Contemporary Educational Psychology 53 (2018) 123–134

Contents lists available at ScienceDirect

Contemporary Educational Psychology

journal homepage: www.elsevier.com/locate/cedpsych

Predictive inference generation and story comprehension among children T

with ADHD: Is making predictions helpful?
⁎
Angela Hayden , Elizabeth P. Lorch, Richard Milich, Cristina Cosoreanu, Jessica Van Neste
University of Kentucky, 125 Kastle Hall, Lexington, KY 40506 USA

A R T I C LE I N FO A B S T R A C T

Keywords: Children with ADHD exhibit narrative comprehension difficulties relative to typically developing peers. One
ADHD unexplored comprehension area for this population is the generation of explanatory predictive inferences.
Comprehension Plausible explanatory predictive inferences allow for smooth integration of new information, and are targeted in
Predictive inferences many comprehension interventions. The current study examined three questions: (1) Is there a difference be-
Predictions
tween children with ADHD and typically developing children in the creation of explanatory predictive inferences
during a predictive prompt task? (2) Does the group difference in the generation of these inferences mediate
group differences in recall measures? (3) Does the act of making predictions during story presentation improve
recall relative to a control condition? Seven- to 11-year-old children with and without ADHD listened to two
audio-taped stories, and made predictions at several pauses during one of the stories (predictive prompt task).
The recording of the other story was paused without asking for predictions in the control task. Children recalled
the story following the task. During the predictive prompt task, children with ADHD generated fewer plausible
explanatory predictive inferences than typically developing peers and this group difference mediated a group
difference in the recall of highly important events. Further, children in both diagnostic groups included more
plausible explanatory backwards inferences during recall following the predictive prompt task than following
the control task. Given the importance of explanatory inferences in constructing coherent story representations,
these findings are encouraging for educators using predictive tasks but indicate that children with ADHD may
need further instruction on generating appropriate predictive inferences.

1. Introduction critical part of building an accurate representation of a narrative is the

creation of reasonable predictive inferences that serve to connect events
Attention Deficit Hyperactivity Disorder (ADHD) is a developmental in a story (Klin, Guzmán, & Levine, 1999). Little is known about the
disorder affecting between 5% and 10% of school-aged children in the ability of children with ADHD to create and use predictive inferences
United States today (Barkley, 2015). ADHD is characterized by high when comprehending and recalling a narrative, or the relation of pre-
levels of inattention, hyperactivity, and impulsivity. These symptoms dictive inference generation to these children’s narrative comprehen-
contribute to significant academic problems; many children with ADHD sion.
fail grades, perform more poorly on standardized tests than peers, and Narrative comprehension is an important area of study because the
are held back (Roberts, Milich, & Barkley, 2015). Much of the research ability to comprehend is a significant contributor to academic success
investigating the academic problems of children with ADHD focuses on (Kendeou et al., 2005). Early pre-reading narrative comprehension
these children’s problems with staying on task and avoiding distraction skills predict later reading comprehension skills above and beyond
(Barkley, 2015). However, a body of work is emerging that examines fluency and vocabulary skills (Kendeou, Bohn-Gettler, White, & van den
these children’s issues with higher-order cognitive skills such as pro- Broek, 2008; Kendeou et al., 2005). In comparison to typically devel-
blem solving, knowledge acquisition, and strategy use. Investigating oping peers, young children with ADHD exhibit significant difficulties
how these children build and comprehend complex narratives has re- in narrative comprehension skills. These difficulties are maintained or
vealed several higher-order cognitive skill difficulties (e.g., Flake, even increase over time (Bailey, Lorch, Milich, & Charnigo, 2009;
Lorch, & Milich, 2007; Lorch, Eastham et al., 2004; Lorch, O’Neil et al., Lorch, Milich, Flake, Ohlendorf, & Little, 2010). These problems are not
2004; Renz et al., 2003; Van Neste, Hayden, Lorch, & Milich, 2015). A ameliorated with medication (Bailey, Derefinko, Milich, Lorch, &

⁎
Corresponding author.
E-mail addresses: anhayd1@uky.edu (A. Hayden), elorch@email.uky.edu (E.P. Lorch), milich@uky.edu (R. Milich), cristina.cosoreanu@uky.edu (C. Cosoreanu),
jessica.s.vanneste@uky.edu (J. Van Neste).

https://doi.org/10.1016/j.cedpsych.2018.02.003

Available online 21 February 2018

0361-476X/ © 2018 Elsevier Inc. All rights reserved.
A. Hayden et al. Contemporary Educational Psychology 53 (2018) 123–134

Metze, 2011; Derefinko, Bailey, Milich, Lorch, & Riley, 2009). Lorch are created by making connections or conclusions about previous
and colleagues have found that children with ADHD have difficulty events as comprehenders incorporate new information into their story
with four important story comprehension skills: (1) sensitivity to causal representations. Explanatory predictive inferences (similar to the causal
relations among story events (e.g., Lorch, Eastham et al., 2004; Lorch, forward inferences described by Klin et al., 1999) are created when
O’Neil et al., 2004); (2) using goal structure to build complex story comprehenders make causal, often goal-based predictions based on the
representations (e.g., Flory et al., 2006); (3) recognizing important ongoing story representation they are constructing. Specifically, com-
story information and using it to guide recall (e.g., Lorch et al., 2010); prehenders may use narrative events already described, background
(4) creating coherent stories (e.g., Freer, Hayden, Lorch, & Milich, knowledge, and knowledge of story structure to predict causal path-
2011). ways to future events (Klin et al., 1999). The likelihood and accuracy of
The four areas of difficulty experienced by children with ADHD that such predictions increases with the predictability of the narrative and
are listed above are interrelated and require higher order cognitive the motivations of the reader (Clinton, 2015; Murray, Klin, & Myers,
processing in order to create and maintain a coherent understanding of 1993). Predictions that are causally related to previously experienced
a story. Integral to creating a coherent story is the ability to identify story information allow comprehenders to anticipate future events on
cause and effect relations among the events in the story (Trabasso & van the causal chain and integrate these events more readily if and when
den Broek, 1985). Events that are highly causally connected to other they occur. If comprehenders make predictions that allow them to in-
events fall on a causal chain that drives the story action from beginning tegrate information more easily, they will build more coherent narra-
to end. Events on this causal chain generally are rated as highly im- tive representations (Klin et al., 1999). In situations in which ex-
portant by adults, and therefore often are the most important events in planatory predictions are likely to be made but are met with
the narrative (Trabasso & Sperry, 1985). In addition, many of these contradictory information (e.g., a vase that is inferred to have been
highly causally connected events are part of the goal structure of the broken when someone throws it at the wall is instead picked up whole),
story (Mandler & Johnson, 1977). Events with multiple causal con- adults significantly slow their reading speed (Klin et al., 1999; van den
nections (i.e., events that are important and oftentimes goal-based) are Broek, Beker, & Oudega, 2015). This suggests that building a coherent
recalled more than other events in a narrative (van den Broek, 1989). narrative representation is disrupted when adults experience conflicting
However, children with ADHD have more trouble distinguishing un- information related to explanatory predictive inferences. Taken to-
important from important events (Lorch, Milich, Astrin, & Berthiaume, gether, these findings suggest that explanatory predictive inferences are
2006) and identifying and recalling events that are on the causal chain important and useful to the coherent construction of a narrative re-
than do typically developing children (Flake et al., 2007; Lorch, Diener presentation.
et al., 1999; Lorch, Sanchez et al., 1999). These issues likely contribute
to group differences in the global coherence of recall (Lorch et al., 2010), 1.1. Inference generation by children with ADHD
which is the logical flow and comprehensibility of a recalled narrative
as a whole (Albrecht & O’Brien, 1993; Cook, Limber, & O’Brien, 2001; Very few studies have explicitly examined inference generation for
Habermas & de Silveira, 2008; Myers, O’Brien, Albrecht, & Mason, children with ADHD. However, often causal connections are not explicit
1994). Group differences in answering questions about causal connec- in stories, requiring inferences to make causal connections among
tions and creating a coherent recall increase across 21 months, and events. As many studies do not distinguish between explicit and implicit
other recall difficulties are maintained across the same period (Bailey causal connections, well-documented diagnostic differences in sensi-
et al., 2009; Lorch et al., 2010). Even when the focus shifts from recall tivity to causal connections may indicate that children with ADHD have
to ongoing comprehension by asking children to narrate a wordless difficulties generating explanatory inferences (Bailey et al., 2009;
picture book, children with ADHD have difficulty maintaining a co- Lorch, Diener et al., 1999; Lorch, Sanchez et al., 1999; Lorch et al.,
herent goal structure (Flory et al., 2006; Renz et al., 2003). 2000; Lorch, Eastham et al., 2004; Lorch, O’Neil et al., 2004). Only
Previous research examining narrative comprehension difficulties three studies have explicitly examined inference generation in children
among children with ADHD generally has focused on comprehending with ADHD (Berthiaume, Lorch, & Milich, 2010; Van Neste et al., 2015;
and recalling explicit story information. However, to build a coherent Yeari, Avramovich, & Schiff, 2017), and these provide only limited
representation of a narrative, higher order cognitive abilities also are information about diagnostic group differences in generating ex-
required to infer important information such as character goals and planatory predictive inferences.
causal connections between events. Inferences are created when a In one study (Van Neste et al., 2015), diagnostic group differences in
comprehender uses explicit story information or background knowl- the generation of explanatory backwards inferences by 7-to-11-year-old
edge to form implicit connections among events or conclusions about boys and girls were examined in a think-aloud procedure (Ericsson &
the story (Trabasso & Magliano, 1996). Inferences can be either ex- Simon, 1993; Fox, Ericsson, & Best, 2011; Laing & Kamhi, 2002; Lynch
planatory or elaborative in nature. Inferences that are both plausible and & van den Broek, 2007). Although the Van Neste et al. study focused on
explanatory are vital to the construction of a coherent representation explanatory backwards rather than explanatory predictive inferences,
(Trabasso & Magliano, 1996). Plausible explanatory inferences, which its design, analyses, and findings have direct relevance to the present
are similar to Thurlow and van den Broek’s (1997) definition of co- study. Children participated in two conditions, a think-aloud condition
herence inferences, can include implicit causal connections among in which they stated their thoughts about a 21-min episode of a family
events, character motivations, and inferences concerning goal structure situation comedy during 5 pre-planned story pauses, and a counter-
(Kendeou et al., 2008; van den Broek, Tzeng, Risden, Trabasso, & balanced control condition in which they watched a different episode
Basche, 2001). These are in contrast to plausible elaborative inferences, with 5 pauses in which they did not think aloud. Children with ADHD
which serve to enrich or embellish the story but are not necessary for stated significantly fewer plausible explanatory backwards inferences
comprehension (Trabasso & Magliano, 1996). Children as young as four than did typically developing peers during both the think-aloud pauses
include inferences in story recall, and with increasing age, causal ex- while viewing the episode and when completing a free recall task fol-
planatory inferences become integral to story comprehension and lowing the video. The group difference in generation of explanatory
subsequent recall (Kendeou et al., 2008). Between the two types of backwards inferences while thinking aloud significantly mediated
inferences, plausible explanatory inferences are significantly more im- group differences in free recall measures, including recall as a function
portant to narrative comprehension (Trabasso & Suh, 1993) and thus of importance and causal chain status, as well as global coherence
are the focus of the current study. ratings. Importantly, these mediation effects were observed for both
Explanatory inferences can be backwards or predictive (i.e., for- recall of the televised story presented in the think-aloud condition and
ward) in nature. Explanatory backwards inferences are inferences that of the televised story presented in the control condition. The fact that

124
A. Hayden et al. Contemporary Educational Psychology 53 (2018) 123–134

mediation effects were not restricted to the story during which in- Strode, Raphael, Corsi, & Au, 2012: Question-Answer Relationships
ferences were generated aloud suggests that children who stated many (QAR) Read-Aloud Program; TeacherVision; Woolley, 2007: The Com-
plausible explanatory inferences may have been relatively skilled at prehension of the Narrative Intervention Program). These include the
making causal links between events and recounting logically connected Reciprocal Teaching Program (Palincsar & Brown, 1984), which has
narratives. Thus, the Van Neste et al. study revealed diagnostic group been incorporated into many other reading programs and reading in-
differences in the generation of explanatory backwards inferences that terventions (e.g., Klingner & Vaughn, 1998: Collaborative Strategic
helped to explain group differences in several recall measures, but did Reading program; Stricklin, 2011: Hands-On Reciprocal Teaching).
not address the question of group differences in predictive inferences. However, these programs are not designed specifically for children with
In another study using a think-aloud procedure that did include ADHD, and instruction on making predictions within these programs
examination of predictive inferences (Berthiaume et al., 2010), 7- to 12- generally does not focus on strategies for making plausible explanatory
year-old boys with and without ADHD told what they were thinking predictive inferences. Thus, if generating plausible explanatory in-
following each sentence of a story that they were read aloud. No sig- ferences is helpful to these children, examining the relation of plausible
nificant diagnostic group differences were observed in generating explanatory predictive inferences to the recall of children with ADHD
plausible explanatory backwards or plausible predictive inferences. It is could provide an initial step in the development of interventions tai-
possible that including only boys may have limited diagnostic differ- lored to this population. Therefore, in addition to examining diagnostic
ences in inference generation (Clinton, Seipel, van den Broek, & Rapp, differences in generation of explanatory predictive inferences, the
2014; Hannon, 2014). Further, for the purpose of this study explanatory second question addressed in the current study was whether making
and elaborative predictive inferences were not distinguished, possibly plausible explanatory predictive inferences mediated potential diag-
masking any diagnostic differences in the generation of explanatory nostic group differences in recall. To these ends, children listened to an
predictive inferences. audiotaped story and were prompted to make predictions during story
In a third study (Yeari et al., 2017), a probe naming task was used to pauses, then recalled the story when presentation was complete. Much
examine diagnostic group differences in adolescents’ (mean like Berthiaume et al. (2010), children in the current study were not
age = 15 years) spontaneous activation of explanatory inferences, asked to read the story themselves to avoid issues with decoding, as
which could be predictive or backwards inferences depending on the comprehension was the focus.
position of probes within “inference” versions of short narrative texts. It is possible that the opportunity to verbalize online predictions
“Control” versions of the texts changed a key sentence to eliminate could improve or disrupt comprehension differentially for children with
likely inferences. Naming facilitation (faster naming times for inference ADHD relative to typically developing peers. Procedures in which
texts than control texts) was greater for adolescents without ADHD than children are asked to verbalize thoughts about a story online (including
for those with ADHD, suggesting that activation of explanatory in- predictions of future events, as indicated above) have generally been
ferences was weaker in adolescents with ADHD. The authors inter- shown to be effective in comprehension interventions for young
preted a lack of significant interaction of this effect with probe position struggling readers (e.g., Bereiter & Bird, 1985; Brown, Pressley, Van
as evidence that weaker activation of inferences applied both to pre- Meter, & Schuder, 1996; Duke & Pearson, 2008; Kucan & Beck, 1997).
dictive and explanatory backwards inferences, but the group difference In the Van Neste et al. (2015) study described earlier, both children
in activation of predictive inferences was not examined separately. with ADHD and typically developing peers recalled more of the most
Taken together, these studies suggest group differences both in the important story information after an open-ended opportunity to think
spontaneous activation of explanatory inferences (Yeari et al., 2017) aloud during story pauses than when no thinking aloud took place
and the generation of backwards explanatory inferences during think- during story pauses. The act of voicing thoughts periodically over the
aloud pauses within presentation of a televised story (Van Neste et al., course of a narrative may allow children to reflect on and consolidate
2015), but do not address group differences in children’s ability to make past narrative events in smaller chunks. However, under certain cir-
predictive inferences when explicitly prompted to do so. Thus, using cumstances comprehension can be disrupted by responding during a
pre-planned story pauses similar to the Van Neste et al. design, the first story: typically developing children who were asked specific questions
question addressed by the current study was whether there a difference during frequent story pauses experienced reduced comprehension (van
between children with ADHD and typically developing peers in the den Broek et al., 2001). Thus, it appears that the experience of re-
number of plausible explanatory inferences created when asked to sponding during pauses in story presentation can either help or hurt
make predictions during story presentation. comprehension, depending on the particular circumstances under
The role of plausible explanatory predictive inferences in the nar- which children complete the task. The third question examined in the
rative comprehension of children with ADHD relative to typically de- current study was whether children’s recall would improve more after
veloping peers remains an unexamined question. This question is im- children made predictions during pauses in story presentation than in
portant because predictive inferences may not play the same role for the the absence of such predictions, and whether such an effect would differ
comprehension of children with ADHD as that of typically developing for children with and without ADHD. If measures of recall of children
peers. Predictive inferences may be particularly challenging for these with ADHD are better after making predictions, this would suggest that
children. Combined with lack of story structure knowledge and limited generating predictions during story presentation could be a valuable
ability to make use of narrative strategies which are generally experi- tool for addressing the comprehension difficulties experienced by these
enced by struggling readers, working memory difficulties among chil- children.
dren with ADHD (Barkley, 2015) may limit the degree to which chil-
dren with ADHD are able to engage in using past story events and 1.2. Summary of current study
causal relations to generate predictions of future causally related events
while experiencing a narrative. Further, these children have difficulty The current study addressed three central questions: (1) Is there a
thinking ahead to the consequences of their actions (Barkley, 2015), group difference between 7-to-11-year-old children with ADHD and
which may apply to predicting the actions of protagonists in narratives. typically developing peers in the number of plausible explanatory
Not only does this question have theoretical importance regarding predictive inferences created during a predictive prompt procedure in
the role of inferences in narrative comprehension among children with which children are specifically directed to generate predictive in-
ADHD, it has practical importance as well. Teaching children to make ferences? Given that children with ADHD exhibit several narrative
plausible predictions while reading is integral to multiple reading comprehension difficulties (e.g., Flake et al., 2007; Lorch, Eastham
strategy programs meant to improve children’s comprehension (e.g., et al., 2004; Lorch, O’Neil et al., 2004; Renz et al., 2003) as well as
Palincsar & Brown, 1984: Reciprocal Teaching; Reading Rockets; working memory problems (Barkley, 2015), it was hypothesized that

125
A. Hayden et al. Contemporary Educational Psychology 53 (2018) 123–134

Table 1 examine the extent to which children can use narrative information to
Relevant demographic information for children with ADHD and comparison peers. create predictive inferences in response to prompts. Further, comparing
this predictive prompt task to a control condition in which participants
Factor ADHD Comparison t
(n = 23) (n = 36) were not asked to make predictions may serve to inform educators on
M (SD) M (SD) whether predictive prompt techniques could help to improve narrative
comprehension in children with ADHD (Bereiter & Bird, 1985; Brown
Child Age/Parent’s Education
et al., 1996).
Age in months 107.87 (14.61) 110.39 (13.95) .51
Mother’s Education (Years) 14.72 (2.23) 15.97 (2.36) 2.03*
Father’s Education (Years) 15.05 (2.61) 16.25 (2.75) 1.53 2. Method
IQ/Expressive Language
KBIT – Total IQ 109.43 (12.99) 120.69 (10.95) 3.58*** 2.1. Participants
KBIT – Matrices IQ 104.87 (18.01) 113.86 (10.98) 2.39*
KBIT – Verbal IQ 111.78 (11.96) 121.78 (13.10) 2.96** Fifty-nine children ranging in age from 7 to 11 years (M
OWLS Expressive Language 90.23 (10.09) 109.06 (18.46) 4.39***
age = 8.71 years) from a mid-sized city in the east south-central region
ADHD Symptoms of the United States participated in the current study. Participants in-
DSM Inattentive Symptoms 7.00 (1.68) .14 (.35) 23.83***
cluded 36 typically developing children (52.78% males), and 23 chil-
DSM Hyperactive 6.83 (1.67) .53 (.88) 18.96***
Symptoms dren with ADHD/Combined type (69.57% males). The tasks described
DSM ODD Symptom 3.87 (2.56) .56 (1.00) 7.00*** in the current study were part of a larger examination of the narrative
Conners ADHD T-Score 73.13 (9.80) 45.28 (5.03) 14.39*** comprehension of children with ADHD; data from separate tasks ad-
dressing different research questions appear in Van Neste et al. (2015).
* p ≤ 0.05.
Reflecting the city’s population distribution, 74.6% of participants were
** p value ≤0.01.
*** p value ≤0.001.
Caucasian, 16.9% were African-American, 1.7% were Asian, 1.7% were
Hispanic/Latino, and 5.1% were Biracial. An additional 6 children (2
these children would generate fewer plausible explanatory predictive typically developing children) were excluded for the following reasons:
inferences than typically developing children. (2) Are there group dif- In the ADHD group, 2 children were excluded due to IQ scores below
ferences in measures of story recall (i.e., differential recall of important 70, 1 child exhibited symptoms of Asperger’s Syndrome and severe
events, inferences included during recall, and recall coherence) and, if speech difficulties, and 1 child exhibited only inattentive symptoms. In
so, does a group difference in the number of plausible explanatory the comparison group of typically developing children, 2 children were
predictive inferences created during the predictive prompt task mediate excluded because they exceeded inattentive symptom criteria (i.e.,
these group differences in story recall?1 Similar to the design and more than three inattentive symptoms on the Children’s Interview for
analytic strategy of Van Neste et al. (2015), these mediation analyses Psychiatric Syndromes [P-ChiPS], described below). See Table 1 for
were conducted for recall of stories experienced both in the predictive relevant demographic and diagnostic information.
prompt and control conditions, in order to determine whether effects Children with ADHD were recruited through a university psychiatric
were specific to processing the story presented in the predictive prompt center, from a list of children who had participated in previous studies,
condition or reflective of group differences in the overall quality of and through public advertisements targeted to recruit children with
narrative comprehension. Based on the mediation results from Van ADHD. These children were required to be formally diagnosed with
Neste et al., it was hypothesized that a group difference in generation of ADHD/Combined type (Milich, Balentine, & Lynam, 2001). Eligibility
plausible explanatory predictive inferences during the predictive for the ADHD group was determined based on a three-step process.
prompt task would mediate group differences in measures of story re- First, when parents expressed interest in having a child with ADHD
call. Because the predictive prompt task may elicit differences in pro- participate, a screening call confirmed diagnostic information and that
cessing from the Van Neste et al. think-aloud task, it is possible that the child met other inclusion criteria (e.g., age) and did not meet ex-
mediation effects will be found for both conditions or will be specific to clusion criteria (e.g., if prescription ADHD medication would not be out
the prompt condition. (3) Finally, does the predictive prompt procedure of the system within 24 h). If the child remained eligible following this
affect recall relative to a control condition in which children paused screening call, the child was scheduled to participate.
during the story but were not asked to make predictions, and if so, is For the second step in confirming the ADHD diagnosis, a trained
there a group difference in this effect? This is an open research ques- graduate student or post-doctoral scholar conducted a semi-structured
tion, as there is evidence of both enhancement (Van Neste et al., 2015) interview with the child’s parent or caregiver. The interview was si-
and disruption (van den Broek et al., 2001) of comprehension for pre- milar to the Children’s Interview for Psychiatric Syndromes – Parent
vious studies that asked children to pause during story processing to Version (P-ChIPS; Weller, Weller, Rooney, & Fristad, 1999), but was
think aloud or respond to questions. limited to verbatim DSM-IV criteria for ADHD and Oppositional Defiant
To address these questions, children listened to and recalled two Disorder (ODD). Parents were asked to endorse the behavioral criterion
short stories in separate sessions. In one session, children were asked to exhibited by their child and were probed to provide behavioral ex-
predict future story events at five pre-planned pause points during story amples for any endorsed criterion. They also were asked if endorsed
presentation. In a separate session with another story, the same chil- behaviors seemed age-inappropriate, and whether these behaviors im-
dren completed a control task in which the story was paused but chil- paired their child’s social or academic functioning. This interview
dren were not asked to predict future events. procedure has been employed successfully in previous studies and has
This predictive prompt procedure therefore allowed for examination achieved inter-rater reliability above 95% for the number of ADHD
of diagnostic differences in generating explanatory predictive in- symptoms endorsed by the parent (Lorch, Diener et al., 1999; Lorch,
ferences as the story progressed. This procedure was not designed to Sanchez et al., 1999). Parents also completed the Conners Parent Rating
reflect children’s naturally occurring generation of predictions, but to Scale-Revised (CPRS-R:S; Conners, 1997). This standardized and reli-
able rating form (Cronbach’s alpha scores range from .69 to .97; test-
retest reliability coefficients range from .56 to .96) is an instrument for
1
assessing and screening behavioral issues across cultures. It provides
The current study focused on plausible explanatory inferences because these in-
ferences are vital to building a coherent narrative representation. Other inference types
information regarding a child’s internalizing and externalizing beha-
(elaborative inferences; implausible explanatory inferences) were coded but not ex- viors, is the most widely used and validated rating scale for ADHD
amined for the current study. (Catale, Geurten, Lejeune, & Meulemans, 2014; Fumeax et al., 2016),

126
A. Hayden et al. Contemporary Educational Psychology 53 (2018) 123–134

and has more research than any other measure (Barkley, 2015). riddles, and matrices. Only the verbal knowledge and matrices sections
In the third step of the diagnostic process, one of the authors (RM, were administered in this study, which allowed an estimate of IQ level.
who has 30+ years of expertise in ADHD) examined all diagnostic data The KBIT-2 has a test re-test reliability of .88 for the composite IQ, and
collected (i.e., available diagnostic history data, symptoms reported on has a .77 correlation to a full-scale IQ score from the WISC-IV (Kaufman
the ADHD interview, Inattentive and Hyperactive Scale T-Scores from & Kaufman, 2004).
the Conners Parent Rating Scale) and confirmed assignment to the Children were given the Expressive Language subtest of the OWLS,
ADHD group. Thus, multiple sources were used as convergent evidence for which they were shown pictures as cues and asked to complete or
of ADHD/Combined type diagnosis. create sentences. Their responses were scored for grammar, semantic
All children with ADHD were required to be off of their medication errors, and pragmatic errors. The test re-test reliability for 8–11 year old
for at least 24 h prior to both appointments in order to study the impact children is .80 for the expressive language scale, and the inter-rater
of ADHD symptoms on inferencing without the influence of medication. reliability for scoring of protocols is .96 for 7-year-old children, and .93
This procedure is a standard way of ensuring that medication does not for 10-year-old children (Carrow-Woolfolk, 1995).
impact the child’s behavior. Parents were reminded of this a day in Typically developing children scored significantly higher on both
advance, and were asked on the days of the study whether their child the KBIT-2 and the OWLS (see Table 1), but both groups were above
had taken medication that day. Participants with ADHD who had taken average in IQ (ADHD: M = 109; Comparison: M = 120).
medication within 24 h of participating were rescheduled.
Typically developing children were recruited through flyers, online 2.3. Procedure
public advertisements, and word-of-mouth. Interested parents of typi-
cally developing children were contacted and screened via a phone call. As part of a separate comprehension study spanning two testing
Children were excluded if parents reported their child as having major sessions (Van Neste et al., 2015), children first completed an unrelated
behavioral or learning disorders. During the first of two sessions, par- comprehension task and the KBIT-2 (first session) or the OWLS (second
ents of typically developing children completed the semi-structured session) prior to beginning the procedure for the current study.
interview that was given to parents of children with ADHD. These In session one, children were randomly assigned to a condition
children were excluded from the study if parents endorsed more than (control or predictive prompt) and to one of the two audio-recorded
three items on the hyperactivity/impulsivity or inattentive scales, re- stories. As this was a within-subjects design, children experienced the
spectively. Thus, typically developing children were not required to be counterbalanced condition and the other audio-recorded story during
symptom-free, but only children with three or fewer symptoms of hy- session two. Children were tested individually. If the child was com-
peractivity/impulsivity, inattention or ODD were included in the study. pleting the predictive prompt condition, the experimenter introduced
The comparison group exhibited significantly fewer symptoms of ADHD the task by saying, “For the next few minutes, you will be listening to
and ODD than the ADHD group based on DSM-IV-TR criteria (see someone tell a story. Every once in a while the story will stop playing.
Table 1), and scored in the normal range on the Conners Parent Rating When the story stops playing, I am going to ask you what you think will
Scale. happen next. After the story is finished, I’ll ask you about what you
The current study protocol was approved by the Institutional remember from the story. Do you understand?” During the control
Review Board. Experimenters acquired both written parental consent condition, the experimenter said, “For the next few minutes, you will be
and oral assent from the child prior to testing. All children were com- listening to someone tell a story. Every once in a while the story will
pensated $40 for participation in both sessions. stop playing. When the story stops playing, we are just going to wait
until the story starts again. After the story is finished, I’ll ask you about
2.2. Materials what you remember from the story. Do you understand?” During both
the predictive prompt and control conditions, the experimenter stopped
2.2.1. Audio-recorded stories the story a total of five times, with each stop lasting approximately 20 s.
Children listened to 2 audiotaped stories, entitled “Silly Richard” These pauses were strategically placed directly before an important
(676 words, 84 story events) and “The Brave Knight” (623 words, 91 causal connection would need to be made (See Appendix A for stories
story events). Each story is approximately 4-1/2 min long and has a with narrative pauses indicated). During the predictive prompt condi-
main character who makes multiple failed attempts at trying to reach a tion, the experimenter stopped the story and asked, “What do you think
goal before achieving it. Silly Richard’s goal is to make enough money is going to happen in this story?” Following the response, the experi-
to buy a skateboard through completing chores, and the Brave Knight’s menter prompted once for additional information, saying “Is there
goal is to protect the kingdom from a threatening Dragon (See full anything else you think is going to happen?” The experimenter waited
transcript of these stories in Appendix A). Prior to the current study, until a full 20 s had elapsed from the beginning of the first prompt, and
importance ratings for each story event were obtained from college then continued the story. In the control condition the experimenter
undergraduate students who independently rated the importance level halted the story for 20 s before resuming it. If children began to discuss
of each event on a scale of 1 (not important) to 7 (very important). the story during a control condition stop, they were reminded that they
Events with average ratings in the highest quartile were defined as ‘high would be able to talk about the story at the end. Once the story was
important events,’ and events with average ratings in the lowest quar- over, the experimenter said to all children, “Ok, now it is time for you to
tile were defined as ‘low important events.’ There were 21 events in tell the story from beginning to end, even if you have talked about it
both the high and low quartiles for Silly Richard, and 23 events in both before. Think really hard and tell me everything you can remember from
the high and low quartiles for The Brave Knight. Events from the high the story.” Children were prompted twice to recall any additional in-
and low quartiles were examined because these quartiles were the most formation (Prompt 1: Okay. Think again about the story and tell me
conceptually distinct from one another and most reliably rated based on some more things about the story. Prompt 2: Okay, now think really hard
adult ratings of importance. Additionally, high and low quartiles have and tell me something else from the story.). The experimenters ad-
been employed successfully by previous studies examining recall as a ministering the tasks were blind to diagnostic status and purpose of the
function of importance (e.g., Van Neste et al., 2015). study.

2.2.2. Standardized testing 2.4. Coding

Standardized tests were used to estimate intelligence (Kaufman Brief
Intelligence Test; KBIT-2) and oral expressive language (Oral and Written 2.4.1. Predictive prompt task
Language Scale; OWLS). The KBIT-2 includes verbal knowledge items, Responses made during the five story stops were coded for plausible

127
A. Hayden et al. Contemporary Educational Psychology 53 (2018) 123–134

explanatory predictive inferences.2 These inferences were defined as transitions to new ideas and connections within an idea with good
reasonable predictions that were causally connected to previous events overall flow. The child explained the sequence of events clearly but
in the story. Coding for these inferences was inspired by Klin et al.’s with some ambiguities. Finally, a score of “4” signified appropriate
(1999) description of causal forward inferences. Most often, these in- transitions to new ideas and connections within an idea with good
ferences were coded when children made predictions that were asso- overall flow; the child explained the sequence of events clearly with no
ciated with the main character’s goals. For example, the prediction that or very few ambiguities.
Silly Richard would try another task to make money to buy the skate-
board was coded as an explanatory predictive inference. There was no 2.4.2.4. Inter-rater reliability. To estimate inter-rater reliability, two
maximum number of inferences that children could make during this coders who were blind to diagnostic status coded all prompt
task, but they were restricted by the 20-s interval during each pause. protocols for plausible explanatory predictive inferences and 25% of
recall protocols for events, global coherence, and plausible explanatory
2.4.2. Coding of recalls backwards inferences. The Pearson correlation coefficient (r) was used
2.4.2.1. Events. Story recalls were transcribed and parsed into separate to calculate inter-rater reliability for the continuous variable of
idea units, defined as a single idea or event in the story recall (e.g., inferences and Spearman’s Rank Order correlation for the ordinal
Bailey et al., 2011; Flake et al., 2007). Idea units can include a single variable of coherence ratings. Cohen’s Kappa (K) was used to
character action, a description of a character’s emotion, a portion of the calculate inter-rater reliability for recall of events because inter-rater
setting, etc. The idea units from recall protocols were compared to a list agreement was categorical for this variable. The inter-rater reliability
of events that occurred in the original story and each idea unit was for explanatory predictive inferences in the prompt protocols was
coded for the original story event or events that most closely matched r = .98. For the recall protocols, the inter-rater reliability was
the recalled information. Each story event was therefore coded as either K = .82 for coding events, rs = .85 for coherence, and r = .85 for
0 (not recalled) or 1 (recalled). Verbatim recall was not required; recall explanatory backwards inferences.
was credited when the gist of the unit was captured by the child’s
statement. 3. Results

2.4.2.2. Inferences. Some idea units in recall protocols contained both 3.1. Preliminary analyses
explicit story information as well as inferred information related to an
explicit event. These idea units first were coded for recall of the explicit Preliminary repeated measures analyses of covariance (ANCOVA)
event, and then coded for the inference. Some statements contained were conducted to determine whether any of the dependent variables
inferred information only. These statements were evaluated for (predictive prompt task: explanatory predictive inferences; recall:
plausible explanatory inferences. Children included very few events recalled, explanatory backwards inferences, and coherence rat-
predictive inferences in their recalls (participant recalls averaged less ings) were affected by the particular story that children experienced
than 1 predictive inference across control and predictive prompt during the predictive prompt task (Brave Knight, Silly Richard), or the
conditions) because the narrative was complete. Thus, only order in which the prompt task was experienced (first or second).
explanatory backwards inferences were coded in recall protocols; that Neither story nor prompt task order was a significant covariate for any
is, when children stated reasonable connections or explanations of analyses (all ps > .05; see Table 2). Given that it would be in-
events that had not been explicitly stated in the story (Berthiaume et al., appropriate to include non-significant covariates in the primary ana-
2010; Van Neste et al., 2015). These inferences often involved implicit lyses (Vik, 2014), story and order were not included in the primary
causal connections as well as goal-related information (e.g., Silly analyses.
Richard thought of ways to make money because he wanted to buy Additional repeated-measures ANCOVAs were conducted with each
the skateboard that his mom couldn’t afford). Explanatory backwards dependent variable to evaluate whether gender, IQ, expressive lan-
inferences are important to incorporate into recall because they are guage, or the mother’s years of education was a significant covariate.
vital to a coherent story representation. These analyses were conducted because typically developing children
scored significantly higher on these measures, and it was possible that
2.4.2.3. Global coherence. Recall protocols prior to any prompts were these measures would be related to verbal narrative production (Freer
coded for global coherence. Global coherence represents the general et al., 2011). However, none of these covariates were significant (all
logical flow of the entire recall as perceived by adult evaluators ps > .05; see Table 2) and therefore they were not included in primary
(Albrecht & O’Brien, 1993; Cook et al., 2001; Habermas & de Silveira, analyses (Vik, 2014).
2008; Myers et al., 1994). A scoring rubric was derived from Habermas
and de Silveira (2008), and inter-rater reliability has been consistently 3.2. Primary analyses
high for previous studies employing this coding system (ranging from
r = .86 to r = .98; Freer et al., 2011; Lorch et al., 2010). Each recall To examine whether there was a group difference in explanatory
transcript was scored for global coherence on a scale of 1 to 4, with 1 predictive inferences generated during the predictive prompt task, the
being not at all coherent, and 4 being very coherent. A score of “1” primary analysis was a one-way analysis of variance (ANOVA). For
reflected poor transitions from one idea to the next, poor connections dependent variables from the recall task, primary analyses were mixed
while talking about the same idea, and poor overall flow. The child had ANOVAs (group × prompt condition × importance for recall of idea
significant difficulty explaining the sequence of events, with little or no units; group × prompt condition for global coherence and explanatory
story line maintained. A score of “2” signified some appropriate backwards inferences), with follow-up planned contrasts for significant
transitions to new ideas and connections within an idea, but difficulty interactions involving group. The Bonferroni procedure was used to
explaining the sequence of events, with some parts of storyline control family-wise error rate with alpha p < .05. To evaluate whether
maintained but little substance. A score of “3” reflected appropriate the group difference in generating explanatory predictive inferences
during the predictive prompt task helped to explain group differences in
story comprehension measures, mediation analyses were conducted
2
Implausible inferences also were coded, but were not examined due to infrequency of
using bootstrapping procedures at the 95% confidence level with 1000
occurrence. Specifically, in the predictive prompt task, participants generated an average
of .44 implausible explanatory predictive inferences. Further, 39 of these participants
resamples (Preacher & Hayes, 2008). The regression analyses met the
made no implausible explanatory predictive inferences. ‘Explanatory inferences’ refer assumptions of normality of residuals, homoscedasticity, multi-
only to plausible explanatory inferences. collinearity, leverage, discrepancy, and influence.

128
A. Hayden et al. Contemporary Educational Psychology 53 (2018) 123–134

Table 2 group differences on recall measures, relevant mediation analyses

Statistical information for the preliminary analyses. follow.
Factor F df p
3.2.2.1. Are there group differences in measures of narrative recall?. To
Recall Importance examine recall as a function of importance, a group (ADHD,
Story Variables comparison) × condition (prompt, control) × importance (low, high)
Story Order .00 (1, 56) .99
mixed ANOVA was conducted. Children recalled a greater proportion of
Predictive Prompt Story .08 (1, 56) .78
Sample Variables events of high importance (M = 48%; SD = 13.63) than events of low
Mother’s Education (Years) 1.83 (1, 56) .23 importance (M = 24%; SD = 17.16), F(1, 57) = 369.33, p < .001,
KBIT – Total IQ .05 (1, 56) .82 d = 5.22. Replicating previous findings (e.g., Flake et al., 2007; Lorch
Gender .14 (1, 56) .71 et al., 2010), this main effect was qualified by a significant
Inferences – Predictive Prompt group × importance interaction, F(1, 57) = 5.34, p = .03, d = .63,
Story Variables indicating that the effect of importance on recall was larger for
Story Order .11 (1, 56) .74
typically developing children (Low: M = 26%; SD = 12.26; High:
Predictive Prompt Story 1.40 (1, 56) .24
Sample Variables M = 53%; SD = 14.09) than for children with ADHD (Low: M = 20%;
Mother’s Education (Years) .26 (1, 56) .61 SD = 14.73; High: M = 40%; SD = 19.00). Global coherence of recall
KBIT – Total IQ 2.66 (1, 56) .11 was examined in a group (ADHD, Comparison) × condition (prompt,
OWLS Expressive Language .01 (1, 56) .94 control) mixed ANOVA. Also replicating previous findings (Freer et al.,
Gender .03 (1, 56) .87
2011; Lorch et al., 2010), global coherence was rated higher for
Inferences – Recall typically developing children (M = 2.82, SD = .84) than for children
Story Variables
with ADHD (M = 2.37, SD = .98; F[1, 57] = 4.30, p = .04, d = .56).
Story Order .001 (1, 56) .98
Predictive Prompt Story .17 (1, 56) .68 Explanatory backwards inferences included in recall were examined
Sample Variables with a group (ADHD, comparison) × condition (prompt, control) mixed
Mother’s Education (Years) 3.64 (1, 56) .06 ANOVA. The inferences included in recall also were non-normally
KBIT – Total IQ .001 (1, 56) .97
distributed and therefore transformed using the log10 function.
OWLS Expressive Language 1.14 (1, 56) .29
Gender .60 (1, 56) .44
Replicating Van Neste et al. (2015), children with ADHD made sig-
nificantly fewer explanatory backwards inferences (M = 4.17;
Coherence
Story Variables
SD = 3.15) than did typically developing children (M = 5.38;
Story Order .56 (1, 56) .46 SD = 3.10; F[1, 57] = 4.99, p = .03, d = .61).
Predictive Prompt Story .66 (1, 56) .42 These findings indicate that the conditions necessary to address the
Sample Variables second research question were obtained: children with ADHD per-
Mother’s Education (Years) .62 (1, 56) .44
formed more poorly than typically developing peers on all important
KBIT – Total IQ .05 (1, 56) .82
OWLS Expressive Language 2.95 (1, 56) .09 measures of inference-making and recall for both the online predictive
Gender .30 (1, 56) .59 prompt task and the post-narrative free recall task.

3.2.2.2. Are group differences in narrative recall mediated by group

3.2.1. Is there a diagnostic group difference in generating predictive differences in the number of explanatory predictive inferences made during
inferences? the predictive prompt task?. Given the group differences on all measures
The first research question was whether there was a diagnostic of inference-making and recall, the major part of the second research
group difference in the number of explanatory predictive inferences question was whether these group differences in the recall measures
stated during story stops.3 The predictive inference data were non- (recall of important events, global coherence of recall, and explanatory
normally distributed, with significant skew and kurtosis. Thus, all backwards inferences) were mediated by the group difference in the
predictive inference data were normalized using a log10 function. Fol- number of explanatory predictive inferences that children made during
lowing data normalization, a one-way ANOVA was conducted to ex- the predictive prompt task. To address this part of the research
plore the diagnostic difference in the number of explanatory predictive question, bootstrapping mediation analyses (Preacher & Hayes, 2008)
inferences that children made during the predictive prompt task. As were conducted. Because highly important events are critical to
predicted in the first hypothesis, children with ADHD created fewer narrative comprehension, the first recall measure was recall of events
explanatory predictive inferences (M = 12.33; SD = 6.65) than did ty- at the highest level of importance. The second recall measure was the
pically developing peers (M = 15.33; SD = 6.93; F[1, 57] = 4.76, rated global coherence of the child’s recall protocol. The third recall
p = .03, d = .59). measure was the number of explanatory backwards inferences the child
stated during narrative recall. The proportion of the effect of diagnostic
status on the outcome variable accounted for by the mediator was
3.2.2. Does the group difference in generating explanatory predictive calculated as the inverse of the indirect effect of the diagnostic group on
inferences mediate group differences in recall? the recall measure (c path) divided by the direct effect (c′ path). Six
The second main question was composed of two parts: (a) whether separate mediation models were tested: 3 recall measures (importance,
there were group differences in measures of narrative recall and (b) coherence, explanatory backwards inferences) × 2 conditions (prompt,
whether such differences were mediated by the group difference in the control). Both the predictive prompt and control conditions were
number of predictive inferences created during the predictive prompt included in these analyses to determine whether any mediation
procedure. In order to address this question, it was necessary to first effects were specific to the story for which explanatory predictive
examine whether there were group differences on three measures of inferences were stated during the prompt condition, or if mediation
narrative recall: recall as a function of importance, global coherence, effects were a general indicator of narrative comprehension ability and
and inclusion of explanatory backwards inferences. After documenting so observed in both conditions (Van Neste et al., 2015). See Fig. 1 for an
illustration of the mediation model.
3
All children produced at least one predictive inference overall, and all but one child
Group differences in recall of highly important events in both the
produced a prediction on each story pause (this child failed to produce a predictive in- prompt and control conditions were mediated by the number of ex-
ference on the first pause, but produced predictive inferences on each subsequent pause). planatory predictive inferences stated in the predictive prompt task

129
A. Hayden et al. Contemporary Educational Psychology 53 (2018) 123–134

Explanatory Plausible
Inferences
(Predictive Prompt Task)

Prompt: -.12* (.06) Prompt: 42.30* (8.31)

Control: -.12* (.06) a path b path Control: 24.46* (10.33)

c path, c’ path

Diagnostic Group Highly Important Events

(Recall)

Prompt: c = - 12.28* (2.77); c’ = -7.18* (3.61)

Control: c = 12.53* (4.48); c’ = -9.58* (4.48)

Fig. 1. Diagram of mediation analysis examining whether group differences in generation of explanatory predictive inferences mediate group differences in recall of highly important
events.

Table 3
Mediation of diagnostic group differences for explanatory predictive inferences generated during predictive think-aloud on diagnostic group differences for recall of high importance
units, explanatory backwards inferences, and global coherence during recall.

A Patha B Pathb C Pathc C′ Pathd 95% Confidence Interval Effect of Mediator Adj R2 df
DV B (SE) B (SE) B (SE) B (SE) (Bias correct and accelerated) 1 − (C/C′)

Think-Aloud Condition
High Importance −.12 (.06)* 42.30 (8.31)* −12.28 (2.77)* −7.18 (3.61)* −10.15 to −.96 .4153 .39* 2, 56
Explanatory Backwards Inferences −.12 (.06)* .39 (.16)* −.11 (.07) −.06 (.07) −.12 to −.0001 .4412 .10* 2, 56
Global Coherence −.12 (.06) .56 (.59) −.61 (.24) −.54 (.25) −.35 to .06 .1106 .08* 2, 56

Control Condition
High Importance −.12 (.06)* 24.46 (10.33)* −12.53 (4.48)* −9.58 (4.48)* −8.20 to −.25 .2354 .17* 2, 56
Explanatory Backwards Inferences −.12 (.06)* .33 (.17) −.17 (.07)* −.13 (.07) −.10 to −.004 .2309 .11* 2, 56
Global Coherence −.12 (.06)* .70 (.58) −.29 (.24) −.20 (.25) −.27 to .02 .2926 .01 2, 56

* p value ≤.05.
a
A path = relation between diagnostic status and mediator (explanatory predictive inferences).
b
B path = relation between mediator and dependent variable.
c
C path = total effect of diagnostic status on the dependent variable.
d
C′ = relation between diagnostic status and the dependent variables.

(41.5% of the effect accounted for by the mediator in the prompt condition (M = 5.32; SD = 3.31) than following the control condition
condition; 23.5% of the effect accounted for by the mediator in the (M = 4.49; SD = 3.03). There were no condition main effects for recall
control condition). Thus, the group difference in creating these in- as a function of importance or coherence (Fs < 1), respectively. Thus,
ferences did help to explain group differences in recall of the story thinking aloud about the story by making predictions during story
presented in both the predictive prompt and control conditions (see presentation improved the number of explanatory inferences that
Table 3 and Fig. 1). Production of these inferences did not mediate children in both diagnostic groups included in their narrative recall.
group differences in global coherence or number of explanatory in-
ferences stated during recall. Thus, the second hypothesis that group
4. Discussion
differences in generation of explanatory predictive inferences mediate
group differences in recall was partially confirmed by these results.
The current study examined diagnostic group differences in the
generation of explanatory predictive inferences during a predictive
3.2.3. Does predicting future events during story presentation affect prompt task, as well as the relation between predictive inference gen-
narrative recall? eration and narrative recall. Three questions were addressed: (1) Is
The third question was whether the activity of predicting future there a difference between children with ADHD and typically devel-
events during story presentation affected narrative recall relative to the oping children in the creation of explanatory predictive inferences
control condition in which children did not predict events. Condition during a predictive prompt task? (2) Does the group difference in the
(predictive prompt, control) was included in the ANOVAs described generation of these inferences mediate group differences in recall
above that examined recall as a function of importance, global co- measures? (3) Does the act of making predictions during story pre-
herence of recall, and inclusion of explanatory backwards inferences sentation improve recall? Regarding the first question, children with
during recall. Condition did not interact significantly with diagnostic ADHD created fewer explanatory predictive inferences during the pre-
group for any measure of recall (recall as a function of importance and dictive pauses than did typically developing peers. Further, in terms of
explanatory backwards inferences: Fs < 1; coherence: F[1, 57] = 2.21, the second question, the ability to generate explanatory predictive in-
p = .14). More importantly, there was a significant condition main ef- ferences appeared to play a role in narrative recall. For both predictive
fect for the number of explanatory backwards inferences included in prompt and control conditions, group differences in the creation of
recall, F(1, 57) = 5.40, p = .02, d = .63, such that children included a these inferences during the predictive prompt task mediated group
greater number of these inferences following the predictive prompt differences in free recall of highly important events. Finally, addressing

130
A. Hayden et al. Contemporary Educational Psychology 53 (2018) 123–134

the third question, both children with ADHD and typically developing events. Further, group differences in generation of explanatory pre-
children benefited from creating predictive inferences during the dictive inferences in the predictive prompt condition mediated group
pauses in story presentation. Children in both diagnostic groups in- differences in recall of highly important events for both the predictive
cluded a greater number of explanatory backwards inferences during prompt and the control conditions. Consistent with the conclusions of
free recall if they had been asked to make predictions during story Van Neste et al. (2015), this suggests that the mediation effects may
presentation. reflect group differences in general comprehension ability rather than
There is substantial evidence that children with ADHD have diffi- group differences in the direct contribution of generating explanatory
culties with narrative comprehension relative to typically developing predictive inferences to recall during the predictive prompt task. The
peers (e.g., Flake et al., 2007; Lorch et al., 2004, 2010; Renz et al., fact that typically developing children generated a greater number of
2003). The current study replicated two of these findings concerning explanatory predictive inferences than children with ADHD during the
narrative recall: (1) Children with ADHD in the current study recalled predictive prompt procedure may reflect a related group difference in
fewer important events relative to unimportant events when compared the construction of an integrated story representation that enables re-
to typically developing children (Flake et al., 2007). This well-docu- call of more highly important events, regardless of whether children
mented finding for children with ADHD is consistent with the centrality made predictions aloud during story presentation.
deficit identified by Miller et al. (2013) for struggling comprehenders, Making predictions aloud during the predictive prompt task im-
in which these readers recall fewer events than typical readers overall, proved children’s generation of explanatory backwards inferences
but struggle with recall of important events in particular. (2) The recall during recall for both diagnostic groups, although group differences
protocols of children with ADHD were less coherent than those of ty- were not eliminated. Predictive prompts may encourage anticipation of
pically developing children (e.g., Lorch et al., 2010). Further, the cur- connections between current and future story events that in turn could
rent study’s finding adds to the documented list of narrative compre- make it more likely that children include explanations of events in their
hension difficulties among children with ADHD: when prompted for recalls. Asking children to make predictions could have benefits for
predictions, these children made fewer explanatory predictive in- children in general. Such benefits may be especially important for
ferences than did typically developing peers. This result also adds to children with ADHD, who make fewer of these inferences.
growing evidence that there are diagnostic group differences in the As making predictions is central to both common core curricula as
production and activation of explanatory inferences of any type well as many supplemental interventions that combine multiple com-
(Berthiaume et al., 2010; Van Neste et al., 2015; Yeari et al., 2017). prehension strategies (e.g., Palincsar & Brown, 1984; Strode et al.,
Further, difficulties with explanatory inference generation have been 2012; Woolley, 2007), it was encouraging to find that actively gen-
observed across a wide range of ages, tasks, and media. erating predictive inferences during the predictive prompt task im-
The current study is notable because it is the first study which ex- proved generation of explanatory backwards inferences during recall
amined whether children with ADHD can make explanatory predictive for both groups. Although the task did not normalize the recall per-
inferences when prompted to do so. That is, because the task was to formance of children with ADHD, it is promising that these children
predict future story events, by definition inferences were required for improved. Given the group difference in generation of explanatory
successful task completion. In contrast, the open-ended prompts em- predictive inferences, it is possible that specific instruction on how to
ployed by Berthiaume et al. (2010) and Van Neste et al. (2015) made it create plausible explanatory predictive inferences would further im-
as acceptable for children to discuss explicit story information or their prove the narrative comprehension of children with ADHD. If future
responses to the story as to make inferences. Similarly, the naming studies find that children with ADHD show comprehension improve-
probe procedure utilized by Yeari et al. (2017) assessed adolescents’ ment with specific instruction on generating explanatory predictive
immediate activation of inferences but was not designed to evaluate the inferences, strategies for creating these central inferences could be in-
quality of their inferences in supporting text comprehension. Thus, the corporated into existing interventions. This would further increase the
previous studies addressed questions about the inferences that children chance of normalizing comprehension for these children.
and adolescents with ADHD do make in open-ended tasks, whereas the
predictive prompt procedure of the current study complements the 4.1. Limitations
earlier findings by examining the predictive inferences that children can
generate when specifically asked to do so. This study therefore is the There were several sampling limitations to consider for the current
first to specifically focus on the ability of children with ADHD to create study. As in many studies composed of both clinical and community
inferences relative to typically developing children. samples, there were significant group differences in IQ, expressive
Explanatory predictive inferences are important for building a language, and maternal education. However, it is important to note that
complete story representation (Klin, Murray, Levine, & Guzmán, 1999). these variables were not significant covariates and thus unrelated to the
To generate explanatory predictive inferences, children must be making pattern of group differences in the narrative comprehension measures.
reasonable implicit and explicit causal connections (Trabasso & Sperry, Additionally, teacher reports were not collected to confirm diagnosis of
1985) and maintaining goal structure information (Mandler & Johnson, ADHD. Thus, it is possible that some children in the ADHD group would
1977) as they experience the story. They must use the representation not meet full DSM criteria for ADHD. However, the findings of the
they are building, as well as both prior background knowledge and current study were consistent with those of previous studies, both re-
general knowledge of story structure, to make predictions concerning plicating prior findings and providing new information on the diffi-
future events (Klin et al., 1999). These explanatory predictive in- culties with narrative comprehension experienced by children with
ferences allow comprehenders to smoothly integrate new information ADHD. Finally, group sizes were relatively small, as is typical for stu-
into their story representations. Relative to typically developing chil- dies with children diagnosed with ADHD. This may limit the ability to
dren, children with ADHD struggle to make causal connections among find group differences. However, group differences clearly emerged
story events (Bailey et al., 2009) and to maintain goal structure across a during both the predictive prompt task as well as during recall. Thus,
story (Flory et al., 2006; Renz et al., 2003). It therefore is likely that a sample size appears to have had limited effects on the results of the
less integrated story representation contributed to the generation of current study.
fewer explanatory predictive inferences for these children. The current study also had potential procedural limitations. First,
Not only were there group differences in the number of explanatory pauses may have disrupted typical story processing (van den Broek
predictive inferences generated by children with ADHD and typically et al., 2001). However, patterns of recall replicated prior studies in
developing peers during the predictive prompt task, these group dif- which pauses during stories were not used. Thus, it is likely that chil-
ferences mediated group differences in recall of highly important dren experienced limited to no disruption to typical story processing. As

131
A. Hayden et al. Contemporary Educational Psychology 53 (2018) 123–134

suggested above, story processing even appeared to be enhanced when Appendix A. Stories used in the current study
children had made predictions. Second, the brevity of the audiotape file
(4 min) may have restricted the number of predictive inferences that A.1. Silly Richard
could be generated and thus the benefit that generating predictive in-
ferences could provide. Van Neste et al. (2015) employed 21-min One Saturday morning, a little boy named Richard was walking
televised stories, which allowed for multiple minutes to pass between downtown. As he passed all the stores, he saw a shiny, green skateboard
open-ended story pauses. Children in both groups in the Van Neste et al. on sale in the Skate World window. He really wanted the skateboard, so
study showed an improvement on recall of events high in importance, he went to his mom and said, “Mom, I saw this awesome skateboard in
whereas children in both groups in the current study included a greater the window at Skate World. Will you please buy it for me?”
number of explanatory backwards inferences in their recall but showed “Skateboards are expensive, honey,” his mom said. “You have to
no significant improvement in recall of events high in importance. Fi- earn the money by yourself.” So Richard tried to think of some ways to
nally, it is possible that children generated predictive inferences during make money. (Stop 1) He asked his mom if she would give him the
the silent pauses in the control condition, and that there were group money for doing chores. She agreed to give him five dollars for washing
differences in generation of predictive inferences during these story the dishes.
pauses that were not measured. However, given the purpose of the Richard began washing the dishes, but the suds grew and grew until
control condition, we have no record of what children were thinking he was covered from head to toe in bubbles! His mom pulled him out of
during the control story pauses. Therefore, we can only report on group the suds, dried him off and said, “Silly Richard, didn’t you know you
differences that were observed during the predictive prompt condition. weren’t supposed to use the entire bottle of soap to wash the dishes?”
This discrepancy between the current study and Van Neste et al. She smiled and gave him two dollars. (Stop 2) Richard whined, “But
(2015) in the specific benefits of completing a task in which children Mom, you said you’d give me five dollars.” “Yes, I did, Richard. And I
are given prompts during pauses may be related to other differences know you tried your best, but you made a big mess. You used all of our
between the investigations. In addition to differences in length and dish soap, so I have to buy more soap, and you only get two dollars.”
associated reduction in the number of events of high and low im- “But I still don’t have enough money. What else can I do?” His mom
portance, the two stories used in each study differed in medium of helped him think of some other ways to make money.
presentation (audiovisual vs. audiotape) and complexity of the story Richard decided to set up a lemonade stand. Unfortunately he forgot
structures. The two prompt tasks also differed in how directive they to put sugar in the lemonade. It tasted yucky! He only sold one cup.
were: Children in the Van Neste et al. study were asked to verbalize Richard made a little money, but it still wasn’t enough. Mr. Jones felt
anything that they were thinking about the story, whereas children in sorry for Richard. He said, “Maybe the lemonade business isn’t for you.
the current study were asked specifically to verbalize what they thought Why don’t you try mowing lawns? You can start with mine.” Richard
would happen next. Thus, the current task that focused on predictions got out the lawn mower and started it up. But he had never mowed a
may have limited children’s ability to integrate information to form a lawn before. (Stop 3) He didn’t know you were supposed to go in
more coherent narrative for recall purposes. However, even with fewer straight lines. Mr. Jones walked out to find a yard full of zig-zags! Silly
events of high and low importance and more directive prompts in the Richard! But Richard had worked hard, so Mr. Jones paid him half of
current study, children incorporated a greater number of important the money anyway.
explanatory backwards inferences into their recall. This indicates that Richard felt good about earning some money, but he still didn’t have
the predictive prompt task facilitated a more integrated understanding enough to buy his skateboard. As he started home, he noticed Mrs.
of the narrative. Smith, his neighbor, struggling to walk her dog. Richard went over to
her and said, “Excuse me. I’m trying to earn some money to buy a new
4.2. Conclusions and future directions skateboard. I’ll walk your dog for five dollars.” Mrs. Smith happily
agreed. (Stop 4) When Richard returned from the walk, she gave him
The current study provides further support for significant compre- the money. Richard decided since that it went so well, he’d offer to walk
hension problems for children with ADHD relative to typically devel- some more dogs. “In fact”, he thought, “I can walk several dogs at
oping children in a task designed specifically to measure their ability to once!” He went around offering to walk his neighbors’ dogs. Before he
generate predictive inferences. The study also suggests a promising knew it, he had 5 leashes in his hands! Richard didn’t realize that five
avenue for improving comprehension, in that both groups of children dogs put together were much stronger than he, and he quickly lost
exhibited improvements in recall after engaging in the predictive control. As the dogs dragged him around the neighborhood, Richard
prompt task. This result indicates that making predictions can be an realized that he was not walking them, but they were walking him! He
effective strategy for improving the quality of narrative recall among finally managed to get all of the dogs back to their owners and collect
both typically developing children and children with ADHD. his money. (Stop 5)
Future studies could enact a direct comparison of a predictive He returned home very tired and sat down on the couch. “How
prompt task versus an open-ended think-aloud task using the same much money did you make?” his mom asked. “Oh!” he said. “After
medium and stories to see whether one is more beneficial to compre- chasing those dogs around, I forgot all about my skateboard!” He
hension. This would inform educators on the most effective method of emptied out his pockets and counted his money. He had just enough to
teaching inferences. In addition, designing interventions that include buy his skateboard. “I’m too tired tonight,” he said. “I’ll go tomorrow.”
teaching children with ADHD how to create explanatory predictive And with that he fell into his bed for a long night’s rest.
inferences may significantly improve narrative comprehension. As THE END
suggested above, making predictions is a major element of general
school curricula as well as many interventions, and thus integrating A.2. The Brave Knight
information on how to generate important inferences that are vital to
narrative comprehension may substantially improve narrative com- A long, long time ago there was a kingdom high in the mountains
prehension for all children, but particularly for children with ADHD. called Castle Kingdom. The people of the kingdom were afraid of a
large, scary Dragon who lived on the highest mountain peak. “We’re in
Funding danger,” the King said. So the King asked his best knight to save the
kingdom from this Dragon. The Knight tried to think about how he
This research did not receive any specific grant from funding would defeat the Dragon. (Stop 1) The Knight said, “I will go to the
agencies in the public, commercial, or not-for-profit sectors. Dragon’s cave and fight him so that the kingdom will be safe.” The King

132
A. Hayden et al. Contemporary Educational Psychology 53 (2018) 123–134

promised the knight a reward if he could save the kingdom from the van den Broek, P., Beker, K., & Oudega, M. (2015). Inference generation text compre-
Dragon. hension: Automatic and strategic processes in the construction of a mental re-
presentation. In E. J. O’Brien, A. E. Cook, & R. F. Lorch (Eds.). Inferences during
First, the Knight got out his finest armor and began to shine it. He reading (pp. 94–121). Cambridge, UK: Cambridge University Press.
gathered his sword and shield and went off to challenge the Dragon. van den Broek, P., Tzeng, Y., Risden, K., Trabasso, T., & Basche, P. (2001). Inferential
The Knight began to climb the Dragon’s mountain, but just as he got questioning: Effects on comprehension of narrative texts as a function of grade and
timing. Journal of Educational Psychology, 93(3), 521–529.
close to the top, it started to rain. “Rain can’t hurt me,” said the Knight. Brown, R., Pressley, M., Van Meter, P., & Schuder, T. (1996). A quasi-experimental va-
But then he realized it was getting hard to move his arms and legs. He lidation of transactional strategies instruction with previously low-achieving second-
looked down and saw that his suit of armor had begun to rust! (Stop 2) grade readers. Journal of Educational Psychology, 88, 18–37.
Carrow-Woolfolk, E. (1995). Oral and written language scales manual. Circle Pines, MN:
So he took off his armor and dropped it on the side of the mountain. American Guidance Service Inc.
Then he continued on his way towards the Dragon’s cave. Catale, C., Geurten, M., Lejeune, C., & Meulemans, T. (2014). The Conners Parent Rating
Soon the Knight was close to the cave, and it was almost dark Scale: Psychometric properties in typically-developing 4- to 12-year-old Belgian
French-speaking children. European Review of Applied Psychology, 64(5), 221–227.
outside. “I should surprise the Dragon, by attacking quickly,” said the
Clinton, V. (2015). Examining associations between reading motivation and inference
Knight. The Knight began to run along the path to the Dragon’s cave. generation beyond reading comprehension skill. Reading Psychology, 36(6), 473–498.
Suddenly, the knight was falling. He had fallen into a giant hole on the Clinton, V., Seipel, B., van den Broek, P., & Rapp, D. N. (2014). Gender differences in
way to the cave! The hole was very deep and muddy, and the sides were inference generation by fourth-grade students. Journal of Research in Reading, 27(4),
356–374.
slippery. Very carefully, the Knight began to climb. After trying for a Conners, C. K. (1997). Conners’ parent rating scales revised-short version. North Tonawanda:
very long time, he finally made it out, but he soon realized that his Multi Health Systems.
sword and shield had been lost at the bottom. Cook, A. E., Limber, J. E., & O’Brien, E. J. (2001). Situation-based context and the
availability of predictive inferences. Journal of Memory and Language, 2, 220–234.
“Oh no, how will I protect the kingdom now?” said the Knight, but Derefinko, K., Bailey, U., Milich, R., Lorch, E., & Riley, E. (2009). The effects of stimulant
he was very brave so he approached the save carefully. As he was medication on the online story narrations of children with ADHD. School Mental
standing outside the entrance to the cave he heard something snoring Health, 1(4), 171–182.
Duke, N. K., & Pearson, D. (2008). Effective practices for developing reading compre-
inside. He was not sure how he would defeat the Dragon, but then he hension. Journal of Education, 189(1/2), 107–122.
looked around and saw rocks and boulders. (Stop 3) He had an idea, he Ericsson, K. A., & Simon, H. A. (1993). Protocol analysis: Verbal reports as data (revised
would block the entrance of the cave. He carefully stacked all the rocks ed.). Cambridge, MA: Bradford books/MIT Press.
Flake, R., Lorch, E., & Milich, R. (2007). The effects of thematic importance on story recall
he could move in front of the entrance. As he used all his strength to among children with Attention Deficit Hyperactivity Disorder and comparison chil-
move a huge rock, rocks started to fall all around him, so he darted dren. Journal of Abnormal Child Psychology, 35(1), 43–53.
away from the rock slide. When the rocks stopped falling, the knight Flory, K., Milich, R., Lorch, E., Hayden, A., Strange, C., & Welsh, R. (2006). Online story
comprehension among children with ADHD: Which core deficits are involved?
found himself inside the cave with the entrance sealed! The Dragon
Journal of Abnormal Child Psychology, 34(6), 853–865.
would not be able to escape, but the Knight was stuck in the cave as Fox, M. C., Ericsson, K. A., & Best, R. (2011). Do procedures for verbal reporting of
well. thinking have to be reactive? A meta-analysis and recommendations for best re-
The snoring noises had been replaced by growls and rumbles. The porting method. Psychological Bulletin, 137(2), 316–344.
Freer, B. D., Hayden, A., Lorch, E. P., & Milich, R. (2011). The stories they tell: Story
rock slide must have woken the Dragon! The knight tried to hide but production difficulties of children with ADHD. School Psychology Review, 40(3),
could not see anything because the cave was so dark. (Stop 4) The 352–366.
Dragon smelled the presence of a human and began to breathe fire in Fumeax, P., Mercier, C., Roche, S., Iwaz, J., Bader, M., Stephan, P., ... Revol, O. (2016).
Validation of the French version of Conners’ Parent Rating Scale revised, short ver-
the Knight’s direction. In the light of the dragon’s fire, the Knight could sion: Factorial structure and reliability. The Canadian Journal of Psychiatry, 61(4),
see his fierce face and large claws, and knew the Dragon was about to 236–242.
attack. As the Knight tried to escape the dragon’s fire, he spotted a hole Habermas, T., & de Silveira, C. (2008). The development of global coherence in life
narratives across adolescence: Temporal, causal, and thematic aspects. Developmental
in the wall of the cave. The Knight dived for the hole and was able to get Psychology, 44, 707–721.
out of the dragon’s reach. As the Dragon roared in anger, the knight Hannon, B. A. (2014). Are there gender differences in the cognitive components of adult
noticed he could see daylight. “Maybe there is a way out big enough for reading comprehension? Learning and Individual Differences, 32, 69–79.
Kaufman, A. S., & Kaufman, N. L. (2004). Kaufman brief intelligence test (2nd ed.). MN,
me to crawl through,” thought the Knight. The tunnel was just big
Pearson Inc: Bloomington.
enough for him, and soon he had wiggled through to the outside. Kendeou, P., Bohn-Gettler, C., White, M., & van den Broek, P. (2008). Children's inference
The Knight was safe from the Dragon. (Stop 5) The knight was tired generation across different media. Journal of Research in Reading, 31(3), 259–272.
Kendeou, P., Lynch, J. S., van den Broek, P., Espin, C., White, M., & Kremer, K. E. (2005).
and sat down to think. “That hole is too small for the Dragon to get
Developing successful readers: Building early comprehension skills through television
through, and the Dragon is trapped by the rock slide. The kingdom is viewing and listening. Early Childhood Education Journal, 33, 91–98.
safe!” The Knight said, “I will go home tomorrow to give the people the Klin, C. M., Guzmán, A. E., & Levine, W. H. (1999). Prevalence and persistence of pre-
good news and collect the reward the king promised me.” dictive inferences. Memory and Language, 40, 593–604.
Klin, C. M., Murray, J. D., Levine, W. H., & Guzmán, A. E. (1999). Forward inferences:
THE END From activation to long-term memory. Discourse Processes, 27(3), 241–260.
Klingner, J. K., & Vaughn, S. (1998). Using Collaborative Strategic Reading. The Council
References for Exceptional Children, 33, 32–37.
Kucan, L., & Beck, I. L. (1997). Thinking aloud and reading comprehension research:
Inquiry, instruction, and social interaction. Review of Educational Research, 67(3),
Albrecht, J. E., & O’Brien, E. J. (1993). Updating a mental model: Maintaining both local 271–299.
and global coherence. Journal of Experimental Psychology: Learning, Memory, and Laing, S. P., & Kamhi, A. G. (2002). The use of think-aloud protocols to compare in-
Cognition, 19(5), 1061–1070. ferencing abilities in average and below-average readers. Journal of Learning
Bailey, U. L., Derefinko, K. J., Milich, R., Lorch, E. P., & Metze, A. (2011). The effects of Disabilities, 35(5), 436–447.
stimulant medication on free recall of story events among children with ADHD. Lorch, E., Diener, M., Sanchez, R., Milich, R., Welsh, R., & van den Broek, P. (1999). The
Journal of Psychopathology and Behavioral Assessment, 33(4), 409–419. effects of story structure on the recall of stories in children with Attention Deficit
Bailey, U. L., Lorch, E. P., Milich, R., & Charnigo, R. (2009). Developmental changes in Hyperactivity Disorder. Journal of Educational Psychology, 91(2), 273–283.
attention and comprehension among children with attention deficit hyperactivity Lorch, E. P., Sanchez, R. P., van den Broek, P., Milich, R., Murphy, E. L., Lorch, R. F., Jr., &
disorder. Child Development, 80(6), 1842–1855. Welsh, R. (1999). The relation of story structure properties to recall of television
Barkley, R. A. (2015). Attention deficit hyperactivity disorder: A handbook for diagnosis and stories in young children with Attention-Deficit Hyperactivity Disorder and non-
treatment (4th ed.). New York: Guilford. referred peers. Journal of Abnormal Child Psychology, 27(4), 293–309.
Bereiter, C., & Bird, M. (1985). Use of thinking aloud in identification of teaching of Lorch, E. P., Eastham, D., Milich, R., Lemberger, C. C., Polley Sanchez, R., Welsh, R., &
reading comprehension strategies. Cognition and Instruction, 2(2), 131–156. van den Broek, P. (2004). Difficulties in comprehending causal relations among
Berthiaume, K., Lorch, E., & Milich, R. (2010). Getting clued in: Inferential processing and children with ADHD: The role of cognitive engagement. Journal of Abnormal
comprehension monitoring in boys with ADHD. Journal of Attention Disorders, 14(1), Psychology, 113(1), 56–63.
31–42. Lorch, E., O'Neil, K., Berthiaume, K., Milich, R., Eastham, D., & Brooks, T. (2004). Story
van den Broek, P. (1989). The effects of causal structure on the comprehension of nar- comprehension and the impact of studying on recall in children with Attention Deficit
ratives: Implications for education. Reading Psychology: An International Quarterly, 10, Hyperactivity Disorder. Journal of Clinical Child and Adolescent Psychology, 33(3),
19–44. 506–515.

133
A. Hayden et al. Contemporary Educational Psychology 53 (2018) 123–134

Lorch, E. P., Milich, R., Astrin, C. C., & Berthiaume, K. S. (2006). Cognitive engagement hyperactivity disorder (pp. 50–80). New York, NY: The Guilford Press.
and story comprehension in typically developing children and children ADHD from Stricklin, K. (2011). Hands-on reciprocal teaching: A comprehension technique. The
preschool through elementary school. Developmental Psychology, 42(6), 1206–1219. Reading Teacher, 64(8), 620–625.
Lorch, E. P., Milich, R., Flake, R. A., Ohlendorf, J., & Little, S. (2010). A developmental Strode, E. V., Raphael, T. E., Corsi, K., & Au, K. H. (2012). QAR for the early primary grades:
examination of story recall and coherence among children with ADHD. Journal of A read-aloud program to improve comprehension. Chicago, IL: SchoolRise LLC.
Abnormal Child Psychology: An Official Publication of the International Society for Thurlow, R., & van den Broek, P. (1997). Automaticity and inference generation during
Research in Child and Adolescent Psychopathology, 38(3), 291–301. reading comprehension. Reading & Writing Quarterly: Overcoming Learning Difficulties,
Lynch, J. S., & van den Broek, P. (2007). Understanding the glue of narrative structure: 13(2), 165–181.
Children’s on-and-off-line inferences about characters’ goals. Cognitive Development, Trabasso, T., & van den Broek, P. (1985). Causal thinking and the representation of
22, 323–340. narrative events. Journal of Memory and Language, 24, 612–630.
Mandler, J. M., & Johnson, N. S. (1977). Remembrance of things parsed: Story structure Trabasso, T., & Magliano, J. P. (1996). Conscious understanding during comprehension.
and recall. Cognitive Psychology, 9(1), 111–151. Discourse Processes, 21(3), 255–287.
Milich, R., Balentine, A., & Lynam, D. R. (2001). ADHD combined type and ADHD pre- Trabasso, T., & Sperry, L. L. (1985). Causal relatedness and importance of story events.
dominately inattentive type are distinct and unrelated disorders. Clinical Psychology: Journal of Memory and Language, 24, 595–611.
Science and Practice, 8, 463–488. Trabasso, T., & Suh, S. Y. (1993). Understanding text: Achieving explanatory coherence
Miller, A. C., Keenan, J. M., Betjemann, R. S., Willcutt, E., Pennington, B. F., & Olson, R. through on-line inferences and mental operations in working memory. Discourse
K. (2013). Reading comprehension in children with ADHD: Cognitive underpinnings Processes, 16, 3–34.
of the centrality deficit. Journal of Abnormal Child Psychology, 41(3), 473–483. Van Neste, J., Hayden, A., Lorch, E., & Milich, R. (2015). Inference generation and story
Murray, J. D., Klin, C. M., & Myers, J. L. (1993). Forward inferences in narrative text. comprehension among children with ADHD. Journal of Abnormal Child Psychology,
Journal of Memory and Language, 32, 464–473. 43(2), 259–270.
Myers, J. L., O’Brien, E. J., Albrecht, J. E., & Mason, R. A. (1994). Maintaining global Vik, P. (2014). Regression, ANOVA, and the general linear model: A statistics primer.
coherence during reading. Journal of Experimental Psychology: Learning, Memory, & Thousand Oaks, CA: SAGE Publications Inc.
Cognition, 20, 876–886. Weller, E., Weller, R., Rooney, M., & Fristad, M. (1999). Children's interview for psychiatric
Palincsar, A. S., & Brown, A. L. (1984). Reciprocal teaching of comprehension-fostering syndromes—Parent version (P-ChIPS). Washington, DC, US: American Psychiatric
and comprehension-monitoring activities. Cognition and Instruction, 1(2), 117–175. Association.
Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for assessing Woolley, G. (2007). A comprehension intervention for children with reading compre-
and comparing indirect effects in multiple mediator models. Behavior Research hension difficulties. Australian Journal of Learning Disabilities, 12(1), 43–50.
Methods, 40(3), 879–891. Yeari, M., Avramovich, A., & Schiff, R. (2017). Online inferential and textual processing
Renz, K., Lorch, E. P., Milich, R., Lemberger, C., Bodner, A., & Welsh, R. (2003). On-Line by adolescents with attention-deficit/hyperactivity disorder during reading compre-
story representation in boys with Attention Deficit Hyperactivity Disorder. Journal of hension: Evidence from a probing method. Journal of Clinical and Experimental
Abnormal Child Psychology, 31(1), 93–104. Neuropsychology, 39(5), 485–501. http://dx.doi.org/10.1080/13803395.2016.
Roberts, W., Milich, R., & Barkley, R. A. (2015). Primary symptoms, diagnostic criteria, 1236906.
subtyping, and prevalence of ADHD. In R. A. Barkley (Ed.). Attention deficit

134