Full Terms & Conditions of access and use can be found at
Oxford Review of Education
ISSN: 0305-4985 (Print) 1465-3915 (Online) Journal homepage: https://www.tandfonline.com/loi/core20
Effects of the TutorBright tutoring programme on
the reading and mathematics skills of children in
foster care: a randomised controlled trial
Andrea J. Hickey & Robert J. Flynn
To cite this article: Andrea J. Hickey & Robert J. Flynn (2019) Effects of the TutorBright tutoring
programme on the reading and mathematics skills of children in foster care: a randomised
controlled trial, Oxford Review of Education, 45:4, 519-537, DOI: 10.1080/03054985.2019.1607724
To link to this article: https://doi.org/10.1080/03054985.2019.1607724
Published online: 26 Jul 2019.
Submit your article to this journal
Article views: 14
View Crossmark data
Eﬀects of the TutorBright tutoring programme on the
reading and mathematics skills of children in foster care: a
randomised controlled trial
Andrea J. Hickey and Robert J. Flynn
Centre for Research on Educational and Community Services, University of Ottawa, Ottawa, Canada
We evaluated the eﬀects of TutorBright tutoring on the reading and
mathematics skills of children in family foster care, examined several
potential moderators of the impact of tutoring, and explored pos-
sible ‘spill-over’eﬀects on the children’s executive functioning and
behavioural diﬃculties and on their caregivers’level of involvement
in schoolwork in the home. The sample consisted of 70 children in
care in Ontario, Canada. At the pre-test, the children were aged
5–16 years (M= 10.41, SD = 2.94) and enrolled in school grades 1–11
(M= 5.53, SD = 2.90). Thirty-four children were randomly assigned
to tutoring and 36 to a waiting list control condition. Seven subtests
from the Woodcock-Johnson III (WJ-III) achievement test served as
outcome measures. The tutored children made statistically greater
gains than those in the control group on the WJ-III subtests of
Reading Fluency, Reading Comprehension, and Mathematics
Calculation, but not on Word Reading, Spelling, Math Fluency, or
Applied Math Problems. Age, executive functioning, caregiver con-
trolling involvement in schoolwork, and self-reported post-trau-
matic stress disorder symptoms were found to moderate the
eﬀectiveness of tutoring. There were no spill-over eﬀects of tutor-
ing. The implications of the results for improving foster children’s
reading and mathematics skills were discussed.
Children in care; foster care;
tutoring; academic skills;
Many children in care have higher rates of learning disabilities, grade retention, and
below-average performance in mathematics and reading, compared with children in the
general population (McGuire & Jackson, 2018;O’Higgins, Sebba, & Gardner, 2014; Sebba
et al., 2015; Tessier, O’Higgins, & Flynn, 2018; Trout, Hagaman, Casey, Reid, & Epstein,
2008). They also often have cognitive and language deﬁcits and poor problem-solving or
reasoning skills (Kavanaugh, Dupont-Frechette, Jerskey, & Holler, 2016). Approximately
30–50% of children in care are eligible for special education (Lightfoot, Hill, & LaLiberte,
2011), compared to 10% of children in the general population (Zetlin, Weinberg, & Shea,
2006). Without mastery of reading or mathematics, children in care are at an increased
risk of academic failure or school dropouts (Berlin, Vinnerljung, & Hjern, 2011; Trout
et al., 2008; Viner & Taylor, 2005). Based on national longitudinal register data in Sweden,
CONTACT Andrea J. Hickey firstname.lastname@example.org
OXFORD REVIEW OF EDUCATION
2019, VOL. 45, NO. 4, 519–537
© 2019 Informa UK Limited, trading as Taylor & Francis Group
Forsman, Brännström, Vinnerljung, and Hjern (2016) found that among young adults
who had previously been in care, poor school performance at the end of obligatory
schooling had a causal and severe negative impact, including economic hardship, illicit
drug use, and poor mental health. Brännström, Vinnerljung, Forsman, and Almquist
(2017) recently extended their analyses in Sweden, demonstrating that the serious
negative psychosocial consequences of poor educational performance in childhood
persist until well into middle age (i.e. 39–55 years).
Poor educational achievement is among the strongest risk factors for the future well-
being of children in care. As the Swedish research has shown, the links between poor
academic achievement and unfavourable psychosocial outcomes remain strong, even
after controlling for the socioeconomic status of the family when the child was still very
young, the length of time in care, and age at ﬁrst placement (Brännström et al., 2017;
Forsman et al., 2016).
Several meta-analyses have been conducted on interventions for at-risk students.
Lauer et al. (2006) found that although out-of-school-time programmes for low-
achieving students had only a small overall eﬀect, one-to-one tutoring to improve
reading skills had a moderately large average eﬀect (Hedges g= .50). Dietrichson,
Bøg, Filges, and Klint Jørgensen (2017), in a meta-analysis of approximately 100 aca-
demic interventions for children of low socioeconomic status, found that, on average,
the most eﬀective academic interventions were, in order, one-to-one or small-group
tutoring (ES = .36), progress monitoring (in which teachers or students received infor-
mation about development; ES = .32), and cooperative or peer-assisted learning
(ES = .22). Slavin, Lake, Davis, and Madden (2009) found that one-to-one tutoring was
more eﬀective than small-group tutoring or computer-assisted instruction programmes
for struggling readers in grades K-5. Finally, Pellegrini, Lake, Inns, and Slavin (2018)
recently completed a best-evidence synthesis of eﬀective programmes in elementary
mathematics. In their review of 78 high-quality evaluations (83% randomised, 17% quasi-
experimental) of 61 programmes in grades K-5, they found that tutoring programmes
were found to have especially positive outcomes. One-to-one and one-to-small group
programmes had equal impacts, as did teachers and paraprofessionals as tutors. Overall,
tutoring (particularly in a one-to-one format) seems a promising means of improving the
academic skills of children who need help in reading or mathematics.
At present, there are relatively few evaluated interventions aimed at remediating the
educational challenges of young people in care. In a scoping review, Forsman and
Vinnerljung (2012) identiﬁed 11 studies that were either randomised controlled trials,
quasi-experiments, or pre-test-post-test studies. Overall, nine of the 11 evaluations
reported positive outcomes, with four out of ﬁve tutoring evaluations yielding positive
results. Evans, Brown, Rees, and Smith (2017) reviewed the controlled research –12
projects, including 15 randomised controlled trials –on academic interventions for
young people in care under the age of 19 years. Of the 12 projects, nine had demon-
strated some evidence of eﬀectiveness. Three had employed academic tutoring, either
one-on-one (Flynn, Marquis, Paquet, Peeke, & Aubry, 2012; Marquis, 2013) or small
group-based (Harper & Schmidt, 2016). Evans et al. (2017) called for the continued use
of randomised trials but with greater attention to methodological quality, in order to
generate more convincing evidence about which academic interventions work well and
520 A. J. HICKEY AND R. J. FLYNN
To our knowledge, only two randomised controlled trials of tutoring have been published
that have had positive results with children in care. Flynn et al. (2012) assessed the eﬀective-
ness of one-to-one, direct-instruction tutoring (Maloney, 1998), delivered by foster carers, on
the reading and mathematics skills of 64 foster children, aged 6 to 13 years and in primary-
school grades 2–7. Thirty were in the tutoring group and 34 in a waiting list control group. The
tutored children made signiﬁcantly greater gains than those in the control condition on
sentence comprehension (Hedges’g= 0.38, p< .05), reading composite (g= 0.29, p< .10), and
mathematics computation (g= 0.46, p< .01), but not on word reading or spelling. Harper and
Schmidt (2016) also evaluated Maloney’s(1998) version of direct-instruction tutoring, using
a small-group format and university students as tutors. The foster children were 6–13 years
old, in primary grades 1–8, and mainly of Indigenous origin. Forty-ﬁve were randomly
assigned to the tutoring group and 46 to a waiting list control group. The tutoring group
made signiﬁcant gains on word reading (Hedges’g= 0.40, p< .001), spelling (g= 0.25, p< .02),
and mathematics computation (g= 0.34, p< .04), but not on sentence comprehension.
Overall, as Evans et al. (2017) have advocated, more high-quality research is needed.
The primary objective of the present study was to evaluate the impact of an individua-
lised, home-based tutoring programme, known as TutorBright, on the reading and
mathematics skills of children in family foster care. We hypothesised that the children
who had received academic tutoring would improve signiﬁcantly more on their mathe-
matics and reading skills than those in a waiting list control group. We also examined
whether there was a dose-response relationship between the amount of exposure to
tutoring and gains in reading and mathematics.
The second objective involved exploring whether selected covariates –child age,
gender, executive functioning, behavioural diﬃculties, post-traumatic stress disorder
(PTSD) symptoms, or caregiver involvement in academic activities in the home –mod-
erated the impact of tutoring on reading and mathematics skills. The third objective
consisted of determining whether tutoring produced positive ‘spillover eﬀects’on the
children’s executive functioning or behavioural diﬃculties or on their caregivers’invol-
vement in academic activities in the home.
The present study was a randomised eﬀectiveness trial, conducted under real-world
conditions, rather than an eﬃcacy trial, carried out under near-ideal, laboratory condi-
tions. The ﬁrst author (AJH) undertook the study as part of her doctoral thesis research in
clinical psychology (Hickey, 2018), of which the second author (RJF) was the research
supervisor. The study was approved by the Social Sciences and Humanities Research
Ethics Board of the University of Ottawa. This trial was not registered.
Recruitment and participants
Child welfare workers nominated foster children who had met several eligibility
criteria: enrolled in grades 1–11, ﬂuent in English (the language of the TutorBright
materials), currently living in a foster-family setting (including kinship-care and
OXFORD REVIEW OF EDUCATION 521
adoption-probation), and judged likely to remain in care for the duration of the study.
Foster children were excluded if they were living in a group home, were rated by their
child welfare workers as either strong students (and thus not likely to need tutoring),
or else were either intellectually disabled or very behaviourally disturbed (and thus not
likely to complete the programme). Seventy-ﬁve children and their caregivers assented
or consented to participate and were enrolled in the study and randomised to either
TutorBright tutoring or a waiting list control condition.
At the pre-test, the 75 foster children were aged 5–16 years (M= 10.59, SD = 2.98;
see Table 1)andingrades1–11 in school (M= 5.68, SD = 2.95). The 70 foster parents
(64 females, 6 males) had been providing care to their respective children in care for
approximately 2 years (M=2.05years,SD = 1.03). Thirty-seven children were
randomly assigned to the TutorBright (experimental) group (20 males, 17
females; Mage = 10.32 years; range = 6–15 years) and 38 to the waiting list control
group (19 males, 19 females; Mage = 10.50 years; range = 5–16 years). Attrition was
low: 3 in the TutorBright group, 2 in the waiting list control group. (See CONSORT
diagram in Figure 1.)
Research design and random assignment
In our pre-test/post-test, waiting list-controlled, two-group design, random assignment to
conditions took place at the pre-test, immediately after the foster parents and foster
children had signed their consent or assent forms and had completed their initial assess-
ments. The children were assessed at two time points: Time 1 (pre-test) and Time 2 (post-
test, approximately 10 months after the pre-test). Unfortunately, given time constraints, no
longer term follow-up was conducted. The assessor was masked to the child’s condition. At
Times 1 and 2, the children in both groups were assessed on their mathematics and reading
skills with selected subtests from the Woodcock-Johnson III (Woodcock, McGrew, & Mather,
2001). The foster parents were asked to complete a questionnaire package (which included
all of the measures noted below) while their child in care was being assessed. The ques-
tionnaire package took foster parents approximately 45 minutes to complete.
Experimental and waiting list control conditions
TutorBright tutoring programme
TutorBright was developed in 2007 in Toronto, Canada, by Mr. Sonny Verma. He stated
that the programme is based on Direct Instruction, which uses a well-organised instruc-
tional methodology and clearly structured teaching materials (Adams & Engelmann,
Table 1. Number of participants in each age group
x experimental group.
Age Total Number of Participants
waiting list Experimental
5–7 years 5 7
8–10 years 12 12
11–13 years 13 7
14–16 years 6 8
522 A. J. HICKEY AND R. J. FLYNN
1996). TutorBright had not been evaluated prior to the present research. However, the
eﬀectiveness of Direct Instruction tutoring has been demonstrated both for children in
the general population (Flores & Ganz, 2009) and for children in care (Flynn et al., 2012;
Marquis, 2013). TutorBright aims to accelerate students’learning in reading, language,
and mathematics and to create conﬁdence via mentoring relationships (S. Verma,
personal communication, 7 June 2017).
TutorBright uses one-on-one, in-home tutoring, with detailed instructor’s manuals
and customised student workbooks. The reading and mathematics curricula each have
10 progressive levels, with students moving to the next level after showing mastery of
the previous level. TutorBright also incorporates homework help, in which the tutor
provides aid in any academic subject of need (e.g. mathematics, reading, science,
geography, etc.). Requirements for TutorBright tutors include an undergraduate or
master’s degree (completed or in progress), experience with teaching or mentoring,
strong communication skills, and a positive attitude. Each tutor-child match is based on
the academic needs of the child, geographic location, and mutual interests. Besides
facilitating tutoring, the one-to-one relationship with the tutor is thought to help the
child develop healthy relationships with other adults.
Throughout the current study, the TutorBright programme provided trained TutorBright
tutors, ongoing consultation to the research team, and regular monitoring of student pro-
gress. After being matched with a child, the tutor conducted home visits to assess the child’s
reading, mathematics, and writing and determine the curriculum levels to use with the child.
The tutor then met individually with the child in the child’s home for two one-hour sessions
51 potential participants
declined to participate due to:
unstable foster home, or not
wanting to be randomised to the
TutorBright tutoring group
Eligible sample (N =126)
Randomisation & pre-test
Waiting list control group
Attrition: n= 3 (1 moved, 1
refused to continue to
participate, 1 withdrew due
to mental health difficulties)
N = 38
Attrition: n= 2 (1 moved; 1
refused to return for the
Analysed (n= 36) Analysed (n= 34)
Figure 1. CONSORT diagram (Schulz, Altman, & Moher, 2010).
OXFORD REVIEW OF EDUCATION 523
per week, on designated days of the week (e.g. every Monday and Wednesday, days available
to the families and tutors), for up to 50 hours of tutoring. The tutoring was aligned with the
school year, from October until the end of June or early July. Except for a two-week break for
Christmas holidays and a one-week break for March break (all of the children took this time oﬀ
from tutoring), the children did not miss more than two weeks of tutoring due to illness or
scheduling conﬂicts. Any missed sessions were added onto the end of the tutoring period to
help ensure each child met the 50-hour goal. On average, the children in the present study
received a total of 47 tutoring sessions (48.66 hours of tutoring; SD = 4.06), over the course of
nine months. More speciﬁcally, 32.4% of the sample completed 23–46 sessions, 32.3%
completed 47–49 sessions, and 35.3% completed 50–55 sessions. During the study, they
received an average (M) of 14.10 hours of tutoring in reading (SD = 4.48), 13.97 hours in
mathematics (SD = 8.11), 10.94 hours in homework help (SD = 9.78), and 8.13 hours in ‘other’
tutoring (e.g. relationship-building, managing behaviour; SD = 3.80).
Waiting list control condition
Children in the waiting list control condition were asked to continue their schooling ‘as
usual’and not seek additional tutoring or academic support during the school year. They
were oﬀered the tutoring intervention at the end of the school year. The participants’
academic skills in the TutorBright and waiting list groups were assessed with the WJ-III
on two occasions, before the study started and at the end of the school year. To ensure
an equivalent amount of time between the pre-test and post-test in the two groups,
a waiting list child was randomly selected to be post-tested whenever a participant in
theTutorBright group reached 50 sessions of tutoring or came to the end of the
school year, whichever came ﬁrst.
Assessment of implementation ﬁdelity and tutoring dosage
The ﬁdelity of implementation by the tutors of each major TutorBright component, in
reading, mathematics, and homework help, was evaluated weekly by the ﬁrst author and
project coordinator (AJH). Another person, an experienced TutorBright coordinator,
oversaw the selection, training, deployment, and ﬁdelity of implementation of the
TutorBright model, while making no research-related decisions. The assessment of
ﬁdelity was based on the weekly performance data sent by the tutors to both coordi-
nators. These weekly data included the lessons covered and the average number of
minutes spent overall on tutoring, on reading and mathematics tutoring (based on the
TutorBright lessons), on homework help, and on oﬀ-task activities (e.g. relationship
building). The children in the tutoring condition completed an average of 47.56 sessions
(SD = 4.23, range = 35–56) over a nine-month period and were judged by the project
and TutorBright coordinators as having received a high level of implementation.
Woodcock-Johnson III tests of achievement –third edition (WJ-III; Woodcock et al., 2001)
The WJ-III is a norm-referenced, standardised series of tests that assess basic reading and
mathematics skills in individuals who are 2 to 90+ years of age or in grades K though
graduate school. The following subtests were administered in the current study: Letter-
Word Identiﬁcation, Reading Fluency, Story Recall, Understanding Directions,
524 A. J. HICKEY AND R. J. FLYNN
Calculation, Math Fluency, Spelling, Passage Comprehension, Applied Problems, Story
Recall-Delayed, Picture Vocabulary, and Oral Comprehension. These subtests were
selected because they enable the calculation of an ‘intra-achievement’discrepancy
score. That is, an Oral Language score (i.e. a measure of oral language development,
derived from the Understanding Directions, Picture Vocabulary, and Oral
Comprehension subtests) can be used to predict an individual’s level of reading and
mathematics achievement. According to Wilkinson and Robertson (2001), a signiﬁcant
discrepancy between a person’s oral language ability and academic performance may
be indicative of a speciﬁc reading or mathematics learning disability.
A Reading Composite (‘Broad Reading’) score is obtained by combining the Word
Reading, Reading Fluency, and Sentence Comprehension standard scores.
A Mathematics Composite (‘Broad Mathematics’) score is obtained by combining the
Calculation, Math Fluency, and Applied Problems subtests.
Comprehensive executive function inventory –parent version (CEFI; Naglieri &
The CEFI is a norm-referenced, standardised, 100-item measure of executive functioning in
children ages 5 to 18 years. The CEFI was completed by the child or youth’sfosterparent.
We used the CEFI total standardised score (M=100,SD = 15), with a higher score indicating
greater executive functioning (Naglieri & Goldstein, 2012). For the alpha coeﬃcients for this
sample for the CEFI and other instruments except the WJ-III, see Table 1.
Strengths and diﬃculties questionnaire (SDQ; Goodman, 1997)
The SDQ uses parent or caregiver ratings to assess mental health problems in children
and youth aged 4–17 (Goodman & Goodman, 2009) on a 20-item Total Behavioural
Diﬃculties scale, on which a higher score indicates a greater level of behavioural
Trauma symptom checklist for children –alternate form (TSCC-A; Briere, 1996)
The TSCC (Briere, 1996) is a self-report, 44-item instrument that assesses a broad range
of traumatic symptoms in children and adolescents, ages 8–17. Because of the reading
level, the TSCC-A was used with children aged 10 and older.
Trauma symptom checklist for young children (TSCYC; Briere, 2001)
The TSCYC is a parent-report, 90-item instrument that assesses traumatic symptoms in
young children, aged 3–12 years. The caregivers of children aged 5–9 years completed
the instrument, with elevated T-scores indicating greater symptoms of posttraumatic
stress (Briere et al., 2001).
Parental support for learning scale (PSLS; Rogers, Markel, Midgett, Ryan, & Tannock,
The 48-item PSLS assesses the extent of caregiver support for educational activities in
the home. Instrumental parental involvement measures the degree of warmth, patience,
independence (i.e. more eﬀective involvement) regarding the child’s schoolwork, while
controlling parental involvement assesses the use of commands, punishment, nagging,
and disapproval (i.e. less eﬀective involvement).
OXFORD REVIEW OF EDUCATION 525
Child welfare worker background information form
With this questionnaire, created by the research team, the child welfare worker provided
background information on the child (e.g. age of entry into care, maltreatment history, etc.).
Foster parent questionnaire
Also developed by the research team and administered at the pre-test and post-test, this
instrument gathered background information on the caregiver (e.g. number of children
in the home, length of time as a foster parent, etc., and supplementary information on
the child (medication, Individualized Education Plans, etc.).
The data were screened for missing data, univariate outliers, skewness, and kurtosis
(Tabachnick & Fidell, 2007; see Table 2). With a missing-data rate of less than 5% on most
variables, we used the expectation maximization (EM) algorithm in SPSS to address
missing data at the item level (Schafer & Graham, 2002).
To answer our main and exploratory research questions, we used ANCOVA via multiple
regression, the most powerful approach for analysing pre-test/post-test control-group
designs (Gliner, Morgan, & Harmon, 2003). We employed one-tailed tests to evaluate our
main hypothesis because of its clearly directional nature. That is, we expected the tutored
children to have higher adjusted mean scores at the post-test than the control children.
This choice aﬀorded us greater statistical power, given the likelihood at the outset of the
study that we would end up with a relatively small research sample, even with data-
collection over two school years. Our experience in a previous tutoring RCT (Flynn et al.,
2012), in which we had also used one-tailed tests, had taught us to be conservative in
predicting our ultimate sample size. We also used correlational analyses to determine
whether a dose-response relationship existed between the amount of tutoring received
(measured in terms of the number of tutoring sessions completed or the total number of
tutoring hours and minutes received) and gains in WJ-III reading and reading.
We also used ANCOVA via multiple regression to answer our exploratory questions
regarding possible moderators of the impact of tutoring and potential spillover eﬀects
from tutoring on the educationally relevant domains of child executive functioning or
behavioural diﬃculties and caregiver involvement in schoolwork in the home. Given the
exploratory nature of these questions, we used two-tailed rather than one-tailed statistical
Eﬀect size index
We used Hedges’gto assess the size of the eﬀect of TutorBright on the foster children’s
reading and mathematics skills, as recommended by the What Works Clearinghouse
(WWC), 2008. We also adopted the WWC criterion according to which an eﬀect size of
0.25 or greater should be considered ‘substantively important’, even if not statistically
signiﬁcant because of small sample size. A Hedges’g of 0.25 or larger reﬂects a 10-
percentile or greater diﬀerence between the means of the experimental and control
groups in a normal distribution.
526 A. J. HICKEY AND R. J. FLYNN
Table 2. Means (or percentages), standard deviations, Cronbach’s alphas, theoretical range, and skewness for study variables at the pre-test, after attrition, for
the sample as a whole (combined TutorBright and waiting list control groups).
Pre-test (N= 70) Post-test (N= 70)
Variable Mean (or %) SD Cronbach’s alpha
Range Skew Mean (or %) SD Cronbach’s alpha
Broad Reading 82.33 19.43 50–150 −1.00 84.43 18.50 50–150 −1.25
Letter-Word Identiﬁcation 87.66 17.54 50–150 −.91 87.20 17.35 50–150 −1.07
Passage Comprehension 81.63 17.15 50–150 −1.16 84.43 14.20 50–150 −1.89
Reading Fluency 84.22 18.03 50–150 −.25 88.04 17.52 50–150 −.29
Spelling 84.27 19.10 50–150 −.58 84.73 19.61 50–150 −.52
Broad Mathematics 74.01 18.97 50–150 −.59 75.67 16.81 50–150 −1.25
Calculation 73.34 22.10 50–150 −.28 74.24 19.02 50–150 −.43
Applied Problems 83.16 15.28 50–150 −.97 85.20 13.88 50–150 −.53
Math Fluency 74.10 13.15 50–150 −.05 75.67 13.10 50–150 −.19
Executive Functioning (CEFI total score) 81.89 13.12 0.92 50–150 −.05 83.63 13.05 50–150 .61
Behaviour (SDQ total behavioural diﬃculties) 16.19 6.59 0.81 0–40 −.11 15.54 7.20 0–40 −.09
PSLS caregiver involvement
Symptoms of PTSD
TSCC (n= 42)
TSCYC (n= 28)
–– – – –
Age 10.41 2.94 –5–16 .07 11.27 2.89 5–16 .12
Gender (male: female %) 54.3:45.7 –– ––54.3:45.7 –– – –
Skewed variables were reﬂected and transformed by means of square root transformations. However, the transformed data unexpectedly yielded fewer statistically signiﬁcant results, such
that the analyses conducted with the untransformed data were retained.
OXFORD REVIEW OF EDUCATION 527
Equivalence of groups at pre-test
Independent-samples t-tests and chi square analyses showed that there were no statis-
tically signiﬁcant pre-test diﬀerences (p< .05) between the tutoring and control groups
on gender, age, school grade, total behavioural diﬃculties, trauma symptoms, or WJ-III
subtests, indicating that randomisation had been eﬀective (see Table 3). However, there
were signiﬁcant group diﬀerences on the child’s pre-care experience of abandonment,
with a higher rate in the control group (χ
[1, N= 70] = 5.90, p< .05), and on controlling
caregiver involvement in the child’s schoolwork, with a higher mean score for caregivers
in the control group (t = 2.46, p< .05).
From the pre-test to the post-test a total of ﬁve children dropped out of the study, for an
overall attrition rate of 6.67%. More speciﬁcally, from the tutoring group, three children
(8.11%) dropped out, dueto mental health problems, a change in caregiver and subsequent
placement move, and a change in school programme that no longer included mathematics
and rendered tutoring less relevant. From the waiting list control group, two children
(5.26%) withdrew due, respectively, to reuniﬁcation with the biological parents (with non-
return for the post assessment) and to refusal to return for the post-test.
Table 3. Sample means (SDs) or percentages at pre-test, after attrition.
Waiting list control group
TutorBright tutoring group
Sex (M:F) 19:17 19:15
Age 10.50 (2.91) 10.32 (3.01)
School grade 5.69 (3.03) 5.35 (2.80)
Age of ﬁrst placement 5.68 (3.91) 5.24 (3.69)
Number of previous placements 1.85 (2.13) 1.94 (1.69)
Number of unplanned school changes 2.70 (2.18) 1.96 (1.62)
Does the child have an IEP at school Yes: 24 (66.7%) Yes: 22 (64.7%)
Long term health conditions:
Autism Spectrum Disorder
Reason for entry into care:
Other: Parental Mental Illness
Trauma Symptoms (PTSD T score):
SDQ Total Diﬃculties 15.50 (6.56) 16.91 (6.65)
CEFI Total Standard Score†84.56 (11.47) 79.06 (14.29)
*signiﬁcant diﬀerence between groups, p< .05
†trend, p< .10
528 A. J. HICKEY AND R. J. FLYNN
Primary research question: eﬀects of tutoring
Table 4 displays the means and SDs of the pre-test scores and the adjusted post-test
scores for the TutorBright (experimental) and waiting list control groups, on the seven
WJ-III subtests and overall Broad Mathematics and Broad Reading composite scores.
Table 4 also displays the t values, signiﬁcance level, eﬀect size (Hedges g), 95% con-
ﬁdence intervals, and improvement index for each of the ANCOVAs.
ANCOVA via multiple regression was used to assess whether TutorBright tutoring reduced the
discrepancy between the children’s actual and predicted Broad Mathematics or Broad Reading
score, with a decrease suggesting an increase in skills (Woodcock et al., 2001). A pre-test-post-
test diﬀerence score was ﬁrst calculated by subtracting the predicted Broad Reading and Broad
Mathematics scores (based on each participant’s Oral Language score) from the actual Broad
Reading and Broad Mathematics scores. The diﬀerence between the adjusted post-test group
means was non-signiﬁcant for both Broad Reading (t = .62, p= .54, 1-tailed) and Broad
Mathematics (t = .58, p= .56, 1-tailed).
Results for primary research question 2: dose-response relationship
The calculation of Pearson correlations between the tutoring ‘dosage’measures and the
WJ-III outcome measures in the TutorBright group of 34 children in care revealed
a positive relationship, at the level of a trend, between Letter-Word Identiﬁcation and
the total number of tutoring sessions received (r = .30, p= .08). There were also
statistically signiﬁcant and negative relationships, at the level of a trend, between the
number of minutes spent on other activities during the tutoring sessions, namely,
behavioural management or relationship-building, and Broad Reading (r = −.30,
p= .09), Reading Fluency (r = −.32, p= .06), and Math Fluency (r = −.32, p= .07).
Results for exploratory research question 1: moderators of tutoring
Given the exploratory nature of these research questions, two-tailed tests were used in
these regression-based ANCOVAs.
Age negatively moderated the eﬀect of tutoring, at the level of a trend, on Broad
Mathematics (t = −1.80, p= .08, 2-tailed), with the younger children beneﬁtting
relatively more than the older children.
Gender did not moderate the eﬀect of tutoring (all ps > .20) but did have a signiﬁcant
main eﬀect, at the level of a trend, on Spelling (t = −1.75, p= .09, 2-tailed). The boys
made relatively greater gains than the girls.
OXFORD REVIEW OF EDUCATION 529
Table 4. Means and standard deviations on the pre-test and adjusted post-test scores and the ANCOVA results for the primary research question (t score, eﬀect
size (Hedges g), 95% conﬁdence interval (C.I.) and improvement index) for the 7 WJ-III subtests and overall broad mathematics and broad reading scores.
Wait-list control group
TutorBright tutoring group
WJ-III sub-test Pre-test Adjusted post-test Pre-test Adjusted post-test tHedges g95% C.I. Improvement Index
Letter word identiﬁcation 87.42 (16.78) 86.80 (15.60) 87.91 (18.57) 87.61 (17.27) t = 0.25 0.05 −0.42, +0.52 2% (i.e. 52.0
Reading ﬂuency 85.18 (16.08) 86.48 (13.88) 83.29 (19.95) 89.35 (17.21) t = 2.21** 0.16 −0.31, +0.63 6.4% (56.4
Passage comprehension 80.72 (18.80) 82.11 (13.22) 82.59 (15.44) 86.94 (10.86) t = 2.03** 0.34 −0.13, +0.81 13.3% (63.3
Spelling 85.58 (19.92) 86.66 (18.97) 82.88 (18.38) 82.67 (17.50) N.A. −0.20 −0.67-, +0.27−8.0% (42.0
Broad Reading 82.19 (19.72) 83.17 (17.71) 82.47 (19.42) 85.77 (17.44) t = 1.95* 0.14 −0.33, +0.61 5.6% (55.6
Calculation 71.86 (21.63) 70.85 (15.73) 74.91 (22.81) 77.81 (16.58) t = 2.03** 0.39 −0.08, +0.86 15.2% (65.2
Math ﬂuency 72.58 (12.72) 73.96 (11.07) 75.71 (13.59) 77.46 (11.82) t = .51 0.27 −0.31, +0.85 10.6% (60.6
Applied problems 81.19 (14.79) 83.84 (10.03) 85.24 (15.74) 86.71 (10.67) t = .06 0.21 −0.26, +0.68 8.4% (58.4
(Table 4 continued)
Broad Mathematics 72.11 (18.81) 73.09 (15.00) 76.03 (19.20) 78.42 (15.30) t = 1.29 0.32 −0.15, +0.79 12.7% (62.7
N.A. = Not applicable, as we used only the positive tail in these 1-tailed hypothesis tests.
**p< .05, 1-tailed test. * p< .06, 1-tailed test.
530 A. J. HICKEY AND R. J. FLYNN
On Letter Word Identiﬁcation, children with higher CEFI scores performed relatively
worse than their lower-scoring counterparts (t = −2.29, p= .03, 2-tailed).
Total behavioural diﬃculties
SDQ behavioural diﬃculties did not have a signiﬁcant moderating or main eﬀect on any
of the WJ-III subtests (all ps > .20, 2-tailed).
Parental support for learning
Instrumental involvement had no signiﬁcant moderating or main eﬀects on any of the
WJ-III subtests (all ps > .10, 2-tailed). Controlling involvement, however, did moderate
the impact of tutoring, at the level of a trend, on Reading Fluency (t = −1.70, p= .09,
2-tailed) and on Broad Mathematics (t = −1.68, p= .09, 2-tailed), with higher levels of
caregiver controlling involvement associated with poorer child schoolwork performance.
On the other hand, controlling involvement had signiﬁcant and positive eﬀects on
Applied Problems (t = 2.38, p= .02, 2-tailed), Passage Comprehension (t = 3.14,
p< .01, 2-tailed), Reading Fluency (t = 2.78, p= .01, 2-tailed), and Broad Reading (t
 = 3.02, p< .01, 2-tailed); higher levels of controlling involvement were related to
better academic performance.
For the younger children (aged 5–9 years, n= 28), their total PTSD symptoms score did
not moderate the eﬀect of TutorBright tutoring on any of the WJ-III subtests (all ps > .15,
2-tailed). However, total PTSD symptoms did have a signiﬁcant main eﬀect on Spelling (t
 = −2.08, p= .05, 2-tailed), such that young children with higher total PTSD
symptoms performed more poorly.
For children aged 10 and older (n= 42), total self-reported PTSD symptoms negatively
moderated the impact of tutoring on math ﬂuency (t = −2.15, p= .04, 2-tailed), such
that children with higher total PTSD symptoms performed more poorly than those with
lower total PTSD symptoms. PTSD symptoms also negatively moderated the impact of
tutoring on Mathematics Calculation, at the level of a trend, with children with greater
PTSD symptoms performing more poorly than those with lower PTSD scores (t
 = −1.72, p= .09, 2-tailed). A signiﬁcant main eﬀect of total PTSD symptoms was
also found on Mathematics Calculation (t = −3.32, p< .01, 2-tailed) and, at the level
of a trend, on Broad Mathematics (t = −1.70, p= .09, 2-tailed), with higher levels of
PTSD associated with poorer performance on both measures.
Results for exploratory research question 2: positive ‘spillover eﬀects’of tutoring
A series of ANCOVAs conducted via multiple regression revealed no signiﬁcant diﬀer-
ences between the adjusted post-test group means on any of the educationally relevant
domains examined, namely, child executive functioning, behavioural diﬃculties, or
caregiver involvement in academic activities in the home (all ps > .10).
OXFORD REVIEW OF EDUCATION 531
Overall, the children in care who participated in the study possessed below-average academic
skills, compared with children from the general population. At the pre-test, the children were
performing approximately 1 SD below the population mean on reading and 2 SDs below the
population mean on mathematics (see Table 4), an observation consistent with previous
ﬁndings (Trout et al., 2008). At the post-test, we found that the TutorBright programme had
had statistically signiﬁcant and positive eﬀects (p< .05) on Reading Fluency (g= .16), Reading
Comprehension (g= .34), and Mathematics Calculation (g= .39), marginally signiﬁcant eﬀects
on Broad Reading (p< .06) and Applied Problems (p< .06), and statistically non-signiﬁcant but
‘substantively important’eﬀects (i.e. above the WCC  criterion of a Hedges gof 0.25 or
more)onMathFluency(g= .27) and Broad Mathematics (g= 0.32). At the same time,
TutorBright did not signiﬁcantly close the children’s‘intra-achievement gap’between their
actual and potential achievement in mathematics and reading. Moreover, despite their gains
in mathematics and reading, the children in the tutoring group continued to perform at the
post-test at below-average levels in reading and mathematics (see Table 4).
The eﬀects of tutoring on children in care seen in the current randomised study are
reasonably consistent with the meta-analysis by Ritter, Barnett, Denny, and Albin (2009)of
randomised evaluations of tutoring with primary-schoolagedchildreninthegeneralpopula-
tion. Ritter et al. found small to moderate eﬀect sizes for one-to-one tutoring provided by
adult volunteers, with Hedge’sgs ranging from 0.18 (Reading Comprehension) to 0.41
(Reading Words and Letters). Our statistically signiﬁcant eﬀect on Mathematics Calculation
(g= .39) was larger than the non-signiﬁcant mean eﬀect (g= .26) of Ritter et al. for the same
domain (which, however, was based on a small number of eﬀect sizes).
The primary ﬁndings from the current study are also consistent with previous aca-
demic tutoring studies carried out with children in care. With individualised direct-
instruction tutoring by foster parents, Flynn et al. (2012) found that the children in the
tutoring group had made signiﬁcant gains on sentence comprehension, reading com-
posite, and mathematics computation, but not on word reading or spelling. Similarly,
using a small-group direct-instruction tutoring programme for foster children in care of
mainly Indigenous ethnic background, Harper and Schmidt (2016) found that tutoring
produced signiﬁcant gains on word reading, spelling, and mathematics, but not on
sentence comprehension. Together, these ﬁndings suggest that tutoring does improve
the mathematics and reading skills of children in care.
No signiﬁcant dose-response relationship was found between reading or mathe-
matics skills and the number of tutoring sessions (i.e. duration) or the number of
tutoring hours and minutes (i.e. intensity) experienced by the children, except for
a trend in the case of the number of tutoring sessions and letter-word identiﬁcation.
This result is consistent with previous meta-analyses in the general population that have
suggested that the length (duration) of tutoring is seldom a signiﬁcant predictor of the
size of intervention eﬀects (Elbaum et al., 2000; Suggate, 2010). On the other hand, it is
of potential pedagogical importance that negative relationships were found, at the level
of a trend, between the number of hours and minutes spent on the non-academic
activities of behaviour management or relationship building and the outcomes of
Reading Fluency, Math Fluency, and Broad Reading Composite. ‘Oﬀ-task’behaviour
may thus be associated with lesser improvement in reading and mathematics.
532 A. J. HICKEY AND R. J. FLYNN
Regarding potential moderating variables, Marquis (2013), like us, found that gender
moderated the eﬀectiveness of direct-instruction tutoring with foster children, with girls
tending to beneﬁt somewhat more than boys. In addition, we also identiﬁed two other
moderators; executive functioning and caregiver involvement. We suggest that the role
of potential moderators be investigated more systematically in future research. At the
same time, our examination of possible moderators involved a total of 72 ANCOVAs,
with a modest yield of only eight statistically signiﬁcant ﬁndings (p< .05, 2-tailed) and 6
results at the level of a trend (i.e. p< .10, 2-tailed), which included both moderators and
In other exploratory analyses, we found no evidence that TutorBright has a spillover
eﬀect on child executive functioning or behavioural diﬃculties or on caregiver involve-
ment in schoolwork in the home. These results are consistent with research by Marquis
(2013), Harper (2012), and Tideman et al. (2011), in which children in care experienced
little or no short-term improvement in mental health from tutoring. In children in the
general population, however, tutoring has been found to positively inﬂuence non-
academic outcomes such as disruptive behaviour or social skills (Bowman-Perrott,
Burke, Zhang, & Zaini, 2014). Future tutoring research with children in care should
thus continue to look for positive spillover eﬀects, while also including a qualitative
component in which caregivers and children are interviewed about such eﬀects.
Overall, TutorBright emerged as relatively eﬀective in improving the reading and
mathematics calculation skills of children in care. Also, the programme oﬀers important
logistical advantages to local child welfare organisations in that it recruits, trains,
deploys, and supervises its tutors in the latters’work with children in the home.
Moreover, the emphasis of TutorBright on creating a positive relationship between
tutor and tutee aligns well with the stated preference of children in care for academic
interventions that have a clear relational component (Evans et al., 2017).
The positive results that we found for TutorBright emerged despite several limitations of
our study. First, although we recruited participants for two full years, we were ultimately
able to enroll and retain only 70 children and their caregivers in the research sample, 34 in
the TurotBright and 36 in the waiting list control group. These numbers correspond closely
to the criterion that Coyne and Kwakkenbos’s(2013) consider to be the minimum require-
ment for randomised trials, namely, 35 participants in each of the intervention and control
groups. Larger samples would obviously be desirable in future studies. Second, the con-
siderable geographic distance (412 kilometres, about 5 hours by car) between our uni-
versity research centre and the county in Ontario where the collaborating child welfare
organisation is located, as well as limited project funding, made the task of assessing
implementation ﬁdelity relatively diﬃcult. With greater ﬁnancial resources, future research
could conduct more frequent and detailed ﬁdelity assessments (e.g. using video recording
with randomly selected tutored foster children). Future studies should also include
a process evaluation to assess ﬁdelity as well as programme implementation and greater
information regarding the caregiver and child experience of the programme. Finally, the
lack of previous evaluations of the eﬀectiveness of TutorBright provided us with little
programme-speciﬁc research guidance on how to proceed in the present study.
We believe that the eﬀectiveness of TutorBright could be enhanced, in several ways.
Most important, the programme creators should strengthen the theoretical and empiri-
cal links between TutorBright and the research that has already been conducted on
OXFORD REVIEW OF EDUCATION 533
tutoring in the general population and with children in care. The Best Evidence
Encyclopedia (http://www.bestevidence.org/) contains useful research syntheses on
mathematics and reading tutoring and related interventions that could easily be con-
sulted when the next version of TutorBright is being prepared.
TutorBright could also beneﬁt from placing a more explicit emphasis on direct-
instruction tutoring. Direct instruction has been found to be eﬀective in improving
educational outcomes in low-performance urban schools (Broman, Hewes, Overman, &
Brown, 2003) and, when combined with contingency management, enhances academic
outcomes for children at risk of school failure (Dolezal, Weber, Evavold, Wylie, &
McLaughlin, 2007). Although part of the TutorBright programme, direct instruction is
not its primary component. Instead, TutorBright emphasises ‘homework help’to
improve in-class performance (S. Verma, personal communication, June 2017). Little
research, however, has been conducted on the eﬀectiveness of homework help.
In addition, TutorBright could exploit the ﬁnding in the general population that parental
involvement in academics is often an important predictor of educational success (for meta-
analyses, see Hill & Tyson, 2009). Although little research on this speciﬁc topic has been
conducted to date with children in care, the latter do have a preference for interventions
delivered by caregivers (Evans et al. 2017), and Jackson (2007) has long been a strong
advocate of a much greater role for foster parents in the academic education of their foster
children. Thus, TutorBright would do well to promote the involvement of caregivers (e.g. by
having them read regularly with their foster children).
Finally, future research on TutorBright could seek to obtain teacher ratings of aca-
demic performance in the classroom, as programme components such as homework
help or relationship-building may improve the child’s in-class behaviour. Future research
could also try to answer two important questions that tutoring research, even in the
general population, has yet to address adequately: are gains in reading or mathematics
maintained, and how much tutoring is needed to bring children up to the ‘average’
range? Unfortunately given that this study was part of a doctoral thesis, a longer term
follow-up to assess the ‘staying power’of the intervention was not possible.
In conclusion, our ﬁndings are broadly consistent with the ﬁnding of Forsman and
Vinnerljung (2012) that educational interventions for children in care, and especially
tutoring, tend to produce positive results. However, many children in care who have
received tutoring still remain at risk of poor academic functioning. More research is
needed to discover which academic interventions work best for children in care, in both
the short and long term.
The authors would like to acknowledge the families who participated in the study as well as the
collaborating staﬀat the participating Children’s Aid Society and TutorBright.
No potential conﬂict of interest was reported by the authors.
534 A. J. HICKEY AND R. J. FLYNN
This research was supported by the secondary author’s (Robert J. Flynn) university research fund.
Notes on contributors
Andrea J. Hickey, PhD, C. Psych is a psychologist in supervised practice in Ottawa, Ontario. She
recently completed her PhD thesis under the supervision of Dr. Robert J. Flynn at the University of
Ottawa. Her thesis assessed the eﬀectiveness of three academic interventions for children in care.
She is interested in the evaluation of interventions that try to improve academic outcomes for
children in care as well as research that examines the factors that predict academic outcomes for
children at risk of academic diﬃculties.
Robert J. Flynn is an emeritus professor in the School of Psychology and a senior researcher at the
Centre for Research on Educational and Community Services at the University of Ottawa (Ontario,
Canada). Each year, he and his research team evaluate the service needs and psychological, social,
educational, and health outcomes of some 5,000 children, adolescents, and young adults residing
in foster, kinship, customary, or group care in Ontario. Annual feedback is provided to the province
and local Children’s Aid Societies. He and his students have also carried out several randomised
trials of academic tutoring that have shown that tutoring produces gains in reading and mathe-
matics in children in care.
Adams, G. L., & Engelmann, S. (1996). Research on direct instruction: 25 years beyond DISTAR. Seattle,
WA: Educational Achievement Systems.
Berlin, M., Vinnerljung, B., & Hjern, A. (2011). School performance in primary school and psycho-
social problems in young adulthood among care leavers from long term foster care. Children
and Youth Services Review,33, 2489–2497.
Borman, G. D., Hewes, G. M., Overman, L. T., & Brown, S. (2003). Comprehensive school reform and
achievement: A meta-analysis. Review of Educational Research,73, 125–230.
Bowman-Perrott, L., Burke, M. D., Zhang, N., & Zaini, S. (2014). Direct and collateral eﬀects of peer
tutoring on social and behavioral outcomes: A meta-analysis of single-case research. School
Psychology Review,43(3), 260–285.
Brännström, L., Vinnerljung, B., Forsman, H., & Almquist, Y. B. (2017). Children placed in out-of-
home care as midlife adults: Are they still disadvantaged or have they caught up with their
peers? Child Maltreatment,22(3), 205–214.
Briere, J. (1996). Trauma symptom checklist for children. Odessa, ﬂ: Psychological assessment
Briere, J., Johnson, K., Bissada, A., Damon, L., Crouch, J., Gil, E., & Ernst, V. (2001). The trauma
symptom checklist for young children (TSCYC): Reliability and association with abuse exposure
in a multi-site study. Child Abuse & Neglect,25, 1001–1014.
Coyne, J., & Kwakkenbos, L. (2013). Triple P-positive parenting programmes: The folly of basing
social policy on underpowered ﬂawed studies. BMC Medicine,11, 11.
Dietrichson, J., Bøg, M., Filges, T., & Klint Jørgensen, A. M. (2017). Academic interventions for
elementary and middle school students with low socioeconomic status: A systematic review
and meta-analysis. Review of Educational Research,87(2), 243–282.
Dolezal, D. N., Weber, K. P., Evavold, J. J., Wylie, J., & McLaughlin, T. F. (2007). The eﬀects of a
reinforcement package for on-task and reading behavior with at-risk and middle school stu-
dents with disabilities. Child & Family Behavior Therapy,29,9–25.
Elbaum, B., Vaughn, S., Tejero Hughes, M., & Watson Moody, S. (2000). How eﬀective are one-to-
one tutoring programmes in reading for elementary students at risk for reading failure? A meta-
OXFORD REVIEW OF EDUCATION 535
analysis of the intervention research. Journal of Educational Psychology,92, 605–619.
Evans, R., Brown, R., Rees, G., & Smith, P. (2017). Systematic review of educational interventions for
looked-after children and young people: Recommendations for intervention development and
evaluation. British Educational Research Journal,43(1), 68–94.
Flores, M. M., & Ganz, J. B. (2009). Eﬀects of direct instruction on the reading comprehension of
students with autism and developmental disabilities. Education and Training in Developmental
Flynn, R. J., Marquis, R. A., Paquet, M. P., Peeke, L., & Aubry, T. D. (2012). Eﬀects of individual direct-
instruction tutoring on foster children’s academic skills: A randomized trial. Children and Youth
Services Review,34, 1183–1189.
Forsman, H., Brännström, L., Vinnerljung, B., & Hjern, A. (2016). Does poor school performance
cause later psychosocial problems among children in foster care? Evidence from national
longitudinal registry data. Child Abuse & Neglect,57,61–71.
Forsman, H., & Vinnerljung, B. (2012). Interventions aiming to improve school achievements of
children in out-of-home care: A scoping review. Children and Youth Services Review,34(6),
Gliner, J. A., Morgan, G. A., & Harmon, R. J. (2003). Pretest-posttest comparison group designs:
Analysis and interpretation. Journal of the American Academy of Child & Adolescent Psychiatry,42,
Goodman, A., & Goodman, R. (2009). Strengths and diﬃculties questionnaire as a dimensional
measure of child mental health. Journal of the American Academy of Child & Adolescent
Harper, J., & Schmidt, F. (2016). Eﬀectiveness of a group-baed academic tutoring programme for
children in foster care: A randomized controlled trial. Children and Youth Services Review,67,
Harper, J. (2012). The eﬀectiveness of a group-based tutorial Direct Instruction programme for long-
term fostercare children: A randomized controlled trial. School of Psychology, Lakehead
University. Retrieved from http://knowledgecommons.lakeheadu.ca:7070/handle/2453/385
Hickey, A. J. (2018). Improving academic outcomes for children in foster care through tutoring or
working memory training: Three randomized trials (Unpublished doctoral thesis). School of
Psychology, University of Ottawa. Retrieved from https://ruor.uottawa.ca/simple-rearch?query=
Hill, N. E., & Tyson, D. F. (2009). Parental involvement in middle school: A meta-analytic assessment
of the strategies that promote achievement. Developmental Psychology,45(3), 740.
Jackson, S., & Ajayi, S. (2007). Foster care and higher education. Adoption & Fostering,31(1), 62–72.
Kavanaugh, B. C., Dupont-Frechette, J. A., Jerskey, B. A., & Holler, K. A. (2016). Neurocognitive
deﬁcits in children and adolescents following maltreatment: Neurodevelopmental conse-
quences and neuropsychological implications of traumatic stress. Applied Neuropsychology:
Lauer, P. A., Akiba, M., Wilkerson, S. B., Apthorp, H. S., Snow, D., & Martin-Glenn, M. L. (2006). Out-of
-school-time programmes: A meta-analysis of eﬀects for at-risk students. Review of Educational
Lightfoot, E., Hill, K., & LaLiberte, T. (2011). Prevalence of children with disabilities in the child
welfare system and out of home placement: An examination of administrative records. Children
and Youth Services Review,33(11), 2069–2075.
Maloney, M. (1998). Teach your children well: A solution to some of North America’s educational
problems. Belleville, ON: QLC Educational Services.
Marquis, R. A. (2013). The gender eﬀects of a foster parent-delivered tutoring programme on foster
children’s academic skills and mental health: A randomized ﬁeld trial (Unpublished doctoral
thesis). School of Psychology, University of Ottawa. doi:10.20381/ruor-3093
McGuire, A., & Jackson, Y. (2018). A multilevel meta-analysis on acaadmic achievement among
maltreated youth. Clinical Child and Family Psychology Review. doi:10.1007/s10567/-018-0265-6
536 A. J. HICKEY AND R. J. FLYNN
Naglieri, & Goldstein. (2012). Comprehensive executive function inventory. Toronto: Multi Health
O’Higgins, A., Sebba, J., & Gardner, J. (2014). What are the factors associated with educational
achievement for children in kinship or foster care: A systematic review. Oxford, UK: Rees Centre for
Fostering and Education, University of Oxford.
Pellegrini, M., Lake, C., Inns, A., & Slavin, R. E. (2018). Eﬀective programmes in elementary mathematics:
A best-evidence synthesis. Best Evidence Encyclopedia. Retrieved from http://www.bestevidence.
Ritter, G. W., Barnett, J. H., Denny, G. S., & Albin, G. R. (2009). The eﬀectiveness of volunteer tutoring
programmes for elementary and middle school students: A meta-analysis. Review of Educational
Schafer, J. L., & Graham, J. W. (2002). Missing data: Our view of the state of the art. Psychological
Schulz, K., Altman, D., & Moher, D. (2010). CONSORT 2010 Statement Updated guidelines for
reporting parallel group randomised trials. BMJ, 340–332.
Sebba, J., Berridge, D., Luke, N., Fletcher, J., Bell, K., Strand, S., . . . O‘Higgins, A. (2015). The
educational progress of looked after children in England: Linking care and educational data.
Oxford,: Rees Centre for Research in Fostering and Education and University of Bristol.
Slavin, R. E., Lake, C., Davis, S., & Madden, N. A. (2009). Eﬀective programmes for struggling readers:
A best-evidence synthesis. Best Evidence Encyclopedia, School of Education, Johns Hopkins
Suggate, S. P. (2010). Why what we teach depends on when: grade and reading intervention
modality moderate eﬀect size. Developmental Psychology,46(6), 1556–1579. doi:10.1037/
Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (6th ed.). Boston, MA: Pearson
Tessier, N. G., O’Higgins, A., & Flynn, R. J. (2018). Neglect, educational success, and young people in
out-of-home care: Cross-sectional and longitudinal analyses. Child Abuse & Neglect,75, 115–129.
Tideman, E., Vinnerljung, B., Hintze, K., & Aldenius Isaksson, A. (2011). Improving foster children’s
school achievements: Promising results from a swedish intensive study. Adoption & Fostering,35
Trout, A. L., Hagaman, J., Casey, K., Reid, R., & Epstein, M. H. (2008). The academic status of children
and youth in out-of-home care: A review of the literature. Children and Youth Services Review,30
Viner, R. M., & Taylor, B. (2005). Adult health and social outcomes of children who have been in
public care: Population-based study. Paediatrics,115, 894–899.
What Works Clearinghouse ([WWC]). (2008). Procedures and standards handbook (version 2.0).
Washington, DC: US Institute of Educational Sciences.
Woodcock, R., McGrew, K., & Mather, N. (2001). Woodcock-Johnson tests of achievement manual.
Zetlin, A. G., Weinberg, L. A., & Shea, N. M. (2006). Seeing the whole picture: Views from diverse
participants on barriers to educating foster youths. Children and Schools,28, 165–173.
OXFORD REVIEW OF EDUCATION 537