, and 17 (53%) of these 32 were independently rescored by a
second rater. Interrater agreement as measured by kappa was .95 for PC, .91 for NC, and .96 for OP.
According to ratings by the primary scorer, PC occurred an average of 9.56 times per interview in the
reinforcement conditions versus 1.00 times per interview in the no-reinforcement conditions, F (1, 29) =
39.41, p < .001, 2 = .585. The corresponding numbers were 5.44 versus .25 for NC, F (1, 29) = 19.95, p <
.001, 2 = .416, and 17.0 versus 0.0 for OP, F (1, 29) = 29,131.2, p < .001, 2 = .999. These results

confirm that children in the reinforcement and cowitness-information conditions received a strong "dose"
of the manipulations, but children in the no-reinforcement and no-cowitness-information conditions did
not.
Scoring of responses.
The dependent variable for this study was the percentage of times the children answered "yes" to a
particular type of question (i.e., misleading mundane, misleading fantastic, or leading correct). Each answer
by a child to an interviewer question was scored as "yes," "no," or "other." "Yes" was scored when a child
agreed either verbally or nonverbally with the main concept of the question. "No" was scored when the
child disagreed with the main concept. All ratings were done from videotape so that the child's nonverbal
and verbal responses could be evaluated. One rater scored all 120 interviews, and 34 (28%) of these 120
were independently rescored by a second rater. Neither rater had participated in the manipulation check.
Interrater agreement as measured by kappa was .93 for "yes" answers. "No" and "other" answers were not
included in the present analyses and therefore are not further described here.
Preliminary analysis.
Because interviews took place in five different locations and there were three separate interviewers, a
preliminary 5 (location) _ 3 (interviewer) between-participants analysis of variance (ANOVA) was first
conducted, with the number of "yes" answers as the dependent variable. No significant main effect was
found for location or interviewer. Consequently, the data were collapsed across location and interviewer for
all subsequent analyses.
Misleading mundane questions.
A 2 (reinforcement) _ 2 (cowitness information) between-participants ANOVA was performed on the
percentage of "yes" answers children gave concerning the eight misleading mundane questions. When
reinforcement was present, children answered "yes" to misleading mundane questions 34.68% of the time.
When no reinforcement was present, children answered "yes" 12.50% of the time, F (1, 115) = 16.73, p <
.001, 2 = .127.
The main effect of cowitness information, although not as large as reinforcement, was also statistically
significant, F (1, 115) = 3.93, p = .050, 2 = .033. When cowitness information was present, children
answered "yes" to misleading mundane questions 29.44% of the time. When no cowitness information was
present, the children answered "yes" 18.75% of the time. There was no significant interaction between
reinforcement and cowitness information.
Misleading fantastic questions.
A 2 (reinforcement) _ 2 (cowitness information) between-participants ANOVA was performed on the
percentage of "yes" answers children gave concerning the four misleading fantastic questions. When
reinforcement was present, children answered "yes" to misleading fantastic questions 51.61% of the time.
When no reinforcement was present, children answered "yes" 4.91% of the time, F (1, 114) = 51.12, p <
.001, 2 = .310. There was no main effect for cowitness information or any significant interaction on
misleading fantastic items.
We had predicted that children would make more false allegations for mundane events than for fantastic
events. To test this hypothesis, a within-subjects ANOVA was performed using only children who had
received reinforcement ( n = 62), with type of question (mundane or fantastic) as the within-subjects factor.
Percentage of "yes" answers to each type of question was the dependent variable. The results were the
opposite of what we had predicted: Children answered "yes" significantly more often to fantastic questions

than to mundane questions, F (1, 61) = 24.42, p < .001, 2 = .286. As mentioned above, children answered
"yes" to 51.61% of fantastic questions and to 34.67% of misleading mundane questions.
Leading correct questions.
All children were asked four leading correct questions, in which the information provided by the
interviewer was accurate (e.g., "Did Paco tell you to sit quietly and listen?"). A 2 (reinforcement) _ 2
(cowitness information) between-participants ANOVA was performed on the percentage of "yes" answers
children gave concerning the four leading correct questions. Children gave correct "yes" answers 93.14% of
the time when reinforcement was present and 86.66% of the time when no reinforcement was present, F (1,
115) = 11.18, p < .001, 2 = .089. There was no main effect for cowitness information on leading correct
items. However, it should be noted that children were very accurate in both conditions. Children in the
cowitness information conditions gave correct "yes" answers 88.13% of the time when cowitness
information was present and 86.66% of the time when cowitness information was not present. There was
no significant interaction between reinforcement and cowitness information on leading correct questions.
Individual item results.
Figure 1 shows children's error rates for the 12 misleading items (8 mundane and 4 fantastic) in the first
interview. As may be seen, children in the two groups receiving reinforcement had strikingly higher error
rates from the fourth item onward (i.e., after only 1 to 2 min of questioning). Figure 1 illustrates how
swiftly reinforcement can lower children's accuracy, as they are "shaped" to give answers that please the
interviewer.
Second Interview Manipulation check.
The manipulation check for the second interview was performed exactly as for the first. One rater scored
interview transcripts for 33 (35%) of the 93 interviews, and 18 (54%) of the 33 were independently
rescored by a second rater. Interrater agreement as measured by kappa was .92 for PC, .78 for NC, and 1.00
for OP. According to ratings by the primary scorer, PC occurred an average of 14.13 times per interview in
the reinforcement conditions versus 1.50 times per interview in the no-reinforcement conditions, F (1, 15)
= 43.57, p < .001, 2 = .757. The corresponding numbers were 3.38 versus .13 for NC, F (1, 15) = 3.27, p =
.072, 2 = .189, and 15.22 versus 0.00 for OP, F (1, 14) = 6,762.6, p < .001, 2 = .998. These results
confirm that children in the second interview received a strong dose of positive reinforcement and
cowitness information. However, negative consequences were delivered much less frequently, and the
between-groups difference only approached statistical significance.
Scoring of responses.
There were two dependent variables of interest in the second interview: the percentage of times the children
answered "yes" to questions and the percentage of times the children said "yes" to the challenge question.
The protocol for the second interview varied slightly from the protocol of the first interview. As in the first
interview, children were asked questions concerning 16 items. However, for the second interview, the
interviewer kept track of "yes" answers during the initial portion of an interview and afterward returned to
those items and asked the child a follow-up challenge question (e.g., "You said that Paco stole a pen from
the teacher's desk. Did you see that, or did you hear about that?"). For the misleading fantastic items, if the
child indicated that the event had really happened (i.e., the child had actually flown in a helicopter), the
child was asked to elaborate ("Can you tell me more about that?").
Answers for the core set of 16 items were scored exactly as they had been for the first interview. One rater
scored all 93 interviews, of which 37 (40%) were independently rescored by a second rater. Interrater
agreement as measured by kappa was .99 for "yes" answers. Answers to the challenge questions were
scored by having raters determine whether the child's answer indicated that the child actually observed the
event or only heard about it occurring. Interrater agreement for this scoring as measured by kappa was .85.

Preliminary analysis.
Children from four of the five schools ( n = 93) were interviewed a second time. Because interviews took
place in four different locations and there were two separate interviewers, a preliminary 4 (location) _ 2
(interviewer) between-participants ANOVA was first conducted, with the number of "yes" answers as the
dependent variable. No significant main effect was found for location or interviewer. Consequently, the
data were collapsed across location and interviewer for all subsequent analyses.
Main analyses.
As mentioned previously, in the second interview, half of the children ( n = 47) were questioned using the
same technique as in the first interview. For example, if the child was questioned using reinforcement in the
first interview, he or she was questioned using reinforcement in the second interview as well; this was
named the repeat condition. The other half of the children ( n = 46) were interviewed using only suggestive
questions. That is, if the child had been interviewed in the first interview using reinforcement and
cowitness information, in the second interview the child was interviewed using suggestive questions alone;
this was named the new condition. All children in the suggestive-control condition were interviewed with
suggestive questions in the second interview, although the format was slightly different for the repeat and
new conditions, as described in the Method section.
A 2 (cowitness information in first interview) _ 2 (reinforcement in first interview) _ 2 (repeat or new
interview) ANOVA was performed for each of the question types (misleading mundane, misleading
fantastic, and leading correct). The results of the analyses were very similar to those for the first interview.
For misleading mundane items, there was an effect for reinforcement only, F (1, 85) = 42.12, p < .001, 2
= .331. Children in the reinforcement condition answered "yes" to misleading mundane questions 52.71%
of the time, whereas children in the no-reinforcement condition answered "yes" 6.11% of the time. There
were no main effects for cowitness information or repeat on misleading mundane items and no interactions.
For the misleading fantastic items, there was again an effect for reinforcement only, F (1, 85) = 45.72, p <
.001, 2 = .350. Children in the reinforcement condition answered "yes" to misleading fantastic questions
61.95% of the time, whereas children in the no-reinforcement condition answered "yes" 6.91% of the time.
There were no effects for cowitness information or repeat on misleading fantastic items. There were no
interactions.
As in the first interview, we wished to compare error rates for misleading mundane versus misleading
fantastic questions. A within-subjects ANOVA was performed with type of question (mundane and
fantastic) as the within-subjects factor, using only children who had received reinforcement at the first
interview. As in the first interview, a significant difference was found, but in the direction opposite from
what was predicted, F (1, 45) = 5.16, p = .028, 2 = .103. Specifically, children answered "yes" to mundane
questions 52.71% of the time, but to misleading fantastic questions 61.95% of the time.
For leading correct items, main effects were observed for both reinforcement and repeat. Children in the
reinforcement condition correctly answered "yes" to leading correct questions 96.19% of the time, whereas
children in the no-reinforcement condition answered "yes" to leading correct questions 88.29% of the time,
F (1, 85) = 5.14, p = .026, 2 = .057. Children in the repeat condition correctly answered "yes" to leading
correct questions 96.19% of the time, whereas children in the no-repeat condition answered "yes" 88.29%
of the time, F (1, 85) = 4.94, p = .029, 2 = .055.
The large effect of reinforcement, and the lack of effect of repeat for both the misleading mundane and
fantastic questions, indicates that children who received reinforcement in the first interview were very
likely to answer "yes" again in the second interview, whether reinforcement continued or not. To see
whether the same children were answering "yes" in both interviews, Pearson correlations were calculated
between the number of "yes" answers in the first and second interviews for those children who received
reinforcement at least once. For children in the repeat condition, there were strong correlations between

"yes" answers to misleading mundane questions in the first and second interviews ( r = .772, p < .01) and
between "yes" answers to misleading fantastic questions in the first and second interviews ( r = .825, p <
.01). For children in the new condition, the corresponding correlations were .708 and .871 (both p s < .01).
Thus, the same children who answered "yes" to misleading questions in the first interview were likely to
answer "yes" again in the second interview.
Challenge questions and elaborations.
After the children had answered the 16 core questions in the second interview, the interviewer returned to
those items to which the child had replied "yes" and followed up with mild challenge questions (e.g., "You
said that Paco tore the book while he was reading it. Did you see that or just hear about it?"). A 2
(cowitness information in first interview) _ 2 (reinforcement in first interview) _ 2 (repeat or new
interview) ANOVA was performed for each of the question types (misleading mundane, misleading
fantastic, and leading correct). The dependent variable was the number of times that a child stated that he or
she had personally observed an event. The effect of reinforcement was significant for both misleading
mundane, F (1, 85) = 19.41, p < .001, 2 = .186, and misleading fantastic items, F (1, 85) = 18.00, p <
.001, 2 = .175. Children who received reinforcement stated that they had personally observed 25.00% of
the misleading mundane events and 30.43% of the misleading fantastic events. The corresponding numbers
for children who did not receive reinforcement were 3.72% and 4.25%. No other main effects or
interactions were significant. For leading correct questions the ANOVA revealed no main effects or
interactions: Overall, children correctly stated that they had personally observed an event 73.92% of the
time.
If children said "yes" in response to challenge questions, they were invited to elaborate on their answers
(e.g., "You said Paco took you to a farm. Can you tell me more about that?"). Children responded to 82.6%
of such invitations by elaborating on their answers. Thus, if children falsely insisted they had observed
something, they were also likely to provide additional false information in response to open-ended
invitations.

Discussion
Four findings of the present study are particularly noteworthy. First, reinforcement dramatically increased
the rate of making false allegations by children ages 5 to 7 years against Paco Perez, a classroom visitor. In
contrast, cowitness information yielded weak effects that were rarely significant. Second, the false
allegation rate was high even when children were questioned about extremely implausible or fantastic
events. Third, reinforcement had a strong carryover effect: Children who made false allegations in response
to reinforcement during a first interview tended to repeat the allegations during a second interview, even
when no further reinforcement was given. Fourth, even when challenged, a substantial proportion of
children insisted that the false allegations were based on their own personal observations. Each of these
findings is discussed below.
Reinforcement Versus Cowitness Information
As a follow-up to the study by Garven et al. (1998) , the present experiment examined two specific
interviewing techniques from the McMartin Preschool case, reinforcement and cowitness information. As
predicted, reinforcement had a strong effect: When children received reinforcement in the first interview,
the false allegation rate against classroom visitor Paco Perez was 34.68% for mundane events and 51.61%
for fantastic events, compared with 12.50% and 4.91%, respectively, for controls. Similar effects were
observed when children were reinterviewed a few weeks later. As may be seen in Figure 1 , reinforcement
had a swift effect during the first interview: Children's error rates increased strikingly by the fourth
question in the interview (i.e., within 1 or 2 min).

It is important to note that most misleading questions in the present study involved allegations of
wrongdoing against Paco (e.g., stealing, throwing a crayon at a child), including two questions based on the
McMartin case that involved allegations of abduction ("Did Paco take you on a helicopter ride?" "Did Paco
take you to a farm?"). In addition, two questions involved touching ("Did Paco kiss you on the nose?" "Did
Paco tickle your tummy?") and one question involved a secret ("Did Paco tell you a secret and tell you not
to tell?"). Thus, the results appear relevant to real-life situations in which children are asked about alleged
wrongdoing that involves touching and secrecy (e.g., in sexual abuse cases).
The present findings regarding reinforcement are consistent with well-established theory and research. It
has long been recognized that reinforcement can have a strong shaping influence on both children and
adults ( Ettinger et al., 1994 ). Even in the 1960s there was recognition that reinforcement could affect the
accuracy of children's responses during interviews ( Tharp & Wetzel, 1969 ). Recent research has shown
that positive reinforcement can increase the confidence of adult eyewitnesses in false identifications and
change their retrospective reports in forensically important ways ( Wells & Bradfield, 1998 ).
The present findings do not contradict those of Goodman et al. (1991) and Carter et al. (1996) , who found
that reinforcement increases, rather than decreases, children's accuracy. The kind of reinforcement studied
by Goodman et al. and Carter et al. was different from the kind studied in the present experiment and by
Garven et al. (1998) . Specifically, the reinforcement of Goodman et al. and Carter et al. was noncontingent
and unconditional: The interviewer smiled, used a warm voice, and made eye contact, no matter what
children said or did. In contrast, the reinforcement in the present study and in Garven et al. was contingent
and conditional: Children were given positive reinforcement only when they "remembered" something
negative about Paco Perez and were punished when they could not remember or said "no."
The kind of unconditional reinforcement studied by Goodman et al. (1991) and Carter et al. (1996) is
perhaps more appropriately called social support, and in fact this is a term that these researchers sometimes
have preferred to use. Their studies show that social support can have beneficial effects during child
interviews. Our own findings are consistent with such a conclusion, which is widely accepted by experts on
child interviewing (e.g., Jones, 1992 ; Poole & Lamb, 1998 ; Warren, Woodall, Hunt, & Perry, 1996 ;
Wood, McClure, & Birch, 1996 ).
Contrary to prediction, cowitness information had little effect in the present study. Although cowitness
information had a small, statistically significant effect on accuracy for mundane events in the first
interview, no effect was observed for fantastic events, nor were any effects observed when children were
reinterviewed a few weeks later. Also contrary to prediction, cowitness information failed to have a
synergistic effect when combined with reinforcement. Thus, the present findings suggest that
reinforcement, rather than cowitness information, was the "active ingredient" in the McMartin interviews
and may have induced children in that case to make false allegations.
It is puzzling that cowitness information had so little effect on children's accuracy in the present study,
considering that other researchers have found such an effect with children and adults. Leichtman and Ceci
(1995) found that stereotypes conveyed by a teacher negatively affected children's accuracy over time.
Kassin and Kiechel (1996) found that college students were substantially more likely to make a false
confession to wrongdoing when a "witness" claimed to have seen the act. Hyman and Pentland (1996)
found that college students were more likely to "remember" false events from childhood if the students
were told that family members had reported those events as true. Shaw et al. (1997) found that a witness'
initial and subsequent reports could be influenced by false information from another witness. Finally, Luus
and Wells (1994) found that eyewitnesses became more confident in their false identifications if they were
told that a cowitness had made the same choice.
In light of the results of these five studies, it appears that cowitness information can lead to false reports but
perhaps only under certain conditions. What are those conditions? We suggest four possibilities. First, it
may be that the effectiveness of cowitness information varies according to the characteristics of the
cowitness. Specifically, the effect may be strongest if the cowitness is an authority figure, highly respected,

a close friend or family member, or is physically present. Second, it may be that cowitness information is
most effective if it is followed by repeated recall efforts, with intervening "incubation periods." For
example, in the studies by Leichtman and Ceci (1995) and Hyman and Pentland (1996) , the strongest
effects emerged over time, after repeated questioning. Third, it may be that cowitness information has a
stronger effect if it is given as feedback (e.g., Luus & Wells, 1994 ) and therefore constitutes a form of
positive reinforcement. Finally, perhaps cowitness information has the greatest effect when relevant
memories are weak or old and less influence when memories are clear and recent. Future research may
explore these possibilities.
Fantastic Allegations
In the McMartin Preschool case, children made fantastic allegations that they had been kidnapped from
their school in a helicopter or airplane and taken to a farm where they saw animals tortured ( Nathan &
Snedeker, 1995 ). In the present study, we predicted that some children could be induced to make similar
fantastic allegations but that most children would view such allegations as absurd and reject them. To our
surprise, this prediction was not confirmed. Among children who received reinforcement at the first
interview, the false allegation rate for fantastic events was 51.61% at the first interview and 61.95% at the
second interview, as compared with 4.91% and 6.91%, respectively, for controls. As can be extrapolated
from Figure 1 , calculating a main effect for reinforcement showed that 37% of children receiving
reinforcement reported that Paco Perez had taken them on a helicopter ride and 54% reported that he had
taken them to a farm.
Although the rate of false fantastic allegations was significantly higher than the rate of false mundane
allegations, this difference is probably artifactual. Children in the present study were always questioned
about the mundane items first and the fantastic items second. Because the effect of reinforcement increased
as the interview progressed, the fantastic items were probably disproportionately affected. The safest
conclusion is simply that reinforcement has considerable power to elicit false allegations, both mundane
and fantastic.
Carryover Effects on Children's Subsequent Statements
In the McMartin case, children who made allegations during the initial investigative interview usually made
similar allegations in subsequent interviews and therapy sessions ( Gonzalez, Waterman, Kelly, McCord, &
Oliveri, 1993 ). We hypothesized that the interviewing techniques in the McMartin case might exert a
carryover effect and negatively affect children's accuracy in subsequent interviews. This prediction was
confirmed in the present study. Specifically, when children who received reinforcement during the first
interview were reinterviewed 2 to 3 weeks later without reinforcement, their error rate was 47.02% for
misleading mundane items and 49.76% for fantastic items. These error rates were not significantly different
from the error rates of children who received reinforcement during both interviews. Furthermore, within
both of these groups, the number of errors during the first and second interviews were highly correlated ( r
= .70 to .87). In other words, the same children who made false allegations in response to reinforcement at
one interview were very likely to repeat those allegations at a later interview, whether or not the second
interview contained reinforcement.
Why did children continue to repeat their false allegations, even after several weeks and when
reinforcement had been discontinued? Theories from two different areas of psychology provide
explanations. First, according to basic learning theory, any behavior that has been reinforced and learned is
likely to persist even after reinforcement has stopped ( Ettinger et al., 1994 ). Second, according to social
psychology theory regarding consistency and commitment, individuals tend to stick with their statements,
particularly if those statements are known to other people ( Cialdini, 1993 ).
Response to Challenge

In the McMartin case, some children appeared as witnesses in criminal court and repeated their allegations
under cross-examination. If allegations are false, will children continue to maintain them tenaciously when
pressed? This issue was explored during the second interview of the present study: If a child made a false
allegation during the second interview, the interviewer pressed the child by asking whether the allegation
was based on the child's personal observation or instead was just something that he or she had heard about.
Children who received reinforcement during the first interview claimed that the allegation was true and was
based on their personal observation for 25% of all misleading mundane items and 30% of all fantastic
items. The same findings can be described in a slightly different way: If a child made a false allegation in
response to reinforcement, then the probability was higher than 50% that the child would later claim that
the allegation was based on personal observation, rather than hearsay or secondhand information. In
contrast, when pressed by the interviewer, virtually none of the children interviewed without reinforcement
claimed that their allegations were based on personal observation.
If a child in the reinforcement condition asserted that one of the fantastic events actually happened, he or
she was asked to elaborate (e.g., "Can you tell me more about that?"). Eighty-two percent of the children
who alleged that the fantastic events actually occurred would then proceed to elaborate. Some of the
elaborations were brief and consistent with conventional schemas involving helicopter rides or farms:
"I was looking down from [the helicopter]. We weren't far up."
"There were a lot of buttons and lights [in the helicopter]."
"I saw a farmer milking a cow."
"It was a brown horse."
"He showed us baby animals."
However, several children gave long, detailed elaborations, filled with action. An excerpt from one of these
narratives is given in the Appendix . As may be seen, this 6-year-old girl's spontaneously generated story
included (a) extensive and unusual details about the farm and other places, (b) moderately violent acts (e.g.,
animals being kicked) and scatological imagery (a "gal" being pushed down on "poop"), (c) a clothes
change by Paco that could be interpreted in a negative light, and (d) several event sequences arranged into
logically unfolding narratives. Some scholars have argued that children's true statements are characterized
by inclusion of details and a logical arrangement of events ( Faller, 1988 ; Raskin & Esplin, 1991a , 1991b ;
but see Lamb et al., 1997 ). As the example in the Appendix demonstrates, however, patently false
statements elicited by reinforcement can also contain unusual and elaborate detail and a logical narrative
structure.
Implications for Interviewing Children
The present study, along with the earlier study by Garven et al. (1998) , has two practical implications for
interviewing children. First, these studies show that reinforcement during an interview can swiftly induce
younger children to make false allegations of wrongdoing against an adult. The practical implications for
sexual abuse interviews seem obvious: During the substantive part of an interview, the interviewer should
refrain from promising, implying, or giving any positive reinforcement or punishment to the child (what we
have called Positive or Negative Consequences).
The following interviewing techniques all involve reinforcement and seem highly undesirable: (a) implying
that the child can demonstrate helpfulness, intelligence, or other good qualities by talking with the
interviewer or making allegations; (b) praising or thanking the child for making allegations; (c) giving
tangible rewards (e.g., stickers or food) to reward disclosure; (d) criticizing the child's statements or
suggesting that they are false, inaccurate, or otherwise inadequate; (e) limiting the child's mobility (e.g.,

delaying a visit to the bathroom, the end of the interview, or return to home) until the child has discussed
issues of interest to the interviewer; (f) subjecting the child to physically or verbally stressful stimuli during
the interview (e.g., calling the child a liar); and (g) repeating a question that the child has already answered,
in a way that suggests that the child's first answer was unsatisfactory.
Although we believe that reinforcement is inappropriate during the substantive part of a child forensic
interview, research indicates that noncontingent reinforcement in the form of social support can be
beneficial ( Carter et al., 1996 ; Goodman et al., 1991 ). Specifically, we see no harm in any of the
following interviewing techniques: (a) acting and speaking in a warm, friendly manner; (b) giving the child
compliments during the rapport-building stage of the interview ("My, what pretty eyes you have!"); (c)
praising the child for knowing the difference between truth and falsehood; (d) offering one or two
supportive statements at appropriate places during the interview ("I know this is difficult. Can you tell me
more about that?"); and (e) thanking the child at the end of the interview.
The second practical implication of the present study concerns the possibility of undoing the effects of
improper interviewing. The present findings indicate that if a child makes false allegations in response to
reinforcement at one interview, then he or she is likely to repeat those allegations at later interviews, even if
reinforcement is discontinued. When challenged, a substantial number of children in the present study stuck
to their story and claimed that they had personally witnessed the fictitious events. Some children elaborated
long, detailed narratives. Taken together, these findings indicate that improper interviewing can have a
lasting negative effect on children's accuracy (see also Ceci, Huffman, Smith, & Loftus, 1994 ; Leichtman
& Ceci, 1995 ). To avoid such problems, therefore, it is extremely important that children be interviewed
properly from the very start of an investigation. Furthermore, because a single improper interview may
contaminate the interviews that follow, we recommend that all forensic interviews with children be
videotaped to ensure the integrity of the interviewing process from beginning to end ( McGough, 1994 ;
Myers, 1993 ).
APPENDIX A
Elaborations of a 6-Year-Old Girl Regarding the Visit of Paco Perez to her Classroom
In the dialogue below, I = interviewer and C = child. I
Wow. You said that Paco took you to a farm. Did you see that or hear about it? C
I see that. I
Can you tell me more about the farm? C
Yeah, it had a lot of animals. I could ride on them, I could take milk out. I
Oh, anything else. C
No. I
Can you tell me more about the animals? C
Uhm they kicked one of the animals out of their way. I
Oh wow. What else can you remember? C
Uhm. Well the gal got pushed down on some of the poop. I
Oh yeah. Were they messy? C
Yeah. Got dirty. I
What else can you remember about the farm and the animals? C
Uhm, he kicked one of their, their legs. I
Uh-huh. C
And the animal pushed him down. But I pet one of them, they didn't push they snuggled me. I
Oh wow. C
I'm careful with animals every time when I'm outside. [Material omitted] I
What else can you tell me about the helicopter and the farm? C
The helicopter almost crashed but it was a mountain and it was blocking us. I
Oh. C
So we got out of the way and went across it. I
Oh can you remember anything else? C
Uh-huh. I

Well tell me. C

The helicopter was landing, landing on Earth. I
Uh-huh. C
Somewhere else. I
Oh wow. C
We were in Canada. I
Uh-huh. C
I saw bears that was friendly. I
Oh wow. C
And I almost fell off the mountain but I got a rope. [Material omitted] I
Is there anything else that you remember? C
Yeah we spilled water. I
Oh no a mess. C
Yeah but it didn't get all over me. I
Oh that was lucky. C
I was on a ladder. I
Uh-huh. C
It only got on the ladder but not me. I
Well that was lucky for you huh. C
But Paco got wet all over him. I
Oh no. C
His shirt, his pants, his head. I
Oh no. C
He had to go into the bathroom and put new clothes on. I
Uh-huh. C
And dry his hair. I
Uh-huh. I
Oh. Is there anything else about Paco's visit? C
Yeah. He took me to the North Pole. I
Really. C
Uh-huh. He had a sleigh (inaudible). I
Oh really. C
And he had real reindeers. I
Oh nice. Can you remember anything else or is that all? C
(No audible response) I
Is that all? C
Nope. I
Oh well what else? C
Uhm we saw the reindeers and Rudolph was hiding behind the bush. I
Oh wow. C
That way he wouldn't show his nose. I
Uh-huh. C
But I went behind the bush and I found him. I
Oh well that must have been nice. C
I said "boo." I
Oh. Well what else? C
I scared him and he looked around. I
Uh-huh. C
He came out of the bushes and saw Paco and he kicked. I
Kicked him in the head huh. C
Uh-huh. I
Oh. C
But I was petting him.

References

Asch, S. E. (1956). Studies of independence and conformity: A minority of one against a unanimous
majority. Psychological Monographs, 70, Whole No. 416).
Carter, C., Bottoms, B. & Levine, M. (1996). Linguistic and socio-emotional influences on the accuracy of
children's reports. Law and Human Behavior, 20, 335-358.
Cassel, W. S., Roebers, C. E. M. & Bjorklund, D. F. (1996). Developmental patterns of eyewitness
responses to repeated and increasingly suggestive questions. Journal of Experimental Child Psychology,
61, 116-133.
Ceci, S. J. & Bruck, M. (1993). Suggestibility of the child witness: A historical review and synthesis.
Psychological Bulletin, 113, 403-439.
Ceci, S. J. & Bruck, M. (1995). Jeopardy in the courtroom. (Washington, DC: American Psychological
Association)
Ceci, S. J., Huffman, M. L. C., Smith, E. & Loftus, E. F. (1994). Repeatedly thinking about a non-event:
Source misattributions among preschoolers. Consciousness and Cognition, 3, 388-407.
Cialdini, R. B. (1993). Influence: Science and practice. (New York: HarperCollins)
Eberle, P. & Eberle, S. (1993). The abuse of innocence. (Buffalo, NY: Prometheus Books)
Ettinger, R. H., Crooks, R. L. & Stein, J. (1994). Psychology: Science, behavior, and life. (Fort Worth, TX:
Harcourt Brace)
Faller, K. C. (1988). Criteria for judging the credibility of children's statements about their sexual abuse.
Child Welfare, 67, 389-401.
Faller, K. C. (1996). Interviewing children who may have been abused: A historical perspective and
overview of controversies. Child Maltreatment, 1, 83-95.
Garven, S., Wood, J. M., Malpass, R. S. & Shaw, J. S. (1998). More than suggestion: The effect of
interviewing techniques from the McMartin Preschool case. Journal of Applied Psychology, 83, 347-359.
Gilboa, D. & Greenbaum, C. W. (1978). Adults' warmth, type of verbal reinforcer, and children's learning.
Psychological Reports, 43, 223-226.
Gonzalez, L. S., Waterman, J., Kelly, R. J., McCord, J. & Oliveri, M. K. (1993). Children's patterns of
disclosures and recantations of sexual and ritualistic abuse allegations in psychotherapy. Child Abuse &
Neglect, 17, 281-289.
Goodman, G. S., Bottoms, B. L., Schwartz-Kenney, B. M. & Rudy, L. (1991). Children's testimony about a
stressful event: Improving children's reports. Journal of Narrative and Life History, 1, 69-99.
Green, R. (1992). Sexual science and the law. (Cambridge, MA: Harvard University Press)
Hicks, R. D. (1991). In pursuit of Satan: The police and the occult. (Buffalo, NY: Prometheus Books)
Hyman, I. E. & Pentland, J. (1996). The role of mental imagery in the creation of false childhood
memories. Journal of Memory and Language, 35, 101-117.
Jones, D. P. H. (1992). Interviewing the sexually abused child ((4th ed.). London: Gaskell)
Kassin, S. & Kiechel, K. (1996). The social psychology of false confessions: Compliance, internalization,
and confabulation. Psychological Science, 7, 125-128.
Lamb, M. E., Sternberg, K. J., Esplin, P. W., Hershkowitz, I., Orbach, Y. & Hovav, M. (1997). Criterionbased content analysis: A field validation study. Child Abuse & Neglect, 21, 255-264.
Larsen, K. S. (1991). The Asch conformity experiment: Replication and transhistorical comparisons.(In
James W. Neuliep (Ed.), Replication research in the social sciences (pp. 151156). Newbury Park, CA:
Sage.)
Leichtman, M. D. & Ceci, S. J. (1995). The effects of stereotypes and suggestions on preschoolers' reports.
Developmental Psychology, 31, 568-578.
Loftus, E. F. & Davies, G. M. (1984). Distortions in the memory of children. Journal of Social Issues, 40,
51-67.
Luus, C. A. E. & Wells, G. L. (1994). The malleability of eyewitness confidence: Co-witness and
perseverance effects. Journal of Applied Psychology, 79, 714-724.
Mason, M. (1991). The McMartin Case revisited: The conflict between social work and criminal justice.
Social Work, 36, 391-395.
McGough, L. S. (1994). Child witnesses: Fragile voices in the American legal system. (New Haven, CT:
Yale University Press)
Myers, J. E. B. (1993). Investigative interviews of children: Should they be videotaped? Notre Dame
Journal of Law, Ethics, and Public Policy, 7, 371-386.
Nathan, D. & Snedeker, M. (1995). Satan's silence. (New York: Basic Books)

Poole, D. A. & Lamb, M. E. (1998). Investigative interviews of children: A guide for helping professionals.
(Washington, DC: American Psychological Association)
Poole, D. A. & Lindsay, D. S. (1995). Interviewing preschoolers: Effects of nonsuggestive techniques,
parental coaching and leading questions on reports of nonexperienced events. Journal of Experimental
Child Psychology, 60, 129-154.
Pynoos, R. S. & Nader, K. (1989). Children's memory and proximity of violence. Journal of the American
Academy of Child and Adolescent Psychiatry, 28, 236-241.
Raskin, D. C. & Esplin, P. W. (1991a). Assessment of children's statements of sexual abuse.(In J. L. Doris
(Ed.), The suggestibility of children's recollections: Implications for eyewitness testimony (pp. 153164).
Washington, DC: American Psychological Association.)
Raskin, D. C. & Esplin, P. W. (1991b). Statement validity assessment: Interview procedures and content
analysis of children's statements of sexual abuse. Behavioral Assessment, 13, 265-291.
Reinhold, R. (1990, January 25). How lawyers and media turned the McMartin case into a tragic media
circus. The New York Times, , D1
Shaw, J. S., Garven, S. & Wood, J. M. (1997). Co-witness information can alter witnesses' immediate
memory reports. Law and Human Behavior, 21, 503-523.
Siegler, R. S. (1992). The other Alfred Binet. Developmental Psychology, 28, 179-190.
Spence, J. T. (1973). Factors contributing to the effectiveness of social and nonsocial reinforcers in the
discrimination learning of children from two socioeconomic groups. Journal of Experimental Child
Psychology, 15, 367-380.
Summit, R. C. (1994). The dark tunnels of McMartin. The Journal of Psychohistory, 21, 397-416.
Tavris, C. (1997, April 11). A day-care witch hunt tests justice in Massachusetts. Los Angeles Times, , B9
Tharp, R. G. & Wetzel, R. J. (1969). Behavior modification in the natural environment. (New York:
Academic Press)
Timnick, L. & McGraw, C. (1990, January 19). McMartin verdict: Not guilty. Los Angeles Times, , A1
Warren, A. R., Woodall, C. E., Hunt, J. S. & Perry, N. W. (1996). "It sounds good in theory, but ...": Do
investigative interviewers follow guidelines based on memory research? Child Maltreatment, 1, 231-245.
Wells, G. L. & Bradfield, A. L. (1998). "Good, you identified the suspect": Feedback to eyewitnesses
distorts their reports of the witnessing experience. Journal of Applied Psychology, 83, 360-376.
Wilkerson, T. & Rainey, J. (1990, January 19). Tapes of children decided the case for most jurors. Los
Angeles Times, , A1
Wood, J. M., McClure, K. A. & Birch, R. A. (1996). Suggestions for improving interviews in child
protection agencies. Child Maltreatment, 1, 223-230.
Wood, J. M., Schreiber, N., Martinez, Y., McLaurin, K., Strok, R., Velarde, L. D., Garven, S. & Malpass,
R. S. (1998, March). Child interviewing techniques in the McMartin Preschool and Kelly Michaels cases:
A quantitative comparison. (Paper presented at the biennial conference of the American PsychologyLaw
Society, Redondo Beach, CA)
Zigler, E. & Kanzer, P. (1962). The effectiveness of two classes of verbal reinforcers on the performance of
middle- and lower-class children. Journal of Personality, 30, 157-163.

1
Transcripts of the McMartin interviews and the numbering system for them are archived in the Department
of Psychology, McGill University, Montreal, Quebec, Canada.

2
The scoring categories are summarized here. A complete copy of the scoring rules may be obtained from
James M. Wood, Department of Psychology, University of Texas at El Paso, El Paso, Texas 79968.

We thank Todd Flynn and Luis Natalicio, two scientist-practitioners who astutely called our attention to the

potential effects of reinforcement in child sexual abuse interviews. We also express our gratitude to the
members of Por Nios y Familias who contributed to this project.
Correspondence may be addressed to Sena Garven, Department of Psychology, University of
NebraskaLincoln, Lincoln, Nebraska, 68588-0308.
Electronic mail may be sent to [email protected]
Received: September 28, 1998
Revised: March 11, 1999
Accepted: March 15, 1999

Figure 1. Mean percentage of incorrect "yes" answers as related to order of item presentation, by
experimental condition.