Vous êtes sur la page 1sur 6

Journal of Applied Research in Memory and Cognition 3 (2014) 16

Contents lists available at ScienceDirect

Journal of Applied Research in Memory and Cognition
j our nal homepage: www. el sevi er . com/ l ocat e/ j ar mac
The effect of age and reminders on witnesses responses to
cross-examination-style questioning

Fiona Jack

, Rachel Zajac
Psychology Department, University of Otago, P.O. Box 56, Dunedin 9054, New Zealand
a r t i c l e i n f o
Article history:
Received 29 June 2012
Received in revised form4 December 2013
Accepted 6 December 2013
Available online 14 December 2013
Eyewitness memory
Memory development
a b s t r a c t
Witnesses of all ages struggle with cross-examination questions, often changing their testimony as a
result. In the laboratory, cross-examination-style questioning decreases both childrens and adults accu-
racy. We examined the extent to which this effect varies with age. We interviewed children, adolescents,
and adults (N =128) about a lm clip after a short delay. Eight months later we cross-examined partic-
ipants on their original responses to some questions, and simply repeated other questions. Participants
of all ages were more likely to change their answers when cross-examined than when asked the same
question again. Cross-examination negatively affected accuracy, although this effect decreased as age
increased. Listening to an audio-recording of their original responses before the second interview reduced
the number of changes participants made in response to repeated questions, but not cross-examination
questions. These data give cause for concern about the effect cross-examination has on the accuracy of
child and adolescent witnesses.
2013 Society for Applied Research in Memory and Cognition. Published by Elsevier Inc. All rights
Under an adversarial legal system, every witnesss evidence is
subject to cross-examination. The purpose of cross-examination is
to test the credibility of the witness and help to establish the truth
(Bouvier, 1856). It is a fundamental stepinthe legal process because
it fulls the absolute right of defendants to have witnesses against
themexamined (Spencer, 2012). Some legal scholars have argued,
however, that the primary aimof many cross-examining lawyers is
to discredit the witness, with no regard for the truth (Plotnikoff &
Woolfson, 2012; Spencer, 2012). They argue that the current model
of cross-examination is not an appropriate or effective method for
testing the evidence of vulnerable witnesses, such as children (see
also Pigot et al., 1989), because cross-examination usually occurs
after a long delay, and often involves questions that are leading,
confusing, or credibility-challenging (Zajac & Cannan, 2009; Zajac,
Gross, & Hayne, 2003). Because each of these factors can reduce
the reliability of eyewitness reports (Bruck & Ceci, 1999; Carter,
Bottoms, & Levine, 1996; Read & Connolly, 2007; Zajac & Cannan,

This research was supported by a New Zealand Science & Technology Postdoc-
toral Fellowshipto Fiona Jack andby a grant awardedto Rachel Zajac by the Marsden
Fund Council, fromGovernment funding administered by the Royal Society of New
Zealand. Portions of the data reported in this article were presented at the Soci-
ety for Research on Memory and Cognition, New York, NY, 2011; The American
Psychology-Law Society, San Juan, Puerto Rico, 2012; and KiwiCAM, Wellington,
NZ, 2012.

Corresponding author. Tel.: +64 3 479 5460; fax: +64 3 479 8335.
E-mail addresses: ona@psy.otago.ac.nz (F. Jack), rachelz@psy.otago.ac.nz
(R. Zajac).
2009), we must question the extent to which cross-examination
achieves its goal of uncovering the truth.
During cross-examination, child sexual abuse complainants
often comply with leading and closed questions and frequently
misunderstand questions, with the majority making at least one
change to their testimony (Plotnikoff & Woolfson, 2009; Zajac
et al., 2003). These data are concerning, but they cannot informus
about accuracy. To address this issue, Zajac et al. have conducted
laboratory research in which children experience a unique event
and are then interviewed with analogues of direct examination
and cross-examination (ONeill & Zajac, 2013a, 2013b; Righarts,
ONeill, & Zajac, 2013; Zajac & Hayne, 2003, 2006; Zajac, Jury, &
ONeill, 2009). Across four experiments and a range of manipu-
lations, cross-examination-style questioning dramatically reduced
childrens accuracy.
Like children, adolescent and adult witnesses struggle with
cross-examination in the courtroom, often resulting in inconsis-
tent testimony (Jack, Cannan, &Zajac, 2009; Plotnikoff &Woolfson,
2009; Zajac &Cannan, 2009). Fewempirical studies, however, have
examinedthe effects of cross-examinationinthese older witnesses.
The results of four experiments that did include adults suggest that
their accuracy is compromised by cross-examination-style ques-
tioning (Brimacombe, Jung, Garrioch, &Allison, 2003; Brimacombe,
Quinton, Nance, & Garrioch, 1997; Turtle & Wells, 1988; Valentine
& Maras, 2011), although methodological issues make it difcult to
draw rmconclusions. These issues include the absence of a non-
cross-examination control condition (Valentine & Maras, 2011),
2211-3681/$ see front matter 2013 Society for Applied Research in Memory and Cognition. Published by Elsevier Inc. All rights reserved.
2 F. Jack, R. Zajac / Journal of Applied Research in Memory and Cognition 3 (2014) 16
andtheecological validityof thequestioningprocedures anddelays
utilised(Brimacombeet al., 1997, 2003; Turtle&Wells, 1988). Using
a standardised, ecologically valid paradigm with children, adoles-
cents, and adults, our primary goal was to establish whether the
effect of cross-examination-style questioning on accuracy varies
with age.
Our secondary aimwas to examine the effect of a reminder on
cross-examination performance. Depending on jurisdiction, some
witnesses are shown the video-recording of a previous interview
or asked to read over their prior statements before testifying (Pipe
& Henaghan, 1996; Read & Connolly, 2007). Exposure to stimuli
that encapsulate aspects of the original event can facilitate recall
(Tulving & Thomson, 1973). Although a body of literature demon-
strates thefacilitativeeffect of suchreminders, most of this research
focuses on young children, and involves non-verbal recall and/or
non-verbal reminders (e.g., Hudson & Shefeld, 1999; Morgan
& Hayne, 2007; Priestley, Roberts, & Pipe, 1999). It is therefore
important to establish what effect reminders exert on witnesses
cross-examination performance, and whether this differs as a func-
tion of age.
Our third aimwas to examine additional factors that might pre-
dict cross-examination performance. Zajac et al. (2009) found that
childrens performance under cross-examination was positively
related to their self-esteem, self-condence, and assertiveness. We
examined whether cross-examination performance is related to
cognitive factors, such as participants general memory and intel-
lectual functioning, or their free recall accounts of the event. We
also examined whether metacognitive skills might play a role;
specically, are witnesses less likely to change answers that they
can support with contextual detail? Finally, some researchers have
foundthat whenaskedcross-examination-stylequestions, children
are no less likely to change responses that were initially correct
than those that were initially incorrect (ONeill & Zajac, 2013b;
Valentine & Maras, 2011; Zajac & Hayne, 2003; but see Zajac &
Hayne, 2006; Zajac et al., 2009). We investigatedage-relatedtrends
in this tendency.
1. Method
1.1. Participants
We recruited community members via newspaper advertise-
ments, invitations distributed to schools, a database of participants
fromunrelated studies, and word of mouth. At the time of the rst
interview, the sample comprised 48 children (911 years), 48 ado-
lescents (1416 years), and 48 adults (2560 years). At the time
of the second interview, 16 participants could not be contacted or
were unable to take part. The nal sample comprised 45 children
(M=10.7 years, SD=0.9 at rst interview; 23 females), 41 adoles-
cents (M=15.4 years, SD=0.8; 23 females), and 42 adults (M=44.4
years, SD=0.8; 20 females). All participants (and a caregiver of each
participant aged under 16 years) gave written informed consent.
Participants received $15.00 at the end of each interview.
1.2. Memory event and rst interview
One of two female experimenters saw each participant. First,
the participant watched a brief lm clip depicting a simulated
non-violent crime. The experimenter then administered the Wech-
sler Abbreviated Scale of Intelligence (WASI; Wechsler, 1999) and
the Wide Range Assessment of Memory and Learning (WRAML;
Sheslow& Adams, 2003). Approximately 45min after the lmclip,
the experimenter interviewed the participant about it, following
the protocol that is used by the NewZealand Police for their level 1
witness interviews. This procedure incorporates core aspects of the
cognitive interview, such as report everything, transfer of control,
context reinstatement, and focused retrieval (Fisher & Schreiber,
2007; Milne & Bull, 1999). The interviewcomprised free recall, fol-
lowed by open-ended prompts. Finally, the experimenter asked
8 yesno questions: four about true details and four about false
details. The questions addressed both central details (e.g., Was
the person a man?) and peripheral details (e.g., Did you see a red
car parked on the street?). Interviews were audio-recorded. Addi-
tional details about this session are available elsewhere (Jack, Leov,
& Zajac, 2013).
1.3. Second interview
Approximately 8 months after the rst interview (M=8.0
months, SD=1.2), participants were re-interviewed by one of
three unfamiliar female experimenters. Immediately beforehand,
approximately half of each age group (21 randomly selected chil-
dren, 19 adolescents, and 18 adults) heard the audio-recording of
their rst interview. All participants then gave a new free-recall
account, and were then questioned on the same 8 items as in
the rst interview. For four of the items, participants were sim-
ply asked the same questions again (control questions). For the
remaining four items, participants were challenged on their origi-
nal responses (cross-examination questions; see below). The eight
items were counterbalanced, suchthat eachappearedequally often
in the cross-examination and control conditions.
We adapted Zajac and Haynes (2003) cross-examination
paradigm. The aimof these questions was tochallenge participants
original yesno responses, irrespective of accuracy. A sample set of
questions is provided in the Appendix. For each of the four items,
the interviewer began by re-stating the participants initial answer
(e.g., Last time you said that the person you saw in the clip was a
man). This was followed by a metamemory question designed to
assess participants ability tosupport their initial response withrel-
evant contextual detail (e.g., How did you know that?). The next
three questions were leading, complex, ambiguous, or irrelevant.
Finally, the interviewer provided a reason for disbelieving the par-
ticipants original response (e.g., Most of the people who saw this
clip said that the person was a woman, and I think they might be
right about that, mightnt they?). If the participant agreed that this
versionof events was possible, theinterviewer followedupwiththe
stronger suggestion, I think they are right about that, arent they?
Four different reasons for disbelief were used (see Appendix). The
assignment of these reasons to items, and their order of presenta-
tion, were counterbalanced across participants. Cross-examination
questions were interspersed with control questions; the order of
these was also counterbalanced. Interviews were audio-recorded
and transcribed verbatim.
1.4. Scoring
Free recall. To quantify participants free-recall accounts, the
number of unique details was tallied; each detail was further coded
as correct, incorrect, or possible/subjective (for more detail, see Jack
et al., 2013).
YesNo/control questions. For the yesno questions at both
interviews, correct responses were given positive scores; incorrect
responses were given negative scores. Participants scored 2 or 2
for anunambiguous yes or no response, 1or 1for a hesitant yes
or no response (e.g., I think so), or 0 for a dont know response.
Metamemory questions. These were scored dichotomously.
Participants received credit if they provided relevant contextual
detail consistent with their initial response (e.g., I knew it was a
man because of his body shape).
Cross-examination questions. With the exception of the
metamemory responses (see above), only participants responses
F. Jack, R. Zajac / Journal of Applied Research in Memory and Cognition 3 (2014) 16 3
Table 1
Mean (SE) number of changes made in response to control and cross-examination questions, as a function of age group and reminder condition.
Question type Age group Reminder condition
Children Adolescents Adults Reminder No reminder
Control 1.26 (0.17) 0.93 (0.17) 1.14 (0.15) 0.81 (0.12) 1.37 (0.13)
Cross-examination 3.33 (0.15) 2.98 (0.19) 2.16 (0.16) 2.82 (0.17) 2.84 (0.13)
Note. Maximumpossible number of changes =4; minimumpossible number of changes =0.
to the reason for disbelief cross-examination question and its
follow-up were scored. For any given response, the valence of the
score (positive or negative) reected accuracy. Accurate responses
(agreement with a correct suggestion or disagreement with an
incorrect suggestion) were assigned a positive score. Inaccurate
responses (agreement with an incorrect suggestion or disagree-
ment witha correct suggestion) were assigneda negative score. The
numerical value of the score (0, 1, or 2) reected the certainty with
which the response was given. Responses were assigned a score of
2 or 2 (depending on accuracy) if they indicated complete agree-
ment or disagreement with the interviewers suggestion, a score of
1 or 1 if they indicated tentative agreement or disagreement (e.g.,
I think so), or a score of 0 for a noncommittal response (e.g., dont
know, maybe). Two coders independently scored 36 transcripts
(28%) and achieved 85% agreement (kappa =.79). Disagreements
were resolved through discussion; one coder scored the remaining
2. Results
We replaced any values that were >3 SD from the mean with
the next-highest value plus 1, or the next-lowest value minus 1,
as appropriate (Tabachnick & Fidell, 2007). In two instances, our
results differ from those obtained with the unadjusted data; we
note these in the text. Preliminary analyses showed no signicant
inuence of gender on any measured variable, so we collapsed
across gender for subsequent analyses.
2.1. First interview
Free-recall data are reported elsewhere (Jack et al., 2013).
Accuracy on the yesno questions did not differ signicantly
as a function of age group (children, M=9.01, SE=0.51; adoles-
cents, M=9.39, SE=0.57; adults, M=9.26, SE=0.54; F(2, 122) =0.13,
p=.88) or subsequent reminder condition (reminder, M=9.38,
SE=0.44; no reminder, M=9.08, SE=0.43; F(1, 122) =0.23, p=.63).
2.2. Second Interview
First, we explored whether cross-examination-style question-
ing led participants to change their responses, and whether the
number of changes made varied with age or reminder condition.
Participants were deemed to have changed a response if their score
for a given question changed valence between interviews (e.g.,
from +2 to 1), or changed to or from 0. The maximum number
of changes for each question type (control or cross-examination)
was 4. Because these were count data, we conducted a Poisson
regression. We included age group, question type, and reminder in
a full-factorial model. Participants made more changes in response
to cross-examination questions than control questions (B=0.68,
95% CI =0.390.97, p<.001, R=1.97; Table 1) and participants
who received a reminder made fewer changes overall than those
who did not (B=0.60, 95% CI =1.16 to 0.05, p=.03, R=0.55),
although these main effects were qualied by a 2-way interaction
(B=0.76, 95% CI =0.201.32, p=.008). Two further Poisson regres-
sion analyses, testing the effect of the reminder for each question
type, revealed that the reminder signicantly reduced the number
of changes participants madewhenrespondingtocontrol questions
(B=0.51, 95% CI =0.86 to 0.16, p=.004, R=0.60) but not cross-
examination questions (B=0.01, 95% CI =0.15 to 0.14, p=.94,
R=0.99). Neither the main effect of age group nor any of its inter-
actions were signicant (ps =.27.74).
To examine whether these changes inuenced participants
overall accuracy, we summed each participants scores across the
four items on which they were cross-examined and the four items
about which questions were simply repeated, separately for the
rst and second interviews. To control for accuracy at interview1,
we then calculated accuracy-change scores for each question type
bysubtractingthesumof scores obtainedintherst interviewfrom
that obtained in the second interview. Positive accuracy-change
scores indicated that the participant became more accurate in the
second interview, relative to the rst; negative scores indicated
a decrease in accuracy. Participants responses during the second
interview were less accurate than those given during the rst,
regardless of age and question type (Table 2). This trend was con-
rmed by one-sample t-tests comparing accuracy-change scores
to 0 (children, cross-examination, t(44) =6.17, d=0.92, control,
t(44) =3.54, d=0.53; adolescents, cross-examination, t(40) =5.22,
d=0.82, control, t(40) =3.69, d=0.58; adults, cross-examination,
t(41) =3.82, d=0.59, control, t(41) =3.94, d=0.61; ps .001).
To examine the effect of our manipulated variables on
accuracy-change scores, we conducted a 2(question type) 3(age
group) 2(reminder condition) ANOVA, with repeated measures
across question type. The decrease in accuracy between inter-
views was greater for cross-examination questions than for control
questions, F(1, 122) =21.39, p<.001,
= .15 and for younger par-
ticipants than for older participants, F(2, 122) =3.46, p=.04,
.05, although these main effects were qualied by a signicant 2-
way interaction, F(2, 122) =3.84, p=.02,
= .06. (Note, the main
effect of age only approached signicance when the data were not
adjusted for outliers, F(2, 122) =2.56, p=.08,
= .04). To explore
the interaction, we conducted one-way ANOVAs to test the effect
of age group for each question type. As shown in Table 2, for
cross-examination questions, the decrease in accuracy between
Table 2
Mean (and SE) accuracy-change scores for control and cross-examination questions, as a function of age group and reminder condition.
Question type Age group Reminder condition
Children Adolescents Adults Reminder No reminder
Control 1.57 (0.44) 1.22 (0.33) 1.45 (0.37) 0.81 (0.29) 1.92 (0.32)
Cross-examination 4.64 (0.75) 3.39 (0.65) 1.89 (0.50) 3.57 (0.65) 3.14 (0.45)
Note. Maximumpossible score =16; minimumpossible score =16. Larger absolute values indicate a greater degree of change in overall accuracy. Negative scores indicate
that accuracy decreased frominterview1 to interview2; positive scores indicate that accuracy increased.
4 F. Jack, R. Zajac / Journal of Applied Research in Memory and Cognition 3 (2014) 16
interviews was greater for children than it was for adults, Welchs
F(2, 81.20) =5.00, p=.009,
= .07, Tamhanes p=.009; adolescents
did not differ signicantly fromthe other groups. For control ques-
tions, there was no signicant effect of age group, p=.80. t-Tests
showed that the effect of question type was signicant for chil-
dren (t(44) =3.54, p=.001, d=0.53) and adolescents (t(40) =3.11,
p=.003, d=0.49), but not for adults (t(41) =0.76, p=.46). Neither
the main effect of reminder condition, nor any of the interactions
involving this variable were signicant, although the interaction
between question type and reminder condition approached signif-
icance, F(1, 122) =2.85, p=.09,
= .02; remaining ps =.12.72.
2.3. Central versus peripheral details
We questioned participants about both central and periph-
eral aspects of the event, and it would be informative to know
whether these types of details contributed differentially to the
results reported above. We were able to examine this issue ten-
tatively, and give a brief summary here. More details about this
analysis are available fromthe rst author on request.
To classify items as central or peripheral, we asked 10 adult
raters who are highly familiar with the autobiographical and/or
eyewitness memory literature (ranging fromPhD students to Pro-
fessors) to indicate whether each item was peripheral or central,
based on the denitions given by past authors (Bernsten, 2002;
Christianson, 1992; Heuer & Reisberg, 1992; Peterson & Whalen,
2001; Wessel, Van De Kooy, & Merckelbach, 2000). We used 70%
agreement as a cut-off for reliable classication. Because some
items failed to meet this criterion and because of the way items
were assigned to sets of cross-examination and control questions,
we were able to include only 44 participants (16 children, 14 ado-
lescents, and 14 adults) in these analyses.
We used Poisson regression to examine whether the number of
changes participants made betweenthe rst andsecondinterviews
varied as a function of centrality. There was a signicant interac-
tion between question type (cross-examination/control) and detail
type (central/peripheral), B=0.76, 95% CI =0.201.32, p=.008. In
response to control questions, participants were signicantly more
likely to change answers about peripheral details than those about
central details, B=1.03, 95% CI =0.111.94, p=.028, R=2.79. The
effect of centrality on the number of answers participants changed
under cross-examination-style questioning was not signicant
To examine whether the effect of cross-examination on accu-
racy varied as a function of centrality, we conducted an ANOVA
on accuracy-change scores. Again there was a signicant inter-
action between question type and detail type, F(1, 41) =5.89,
= .13. For control questions, the decrease in accuracy
between interviews was signicantly greater for peripheral details
than for central details, t(43) =2.70, p=.010, d=0.41. For cross-
examination-style questions, there was a non-signicant trend in
the opposite direction (p=.11).
These results must be interpreted with caution given the
reduced sample size and limited power of these analyses. They
do, however, suggest that the effects of our cross-examination-
style questions were not solely due to an inuence on participants
responses about peripheral details.
2.4. Predicting the effects of cross-examination-style questioning
Next, we examined several possible predictors of participants
cross-examination performance. First, we examined whether
response accuracy inthe rst interviewwas relatedto the probabil-
ity that that response would be changed under cross-examination.
For each participant, we calculated the proportion of responses
changed, as a function of initial accuracy. Proportions could not be
calculatedfor 10children, 8adolescents, and9adults because at the
rst interviewthey responded correctly on all four of the questions
onwhichtheywere later cross-examined, andfor one childbecause
she responded incorrectly on all four questions. Separate related-
samples Wilcoxon signed rank tests were conducted for each age
group. Adults changed initially incorrect responses more often
than initially correct responses (median proportion of responses
changed=1.0 and .50, respectively; z =2.10, N Ties =27, p=.04,
d=0.54). In contrast, incorrect responses made by younger partic-
ipants were no more likely to be changed than correct responses
(ps =.25 and .44).
Second, we examined whether participants who could back up
responses with contextual detail were less likely to change those
responses when cross-examined. For each participant, we calcu-
lated the proportionof responses changed, as a functionof whether
contextual detail was provided. Proportions could not be calcu-
lated for 11 children, 9 adolescents, and 9 adults because they
providedcontextual detail for either zeroor all four of the questions
on which they were cross-examined. Separate related-samples
Wilcoxon signed rank tests were conducted for each age group. For
adults (z =2.96, N Ties =30, p=.003, d=0.77) but not younger
participants (ps =.75 and .97) responses supported with relevant
contextual detail were less likely to be changed than those that
were unsupported (median proportion of responses changed=.33
and 1.0, respectively).
Third, we examined whether participants free-recall accounts
or their scores on the individual difference measures could pre-
dict the effect that cross-examination would have on accuracy.
The amount of free-recall information reported at the rst inter-
view (Pearsons r(125) =.26, p=.003) and raw scores on the WASI
Vocabulary subtest (r(126) =.19, p=.03) were positively associated
with cross-examination accuracy-change scores. (Note, the latter
correlation only approached signicance when the data were not
adjusted for outliers, r(126) =.16, p=.06). A positive correlation
between accuracy-change scores and the amount of free-recall
information reported at the second interview approached sig-
nicance (r(126) =.15, p=.09). The accuracy of the information
reported during the two free-recall tasks, raw scores on the WASI
Matrices subtest and the WRAML verbal and nonverbal subtests,
and standardised WASI and WRAML scores were not associated
with cross-examination accuracy-change scores (ps =.411.00).
Finally, we used linear regression to examine the predictive
power of a model comprising a combination of the variables asso-
ciated with cross-examination accuracy-change scores. Analysis
of Cooks D values indicated that the data of one female child
exerted an undue inuence on the overall results (Cooks D=.13).
The data fromthis participant were therefore excluded. There was
good tolerance and lowcollinearity (>.35). Using the simultaneous
method, we entered age at second interview, WASI Vocabulary raw
score, and amount of free recall information reported at the rst
interview. A signicant model emerged: F(3, 122) =4.73, p=.004,
explaining 8% of the variance (Adjusted R
=.08). The amount
of free recall information that participants reported at the rst
interview uniquely predicted 3.3% of the variance in the effect
of cross-examination on accuracy (R
change =.033; standardised
beta weight =.214).
3. Discussion
Our ndings support those of Zajac et al. (ONeill &Zajac, 2013a,
2013b; Righarts et al., 2013; Zajac &Hayne, 2003, 2006; Zajac et al.,
2009), who have repeatedly shown that cross-examination-style
questioning decreases the accuracy of children aged up to 10 years.
This effect decreased with age in the present experiment, although
adolescents accuracy was also signicantly affected. This aspect of
F. Jack, R. Zajac / Journal of Applied Research in Memory and Cognition 3 (2014) 16 5
our results provides animportant extensionto existing experimen-
tal research on the effects of cross-examination-style questioning,
which until nowhas not included the adolescent age group. Given
that adolescence represents a unique developmental stage (Jack
et al., 2013) and that adolescents are just as likely as younger chil-
dren to testify in court (Hanna, Davies, Henderson, Crothers, &
Rotherham, 2010), research on this age group is vital.
Importantly, accuracy varied with age only when participants
were cross-examined on their previous responses; age did not
inuence performance on the yesno questions asked during the
rst and second interviews. Overall, these results are consistent
with research showing that under neutral questioning conditions,
the accuracy of recall does not change with age (Jack et al.,
2013; Marin, Holmes, Guth, & Kovac, 1979; Oates & Shrimpton,
1991; Sutherland & Hayne, 2001; Tustin & Hayne, 2010), although
younger witnesses are typically more suggestible than adults (e.g.,
Oates & Shrimpton, 1991; Sutherland & Hayne, 2001; see Melnyk,
Crossman, & Scullin, 1997, for a review).
Irrespective of age, participants were more likely tochange their
original responses whentheywerecross-examinedthanwhenthey
were simply asked the same question again. Why then did our
cross-examination questions negatively affect childrens and ado-
lescents accuracy, but not that of adults? Our analyses suggest that
when challenged on their initial responses, adults made better
decisions about changing their responses. That is, unlike younger
participants, adults were less likely to change responses if they
could recall supporting event details. It appears that adults, but not
younger participants, used some degree of metacognitive moni-
toring to guide their responses. The fact that adults changed more
incorrect responses than correct responses suggests that this strat-
egy was somewhat effective, although cross-examination did not
exert any effect (negative or positive) on adults overall accuracy.
Across our sample, the variable most closely related to cross-
examination accuracy was the amount of information participants
provided about the event shortly after witnessing it. Giving a
more comprehensive account of the event soon after it occurred
might have enabled participants to store a more robust memory
representation (Peterson, Pardy, Tizzard-Drover, & Warren, 2005;
Poole & White, 1991). Alternatively, both variables might have
been inuenced by a third variable, such as self-condence (see
Zajac et al., 2009). Together, the cognitive and mnemonic vari-
ables in our regression analysis explained only 8% of the variance
in participants cross-examination accuracy, suggesting that other
variables play important roles. As mentioned above, metacogni-
tive factors might be important in determining adults responses.
In contrast, perhaps childrens responses to cross-examination are
better explained by social factors. Further research is needed to
examine predictors of cross-examination performance across the
Exposure to a reminder reduced the number of changes
participants made in response to repeated questions, but not cross-
examination questions. The reminder did not have a signicant
effect on accuracy for either question type; nor did the effect of
reminders vary as a function of age. These data suggest that expo-
sure to a reminder might have no effect on cross-examination
performance, although it is important to note that the nature of
our reminder differed fromthose often used in legal contexts (i.e.,
video recordings and interviewtranscripts).
Methodological issues to be kept in mind when interpreting
our results include our use of change scores to assess the effects
of our manipulations on participants accuracy, and issues relating
to ecological validity, such as our use of young adult interviewers,
the transparency (or lack thereof) of the interviewers intentions,
and the absence of an oath requirement. These latter issues could
have affected younger and older participants differentially. Aspects
of our target event may also limit the generalisability of our
results. First, it was not personally relevant and was arguably less
memorable than the types of events that witnesses are typically
cross-examinedabout. Second, our target event was knownto have
occurred; our results might not generalise to situations in which
allegations have been made about an event that never happened.
3.1. Practical applications
Results we obtained when looking at predictors of cross-
examination performance suggest that, when used with adult
witnesses, cross-examination might help to discriminate between
more and less reliable aspects of a witnesss testimony. Other
aspects of our ndings, however, give cause for concern about
the effect that cross-examination might have on the accuracy of
both child and adolescent witnesses, at least in cases that involve a
truthful witness being questioned about details of an event that all
agree occurred. Our control data suggest that the negative effect of
cross-examination on accuracy can be attributed in part to delay,
but the difference between the cross-examination and control data
demonstrates that the style of the questioning had a particularly
deleterious effect on young witnesses accuracy. Our reminder data
suggest that the common practice of playing children a recording
of their evidential interview prior to appearing in court provides
no buffer against this effect.
Howcan the situation be improved? A number of changes have
beenproposed, most of whichrelate to: (a) whoconducts the cross-
examination of young witnesses, (b) the types of questions they
may ask, and (c) when the cross-examination is conducted (e.g.,
Pigot et al., 1989; see Spencer & Lamb, 2012 for comprehensive
discussion). Our ndings suggest that procedures that challenge
witnesses to elaborate on their initial responses without pres-
suring themto change their accounts are likely to elicit the most
accurate testimony.
Conict of interest statement
No conicts of interest exist in relation to this research.
We thank The New Zealand Police for supplying the lm clips
used in this research; Jessica Leov, Nicola Davis, and Celia Wright
for their assistance with data collection and coding; Chris Perkins,
Peter Herbison, and Brian Niven for statistical advice; and all of the
children, adolescents, and adults who participated in this research.
This research was approved by the University of Otago Human
Ethics Committee.
A sample set of cross-examination questions:
1. Last time, you said that the person you sawin the clip was wear-
ing a hat.
2. Howdid you knowthat? (meta-memory question).
3. The person was wearing a leather jacket, wasnt he? (leading).
4. Do you knowanyone who has a jacket like that? (irrelevant).
5. What colour was it? (ambiguous).
6. Are you SURE that the person you sawin the clip was wearing a
7. I think maybe the person was not wearing a hat, but you didnt
tell the truth about that. That might be what happened, mightnt
it? [if yes or maybe]: I think that is what happened, isnt it?
The four reasons for disbelief:
6 F. Jack, R. Zajac / Journal of Applied Research in Memory and Cognition 3 (2014) 16

I think maybe [the person was not wearing a hat], but you didnt
tell the truth about that. That might be what happened, mightnt
it? [if yes or maybe]: I think that is what happened, isnt it?

I dont think [the person was wearing a hat]. I think you might
have remembered that incorrectly, mightnt you? [if yes or
maybe]: I think that is what happened, isnt it?

Most of the people who saw this clip said that [the person was
not wearing a hat], and I think they might be right about that,
mightnt they? [if yes or maybe]: In fact, I think they are right
about that, arent they?

If the person who remembered the clip the best said that [the
person was not wearing a hat], they might be right about that,
dont you think? [if yes or maybe]: I think they would be right
about that, wouldnt they?
Bernsten, D. (2002). Tunnel memories for autobiographical events: Central details
are remembered more frequently fromshocking than happy experiences. Mem-
ory and Cognition, 30(7), 10101020.
Bouvier, J. (1856). A law dictionary, adapted to the constitution and
laws of the United States of America (6th ed.). Retrieved from:
Brimacombe, C. A., Jung, S., Garrioch, L., & Allison, M. (2003). Perceptions of older
adult eyewitnesses: Will you believe me when Im64? Lawand Human Behavior,
27, 507522.
Brimacombe, C. A., Quinton, N., Nance, N., & Garrioch, L. (1997). Is age irrelevant?
Perceptions of young and old adult eyewitnesses. Law and Human Behavior, 21,
Bruck, M., & Ceci, J. (1999). The suggestibility of childrens memory. Annual Review
of Psychology, 50, 419439.
Carter, C. A., Bottoms, B. L., & Levine, M. (1996). Linguistic and socioemotional
inuences on the accuracy of childrens reports. Law and Human Behavior, 20,
Christianson, S.-A. (1992). Emotional stress and eyewitness memory: A critical
review. Psychological Bulletin, 112(2), 284309.
Fisher, R. P., &Schreiber, N. (2007). Interviewprotocols toimprove eyewitness mem-
ory. In M. P. Toglia, ReadF J.D., D. F. Ross, & R. C. L. Lindsay (Eds.), Handbook
of eyewitness psychology. Volume 1: Memory for events. (pp. 5380). Erlbaum:
Mahwah, NJ.
Hanna, K., Davies, E., Henderson, E., Crothers, C., & Rotherham, C. (2010). Child wit-
nesses in the NewZealand criminal courts: A reviewof practice and implications for
policy. Auckland, NZ: Institute of Public Policy.
Heuer, F., & Reisberg, D. (1992). Emotion arousal and memory for detail. In S.-A.
Christianson (Ed.), Handbook of emotion and memory (pp. 151180). Hillsdale,
NJ: Erlbaum.
Hudson, J. A., & Shefeld, E. G. (1999). The role of reminders in young childrens
memory development. In L. Balter, & C. S. Tamis-LeMonda (Eds.), Child psychol-
ogy: Ahandbook of contemporary issues (pp. 193214). NewYork, NY: Psychology
Jack, F., Cannan, P., & Zajac, R. (2009). Adolescents under cross-examination: A quan-
titative analysis of lawyers questions and witnesses responses. Sorrento, Italy:
European Association of Psychology and Law.
Jack, F., Leov, J., &Zajac, R. (2013). Age-related differences in the free-recall accounts
of child, adolescent, and adult witnesses. Applied Cognitive Psychology, 2951.
http://dx.doi.org/10 1002/acp
Marin, B. V., Holmes, D. L., Guth, M., & Kovac, P. (1979). The potential of children
as eyewitnesses: A comparison of children and adults on eyewitness tasks. Law
and Human Behavior, 3, 295306.
Melnyk, L., Crossman, A. M., & Scullin, M. H. (2007). The suggestibility of childrens
memory. In M. P. Toglia, J. D. Read, D. F. Ross, & R. C. L. Lindsay (Eds.), Handbook
of eyewitness psychology. Volume 1: Memory for events (pp. 401427). Mahwah,
NJ: Erlbaum.
Milne, R., & Bull, R. (1999). Investigative interviewing: Psychology and practice. Chich-
ester, UK: Wiley.
Morgan, K., &Hayne, H. (2007). Nonspecic verbal cues alleviate forgetting by young
children. Developmental Science, 10, 272733.
Oates, K., &Shrimpton, S. (1991). Childrens memories for stressful andnon-stressful
events. Medicine, Science and The Law, 31, 410.
ONeill, S., & Zajac, R. (2013a). Preparing children for cross-examination: Howdoes
intervention timing inuence efcacy? Psychology, Public Policy, and Law, 19,
ONeill, S., & Zajac, R. (2013b). The role of repeated interviewing in childrens
responses to cross-examination-style questioning. British Journal of Psychology,
104, 1438.
Peterson, C., Pardy, L., Tizzard-Drover, T., & Warren, K. L. (2005). When initial inter-
views are delayed a year: Effect on childrens 2-year recall. Law and Human
Behavior, 29, 527541.
Peterson, C., & Whalen, N. (2001). Five years later: Childrens memory for medical
emergencies. Applied Cognitive Psychology, 15, S7S24.
Pigot, T., Kilkerr, A., Parker, R., Rafferty, A., Temkin, J., & Coulson-Gilmer, K. (1989).
Report of the advisory group on video evidence. London, UK: Home Ofce.
Pipe, M.-E., & Henaghan, M. (1996). Accommodating childrens testi-
mony: Legal reforms in New Zealand. Criminal Justice and Behavior, 23,
Plotnikoff, J., &Woolfson, R. (2009). Measuring up? Evaluating implementation of gov-
ernment commitments to young witnesses in criminal proceedings. London, UK:
NSPCC and Nufeld Foundation.
Plotnikoff, J., & Woolfson, R. (2012). Kicking and screaming The slowroad to best
evidence. In J. R. Spencer, & M. E. Lamb (Eds.), Children and cross-examination:
Time to change the rules? (pp. 2142). Oxford, UK: Hart Publishing.
Poole, D. A., & White, L. T. (1991). Effects of question repetition on the eye-
witness testimony of children and adults. Developmental Psychology, 27,
Priestley, G., Roberts, S., &Pipe, M.-E. (1999). Returning to the scene: Reminders and
context reinstatement enhance childrens recall. Developmental Psychology, 35,
Read, J. D., & Connolly, D. A. (2007). The effects of delay on long-term memory for
witnessed events. In M. P. Toglia, J. D. Read, D. F. Ross, & R. C. L. Lindsay (Eds.),
Handbook of eyewitness psychology. Volume 1: Memory for events (pp. 117155).
Mahwah, NJ: Erlbaum.
Righarts, S., ONeill, S., & Zajac, R. (2013). Addressing the negative effect of cross-
examination questioning on childrens accuracy: Can we intervene? Law and
Human Behavior, 37, 354365.
Sheslow, D., &Adams, W. (2003). Wide range assessment of memory and learning (2nd
ed.). Lutz, FL: Psychological Assessment Resources.
Spencer, J. R. (2012). Conclusions. In J. R. Spencer, & M. E. Lamb (Eds.), Children and
cross-examination: Time to change the rules? (pp. 171202). Oxford, UK: Hart
Spencer, J. R., & Lamb, M. E. (Eds.). (2012). Children and cross-examination: Time to
change the rules?. Oxford, UK: Hart Publishing.
Sutherland, R., &Hayne, H. (2001). Age-relatedchanges inthe misinformationeffect.
Journal of Experimental Child Psychology, 79, 388404.
Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5th ed.). Boston,
MA: Pearson.
Tulving, E., & Thomson, D. M. (1973). Encoding specicity and retrieval processes in
episodic memory. Psychological Review, 80, 352373.
Turtle, J. W., & Wells, G. L. (1988). Children versus adults as eyewitnesses: Whose
testimony holds up under cross-examination? In M. M. Gruneberg, P. E. Morris,
& R. N. Sykes (Eds.), Practical aspects of memory: Current research and issues (Vol.
1) (pp. 2733). Chichester, England: John Wiley & Sons.
Tustin, K., & Hayne, H. (2010). Dening the boundary: Age-related changes in child-
hood amnesia. Developmental Psychology, 46, 10491061.
Valentine, T., & Maras, K. (2011). The effect of cross-examination on the accuracy of
adult eyewitness testimony. Applied Cognitive Psychology, 25, 554561.
Wechsler, D. (1999). Wechsler Abbreviated Scale of Intelligence. San Antonio, TX:
Harcourt Assessment.
Wessel, I., Van De Kooy, P., & Merckelback, H. (2000). Differential recall of central
and peripheral details of emotional slides is not a stable phenomenon. Memory,
8(2), 95109.
Zajac, R., & Cannan, P. (2009). Cross-examination of sexual assault complainants: A
developmental comparison. Psychiatry, Psychology and Law, 16, S36S54.
Zajac, R., Gross, J., & Hayne, H. (2003). Asked and answered: Questioning children in
the courtroom. Psychiatry, Psychology and Law, 10, 199209.
Zajac, R., & Hayne, H. (2003). I dont think thats what really happened: The effect of
cross-examination on the accuracy of childrens reports. Journal of Experimental
Psychology: Applied, 9, 187195.
Zajac, R., & Hayne, H. (2006). The negative effect of cross-examination on childrens
accuracy: Older children are not immune. Applied Cognitive Psychology, 20, 316.
Zajac, R., Jury, E., & ONeill, S. (2009). The role of psychosocial factors in young
childrens responses to cross-examination style questioning. Applied Cognitive
Psychology, 23, 918935.