Journal of Experimental Child Psychology 163 (2017) 15–31

Contents lists available at ScienceDirect

Journal of Experimental Child

Psychology
journal homepage: www.elsevier.com/locate/jecp

The role of maternal elaborative structure and

control in children’s memory and suggestibility
for a past event
Gabrielle F. Principe a,⇑, Jacqueline Trumbull a, Grace Gardner a,
Elizabeth Van Horn b, Abigail M. Dean c
a
Department of Psychology, College of Charleston, Charleston, SC 29401, USA
b
Department of Psychology, North Carolina State University, Raleigh, NC 27607, USA
c
Department of Psychology, Swarthmore College, Swarthmore, PA 19081, USA

a r t i c l e i n f o a b s t r a c t

Article history: In this investigation, preschool-aged children experienced a staged

Received 6 March 2017 event about which their mothers received misinformation
Revised 1 June 2017 suggesting that their children witnessed an activity that did not
occur. Later, mothers were asked to talk about this event with their
children. Consistent with previous research, mothers’ provision of
Keywords:
structure (defined as elaborative questions and statements) and
Autobiographical memory
Suggestibility
degree of control (defined in terms of functional control of
Conversation conversational turns) emerged as separate dimensions of maternal
Mother–child communication memory sharing style. When later interviewed by an unfamiliar
Preschool age (2–5 years) examiner about the event, children whose mothers demonstrated
Eyewitness testimony both high structure and high control provided the highest levels
of false reports of the activity suggested to mothers and generously
embellished their accounts of this activity with nonoccurring
details. In contrast, children with mothers who provided low
structure, regardless of their degree of control, made few
false reports and used sparse narrative detail. The implications of
these findings for children’s memory and suggestibility are
discussed.
Ó 2017 Elsevier Inc. All rights reserved.

⇑ Corresponding author.
E-mail address: principeg@cofc.edu (G.F. Principe).

http://dx.doi.org/10.1016/j.jecp.2017.06.001
0022-0965/Ó 2017 Elsevier Inc. All rights reserved.
16 G.F. Principe et al. / Journal of Experimental Child Psychology 163 (2017) 15–31

Introduction

A substantial body of work demonstrates that memory sharing conversations about past events
with parents play a significant role in the development of children’s autobiographical remembering
skills. Through these conversations, children learn how to search their memory for the details of per-
sonal experiences and relay these recollections in a narrative form to others (see, e.g., Nelson & Fivush,
2004). One of the most robust patterns in this literature centers on naturally occurring variations in
the elaborativeness of mothers’ memory sharing style. A high-elaborative style typically is defined
as an amalgam of elements that serve to scaffold children’s accounts such as asking open-ended
wh- questions, introducing new descriptive detail to help children remember, confirming children’s
contributions, and following children’s personal perspectives. Low-elaborative mothers provide less
structure and repeat their own questions with no new information. Longitudinal and experimental
work documents that a high-elaborative style produces mnemonic benefits. For example, when
remembering with their mothers as well as with others, children whose mothers use a high-
elaborative style produce more detailed and coherent event narratives than children with low-
elaborative mothers (Farrant & Reese, 2000; Haden, Haine, & Fivush, 1997; Hendrick, Haden, &
Ornstein, 2009; Leichtman, Pillemer, Wang, Koreishi, & Han, 2000; Reese & Newcombe, 2007). Fur-
thermore, a high-elaborative style is associated with increases in the amount of accurate information
children report about their experiences (Cleveland, Reese, & Grolnick, 2007; Leichtman et al., 2000).
The facilitative effects of mother–child memory sharing conversations notwithstanding, Principe,
DiPuppo, and Gammel (2013) found that mothers can be a potent source of error in children’s remem-
bering when mothers hold false beliefs about their children’s experiences. Their study was motived by
experiments in the suggestibility literature (for a review, see Principe & Schindewolf, 2012) demon-
strating that natural conversations with others about the past can change the way in which children
remember specific personal experiences, especially when conversational partners differ in their beliefs
about what happened. Principe and colleagues asked mothers to talk in a natural manner with their
preschoolers about an earlier nonshared scripted event. Immediately before this conversation, some
mothers received a false suggestion about the event stating that their children may have experienced
a certain activity. Mothers were asked to question their children about whether or not this activity
happened during the event. When later interviewed about the event by an unfamiliar examiner, chil-
dren whose mothers were misinformed were much more likely to claim to have experienced the
suggested-but-nonoccurring activity and more generously embellished their accounts with fabricated
details in line with the suggestion than children whose mothers were not exposed to the misleading
information.
Consistent with the literature on memory sharing conversations, Principe et al. (2013) found that
mothers’ stylistic differences were met with variation in children’s memories. Mothers were charac-
terized as high or low elaborative in terms of the numbers of elaborations (i.e., questions or statements
that introduce new information to the discussion) and evaluations (i.e., comments that confirm or
negate children’s utterances) they made during the post-event conversation with their children
(Reese & Brown, 2000). As expected, children with high-elaborative mothers produced more detailed
accounts of the event during both the mother–child dialogue and the later interview compared with
children with low-elaborative mothers. A high-elaborative maternal style also was linked with greater
accuracy of children’s recall of actually experienced activities during the event. Despite these positive
effects of maternal elaborativeness, during both the mother–child conversation and the interview,
children whose mothers were characterized as high elaborative were more likely than children with
low-elaborative mothers to wrongly report the nonoccurring activity suggested to mothers and to
describe this nonevent with abundant narrative detail.
Why might the high-elaborative maternal style usually associated with more skilled remembering
be linked with higher levels of false reports? Given what is known about the potency of conversation-
ally introduced misinformation on memory (see Principe & Schindewolf, 2012), it seems likely that the
tendency to introduce new detail characteristic of the high-elaborative style might produce greater
memory error when mothers hold false beliefs about their children’s experiences. Examination of
the mother–child dialogues suggests that this happened. Compared with the low-elaborative mothers,
 G.F. Principe et al. / Journal of Experimental Child Psychology 163 (2017) 15–31 17

the high-elaborative mothers provided more than three times the number of novel elaborations con-
sistent with the theme of the suggestion and often went beyond repeating the original suggestion and
offered new and necessarily false details about the suggested activity. Children also behaved differ-
ently as a function of maternal style, with children of high-elaborative mothers appearing especially
deferent. For instance, children with high-elaborative mothers were more likely than children with
low-elaborative mothers to acquiesce to their mothers’ first elaboration about the suggested activity
and acquiesced to proportionally more of their mothers’ questions and statements about this non-
event. Thus, the persistence of elaborations about the misinformation among the high-elaborative
mothers suggests that at least some of these mothers had inadvertently created a memory sharing
environment that would be construed as coercive in the suggestibility literature (see Ceci & Bruck,
1995) and that consequently would be expected to lead to high levels of acquiescence during that con-
versation as well as intrusions of that information during subsequent recalls.
Given these findings demonstrating that high maternal elaborativeness can both help and hinder
children’s memory when mothers and children disagree on the facts of a past event, an important next
step is to examine in a more fine-grained manner which aspects of this style account for its effects on
memory. As described above, a high-elaborative style typically is defined as composed of several ele-
ments: open-ended wh- questions, the addition of new details, praise of children’s memory contribu-
tions, and a willingness to follow in on children’s conversational leads. Most studies in this literature
have not separately examined the various elements that make up maternal style. However, there is
growing evidence (Cleveland & Morris, 2014; Cleveland & Reese, 2005; Cleveland et al., 2007;
Larkina & Bauer, 2010) that elaborative structure (i.e., the provision of elaborative statements and
questions) is independent from autonomy support (i.e., a willingness to talk about what children want
to talk about). For instance, Cleveland and Reese (2005) found that some mothers who used frequent
open-ended elaborative questions regularly followed in on their children’s conversational leads,
whereas other mothers who used this style controlled the remembering agenda. Likewise, some moth-
ers who were low in elaborative structure supported their children’s autonomy in remembering,
whereas others were controlling. Importantly, these maternal differences were met with variation
in children’s memory sharing behavior such that children whose mothers were high in elaborative
structure and autonomy supportive provided the most memory information, whereas children whose
mothers were low in elaborative structure and controlling offered the lowest levels of recall. Further
demonstrating the independence of these two dimensions of style, manipulations designed to increase
the degree of parental control (i.e., telling parents that their children’s memory accuracy would later
be tested) have no effect on parents’ provision of elaborative structure (Cleveland et al., 2007). Finally,
variations in structure and control seem to have independent and differing effects on children’s event
remembering. High structure is linked with increases in the length and coherence of children’s narra-
tives, whereas high autonomy support is associated with increases in children’s motivation to engage
in memory sharing (Cleveland & Morris, 2014).
These findings regarding the effects of structure and control notwithstanding, it is unclear how
these dimensions of memory sharing style may work to shape children’s remembering when mothers’
beliefs about the past differ from their children’s experiences. There are reasons to suspect that the
combination of high structure and control may be associated with the highest levels of inaccuracies
in children’s event reports when mothers and children hold differing beliefs about what happened.
This is because the combined tendencies of high-structure mothers to offer new information consis-
tent with their beliefs and the tendencies of high-control mothers to promote a particular conversa-
tional agenda bear close similarity to questioning behaviors that have been found to distort children’s
recall in the suggestibility literature. Specifically, Ceci and Bruck (1993) introduced the term ‘‘inter-
viewer bias” to characterize interviewers who hold a priori beliefs about what children experienced
and consequently mold the conversation to elicit accounts consistent with their beliefs. This bias
can be communicated to children through a range of interviewer behaviors such as asking questions
that introduce new information in line with beliefs about the event, confirming and praising responses
congruent with those beliefs, negating responses that are inconsistent with beliefs, repeating ques-
tions if the initial response is inconsistent with beliefs, changing the topic to realign the conversation
in the direction of beliefs, and failing to ask questions that might elicit contradictory information. The
18 G.F. Principe et al. / Journal of Experimental Child Psychology 163 (2017) 15–31

concept of interviewer bias is not limited to forensic interviewers and can include any conversational
partners who have an agenda when talking with children about past events.
Dozens of studies in this literature demonstrate that when biased interviewers work to control the
conversational agenda, children come to report nonexperienced details in line with the adults’ beliefs
during both the biased interviews themselves and subsequent interviews with nonbiased adults (for
reviews, see Ceci & Bruck, 1995; Principe, Greenhoot, & Ceci, 2014). This pattern occurs when adults
are given misinformation about an event and asked to question children in any way they see fit
(Thompson, Clarke-Stewart, & Lepore, 1997; White, Leichtman, & Ceci, 1997) and when children are
questioned using structured interviews (e.g., Bruck, Ceci, & Hembrooke, 2002; Leichtman & Ceci,
1995). Furthermore, cross-study comparisons (see Bruck et al., 2002) and experiments that have
manipulated interviewer behavior (e.g., Garven, Wood, & Malpass, 2000; Garven, Wood, Malpass, &
Shaw, 1998; Leichtman & Ceci, 1995) have revealed that the use of multiple biased techniques pro-
duces higher levels of memory error than when only one type of biased utterance is employed. For
example, Garven et al. (1998) found that children’s false reports nearly doubled when misleading
questions (i.e., questions that introduce new information in line with the interviewer’s beliefs) were
paired with other interviewer behaviors that communicated a bias such as selective praise, peer pres-
sure, and encouragement to speculate about occurrences in line with the misinformation. Related
work (Garven et al., 2000) has shown that the addition of mere selective feedback (i.e., praise for fol-
lowing an interviewer’s lead and disapproval for disagreeing) more than doubled the likelihood that
children assented to misleading questions about an earlier event.
These findings suggests that bias might not be very apparent to children when only one technique
(e.g., misleading questions) is used and that multiple techniques more strongly convey bias and con-
sequently lead to higher levels of memory errors. In the vernacular of the mother–child memory shar-
ing literature, when mothers have been exposed to misleading information about their children’s
experiences, the provision of elaborations that introduce event information consistent with the mis-
information in and of themselves might not noticeably communicate mothers’ bias and necessarily
sway children’s reports. However, the addition of techniques to control the conversational agenda
(e.g., negating certain responses, changing the topic of conversation to steer the dialogue in the direc-
tion of mothers’ beliefs) may provide children with sufficient information about their mothers’ bias to
engender high levels of child memory error.

The current study

The purpose of the current research was to contrast the dimensions of naturally occurring maternal
elaborative structure and maternal control when mothers have been exposed to misinformation about
what their children experienced in relation to children’s memory contributions during the mother–
child memory sharing conversation and in a subsequent interview with an experimenter. Replicating
Principe et al.’s (2013) procedure, mothers were exposed to misinformation about an earlier non-
shared event and then asked to talk with their children about this event. To the extent that maternal
elaborative structure and maternal control emerge as separate dimensions of mothers’ memory shar-
ing behaviors and mirror the behaviors of biased interviewers (see Ceci & Bruck, 1993), we expected to
find the highest levels of interference on the accuracy of children’s reports when mothers display a
high-elaborative structure and a high level of control.

Method

Participants

Participants were 108 preschool-aged children (mean = 57 months, range = 50–68) recruited from
one of six preschool programs in suburban areas of the northeastern United States. Approximately 90%
of the children were European American, and approximately half were female. Written consent and
verbal assent were obtained from participating parents and children, respectively. To ensure that
any child reports were consistent with the misinformation given to mothers, an additional 12 children,
 G.F. Principe et al. / Journal of Experimental Child Psychology 163 (2017) 15–31 19

2 from each preschool with a mean age and range comparable to the sample described above, partic-
ipated in the study. These children experienced the same procedures as the full sample, with the only
exception being that their mothers were not exposed to misinformation about the event.

Procedure

The to-be-remembered event

Following Principe et al. (2013), children were visited in their classrooms by a confederate named
Magic Mumfry who performed a scripted magic show made up of 11 tricks such as producing colored
drawings in a seemingly blank book, making lollipops appear in a previously empty bag, and produc-
ing a bouquet of flowers from a single stem. At the end of the show, Mumfry attempted to pull a live
rabbit out of a top hat. After several failed efforts, Mumfry apologized and left the building. Mothers
were not present during the show and were given no prior details about it.

Mother–child conversations
Replicating Principe et al.’s (2013) procedure, the day before the interview, mothers received a let-
ter asking them to question their children about the magic show. The letter stated, ‘‘During the show,
the magician tried to pull a rabbit out of his hat but failed.” It included a suggestion that the magician’s
rabbit may have gotten loose and asked mothers to question their children about this possibility:
‘‘Please ask your child if she or he remembers whether the magician’s rabbit got loose in the school
on the day of the show. We are particularly interested in children’s memory for this event.” Mothers
were instructed to carry out this conversation the morning of the interview and were provided digital
voice recorders to record their interactions. The letter asked mothers to begin by asking a general
question such as ‘‘Tell me what happened during the magic show at your school” and to take their
time and discuss the show in a natural manner.

Memory interview
One week after the show (mean = 7 days, range = 6–8), children were questioned by a condition-
unaware interviewer. Children were asked to talk ‘‘only about things that you remembering happen-
ing to you—things that you really did or remember seeing with your own eyes.” The interview began
with an open-ended question: ‘‘I’d like you to tell me everything that you remember about the day
that Magic Mumfry visited your school. I wasn’t there that day, so I don’t know what happened. Don’t
guess or make anything up. Just tell me what you did or saw the time that Magic Mumfry came to your
school.” After additional open-ended probing (e.g., ‘‘What else happened?”), the interviewer asked a
specific question if the loose rabbit (heretofore referred to as the target activity) had not yet been
reported: ‘‘Did anything happen to Mumfry’s rabbit?” Children who still had not yet mentioned the
loose rabbit were posed a leading question: ‘‘Did Mumfry’s rabbit get loose in the school?” Children
who reported the target activity were asked for the source of their memory, that is, whether they
merely had ‘‘heard” the rumor about the loose rabbit from someone else or actually had seen it them-
selves (i.e., ‘‘Did you see [action vis-à-vis the loose rabbit, as noted by the child, e.g., the rabbit in the
classroom] with your own eyes, or did you hear about it from someone?”). The order of the source
options was counterbalanced. Throughout the interview, children were prompted to elaborate with
general probes (i.e., ‘‘Tell me more about that”).

Maternal questionnaire
Immediately before the mother–child conversation, mothers were asked to rate (on a 5-point Likert
scale) the degree to which they believed that a rabbit had been loose in their children’s school on the
day of the show.

Coding of the mother–child conversations

Mother–child conversations were transcribed verbatim from audiotapes and coded by condition-
unaware raters. For each mother–child pair, conversations were parsed into propositions (subject +
verb + complement structures) about the magic show. Off-topic conversation was not scored.
20 G.F. Principe et al. / Journal of Experimental Child Psychology 163 (2017) 15–31

Maternal elaborative structure

Following Reese and Brown (2000), elaborative structure was characterized by counting mothers’
provision of (a) open-ended wh- questions that asked children to provide new memory information
about the magic show (e.g., ‘‘Where did the rabbit hide?” ‘‘What kind of tricks did Mumfry do?”),
(b) yes/no questions that asked children to either confirm or deny a piece of new information provided
by the mother (e.g., ‘‘Did he pull a rabbit out of his hat?” ‘‘Did Mumfry make anything disappear?”),
and (c) statements that relayed new pieces of information about the magic show that did not require
responses from children (e.g., ‘‘Mumfry must have been sad that he lost his rabbit”). Maternal elabo-
rations that included new details were necessarily guesses, inferences, or based on secondhand infor-
mation because mothers were not present during the show.

Maternal control
Following Kulkofsky (2011), we coded each maternal turn along a 3-point scale in terms of how
functionally controlling or autonomy supportive it was in the context of the memory sharing conver-
sation, such that no turn was considered in isolation from the ongoing dialogue. A turn was considered
complete either when the child responded or when the mother paused long enough for the child to
respond (even if the child did not do so). A turn was scored as a 1 (autonomy supportive) if it func-
tioned to continue or expand on the child’s topic. A 2 (neutral) was coded if the turn did not expand
on, guide, or change the child’s topic. If the mother negated the child or changed the topic of the con-
versation, the turn was scored as a 3 (controlling). A single control score was calculated as the average
score for all turns.

Maternal bias
To characterize the extent to which mothers engaged in conversational behaviors that would be
described as conveying interview bias in the child witness literature (see Bruck et al., 2002; Ceci &
Bruck, 1993), we coded for nine different types of techniques consistent with the idea that the magi-
cian’s rabbit got loose in the school on the day of the magic show. Specifically, we counted the number
of mothers’ utterances that conveyed a bias that the rabbit had gotten loose by (a) introducing new
information (‘‘What did you do to try to catch the rabbit?”), (b) suggesting a particular response (‘‘I
think the rabbit probably just wanted to go outside and eat the grass, don’t you?” and ‘‘The bunny
must have been frightened by all the commotion, wasn’t it?”), (c) limiting the range of answers by
forcing choices (e.g., ‘‘Was the rabbit brown or white like Buggs?”), (d) asking children to imagine, sup-
pose, or consider options (‘‘Where do you think the rabbit was hiding?”), (e) repeating a previous
question if the initial response was inconsistent with the idea that a rabbit had gotten loose, (f) con-
firming or praising responses congruent with the notion of a loose rabbit, (g) negating responses
inconsistent with the idea of a loose rabbit, (h) changing the topic to realign the conversation in
the direction of a loose rabbit, and (i) introducing the possibility that something occurred with a rabbit
without suggesting what happened (e.g., ‘‘What else did the rabbit do?”). Because all mothers were
instructed to ask whether a rabbit got loose during the magic show, we did not code mothers’ query
about this possibility as biased if it was restated verbatim in question form (e.g., ‘‘Did a rabbit get loose
during the show?”). However, if mothers tacked on biased information to this initial query, then it was
scored as biased (e.g., ‘‘The magician’s rabbit got loose, didn’t it?”).

Child codes
First, we scored each proposition (i.e., subject–verb construction) provided by children that
described an original piece of memory information about the magic show as a memory response. Then,
we coded each memory response as accurate, inaccurate, or ambiguous. When it was unclear whether a
statement was consistent with what actually happened, it was coded as ambiguous. For inaccurate
memory responses, we identified those that provided information consistent with the loose rabbit
suggestion.We also coded children’s responses to each maternal turn that offered information about
the suggested loose rabbit as either acquiescence or denial. Following Gilstrap and Ceci (2005),
responses were scored as acquiescence when they expressed agreement with their mothers’ state-
ment or question, whereas denial occurred when children expressed disagreement, reported unrelated
information, said ‘‘I don’t know,” or gave no response.
 G.F. Principe et al. / Journal of Experimental Child Psychology 163 (2017) 15–31 21

Coding of the interviews

Videotaped interviews were transcribed verbatim and coded by condition-unaware raters. First,
children’s narrative responses about the actually occurring magic show were broken down into propo-
sitions, and each proposition that described an original piece of information was scored for accuracy.
Next, transcriptions also were coded for children’s reports of the target activity (i.e., a loose rabbit) and
the degree of prompting that elicited the report. Propositions about the loose rabbit suggestion were
parsed into three categories. Statements were coded as verbatim if they literally repeated information
suggested in the letter such as ‘‘The rabbit was loose in my school.” Constructive propositions were
utterances that were consistent with the notion of a loose rabbit but went beyond the literal content
of the letter (e.g., ‘‘The bunny hopped all over me,” ‘‘I saw his cotton tail in the toy box”). Fantastic
statements described occurrences about a loose rabbit that could not have happened in reality such
as ‘‘The bunny called the cops.”
We also explored the degree to which children’s interview responses describing the suggested
loose rabbit overlapped with things said during the mother–child conversations. Each proposition
about the loose rabbit was assigned to one of four categories. Interview propositions that overlapped
with an idea first expressed by the mother during mother–child conversations were scored as overlap-
ping mother. Propositions that overlapped with any utterances made by the child (but not first sug-
gested by the mother) were coded as overlapping self. Nonoverlapping propositions reflected
information never uttered during the mother–child conversations. Ambiguous propositions were too
vague to be assigned to one of the other categories.
One fifth (20%) of the mother–child conversation transcripts and 20% of the interview transcripts
were selected at random and independently coded by two raters. Interrater agreement was calculated
as the ratio of the number of agreements over the total number of codes given. Reliability across all
mother–child conversation variables was 90% (range = 83–96%), and reliability across all interview
variables was 94% (range = 90–98%). Coding discrepancies were resolved through discussion.

Results

Preliminary analyses

An initial set of analyses indicated no main effects in the dependent variables examined below as a
function of the confederate who played Mumfry, the interviewer, or children’s school, gender, or age in
months. Furthermore, there was no evidence among the 12 children whose mothers did not receive
misinformation about the magic show of exposure to the idea of a loose rabbit. None of these mothers
mentioned a loose rabbit during her conversation with their child, and none of these children reported
anything about a loose rabbit during either the mother–child conversation or the memory interview.
These findings suggest that any mother and child talk about a loose rabbit was prompted by our letter
manipulation. These 12 mothers and children are not included in the analyses below.

Main analyses

All mothers followed the directions in the letter and began by posing a general, open-ended ques-
tion followed by more specific questions and statements. Furthermore, all mothers asked whether the
magician’s rabbit had gotten loose. As expected, mothers varied in their use of elaborative statements
and questions and the degree to which they controlled the direction of the conversation. Mothers’
elaborative statements and questions ranged from 5 to 89 (M = 30.10, SD = 19.10), and their control
scores ranged from 1.00 to 2.78 (M = 1.76, SD = 0.33). In support of the presumed independence of
the dimensions of maternal elaborative structure and maternal control (see, e.g., Cleveland & Reese,
2005), mothers’ provision of elaborative questions and statements and their degree of control were
not correlated (p = .47). With the independence of these two dimensions established, we classified
mothers as high or low on structure and high or low control on the basis of a median split and formed
four post hoc groups: high structure/controlling, high structure/autonomy supportive, low structure/-
22 G.F. Principe et al. / Journal of Experimental Child Psychology 163 (2017) 15–31

controlling, and low structure/autonomy supportive. In line with Cleveland and Reese (2005), rela-
tively equivalent numbers of mothers fell into each of the four groups. Admittedly, post hoc grouping
on the basis of a median score is somewhat arbitrary, and mothers close to the median in both groups
might not be exhibiting statistically different styles. Nonetheless, this sort of division in earlier work
has been associated with differences in children’s event reports (e.g., Cleveland & Reese, 2005; Fivush
& Fromhoff, 1988; Principe et al., 2013).
The general analytic strategy involved a series of analyses of variance (ANOVAs) with maternal
style (four levels) as a between-participants factor to examine the effects of the post hoc maternal
style groups on mothers’ and children’s behaviors during the mother–child conversation as well as
children’s reports during the interview. Significant effects of maternal style were followed up with
Tukey’s honestly significant difference (HSD) tests at p < .05 to test pairwise differences. In cases
where the dependent variable was dichotomous (e.g., children either did or did not recall the target
activity), logistic regression was conducted with maternal style group as the predictor.

Mother–child conversation

First, we explored whether mothers differed as a function of our post hoc groupings in terms of
their provision of utterances typical of individuals who are exhibiting interviewer bias (see Ceci &
Bruck, 1993). Table 1 displays the total number of biased utterances in line with the theme of the tar-
get activity and the number of mothers who used two or more biased techniques as well as the num-
bers for each of the four maternal style groups. Mothers differed as a function of group in the total
number of biased utterances made, F(3, 107) = 20.06, p < .0001, g2p = .37. Follow-up tests indicated that
mothers characterized as high structure and controlling made more biased utterances than mothers
who were high structure and autonomy supportive, who in turn produced more biased utterances
than mothers in either of the low-structure groups. Furthermore, a series of logistic regressions carried
out to predict the likelihood of mothers’ use of two or more biased techniques revealed that mothers
characterized as high structure/controlling were more likely to use two or more biased techniques
than mothers who were high structure/autonomy supportive, who in turn were more likely to use
two or more biased techniques than mothers in either low-structure group, v2s(1, N = 108)  4.23,
ps < .05, Øs  20.
Next, we examined whether children responded differently to their mothers’ elaborative state-
ments and questions about the target activity as a function of maternal style. These data are presented
in Table 2. Children with mothers in the high-structure/controlling group were more likely to acqui-
esce to their mothers’ first elaboration about the target activity than children with mothers in the h
igh-structure/autonomy-supportive group, who in turn were more likely to acquiesce to their moth-
ers’ initial utterance about the target activity than children with mothers in either low-structure
group, v2s(1, N = 108)  4.31, ps < .05, Øs  20. Children also differed as a function of group in terms
of the proportion of responses to all maternal elaborations about the target activity that were coded as
acquiescences, F(3, 107) = 25.83, p < .0001, g2p = .43. Follow-up analyses indicated that children with
high-structure/controlling mothers acquiesced to a higher percentage of their mothers’ utterances
about the loose rabbit than children with mothers in the high-structure/autonomy-supportive group.
Children with mothers in the two low-structure groups had the lowest acquiescence rates.
Next, we examined whether maternal style was associated with children’s provision of narrative
detail during the mother–child conversation. Table 3 shows the average number of accurate, inaccu-

Table 1
Mean numbers (and standard deviations) of maternal utterances that conveyed a bias consistent with the suggested target activity
and the proportions (and counts) of mothers who used two or more different types of biased techniques during the mother–child
conversation.

n Biased utterances Two or more biased techniques

Low structure/autonomy supportive 26 2.65 (2.23) .27 (7)
Low structure/high control 25 3.00 (2.52) .28 (7)
High-structure/autonomy supportive 28 5.86 (4.38) .57 (16)
High-structure/high control 29 9.79 (5.25) .83 (24)
 G.F. Principe et al. / Journal of Experimental Child Psychology 163 (2017) 15–31 23

Table 2
Proportions (and counts) of children who acquiesced to their mothers’ first biased utterance suggesting the target activity and
children’s acquiescence rates (and standard deviations) following all maternal utterances suggesting new information about the
target activity during the mother–child conversation.

First biased utterance Acquiescence rate

Low structure/autonomy supportive .19 (5) .21 (.20)
Low structure/high control .52 (13) .42 (.35)
High structure/autonomy supportive .18 (5) .29 (.23)
High structure/high control .79 (23) .75 (.21)

rate, ambiguous, and total (accurate + inaccurate + ambiguous) memory responses broken down by
maternal style group. Note that we separated inaccurate memory responses that described a loose
rabbit from inaccurate memory responses that described activities irrelevant to the theme of a loose
rabbit. There was a significant effect of maternal style for the total number of memory responses pro-
vided by children, F(3, 107) = 8.36, p < .0001, g2p = .19. Follow-up tests indicated that children with
mothers in the high-structure/autonomy-supportive and high-structure/controlling groups produced
more memory responses than children with mothers in the low-structure/autonomy-supportive and
low-structure/controlling groups. As also shown on Table 3, children relayed a good deal of accurate
information about the magic show to their mothers. The amount of accurate information uttered by
children differed as a function of maternal style, F(3, 107) = 4.44, p < .01, g2p = .11, such that children
whose mothers were in the high-structure/autonomy-supportive group uttered more accurate mem-
ory responses than children with mothers in the high-structure/high-control group, who in turn made
more accurate memory responses than children whose mothers were characterized as low structure/
autonomy supportive or low structure/controlling. There also was an effect of maternal style on the
number of inaccurate memory responses in line with the target activity, F(3, 107) = 13.07, p < .0001,
g2p = .27, such that children whose mothers were in the high-structure/high-control group reported
more memory responses describing a loose rabbit than children with mothers in the high-struc
ture/autonomy-supportive group. Children with mothers in the two low-structure groups reported
the lowest number of memory responses consistent with the theme of a loose rabbit. There were
no group differences in children’s provision of inaccurate information about activities other than a
loose rabbit.
Demonstrating a direct link between maternal bias and children’s accounts is a significant correla-
tion between the number of biased utterances made by mothers and the number of children’s memory
responses in line with the idea of a loose rabbit (r = .44, p < .0001). Additional analyses suggest that the
use of multiple types of biased techniques may be particularly likely to elicit false reports in line with
the bias. Children with mothers who used two or more biased techniques uttered more than five times
the number of memory responses consistent with the notion of a loose rabbit (n = 54, M = 11.33,
SD = 8.82) than children whose mothers used only one or no biased techniques (n = 54, M = 2.03,
SD = 2.79), t(106) = 7.38, p < .001.

Memory interview

Reporting the target activity

The proportions and counts of children who reported the target activity (i.e., that the rabbit was
loose in the school) during the interview are shown in Table 4 by maternal style group and the level

Table 3
Mean numbers (and standard deviations) of memory responses offered by children during the mother–child conversation.

Accurate Inaccurate Inaccurate other Ambiguous Total

rabbit
Low structure/autonomy supportive 9.96 (6.04) 2.77 (4.06) 0.54 (1.07) 1.70 (2.13) 14.96 (8.59)
Low structure/high control 10.52 (7.49) 2.92 (2.71) 0.04 (0.20) 1.12 (1.05) 14.64 (8.38)
High structure/autonomy supportive 18.18 (13.10) 7.25 (7.16) 0.43 (1.37) 1.29 (1.63) 27.14 (16.55)
High structure/high control 12.48 (8.64) 12.90 (10.38) 0.77 (2.05) 1.93 (2.98) 28.07 (16.31)
24 G.F. Principe et al. / Journal of Experimental Child Psychology 163 (2017) 15–31

Table 4
Proportions (and counts) of children who reported the target activity as actually occurring and reported seeing the target activity
during the interview.

Reported the target activity Reported seeing the

target activity
Open-ended Specific Leading Total
Low structure/autonomy supportive .04 (1) .08 (2) .27 (7) .39 (10) .04 (1)
Low structure/high control .12 (3) 0 .16 (4) .28 (7) .04 (1)
High structure/autonomy supportive .25 (7) .14 (4) .32 (9) .71 (20) .21 (6)
High structure/high control .38 (11) .17 (5) .38 (11) .93 (27) .41 (12)

of questioning that elicited the reports: open-ended, specific, leading, and total (open-ended + specific
+ leading). A series of logistic regressions conducted to predict children’s total recall revealed that chil-
dren with high-structure/controlling mothers were more likely to report the target activity than chil-
dren with high-structure/autonomy-supportive mothers, who in turn were more likely to report the
target activity than children with mothers in either of the low-structure groups, v2s(1, N = 108) 
4.00, ps < .05, Øs  20. As also shown in Table 4, a substantial number of children reported the target
activity at the open-ended level of questioning. Children’s open-ended reports differed as a function of
maternal style, such that children with high-structure/controlling mothers were more likely to spon-
taneously report the target activity than children with mothers in either of the low-structure groups,
v2s(1, N = 108)  4.23, ps < .05, Øs  20.
Furthermore, mothers’ use of multiple types of biased techniques during the mother–child conver-
sation was associated with children’s recall of the target activity during the interview. Specifically,
only 9% of children whose mothers used only one or no biased techniques reported the loose rabbit
at the open-ended level of questioning, whereas 32% of children whose mothers used two or more
biased techniques made an open-ended report of the target activity, v2(1, N = 108) = 7.39, p < .01,
Ø = 26. This pattern persisted in children’s total recall, such that 37% of children whose mothers used
only one or no biased techniques reported target activity, whereas 82% of children whose mothers
used two or more biased techniques recalled a loose rabbit, v2(1, N = 108) = 20.03, p < .0001, Ø = 43.

Seeing the target activity

The proportions and counts of children who reported actually seeing (as opposed to merely hearing
about) the target activity when posed the source question are shown in the rightmost column of
Table 4. Children with high-structure/controlling mothers were more likely to report seeing the rabbit
than children with low-structure/autonomy-supportive mothers and those with low-structure/
controlling mothers, v2s(1, N = 108)  6.76, ps < .01, Øs  25. The difference between children with
high-structure/controlling mothers and those with high-structure/autonomy-supportive mothers
approached significance (p = .11). Mothers’ use of multiple types of biased techniques was also asso-
ciated with reports of seeing the target activity, such that 33% of children whose mothers made two or
more biased utterances recalled seeing a loose rabbit, whereas only 4% of children whose mothers
used one or no biased techniques made reports of seeing the target activity, v2(1, N = 108) = 10.92,
p < .001, Ø = 32.

Describing the target activity

Next, we explored children’s narrative accounts of the target activity. Narratives were coded only
for those children who reported a loose rabbit in response to an open-ended, specific, or leading ques-
tion. These data are displayed in Table 5, accompanied by the numbers on which each mean is based.
Consistent with Principe et al.’s (2013) study, many of these children relayed quite detailed descrip-
tions of this nonoccurring event. To illustrate, consider the following excerpts: ‘‘We didn’t catch the
rabbit . . . he was hopping around so fast and fast,” ‘‘I petted him. He was soft and had big giant ears,”
‘‘The bunny was scared because he was lost . . . I said ‘Here bunny bunny, come here, it’s okay.”
Analyses of the total number of propositions describing the target activity yielded an effect of
maternal style, F(3, 60) = 9.88, p < . 0001, g2p = .30. Of those children who reported the target activity,
 G.F. Principe et al. / Journal of Experimental Child Psychology 163 (2017) 15–31 25

Table 5
Mean numbers (and standard deviations) of verbatim, constructive, and fantastic propositions reported about the target activity.

n Verbatim Constructive Fantastic Total

Low structure/autonomy supportive 10 0.30 (0.48) 1.50 (1.96) 0.30 (0.67) 2.10 (2.02)
Low structure/high control 7 0.43 (0.53) 1.00 (1.54) 0.43 (1.14) 1.86 (1.46)
High structure/autonomy supportive 20 0.40 (0.82) 6.30 (3.80) 0.40 (0.82) 7.10 (5.78)
High structure/high control 27 0.63 (0.49) 9.74 (7.10) 1.59 (2.83) 11.96 (7.45)

those with high-structure/controlling mothers provided more voluminous narratives about the nonoc-
curring loose rabbit than children with high-structure/autonomy-supportive mothers. Children with
mothers in the low-structure/autonomy-supportive and low-structure/controlling groups provided
the least amount of narrative detail describing a loose rabbit. To explore the content of children’s nar-
ratives describing the target activity in more depth, separate analyses were conducted on the mean
numbers of verbatim, constructive, and fantastic propositions reported. As can be seen in Table 5,
the majority of children’s descriptions of the target activity were made up of constructive detail con-
sistent with the notion of a loose rabbit. There was an effect of maternal style on constructive propo-
sitions, F(3, 60) = 6.30, p < . 001, g2p = .24, such that children with high-structure/controlling mothers
provided more constructive details about the target activity than children with high-structure/auton
omy-supportive mothers. Children with mothers in the low-structure/autonomy-supportive and low-
structure/controlling groups produced the least constructive propositions. There were no group differ-
ences in the provision of verbatim or fantastic utterances. Maternal bias was also associated with chil-
dren’s provision of details describing the target activity. Of those children who reported the target
activity, those whose mothers used two or more biased techniques (n = 44, M = 9.57, SD = 7.69) fabri-
cated a higher level of propositions describing the suggested loose rabbit than those whose mothers
used one or no types of biased utterance (n = 44, M = 3.90, SD = 3.16).
Another way of exploring the impact of maternal style on children’s narrative descriptions of the
target activity is to examine group differences in the degree of overlap between the content of
mother–child conversation and children’s accounts during the interview. As can been seen in Table 6,
among the subset of children who reported the target activity, there was a significant degree of over-
lap between content shared during the mother–child dialogue and children’s subsequent interview
reports. There was an effect of maternal style on the proportion of overlapping mother propositions,
F(3, 60) = 11.35, p < .0001, g2p = .36, such that children with high-structure/controlling mothers uttered
proportionally more overlapping mother propositions in their accounts of the target activity during
the interview than children with high-structure/autonomy-supportive mothers. Children in both
low-structure groups provided the least overlapping mother propositions. There was also an effect
of overlapping self propositions, F(3, 60) = 6.98, p < . 001, g2p = .27, such that children with high-struc
ture/autonomy-supportive mothers uttered more overlapping self propositions than children with
mothers in the other three groups. Finally, there was an effect of nonoverlapping propositions, F
(3, 60) = 5.21, p < .01, g2p = .21, such that children with mothers in the two low-structure groups
reported proportionately more nonoverlapping propositions than children with mothers in the
high-structure groups. Interestingly, there were also direct links between maternal bias during the
mother–child conversation and the degree of overlap in content between this conversation and the
interview. Specifically, the number of biased utterances made by mothers was positively correlated

Table 6
Proportions of propositions (and standard deviations) reported during the interview that overlapped with utterances shared during
the mother–child conversation.

Overlapping mother Overlapping self Nonoverlapping Ambiguous

Low structure/autonomy supportive .05 (.11) .08 (.18) .79 (.42) .08 (.18)
Low structure/high control .04 (.09) .12 (.14) .74 (.38) .10 (.08)
High structure/autonomy supportive .21 (.24) .38 (.28) .36 (.25) .05 (.06)
High structure/high control .41 (.23) .18 (.13) .39 (.24) .02 (.03)
26 G.F. Principe et al. / Journal of Experimental Child Psychology 163 (2017) 15–31

with the proportion of mother overlapping propositions (r = .53, p < .0001) and negatively correlated
with the number of nonoverlapping propositions (r = .36, p < .01), demonstrating that higher levels
of biased utterances were associated with a greater degree of later repetition of mothers’ suggestions.
Furthermore, the use of multiple biased techniques was associated with the highest level of overlap
between utterances during the mother–child conversation and the subsequent interview. Children
with mothers who used two or more biased techniques uttered proportionately more mother overlap-
ping propositions during the interview (n = 44, M = .34, SD = .26) compared with children with moth-
ers who used only one or no biased techniques (n = 20, M = .05, SD = .10), t(62) = 4.80, p < .0001.
Furthermore, children with mothers who used two or more biased techniques were not particularly
inventive in their false narratives describing the target activity. These children produced proportion-
ately fewer nonoverlapping propositions (n = 44, M = .39, SD = .26) compared with children with
mothers who used only one or no biased techniques (n = 20, M = .65, SD = .38), t(62) = 3.21, p < .01.

Reporting the magic show

Considering these differences in children’s reports of the suggested loose rabbit as a function of
maternal style, we next explored whether mothers’ provision of elaborative structure and autonomy
support was associated with the same sorts of facilitative effects evidenced in other studies of
mother–child memory sharing (e.g., Fivush, Haden, & Reese, 2006; Ornstein, Haden, & Hendrick,
2004). Table 7 displays children’s provisions of propositions for all children during the interview. As
can be seen, there was an effect of maternal style on the total number of propositions reported about
the magic show, F(3, 107) = 11.25, p < .0001, g2p = .25. Children whose mothers were classified as high
structure/controlling and high structure/autonomy supportive provided a higher number of positions
compared with children whose mothers were in the low-structure/autonomy-supportive and low-
structure/controlling groups. There also were group differences in terms of the numbers of accurate
propositions uttered, F(3, 107) = 4.59, p < .01, g2p = .12. Children whose mothers were in the high-struc
ture/autonomy-supportive group produced more accurate propositions than children with mothers in
the high-structure/controlling group, who in turn uttered more accurate propositions than children in
either of the low-structure groups. Likewise, there were group differences in inaccurate propositions, F
(3, 107) = 21.26, p < .0001, g2p = .36, such that children with high-structure/controlling mothers uttered
more inaccurate propositions than children with mothers in the high-structure/autonomy-supportive
group, who in turn generated more inaccurate propositions than children with mothers in the low-
structure groups.

Maternal belief

We used the maternal questionnaire to explore whether degree of belief in the loose rabbit sugges-
tion was associated with differences in mothers’ style or children’s memory performance. It seemed
possible that mothers with a higher level of belief in the errant suggestion might have taken on a more
biased style to try to persuade their children to produce a narrative in line with their beliefs. In line
with Principe et al. (2013), there was a good deal of variation in mothers’ beliefs (M = 3.00,
SD = 1.56), with some rating their belief at the highest level and others noting no belief. Belief level
was not associated with maternal structure; however, it was associated with maternal control
(r = .23, p < .05), indicating that mothers with higher levels of belief in the loose rabbit were also more
controlling during the mother–child conversation. There also was a correlation between mothers’
belief in the loose rabbit and the number of biased utterances used by mothers during the mother–
child conversation (r = .21, p < .05), indicating a link between false belief and the tendency to ask ques-

Table 7
Mean numbers of propositions (and standard deviations) provided about the magic show during the interview.

Accurate Inaccurate Ambiguous Total

Low structure/autonomy supportive 5.61 (5.95) 1.15 (1.85) 0.58 (0.99) 7.35 (6.05)
Low structure/high control 6.72 (4.95) 1.68 (3.64) 1.20 (1.59) 9.60 (7.22)
High structure/autonomy supportive 12.99 (9.67) 5.96 (5.55) 1.29 (2.23) 20.14 (15.08)
High structure/high control 9.24 (5.65) 11.24 (7.83) 1.14 (1.41) 21.62 (13.14)
 G.F. Principe et al. / Journal of Experimental Child Psychology 163 (2017) 15–31 27

tions and make statements in line with that false belief. Finally, mothers who used two or more biased
techniques during the mother–child conversation (n = 54, M = 3.35, SD = 1.47) reported higher levels
of belief in the loose rabbit than mothers who used only one or no biased techniques (n = 54,
M = 2.72, SD = 1.43), t(106) = 2.25, p < .05.

Discussion

Earlier work (Principe et al., 2013) has shown that when mothers encounter misinformation about
an event experienced by their children, mother–child memory sharing conversations can lead to errors
in children’s subsequent recollections, with children of mothers who display a high-elaborative style
more likely than children of low-elaborative mothers to make false reports consistent with the mis-
information and to generously embellish their accounts with nonoccurring details. Given that a
high-elaborative style generally is composed of several elements (e.g., Haden et al., 1997), the aim
of the current study was to examine which features of this maternal style put children most at risk
for remembering errors when their mothers have been misinformed. Specifically, this investigation
contrasted two features that, on the basis of the literature on children’s suggestibility (for a review,
see Bruck et al., 2002), were hypothesized to be linked to false reports, namely mothers’ provision
of structure and the degree to which mothers control the remembering agenda. To examine these
issues, children experienced a scripted event about which mothers received misleading information
suggesting that their children may have experienced an activity that never occurred. Later, mothers
were asked to discuss the event with their children. Finally, children were interviewed about the event
by an unfamiliar examiner.
As expected (Cleveland & Reese, 2005), mothers’ provision of structure (defined as elaborative ques-
tions and statements) and control (defined in terms of functional control of conversational turns) dur-
ing the post-event conversation emerged as separate dimensions of style, with relatively equivalent
numbers of mothers falling into four post hoc groups defined by the combination of high- versus
low-elaborative structure and high versus low control (i.e., controlling vs. autonomy supportive). These
differences in maternal style were met with variation in children’s behavior during the mother–child
dialogue, with children whose mothers were high in structure and control appearing to be the most
suggestible. For example, children with mothers characterized as high structure and controlling were
more likely than children in the other three groups to acquiesce to their mothers’ very first utterance
about the misinformation. They also acquiesced to proportionately more of their mothers’ questions
and statements about the misleading information than children with mothers in the other three
groups. Furthermore, these children fabricated more original detail describing activities consistent
with the misinformation than children with mothers in the other groups, indicating that the influence
of high structure and high control went beyond simple deference to maternal suggestions during the
mother–child conversation and included the generation of original but entirely false narrative detail.
Finally, these children evidenced the highest levels of overlap between things their mothers uttered
in line with the maternal suggestion during the post-event conversation and things children them-
selves said during the interview, demonstrating that a high-structure and controlling maternal style
was not merely associated with a subsequent elevation in children’s reports of details consistent with
the misinformation but also linked with children’s increased proneness to embed the very details sug-
gested by their mothers into their own recollections during the subsequent interview.
This pattern of greater memory error among children with high-structure and controlling mothers
evidenced in the post-event conversation persisted during the subsequent interview. Children with
mothers high in structure and control were more likely than those with mothers in the other three
groups to wrongly report experiencing the nonoccurring activity suggested by the maternal misinfor-
mation. These children also made more reports of the suggested-but-nonexperienced activity at the
open-ended level of questioning and were more likely to claim to have actually seen this nonevent
with their own eyes, as opposed to merely hearing about it, than children with mothers in the two
low-structure groups. Finally, children with high-structure and controlling mothers also provided
the highest levels of narrative detail describing the suggested activity, with the bulk of these accounts
made up of believable constructions rather than fantastic claims.
28 G.F. Principe et al. / Journal of Experimental Child Psychology 163 (2017) 15–31

A suggestibility perspective

Analysis of the mother–child dialogues from a suggestibility framework offers some insight into
why the combination of high structure and control was linked with the highest levels of memory error
among children during both the post-event conversation with mothers and the later interview. Bor-
rowing from the vernacular of the literature on interviewer bias, high-structure and controlling moth-
ers made substantially more statements and questions that would be characterized as conveying a
bias in the veracity of the misinformation compared with mothers in the other three groups. More
important, mothers high in structure and control were more likely than the other mothers to combine
multiple forms of biased utterances when talking about the to-be-remembered event with their chil-
dren, such as introducing new details consistent with the maternal misinformation, asking children to
imagine or suppose occurrences consistent with this suggestion, and praising responses in line with its
theme, as opposed to limiting their utterances about the misinformation to a singular form. This latter
trend is especially noteworthy because experiments in the suggestibility literature have shown that
children are much more prone to make false reports when adults use multiple types of utterances that
convey a bias in the occurrence of nonevents than when adults limit their errant suggestions to a sin-
gular form (e.g., Bruck et al., 2002; Garven et al., 2000; Leichtman & Ceci, 1995).
Demonstrating a direct link between mothers’ provision of multiple biased techniques and chil-
dren’s false reports is the finding that, during the mother–child discussion, children whose mothers
used two or more biased techniques uttered more than five times the number of memory responses
consistent with the misinformation compared with children whose mothers limited themselves to
only one biased technique. This pattern persisted in children’s independent recall during the inter-
view. For instance, children of mothers who used multiple biased techniques were more than twice
as likely as those whose mothers used only one form of biased utterance to report the target activity
as actually occurring and more than eight times as likely to recall seeing the suggested activity as
opposed to merely hearing about it. Furthermore, compared with children whose mothers used only
one type of biased utterance, children whose mothers used multiple biased techniques produced
much more elaborate accounts describing the suggested-but-nonoccurring activity and provided pro-
portionately more narrative details that overlapped with the very things suggested by their mothers
during the earlier mother–child dialogue. This last finding underscores the power of multiple forms of
bias because it demonstrates that mothers’ provision of two or more types of suggestive utterances
not only puts children at increased risk for generating especially elaborate false narratives but also
boosts the extent to which children’s interview reports can become saturated with the very false
details suggested earlier by their mothers.

Memory for experienced details

Children with high-structure and autonomy-supportive mothers recalled the highest levels of
accurate details about the staged event during both the mother–child conversation and the later
interview. This elevated memory for actual experiences among the children with high-structure/au
tonomy-supportive mothers is consistent with prior work demonstrating that these qualities of
mothers’ approach to memory sharing promote children’s abilities to provide detailed and coherent
memories of experience (e.g., Farrant & Reese, 2000; Reese & Newcombe, 2007). Despite this
enhanced memory for experienced details, children with high-structure and autonomy-supportive
mothers, although not as error prone as those with high-structure and controlling mothers, made
more false reports and generated more detail consistent with the misinformation than children with
low-structure mothers. This pattern demonstrates that higher levels of accurate recall of experienced
details do not necessarily translate into lower levels of false reports of nonexperienced activities, at
least not when children have been exposed to erroneous maternal suggestions. Indeed, the frequent
false reports among children with high-structure/autonomy-supportive mothers can be explained by
the considerable bias displayed by their mothers. During the post-event conversation, high-structure
and autonomy-supportive mothers uttered more information in line with the misinformation and
were more likely to use two or more types of biased techniques than mothers in both low-
structure groups.
 G.F. Principe et al. / Journal of Experimental Child Psychology 163 (2017) 15–31 29

A question of interest for future research is why maternal autonomy support was linked with
higher memory for experienced details among children with high-structure mothers given that auton-
omy support generally is associated with children’s interest and engagement in the memory sharing
process and not the completeness or detail of their productions (Cleveland & Morris, 2014; Cleveland
& Reese, 2005). However, earlier work has not explored the influence of autonomy support when
mothers have been misinformed. It may be that mothers’ support of autonomy in remembering plays
some role in buffering children’s memories from the suggestions of others. It seems reasonable to
think that autonomy-supportive mothers’ routine encouragement and appreciation for children’s
own ideas about the past might foster children’s understanding that they have the authority to relay
their own versions of what happened during conversational remembering rather than to follow others’
leads. As such, these children may more readily generate their own representations of personal expe-
riences that are more resistant to intrusions from the suggestions of others.
Consistent with this idea that an autonomy-supportive orientation may foster the development of
qualities that protect memory for experienced details against interference from external suggestion is
the finding that children with mothers in both autonomy-supportive groups more often refrained
from accepting their mothers’ initial suggestions and rejected these suggestions proportionately more
often than those with mothers in the two controlling groups. Furthermore, in their false reports of the
suggested activity during the interview, children with high-structure and autonomy-supportive moth-
ers generated more overlapping self propositions than those with controlling mothers. Thus, even in
children’s reports of nonexperienced events, an autonomy-supportive orientation was associated
more with children’s reliance on their own ideas—even though these details were fabricated during
an earlier conversation—than with a tendency to parrot their mothers’ exact suggestions. Because chil-
dren in this study were exposed to varying levels of bias, future work should equate levels of bias
among children with mothers who commonly use differing degrees of control during memory sharing
to examine whether autonomy support over the long term might bolster children’s tendencies to
resist the false suggestions of others and persist with their own ideas about experience.

Maternal belief in misinformation

The literature on suggestibility demonstrates than when interviewers develop false beliefs about
children’s past, they become prone to shape memory sharing conversations with children to elicit
accounts in line with their beliefs. Mothers’ conversational behaviors in the current study followed
this trend and differed as a function of their belief in the veracity of the maternal suggestion. Replicat-
ing Principe et al. (2013) results, mothers’ degree of belief in the misinformation was not associated
with their provision of structure during the mother–child conversation. This lack of a relationship
between mothers’ belief and the provision of structure was expected given findings suggesting that
this feature of maternal memory sharing behavior is traitlike and not found to vary across tasks or
contexts (see, e.g., Haden, 1998). However, higher levels of belief in the maternal suggestion were
linked to increases in mothers’ conversational control. This relation raises the possibility that mothers
who more strongly believed in the occurrence of the suggested activity chose to exert higher levels of
control in an attempt to elicit reports from their children consistent with their beliefs. Higher levels of
maternal belief also were associated with higher levels of maternal bias, providing direct evidence of a
link between mothers’ degree of belief in the misinformation and their likelihood to mold the memory
sharing conversation to prompt accounts in line with this false information.

Implications and conclusions

The current findings replicate and extend previous research (Principe et al., 2013) by demonstrat-
ing that the very same maternal misinformation can have differing impacts on children’s reports of a
personal experience depending on mothers’ memory sharing style. In particular, this work demon-
strates that when mothers have been misinformed, the combined tendencies of mothers to provide
high structure in the form of elaborative questions and statements and to control the direction of
memory sharing conversations put children at an especially high risk for remembering errors. In fact,
the conversational contributions of a high-structure and controlling approach in the current study
30 G.F. Principe et al. / Journal of Experimental Child Psychology 163 (2017) 15–31

mirrored those typical of a biased interviewing style that is known to produce high levels of false
reports (see Bruck et al., 2002). Furthermore, higher levels of maternal belief in the misinformation
were associated with higher levels of maternal bias, suggesting that false belief may motivate mothers
to shape memory sharing conversations to elicit reports consistent with their beliefs.
Despite this collective evidence of the impact of a high-structure and controlling memory sharing
style when mothers have been misinformed, to establish whether these features have causal effects on
children’s memory, they would need to be experimentally manipulated or examined longitudinally. It
would also be of interest in follow-up work to examine these features outside of a misinformation
paradigm to understand whether mothers may alter their routine memory sharing style if they have
certain knowledge or beliefs about their children’s experiences. The associations between maternal
belief and a controlling orientation suggest that, unlike structure, autonomy support might not be a
stable aspect of a high-elaborative style. Finally, the links in the current investigation between moth-
ers’ high-structure/autonomy-supportive style and children’s accurate recall suggest a potential
mechanism for the development of resistance to the suggestions of others.
The associations in the current study between maternal bias and subsequent child memory errors
notwithstanding, the above analysis is not meant to suggest that all maternal conversational behav-
iors typical of biased interviewers have negative mnemonic consequences. In fact, some of the behav-
iors associated with a biased style might serve to support the development of children’s
autobiographical remembering skills. For instance, the tendency of biased mothers to introduce
new information, to encourage speculation, to ask forced-choice questions, and to suggest a particular
response might serve to effectively scaffold children’s autobiographical remembering skills when chil-
dren are having trouble in recalling details, are unsure how to structure their recounts, or are hesitant
to engage in conversational remembering. Furthermore, even if mothers have been misinformed about
their children’s experiences, such as by a babysitter who wrongly blames the gum on the couch on the
child or by a teacher who mistakenly attributes name calling to the child, the sorts of conversational
disagreements that are likely to ensue might serve as a means for children to learn to trust their own
memories or to develop skills to negotiate disagreements about the past. However, in contexts where
exact accuracy is the goal, such as in legal proceedings, these findings reveal that maternal beliefs and
bias can seriously hinder children’s abilities to accurately recall a specific experience.

Acknowledgments

Special appreciation is expressed to all of the children, parents, and teachers who made this study
possible. Portions of this research were presented at the biennial meeting of the Society for Research in
Child Development, April 2013, Seattle, WA.

References

Bruck, M., Ceci, S. J., & Hembrooke, H. (2002). The nature of true and false narratives. Developmental Review, 22, 520–554.
Ceci, S. J., & Bruck, M. (1993). The suggestibility of children’s recollections: An historical review and synthesis. Psychological
Bulletin, 113, 403–439.
Ceci, S. J., & Bruck, M. (1995). Jeopardy in the courtroom. Washington, DC: American Psychological Association.
Cleveland, E. S., & Morris, A. (2014). Autonomy support and structure enhance children’s memory and motivation to reminisce:
A parental training study. Journal of Cognition and Development, 15, 414–436.
Cleveland, E. S., & Reese, E. (2005). Maternal structure and autonomy support in conversations about the past: Contributions to
children’s autobiographical memory. Developmental Psychology, 41, 376–388.
Cleveland, E. S., Reese, E., & Grolnick, W. S. (2007). Children’s engagement and competence in personal recollection: Effects of
parents’ reminiscing goals. Journal of Experimental Child Psychology, 96, 131–149.
Farrant, K., & Reese, E. (2000). Maternal style and children’s participation in reminiscing: Stepping stones in children’s
autobiographical memory development. Journal of Cognition and Development, 1, 193–225.
Fivush, R., & Fromhoff, F. A. (1988). Style and structure in mother–child conversations about the past. Discourse Processes, 11,
337–355.
Fivush, R., Haden, C. A., & Reese, E. (2006). Elaborating on elaborations: The role of maternal reminiscing style in cognitive and
socioemotional development. Child Development, 77, 1568–1588.
Garven, S., Wood, J. M., & Malpass, R. (2000). Allegations of wrongdoing: The effects of reinforcement on children’s mundane
and fantastic claims. Journal of Applied Psychology, 85, 38–49.
Garven, S., Wood, J. M., Malpass, R., & Shaw, J. S. (1998). More than suggestion: Consequences of the interviewing techniques
from the McMartin Preschool case. Journal of Applied Psychology, 83, 347–359.
 G.F. Principe et al. / Journal of Experimental Child Psychology 163 (2017) 15–31 31

Gilstrap, L. L., & Ceci, S. J. (2005). Reconceptualizing children’s suggestibility: Bidirectional and temporal properties. Child
Development, 76, 40–53.
Haden, C. A. (1998). Reminiscing with different children: Relating maternal stylistic consistency and sibling similarity in talk
about the past. Developmental Psychology, 34, 99–114.
Haden, C. A., Haine, R. A., & Fivush, R. (1997). Developing narrative structure in parent–child reminiscing across the preschool
years. Developmental Psychology, 33, 295–307.
Hendrick, A. M., Haden, C. A., & Ornstein, P. A. (2009). Elaborative talk during and after an event: Conversational style influences
children’s memory reports. Journal of Cognition and Development, 10, 188–209.
Kulkofsky, S. (2011). Characteristics of functional joint reminiscence in early childhood. Memory, 19, 45–55.
Larkina, M., & Bauer, P. (2010). The role of maternal verbal, affective, and behavioral support in preschool children’s
independent and collaborative autobiographical memory reports. Cognitive Development, 25, 309–324.
Leichtman, M. D., & Ceci, S. J. (1995). The effects of stereotypes and suggestions on preschoolers’ reports. Developmental
Psychology, 31, 568–578.
Leichtman, M. D., Pillemer, D. B., Wang, Q., Koreishi, A., & Han, J. J. (2000). When baby Maisy came to school: Mothers’ interview
styles and preschoolers’ event memories. Cognitive Development, 15, 99–114.
Nelson, K., & Fivush, R. (2004). The emergence of autobiographical memory: A social cultural developmental theory.
Psychological Review, 111, 486–511.
Ornstein, P. A., Haden, C. A., & Hedrick, A. M. (2004). Learning to remember: Social-communicative exchanges and the
development of children’s memory skills. Developmental Review, 24, 374–395.
Principe, G. F., DiPuppo, J., & Gammel, J. (2013). Effects of mothers’ conversation style and receipt of misinformation on
children’s event reports. Cognitive Development, 28, 260–271.
Principe, G. F., Greenhoot, A. F., & Ceci, S. (2014). Young children’s eyewitness memory. In T. Perfect & D. S. Lindsay (Eds.), Sage
handbook of applied memory (pp. 633–653). Thousand Oaks, CA: Sage.
Principe, G. F., & Schindewolf, E. (2012). Natural conversations as a source of false memories in children: Implications for the
testimony of young witnesses. Developmental Review, 32, 205–223.
Reese, E., & Brown, N. (2000). Reminiscing and recounting in the preschool years. Applied Cognitive Psychology, 14, 1–17.
Reese, E., & Newcombe, R. (2007). Training mothers in elaborative reminiscing enhances children’s autobiographical memory
and narrative. Child Development, 78, 1153–1170.
Thompson, W. C., Clarke-Stewart, K. A., & Lepore, S. J. (1997). What did the janitor do? Suggestive interviewing and the accuracy
of children’s accounts. Law and Human Behavior, 21, 405–426.
White, T. L., Leichtman, M. D., & Ceci, S. J. (1997). The good, the bad, and the ugly: Accuracy, inaccuracy, and elaboration in
preschoolers’ reports about a past event. Applied Cognitive Psychology, 11, 37–54.