ArticlePDF Available

Abstract and Figures

The Health Committee at The Royal Swedish Academy of Sciences has initiated two State of the Science Conferences to address the following two questions: “Has the mental health of children and adolescents in Sweden changed over time?” and “Which are the causal relationships between mental health and academic achievement?” The two conferences Trends in child and adolescent mental health and School, educational achievement and mental health among children and adolescents, take place in April 2010 and are following the model for State of the Science and Consensus Conferences set up by the NIH, the National Institute of Health at the US Department of Health and Human Services. The NIH-model means that an independent panel of experts is evaluating the current level of knowledge related to questions formulated in advance in order to produce a formal written statement based upon systematic literature reviews, expert statements and open discussions. The present report covers a mapping of and a systematic literature review on the theme: School, learning and the mental health of children and adolescents. A group of experts, appointed by the Academy’s Health Committee, and recruited from a group of highly qualified scientists representing different disciplines related to the topics, has been working with the report since the spring of 2008. The Health Committee and its Planning Group are very grateful to the experts for their excellent achievements. Arne Wittlöv Per-Anders Rydelius Chairman of the Chairman of the Planning Group
Content may be subject to copyright.
, SE-1
5 STOCKHOLM, S WEDEN TEL +46 8 673 95 00, FAX +46 8 15 56 70
School, Learning and
Mental Health
A systematic review
Gustafsson, J.-E., Allodi M. Westling, Alin Åkerman, B.,
Eriksson, C., Eriksson, L. Fischbein, S., Granlund, M.,
Gustafsson, P. Ljungdahl, S., Ogden, T., Persson, R.S.
© Kungl. Vetenskapsakademien 2010 ISBN 978-91-7190-138-5
School, Learning and
Mental Health
A systematic review
Gustafsson, J.-E., Allodi M. Westling, Alin Åkerman, B.,
Eriksson, C., Eriksson, L. Fischbein, S., Granlund, M.,
Gustafsson, P. Ljungdahl, S., Ogden, T., Persson, R.S.
Review group Professor Britta Alin Åkerman
Associate Professor Mara W. Allodi
Professor Charli Eriksson
PhD Lilly Eriksson
Professor Siv Fischbein
Professor Mats Granlund
Professor Jan-Eric Gustafsson
Associate Professor MD Per Gustafsson
PhD Sofia Ljungdahl
Professor Terje Ogden (until 2009-10)
Professor Roland S. Persson (until 2009-09)
Advisory group The Health Committee
Associate professor Curt Hagquist
Professor Per-Anders Rydelius
PhD Sten Anttila, Socialstyrelse
MEng Agneta Pettersson, SBU
Professor Måns Rosén
Chair Professor Jan-Eric Gustafsson
Coordinator Associate Professor Mara W. Allodi
Assistant MLIS Anna Björnberg
Title Barns och ungdomars psykiska hälsa i Sverige
Copyright Kungl. Vetenskapsakademien 2010
ISBN 978-91-7190-138-5
Layout Helena Ledmyr
Datum för publicering April 2010
This report shall be cited as: Gustafsson, J.-E., Allodi M. Westling, Alin Åkerman, B. Eriksson, C.,
Eriksson, L. Fischbein, S., Granlund, M., Gustafsson, P. Ljungdahl, S.,
Ogden, T., Persson, R.S. (2010). School, learning and mental health:
a systematic review. Stockholm: The Royal Academy of Sciences,
The Health Committee
Contact Health Committee Curt Hagquist,
Contact review group Jan-Eric Gustafsson,
Address Kungl. Vetenskapsakademien, Box 50005, 10405 Stockholm
2.1.1 Aspects of mental health 12
2.1.2 Aspects of schooling and learning environments 14
2.1.3 Population 15
2.1.4 Design 15
2.1.5 Criteria for exclusion 15
2.1.6 Criteria for inclusion 16
2.5.1 Quality of the included reviews 23
2.5.2 Age spans of children 24
2.5.3 Countries represented in the review 25
2.5.4 Measures and statistical methods used in included studies
2.5.5 Contents of the reviews 26
2.5.6 Summary of the review of reviews 26
2.6 MAPPING 40
2.6.1 Purpose of the reports 40
2.6.2 Mental health and well being aspects 42
2.6.3 Educational characteristics 45
2.6.4 Background and context 48
2.6.5 Method 48
2.6.6 Age and type of school 49
2.6.7 Relevance 50
2.6.8 Context of the study 50
2.6.9 Publication year 51
2.6.10 Clusters 51
6 7
3.2.1 Description of the studies 60
3.2.2 Data-extraction procedures 62
4.1.1 Trajectories of development of academic achievement and mental health 69
4.1.2 Effects of early academic achievement 76
4.1.3 Effects of academic achievement in adolescence 87
4.1.4 Effects of peers and teacher behaviour 98
4.1.5 Effects of grade retention 104
4.2.1 Effects of internalizing problems 106
4.2.2 Effects of externalizing behaviour problems
4.2.3 Effects of positive mental health factors 126
4.3.1 Effects of academic achievement on mental health 132
4.3.2 Effects of mental health on academic achievement 134
4.3.3 Reciprocal relations between academic achievement and mental health
4.3.4 Stability of problems of academic achievement and mental health 136
4.3.5 Protective factors and risk factors in the school situation 138
4.3.6 Resilience and resources 139
4.3.7 Limitations 141
5.2 AIM 152
5.3 METHOD 153
5.3.1 Criteria for inclusion of studies in the review 154
5.3.2 Search strategy 154
5.3.3 Methods of the review 155
5.6.1 General experiences of mental health and well-being
5.6.2 Protective experiences in educational environments 166
5.6.3 Risky experiences in educational environments 170
5.6.4 Poor relationships with the teacher 180
5.6.5 Important choices and opportunities 181
5.6.6 Specific risks 182
5.7.1 Gaps in research 185
5.7.2 Reflections about the method and the theoretical model
5.7.3 How children and youth talk about their well being at school 187
5.7.4 Correspondences of results from the reviews 188
6.2.1 Pre-school 199
6.2.2 Pre-school class 200
6.2.3 Early school years 200
6.2.4 Final years of comprehensive school 203
6.2.5 Upper secondary school 205
The Health Committee at The Royal Swedish Academy of Sciences has initiated
two State of the Science Conferences to address the following two questions: “Has
the mental health of children and adolescents in Sweden changed over time?” and
“Which are the causal relationships between mental health and academic achieve-
The two conferences Trends in child and adolescent mental health and School,
educational achievement and mental health among children and adolescents, take
place in April 2010 and are following the model for State of the Science and Con-
sensus Conferences set up by the NIH, the National Institute of Health at the US
Department of Health and Human Services.
The NIH-model means that an independent panel of experts is evaluating the
current level of knowledge related to questions formulated in advance in order
to produce a formal written statement based upon systematic literature reviews,
expert statements and open discussions.
The present report covers a mapping of and a systematic literature review on the
theme: School, learning and the mental health of children and adolescents.
A group of experts, appointed by the Academy’s Health Committee, and recruited
from a group of highly qualified scientists representing different disciplines re-
lated to the topics, has been working with the report since the spring of 2008.
The Health Committee and its Planning Group are very grateful to the experts for
their excellent achievements.
Arne Wittlöv Per-Anders Rydelius
Chairman of the Chairman of the Planning Group
Academy’s Health Committee
2 3
The current report describes results from a systematic review of research on school,
learning and mental health, conducted on the initiative of the Royal Swedish Academy
of Sciences.
1.1 Purposes
The eld of research on relations between schooling and mental health is multi-
disciplinary and not very clearly dened. Furthermore, previous systematic reviews have
concluded that the amount of research, which investigates relations between different
aspects of schooling and mental health, is limited. Thus, based on a mapping of research
on relations between schooling and mental health, Harden et al. (2001) concluded that
the research base was too limited to allow an in depth review of anything but young
people’s views on mental health and schooling. However, given that this mapping
was restricted to include British research, and that the research base is likely to have
grown considerably during the decade that has passed since this literature search was
performed, it may be expected that a mapping of the international literature yields more
relevant studies.
Thus, the rst main purpose of the review is to conduct a mapping of research on
relations between schooling and mental health. This involves investigating which
different aspects of schooling and mental health have been focused upon, and which
different combinations have been studied. Other characteristics of the research that
are interesting to map, is to see which age groups have been investigated, and which
research methods have been used.
The second main purpose of the review is to conduct in-depth syntheses of research
concerning one or more issues. While choice of specic issues to be focused upon was to
be determined on the basis of the results of the mapping, the planning committee for the
review determined at an early stage that one set of problems to be analyzed was the effect
of academic achievement on mental health, and the effect of mental health on academic
achievement. Previous research has shown that academic achievement is associated
with well-being and good mental health and that school failure may cause mental health
problems. At the same time, there are ndings which show that mental health may affect
the educational results achieved by the students. Thus, research is available which shows
that there are relations between academic achievement and mental health, but it is not
clear in what way academic achievement affects mental health and vice versa. There also
are reasons to believe that the relationships between school achievement and mental
health are inuenced by individual factors, such as vulnerabilities and abilities, and by
environmental factors, such as parents’ expectations and developmental experiences in
the family environment as well as the functioning of the school system.
The main purposes of the in-depth narrative synthesis are to answer to the following
The project utilized the valuable information and communication resources available through
the Libraries at Stockholm University and Karolinska Institutet.
The group wishes to tank in particular the Faculty of Social Sciences, the Department of
Special Education at Stockholm University, and the Department of Women’s and Children’s
Health at Karolinska Institutet, which made possible the collaboration of reviewers from dif-
ferent universities and agencies that was a prerequisite for this project.
Funding agencies
The Health Committee is funded by grants from the following organisations:
Bristol Myer Squibb
Familjen Erling-Perssons Stiftelse
FAS - Forskningsrådet för arbetsliv och socialvetenskap
Kungl. Vetenskapsakademien
Riksbankens Jubileumsfond
Stiftelsen Clas Groschinskys minnesfond
Stiftelsen Kempe-Carlgrenska Fonden
Stiftelsen Marcus och Amalia Wallenbergs minnesfond
Stiftelsen Sven Jerrings Fond
Svenska Läkaresällskapet
4 5
- Which are the causal relationships between mental health and academic
- How are these relationships inuenced by other factors, both related to the
individuals and their social background, and to factors in the educational environment
(evaluation system, tests, grades, selection procedures, special educational system,
teaching methods, and social climate).
- If the results from the mapping indicate that research is available to support in-
depth reviews of other issues this will also be done if time and resources permit.
The third main purpose of the review is to review research which has investigated
experiences and perceptions of Swedish children and adolescents concerning their
mental health and well being, in particular as this is related to their school attendance
and situation. One of the reasons for this was that the systematic review by Harden et. al
(2001) strongly emphasized the ethical necessity to give space to children’s and youths’
own perspectives and perceptions about their health and well being. Another reason
was that this would give an opportunity to investigate if conclusions from the in-depth
review are applicable to the Swedish context.
The two rst purposes were approached through a systematic search in bibliographic
databases of international peer-reviewed research conducted within a large number
of disciplines. The third purpose was investigated with a review of research using
qualitative methods focussing on Swedish students’ perceptions of school and mental
1.2 Organization and participants
The work has been conducted within a project group that was formed in August 2008
and which nished its work in March 2010. The project group included the following
Jan-Eric Gustafsson (chair), Professor of Education, University of Gothenburg.
Mara Westling Allodi (project coordinator), Associate professor of Special Education,
Stockholm University.
Britta Alin Åkerman, Professor em. of Special education Stockholm University.
Charli Eriksson, Professor of Public health, Örebro University. Fråga Charli vad hans
titel blir på engelska är osäker
Lilly Eriksson, PhD, Public Health Planning Ofcer, Swedish National Institute of
Public Health
Siv Fischbein, Professor em. of Special Education, Stockholm University.
Mats Granlund, Professor of Psychology, Professor of Disability Studies, Jönköping
Per Gustafsson, MD, Associate professor of Child and Adolescent Psychiatry, Linköping
Soa Ljungdahl, PhD, Public Health Planning Ofcer, Swedish National Institute of
Public Health.
Terje Ogden, Professor of Educational Psychology, Director of Research, Atferdssenteret,
Norway (until October 2009).
Roland S. Persson, Professor of Educational Psychology, Jönköping University (until
August 2009).
The group scheduled monthly meetings or video-conferences to plan and discuss the
strategies to employ. The results of the searches and documents produced were made
accessible to the members through Internet resources.
The work on the two rst purposes was conducted in several steps, and in each step the
entire project group participated. In the rst step, from August to December 2008, search
questions and review protocols were developed, and tools for managing references and
protocols were selected and developed. The searches in the bibliographic databases
were performed January to March 2009. The abstract review, in which a rst screening
of references was made, began in March 2009 and was concluded in June 2009. The full
text review started in July 2009 and was concluded in November 2009. In this step a
further screening was made, along with coding of included references for the mapping.
During the period October 2009 to February 2010 data-extraction was conducted of
references potentially relevant for the second purpose. From January to March 2010 the
narrative synthesis was conducted.
The work on the third purpose also was conducted in several steps, and the responsibility
for all of these was taken by the project coordinator, associate professor Mara Westling
In addition to work on these three main purposes, the project group has conducted
several activities, which have not yet been reported in full. Thus, a review of reviews
identied in the searches of bibliographic databases is being conducted by some of
the members of the project group (Mats Granlund, Lilly Eriksson, Roland S. Persson,
and Mara Westling Allodi), supported by PhD Lena Almqvist, Mälardalen University
and Karin Bertills, Jönköping University. A background chapter describing Swedish
educational reforms after 1945 has been written by PhD Mac Murray.
The work of the project group has been supported by a large number of collaborators:
6 7
This chapter describes the search questions used, the searches carried out, and the
mapping of the resulting references.
2.1 Defining the searches
The rst step of the literature search was to dene a set of search terms, and a logic to
combine them. This is described below.
2.1.1 Aspects of mental health
Mental health and illness is a complex area, and all aspects of the constructs are not
well dened. However, the group took its starting point in the denition proposed by
the WHO (2003, 2005a). Child and adolescent mental health is dened as the capacity
to achieve and maintain optimal psychological functioning and well being. This capacity
is directly related to the competence achieved in psychological and social functioning.
Child and adolescent mental health includes a sense of identity and self-worth; sound
social relationships; the ability to be productive and to learn; and the ability to use
developmental challenges and cultural resources to maximize growth.
Child and adolescent mental disorders manifest themselves in many domains and
in different ways. A mental illness or disorder is diagnosed when a pattern of signs
and symptoms is identied that is associated with impairment of psychological and
social functioning, and that meets criteria for disorder under an accepted system of
classication such as the International Classication of Disease, version 10 (ICD-10,
WHO, 1992) or the Diagnostic and Statistical Manual IV (DSM-IV, American Psychiatric
Association, 1994). A major distinction is made between externalizing and internalizing
problem behaviours. Externalizing problem behaviour refers to behavioural problems,
such as conduct disorders, aggressiveness, and antisocial behaviour, or attention decit
and hyperactivity. These kinds of behaviours consist of negative emotions directed
against others, such as anger, aggression, frustration, and fear. The prevalence of
externalizing disorders among children varies between 2% and 15%, depending on
the denitional criteria (Hinshaw, 1992). In some denitions, externalizing problems
include only antisocial behaviour, while in others hyperactivity and attention decit are
also included.
Internalizing problem behaviour, on the other hand, refers to emotional problems, like
depression and anxiety. In internalizing problem behaviour, negative emotions are
directed at oneself rather than others. Psychosomatic symptoms, such as headaches and
abdominal pain, are also often seen as part of internalizing problem behaviour.
It has been recognized that there is a need to explicitly consider positive aspects of
mental health, such as emotional, psychological and social well being. The indicators of
Anna Björnberg, MS Library and Information, project assistant;
Jan Sydolf, MS History, project assistant;
Martin Angeland, project assistant;
Karl Berglund, project assistant;
Viveka Vessberg, librarian, Stockholm University;
Marie Lövgren, librarian, Stockholm University;
Lotten Häggström, librarian, Stockholm University;
Christine Wickman, information scientist , Karolinska Institutet; and
Susanne Gustafsson, librarian, Karolinska Institutet.
Consultants and advisors have generously provided their support:
Agneta Pettersson, MEng and Måns Rosén, Professor of Epidemiology and Public
health,SBU The Swedish Council on Technology Assessment in Health Care; and
Sten Anttila, PhD National Board of Health and Welfare.
8 9
independent work, homework , test, grades, assessment. summative evaluation,
curriculum based assessment, national standards, high stakes, minimum
competency testing.
Factors related to grouping of students: selection, ability grouping, big-sh-little-
pound effect, tracking system, elite classes, age-mixed groups.
Factors related to special provisions: special education, individual plans, special
education, inclusive education, referrals, labelling.
Factors related to individual failure: school failure, drop out, truancy, school
absenteeism, risk factors, perceived stress, adjustment.
Factors related to relationships: Relations teacher-student, relations with peers;
Relations in school (climate, connectedness, belonging); bullying, victimization,
violence, harassment in educational settings.
Factors related to school organisation: organisation, leadership; management,
administration, funding, accountability systems.
Factors related to characteristics of the educational system: rules, reward,
discipline or punitive systems, grade transition.
Factors related to changes in the educational system: reforms, effects of reforms
and changes.
To be included it was required that a reference had terms referring to some aspect of
academic achievement, which restriction was imposed to increase the possibility to
obtain a tractable number of references within the main scope of the review.
2.1.3 Population
The studies included were to investigate children and adolescents in the age interval
2-19 years, in educational contexts from pre-school to upper secondary education.
2.1.4 Design
Studies to be included were meta-analyses and systematic reviews, and empirical
studies. Emphasis was put on studies with longitudinal design, even though this was
not a requirement.
2.1.5 Criteria for exclusion
Studies reporting interventions are not included, since they are the object of another
systematic review.
Studies concerned exclusively or principally with substance abuse were not included.
Studies with a selected population composed of only children and adolescents with a
diagnosis were not included.
positive mental health are not so well established as are the symptoms of mental illness.
Indicators at the individual level include measures of a sense of belonging, self-esteem,
engagement, self-determination and control and quality of life.
WHO (2005b, 159-160) suggests that safe and supportive environments, and social and
learning environment of high quality in which children’s skills and accomplishments are
acknowledged and valued, are indicators of positive mental health at the organizational
level. In our conceptual framework, these factors were seen as aspects of the educational
environment and were kept separate from the mental health aspects.
It was thus decided that search terms would be identied that could locate research
focusing on the following aspects of mental health:
Internalising symptoms: anxiety, depression, self-harm
Externalising symptoms: hyperactivity and concentration problems, conduct
Other psychiatric symptoms
Positive aspects of mental health
2.1.2 Aspects of schooling and learning environments
The complexity of dening relevant aspects of schooling and learning environments is
even greater than the complexity of dening mental health. Previous research has given
indications of characteristics of educational environments that are more successful in
introducing youth into adulthood, and in counteracting the negative effects of other
external or background factors (Rutter, Maughan, Mortimore, & Ouston, 1979; Rutter
& Maughan, 2002). It has been shown that harsh school discipline is not productive;
instead frequent rewards, praise, and appreciation should be used. Schools should be
aware of the importance of a pleasant and comfortable environment. Pupil participation
and possibilities to take responsibility for their school lives as well as emphasis on
academic matters and teacher expectations of student success with clear and attainable
goals also contributes to better results. Teacher practices and school management must
take into consideration mutual values and norms, and consistently work to realize these,
thus the ethos of the school seems to be of great importance.
This research suggests search criteria focusing on aspects of the school and learning
environment such as school climate, and relations among students and teachers. Search
terms reecting different aspects of academic achievement were also included, as were
terms reecting organizational aspects of the schooling.
Eight categories of indicators of the educational situation and learning environment
were identied. They are structured with an order from the micro- to the macro-level
that is broadly inspired by the bio-ecological model of Bronfenbrenner (1999, 2000)
Factors related to individual activity, results, learning: achievement, results;
teaching, instruction, teaching methods; curriculum goals, teacher behaviours,
10 11
1. Search in the databases ASSIA, Sociological abstracts, Social services Abstracts
performed in CSA at Stockholm university (2009-03-13)
Terms Number of References
Population 331163
Mental health 317700
Educational aspects 102564
Method 45415
Combined 4254
Exported (only peer reviewed) 3880
2. Search in EBSCO multiple databases: Academic Search Premier, Business
Source Premier, Communication & Mass Media Complete, EJS E-Journals,
ERIC, Research Starters – Business, Research Starters – Education, only
peer-reviewed (2009-03-25)
Terms Number of References
Population 1.703.189
Mental health 1.523.403
Educational aspects 1.579.480
Method 656.882
Combined (exported) 8494
3. Search in Medline through OVID (2009-02-27)
The database Medline was searched with different and adapted strategies that
resulted in 3493 unique references
4. Search in Psycinfo, Psycarticles etc. through OVID (2009-2-27)
The search in Psycinfo contributed 3013 references.
The search proles required that at least one term from each of the four groups of terms
should be found. We adapted the sensitivity of the search proles to the characteristics
of the databases. For databases lacking a consistent indexing system, we used a broader
search prole. In these cases, we searched the term not only as Descriptor or Subject,
but also in Title and Abstract. When we thought we could trust the index system of the
database, we made more specic search proles.
2.1.6 Criteria for inclusion
Research published in English in peer reviewed journals.
No limitation for publication years was applied in the literature searches. However, in a
second phase, only references published from 1999-2009 were included.
2.2 Literature searches
The methodology of the literature search and the choice of tools was developed on the basis
of previous experiences in the eld. We took advantage of work done internationally, at
the EPPI-Centre (e. g., Harden, et al., 2001), at the Social Care Institute for Excellence
(SCIE; e.g., Coren & Fisher, 2006), and at the Centre for Reviews and Dissemination
(CRD; Rodgers, Arai et al.). We also took advantage of work conducted in Sweden at the
Swedish National Board of Health and Welfare, and at the Swedish Council on Health
Technology Assessment (SBU).
During the fall of 2008, several preliminary literature searches were made in order to
investigate indexing features of the bibliographic databases and the terms employed
in the literature in order to optimize the search strategies. This made it clear that the
search strategies had to be adapted to the logic of the different databases.
In collaboration with librarians at Stockholm University, the multiple databases CSA
and EBSCO were searched. These databases cover research literature in broad social
science elds, including education, psychology, sociology, management, social work,
social psychology, among others.
At the Karolinska Institutet library the Medline database provided access to literature in
the eld of child and adolescent psychiatry, public health, paediatrics, and medicine. The
search in Medline was made through OVID. Searches of the Psycinfo and Psycarticles
databases also were made at Karolinska Institutet through OVID.
The complete search proles are very long and complex. In order to save space they are
not fully reported here, but are available in electronic form via the link xxx, or via the
authors of this report.
In the box below some details of the search results are described. The search proles
build upon the combination of four groups of terms covering a) mental health concepts;
b) educational concepts; c) population and d) methodology.
12 13
2.3 Abstract screening
The 10 717 references were screened for relevance on the basis of title and abstract by
two reviewers for each abstract. This work was conducted during the period March to
June 2009 by a group of ten reviewers belonging to the project group (see section 1.2).
The abstracts were rst sorted by publication year in seven sets (08-09; 07; 06; 05;
04; 02-03; 99-00-01) and then sorted alphabetically by title. This procedure made it
possible to identify duplicates that had not been found previously, for instance because
the author’s name was missing. Each pair of reviewers received an assignment from
every one of the seven sets, each one consisting of approximately 200-300 abstracts.
The protocols, instructions and procedures were rst tested with the abstracts of papers
that were identied as literature reviews (n=792) and a revised protocol was evaluated
on a small sample of references.
Each reviewer lled out a protocol in table format for each abstract reviewed, deciding
about inclusion and exclusion. If the abstract was included, information about mental
health aspects, educational aspects, methodology and sample was entered as well. If
the abstract was excluded, the reviewer entered the code or codes for the applicable
exclusion criterion (Not English, Not peer-reviewed, Not children or adolescents, Not
focus on school and health, intervention study). All the completed protocols from the
two reviewers were stored in a common workplace storage in Mondo/Sakai and the
inclusion codes were registered in user-dened elds in RefWorks.
The agreement between the reviewers in their judgements about inclusion and exclusion
has been calculated for each set of abstracts. The average agreement within pairs of
reviewers was 93.4%.
References that were included by at least one of the reviewer were included in the full
text screening (N = 1033).
2.4 Full text screening
In the next step, the 1033 references remaining after the abstract screening were screened
for relevance on the basis of scrutiny of the full text of the references. This work was
done during the period of July to November 2009, by a group of nine reviewers from
the project group (see section 1.2). The references were divided into seven lists with
about 150-200 in each, on the basis of publication year and were sorted alphabetically
by author name.
The references were available to the reviewers online through RefWorks via links to
Stockholm University library databases and services. The other documents, protocols
and materials produced by the group members were stored and made available online
through the common workplace/storage Mondo/Sakai at Stockholm University.
References that were not available online were ordered as paper copies through the
Karolinska Institutet library or Stockholm University library and sent directly to the
We performed tests and analysed the lists of references in order to understand the
outcomes of our search strategies, and adjusted the search proles on the basis of
these evaluations. In particular, the group decided to improve the educational and
population terms referring to young children, in order to increase the sensitivity of
the literature searches in the databases. The search proles were nally optimized in
cooperation with teams of librarians at Stockholm University (CSA and EBSCO) and at
Karolinska Institutet (OVID). The list of references retrieved were successively exported
to a database developed with RefWorks - an online research management, writing and
collaboration tool which is available through Stockholm University.
The information and reference resources at Stockholm University and Karolinska
institutet are valuable and powerful tools in performing literature searches. However,
the search proles that we developed were too complex for the capacity of the CSA
database and we met several serious difculties, and particularly so with functions to
save searches and with functions to export references. This made it necessary to search
Psycinfo and other databases with OVID instead.
In EBSCO it was possible to perform the complex literature search and to save it
permanently. However, the automatic export function could not easily handle the
amount of references that we wished to put in the databases in RefWorks. The export of
references therefore required more time than planned.
The searches with OVID could be performed effectively and the direct export of
references was successful. There also were several options available for the export that
could be employed. On the basis of these experiences we can conclude that OVID had
the capacity necessary to adequately support our searches, that EBSCO had the capacity
but did not offer all the features that we needed, and that CSA did not have the capacity
to perform the searches that we needed to do. However, we did search some databases
with CSA, since they were not available with other means.
The searches in the bibliographic databases resulted in 20584 references that were
exported to RefWorks (RW) (see the reference ow, Figure 2.1). From this collection
of references, duplicates that could be identied with the duplicate search tools in RW
were excluded (N=4320). We removed also 792 reports that were literature reviews,
systematic reviews, meta-analysis, since these were to be analysed separately.
Some other categories of references were also excluded from the original set: reports
that could be identied as written in another language than English (N= 145); not peer-
reviewed (N=228); or lacking an abstract, even after manual searches for abstracts in
the original databases and in journal sites (N=396).
Considering the large number of remaining references, papers published before 1999
(N=3995) were excluded as well. The exclusion of these references was based on the
expectation that possible interesting studies that had been published before 1999 would
be covered in the literature reviews (N= 792) that were to be reviewed separately.
After these removals, the set included 10717 references.
14 15
The reviewers worked pair wise on the references in the rst group of about 200
references and compared their results in order to calibrate their evaluations. The
members of the pairs then proceeded individually in their review of references, splitting
the groups between them. However, in order to keep evaluation standards calibrated a
subset of references from each group was evaluated jointly. The reviewers evaluated on
average 100 articles each (Min = 33, Max = 198).
On the basis of the protocols completed by the reviewers, more than 50 percent of the
references were excluded, mainly because their content was not judged relevant for
the scope of this review (N= 446) and, to a lesser extent, because of other criteria (see
Figure 2.1). If the reference did not meet at least one of the required inclusion criteria,
it was excluded. All the codes from the protocols, both included and excluded, were
registered in the database in RefWorks and were also imported into SPSS in order to
make possible further analysis.
Characteristics of the included references (N = 471) are described in section 2.6. The
next section presents a short report from the systematic review of the 792 reviews that
were identied among the retrieved references.
2.5 Review of reviews: quality and contents
One aim of the “review of reviews” was to analyze the quality and content of previous
reviews, in order to gain knowledge about aspects of form and content important for
the current review. As has already been mentioned the searches identied 792 titles/
abstracts. One additional review was identied later and added. The abstracts were
reviewed for relevance using an inclusion/exclusion protocol. This step resulted in 148
included reviews. These were read in full-text by a special group of reviewers using
an exclusion/inclusion protocol developed for rating of reviews. Altogether 37 out of
the 148 reviews (25%) were included for further quality rating and mapping. These
studies were categorized after method quality (high, medium, low), type and quality of
information provided and content, i.e. type of relationship between school factors and
mental health investigated.
2.5.1 Quality of the included reviews
Based on an integration of recommendations for method quality indicators for systematic
reviews (Auperin et al, 1997; Schlosser et al, 2007) all 37 reviews included for mapping
were rated for quality according to the criteria displayed in Table 2.1.
Figure 2.1 Flow of literature from the literature searches to the mapping phase
*more than one reason of exclusion is possible, but the majority of the
excluded articles do not have a specific focus on mental health and school
16 17
As seen in Table 2.2 many reviews concerning mental health and school factors cover
the age spans from 10 years and up. Few cover children in the lower age spans.
2.5.3 Countries represented in the review
Values and attitudes toward school as well as school systems show variations between
countries. Reviews may have a certain bias by being based primarily on studies
representing a certain country. For each review included the text has been searched
for information concerning from what countries included studies have collected their
empirical data.
Table 2.3 Number of reviews that have included studies from 1, 2 or 3 or more countries
Number of
3 or more 2 1 No information
Quality rating In
Number 8 8 1 1 4 3 24 15
Twenty four reviews have been assigned to the category “no information”. The reference
lists indicate that these reviews primarily or solely have used North American studies
for their review. In all studies that have named more than one country as the basis for
their review also North American studies are included. The result indicates a strong bias
towards North American studies.
2.5.4 Measures and statistical methods used in included
Less than 50% of the studies provided information on the measures and/or on the
statistical methods used in the included studies. Thus, the reviews in general provided
little opportunities to draw causal conclusions about the relations between school
factors and mental health.
2.5.5 Contents of the reviews
A preliminary analysis of the content of the reviews based on two dichotomies was made.
First, if the focus of the review was on symptoms of problems with mental health or
positive aspects of mental health. Second, if the focus of the review was on information
from individual children or on information from school level. The result revealed that
more reviews focus on data obtained from individual students than on data obtained on
school level. In addition, there was a weak trend for more high method quality reviews
focusing on problems with mental health than on positive aspects of mental health.
The reviewers wrote free text summaries of which relation between school factors
and mental health that were the focus of the review. The summaries were categorized
independently by two judges based on the same criteria that were used in reviewing
individual empirical studies. The categories used for school factors were: achievement,
Table 2.1 Number of included reviews that satisfied each method quality criteria
Criteria Yes No Do not
The authors describe how the quality of the review had been
established 14 22 1
A system for rating quality have been used 7 30
The review has an explicit aim and/or research questions 31 6
Systematic review (yes responses to 1-3 above) 12 24 1
Search strategies are reported with search words and search
strings for each data base 18 19
Inclusion/exclusion criteria are explicitly stated 22 15
A structured extraction protocol was used 12 23 2
The review contains a tabulated display of included studies 15 22
Table 2.1 indicates that the quality of the included studies when judged against these
criteria is relatively low. Few studies describe how review quality has been established,
and very few (7 out of 37) have used a structured system for rating quality of reviewed
studies. In addition relatively few studies (12 out of 37) have used a structured extraction
protocol for included studies. In total 12 out of the 37 reviews were rated as systematic
In addition to rating each review with the quality criteria above the reviewers assigned
an overall quality rating to each review. This overall method quality rating was used to
classify the reviews as high, low or medium in method quality. The overall classication
was in the next step used to compare low method quality reviews with medium and high
method quality reviews in terms of the type and quality of the content presented in the
reviews. For all 37 reviews information was collected regarding time period covered
by the review, age span of target groups of children, number of studies included in
the reviews, countries represented in the reviews, assessment and statistical methods
used in the included studies, and statistical methods used in the included studies. The
proportion of reviews with low versus medium or high method quality providing these
types of information was compared.
2.5.2 Age spans of children
The age spans of children on which each review have their focus are displayed in Table
Table 2.2 Number of reviews including children from different age spans
Agespan 2-5 years 6-9 years 10-15 years 16-19 years
Quality In
total Medium
High In
total Medium
High In
total Medium
High In
total Medium
Number 8 4 23 16 30 21 26 19
18 19
peer relations, teacher relations, bullying/victimization, school organization, school
environment, specic risk groups and demographic factors. Certain summaries could
be assigned to more than one category. The categorization revealed that many reviews
focused on the relations between achievement and mental health. In total 22 of the
37 reviews whereof 16 with a medium or high quality had this focus. Relatively many
reviews also had a focus on the relations between aspects of the school environment and
mental health (9 whereof 6 with moderate to high method quality). Few if any reviews
had their focus on the other categories.
2.5.6 Summary of the review of reviews
The review of reviews shows that the research performed within the scope of the present
study can be grouped around three main themes: the relationship between positive
aspects of mental health (e.g. well being, self-concept) and learning; the reciprocal
relationship between problems with mental health and schooling; and the study of
the indirect relations between school factors, mental health outcomes and learning
outcomes. In the rst case the mental health aspects are seen mainly as determinants
of learning outcomes; in the second case the main concern may be to examine the
consequences of general educational experiences as attainment and failure, but also
of specic features as tests and methods, on negative aspects of mental health of the
students (e.g. anxiety, depression). In the third case the school factors are seen as
mediators or moderators of outcomes both on mental health and achievement. The
conceptualisations emerging in these reviews contributed to an understanding of the
characteristics of this eld of knowledge and these insights have been considered in the
subsequent phases of the main literature review process.
The four reviews displayed in table 2.4. that focused on the relations between self-beliefs
and achievement provide a consistent picture in that positive (high/strong) self-beliefs
predict subsequent achievement in general and in writing and mathematics especially.
The review by Valentine et al. (2004) as well as the one authored by Ma and Kishor
(1997) suggest that the level of specicity at which self-beliefs are measured is a more
important consideration than the particular type of self-system component that such
beliefs resemble. The more the self-belief instrument focuses on specic aspects of self-
beliefs, e.g. skills in mathematics, and/or specic situations, e.g. test taking, the better
self-beliefs predict subsequent achievement. These results indicate that students must
be supported in evaluating their own skills in relation to important individual topics in
school. The mechanisms through which self-beliefs act on subsequent achievement is
not clearly described in the reviews, thus it is difcult to draw any conclusions about how
students’ self-beliefs can be supported and intervened with. The two reviews displayed
in table 2.4. that focused on any impact of forms of summative assessment on students’
motivation for learning (Harlen & Crick, 2003) and the impact of participation in
school-based extracurricular activity on adolescent development (Feldman & Matjasko,
2005) provide more diverse information concerning the relations between positive
aspects of mental health and achievement. Harlen and Crick (2003) do not explicitly
discuss achievement.
Review Purpose Number and
context Result relevant for this review Authors conclusions
Klassen, R. (2002). Writing
in early adolescence.
A review of the role of
self-efficacy beliefs.
Educational Psychology
This review attempts to explain
the role of self-efficacy in a
specific developmental period
and in a specific domain. Another
purpose is to examine the
selected research for differences
in self-efficacy beliefs associated
with grade level, gender and
16 studies
representing the
Many measures of self-efficacy are too general to be
Self-efficacy beliefs play an important role in predicting
writing achievement in early adolescence.
Gender appeared to influence efficacy beliefs, but not
performance, with boys ratings their writing confidence
higher than girls
Adolescence is an important
transition period, with physical,
social and academic challenges
resulting in a sense of loss of
personal control for many young
people. Belief in one’s efficacy
to perform academically is often
depressed in that period, and writing
tasks often suffers. Research
that explores self-beliefs in this
context builds understanding of
the mechanisms that influence
Klassen, R. (2002). A
question of calibration: A
review of the self-efficacy
beliefs of students with
learning disabilities.
Learning Disability
To examine how students with
learning disabilities-who have
been shown to display poor task
analysis and meta-cognitive skills
– calibrate their efficacy beliefs
with criteria tasks
22, primarily
studies from the
Self-efficacy ratings were predictive of subsequent
functioning and increase in conjunction with intervention
and subsequent performance increases. Gender
differences and differences between LD and non-LD
students offered few obvious trends. Investigations into the
domain of writing showed the most consistent problems
with calibration. This is less obvious in mathematics
Some students with LD overestimate
their efficacy to complete writing
Valentine, J., DuBois,
D., & Cooper, H. (2004).
The relationship
between self-beliefs and
academic achievement.
A meta-analytic review.
Educational Psychologist
This article uses meta-analysis to
synthesize findings of longitudinal
investigations of the relation
of self-beliefs to academic
55 whereof
35 USA and
20 Western
Among equally achieving students, having positive self-
beliefs confers a small but noteworthy advantage on
subsequent achievement measures relative to students
who exhibit less favorable self-beliefs. Self-beliefs
pertaining to the academic domain represent a more
important influence on achievement than global or general
beliefs and feelings about self.
Overall there is encouraging
evidence of a contribution of self-
beliefs to achievement as well as
a considerable potential for the
magnitude of this potential to be
underestimated due to various
methodological limitations of extant
Ma, X., & KIshor, N.
(1997). Attitude towards
self, social factors,
and achievement
in mathematics: A
meta-analytic review.
Educational Psychology
Investigating the relationship
between attitude towards self and
social factors with achievement in
143 studies and
2 syntheses.
Different ethnic
groups and some
variations in
The mean effect size was 0.23. It suggests that the
relationship between self-concept and achievement
is statistically reliable. No gender differences, but
age differences, junior high (14-15 years is especially
important). A wide variation of the self-concept –
achievement relationship across ethnic groups.
Self-concept has surfaced as
a critical factor in a person’s
academic success. Self-concept
is content specific in it’s relation to
Table 2.4 Characteristics and results from 6 reviews investigating the relations between positive aspects of mental health and achievement and learning
20 21
Harlen, W., & Crick,
R. (2003). Testing and
motivation for learning.
Assessment for learning
To identify evidence of any impact
of testing and other forms of
summative assessment on
students’ motivation for learning
19 studies, 1
Canada, 1 Israel,
1 Marocco, 8 UK,
Precise/sequential learning disposition is related to high
self-esteem and internal locus of control. Students who
achieve low on test tend to develop lower self-esteem.
Feedback in terms of grades is difficult to understand,
especially for young students. Feedback is a significant
factor influencing willingness to invest effort in a task.
Greater effort is associated with learning goals than with
performance goals. Promoting task involvement by giving
task related, non-ego involving, feedback may promote the
interest ad performance of most student’s. Low achievers
are affected negatively by testing. Girls are more likely than
boys to have high self-esteem in classrooms favouring
sequential learning. Teachers are affected by summative
testing in the direction of spending more time of instruction
in activities related to tests and less time on learning
through inquiry and problem-solving.
Draw attention to the small number
of studies that were found to offer
dependable evidence to address the
question posed in this review. There
are important reason for serious
attention to motivation for learning
as an outcome of education
Feldman, A., &
Matjasko, J. (2005)
The role of school
based extracurricular
activities in adolescent
development: A
comprehensive review and
further directions. Review
of Educational research
In this review we integrate
findings from across sociology,
education and psychology
to provide a comprehensive
picture of rates of school-
based extracurricular activity
participation, the consequences
of such participation, and
the current understanding of
mediators and moderators of the
effects of participation.
45 studies, US
studies only
Activity participation in structured activities has a positive
relationship with both achievement, educational aspirations
and psychological well being. These relationships are
mediated by peer networks and supportive relations with
Activity participation has many
positive influences on adolescent
and young adult outcomes. However,
it is necessary to measure activity
participation in a thorough manner,
to use a comprehensive theoretical
framework to model the influence of
Review and type
of mental problem Purpose Number and
context Result relevant for this review Authors conclusions
Mental Health overall
DeSocio, J, & Hootman, J
(2004) Children’s mental
health and school success. The
Journal of School Nursing
To examine the current
state of children’s mental
health, its impact on
school success, and the
implications for school
Not stated. US
Bidirectional links for most mental health outcomes,
described as degree of symptoms, between mental
health and achievement
The review supports the confluence
of school performance concerns with
emerging and existing mental health
problems in children. Youth with school
problems and those who fail or drop out of
school are at greater risk for poor mental
health outcomes
Trout, A., Nordness, P., Pierce,
C., & Epstein, M. (2003).
Research on the Academic
Status of Children with
Emotional and Behavioral
Disorders. A review of the
literature from 1961 to 2000.
Journal of Emotional and
Behavioral Disorders
A comprehensive
examination of the current
state of the literature on
the academic status of
students with emotional
and behavioural disorders
65, Reference list
indicate primarily
USA studies
Few studies separate ethnicity and gender. No
conclusion about SES can be drawn. Few studies
of EBD student in main stream settings. Outcomes
often measured with tests rather than grades. Little
information about students that perform on grade
level. In the 23 comparisons to students without
disabilities students with EDB generally performed
less well that their peers without disability
This review further supports the notion
that students with EDB are often
academic underachievers.
Hembree, R. (1988) Correlates,
causes, effects and treatment
of test anxiety. Review of
Educational research
To integrate the findings
of the research on test
anxiety, regarding its
nature, effects, and
562 Primarily
US studies but
also studies from
countries with
English as native
Test anxiety and performance are significantly related
at grade 3 and above. The relationships are inverse
and tend to be stronger for worry than emotionality.
No differences appear between males and females.
Test anxiety causes poor performance.
This conclusion follows from the finding
that better performance accompanies TA
Ma, X. (1999) A meta-
analysis of the relationship
between anxiety towards
mathematics and achievement
in mathematics education.
Journal for Research in
What is the relationship
between math anxiety and
math achievement?
26 USA, but also
New Zealand,
Australia, Thailand,
The common populations correlation for the
relationship between anxiety toward mathematics
and achievement in mathematics was-.27. The
relationship between mathematics anxiety and
mathematics achievement is consistent across
gender groups, grade level groups, ethnic groups,
instruments used to measure anxiety and years of
This meta analysis shows support for the
findings of significance of the relationship
between mathematics anxiety and
mathematics achievement for school
students. Reduction in anxiety may cause
improved mathematics achievement
Suicide and depression
Evan, H., Hawton, K., &
Rodham, K. (2004). Factors
associated with suicidal
phenomena in adolescents: A
systematic review of population
based studies. Clinical
Psychology Review
A systematic review of
the evidence for specific
risk and protective factors
for suicidal phenomena
in adolescence based on
community studies
USA and UK
A significant but indirect association between
achievement and suicide attempts. Poor school
attendance was associated with both suicide attempts
and suicide ideation. A negative attitude towards
school and school work was associated with an
increased prevalence of suicidal phenomena. A strong
relationship between poor peer relationships and
suicidal ideation. For suicide attempts there was an
association with problems in relationships with peers
The findings in this review indicate a likely
association between suicidal phenomena
and many school related variables
(achievement, attendance, attitude toward
school). An intervention program aimed at
parents have often focused on children’s
academic performance, they may well
have an impact on suicidal phenomena.
Table 2.5 Characteristics and results of the 10 reviews focusing on problems with mental health and achievement
22 23
Antisocial behaviour
Brier, N. (1995) Predicting
antisocial behaviour in
youngsters displaying poor
academic achievement:
A review of risk factors.
Developmental and
Behavioural Pediatrics
To examine factors that are
believed to effect the risk of
an anti-social outcome in
youngsters who are failing
at school
Not stated
Hyperactivity, distractibility, ridgidity, and impulsivity
linked to both delinquency and poor academic
Low intelligence (especially verbal) affect prosocial
aspirations in regard to education.
Among youngsters failing at school those
with the strongest risk for delinquency are
characterized by overactive temperament,
relatively low intelligence, negative school
attitude and parents who demonstrate
inadequate supervision.
Drug and substance use
Fletcher,A., Bonell, C., &
Hargreaves (2008) School
effects on young people’s drug
use: A systematic review of
intervention and observational
studies. Journal of Adolescent
Examined the hypothesis
that school institutional
factors influence young
people’s use of drugs
USA, Australia, and
Intervention studies provide some evidence that
there is a causal association between modifying the
school environment to increase student participation,
improve relationship, promote a positive school ethos
and address disaffection and truancy and a reduction
in student drug use and other risk behavior. A lack
of effect on girls that might be caused by statistical
power problems. Longitudinal studies show that
disengagement from school and poor teacher –
student relations were associated with subsequent
drug use and other risky behaviors after adjustments
for students demographic characteristics, SES and
prior drug use.
The studies reviewed offer reasonably
consistent evidence in favor of school
effects on rates of drug use among
Dewey (1999) reviewing the
relationship between school
factors and substance use for
elementary, middle school
and high school students. The
Journal of primary Prevention
A summary of school-
related correlates and
risk factors that have
been shown to relate
to substance use in
elementary, middle and
high school populations.
Not explicitly
The majority of reviewed research indicated a causal
sequence whereby substance use preceded and was a
risk factor for negative school related consequences
such as lower grades, higher rates of absenteeism,
lower educational expectations and high rates of
drop out. However a number of studies established a
reverse order of GPA, high school non-completion etc
acting as risk factors for drug use.
The statistical relationships found in
the reviewed analyses, however strong,
suggestive, or consistent, can establish an
antecedent, but not a causal connection..
However, the fact that these relationships
are consistent and have been shown to be
stable over two decades suggests these
variables’ stability as risk factors and their
viability as targets for preventive work
Conduct problems/class room
Gaddy, G. (1988) High
School order and academic
Looking at the evidence on
order and achievement for
high school students.
Number not
specified. Studies
from the USA and
There is no clear causal relationship between order
and achievement but some indications that order,
if combined with an academic focus of the school,
promotes achievement. This tendency is much
stronger on individual level but probably exists also on
school level
More research is need on school level
focused on the causal relationship
between order and achievement. These
studies need to be longitudinal and have
more stringent ways to collect data and
treat data
Bender, W.N, & Smith, J.K
(1990) Class room behavior and
children and adolescent with
learning disabilities. A meta
analysis. Journal of Learning
To compare the classroom
behaviours of children and
adolescents with learning
disabilities to children
without learning disabilities
Not explicitly
stated as deduced
from reference list
North American
For on-task behavior the analysis demonstrated
an average discrepancy of slightly over one SD
between the groups. About the same effect size for
off-task behavior, conduct disorder, shy/withdrawn
and distractibility. Results were about the same
independent of study quality. Differences a bit
larger if based on teacher ratings than if based on
1) Children with LD have a higher
probability to behave disruptively in the
class room
2)The magnitude of the differences
indicate that both teacher ratings and
observations can be used when identifying
these children
3) The ratio of children with learning
disabilities should not be too high in
regular classrooms
Review Purpose Number and
context Result relevant for this review Authors conclusions
Fletcher A., Bonell C.,
Hargreaves J.
(2008). School Effects
on Young People’s Drug
Use: A systematic Review
of Intervention and
Observational Studies
Journal Of Adolescent
Identify effects of school-level
changes on drug use
Explore possible mechanisms by
which school-level influences on
individual drug use might occur.
10 USA, Australia,
The Netherlands,
Interventions studies: there is a causal association
between changes in the school environment (ethos,
engagement, participation) and reduction of risk
behavior and drug use, especially for boys and
especially for early interventions in secondary school.
Observational studies (a large number of):
disengagement from school and poor teacher-student
relationship were associated with subsequent drug
use, and risk behavior. Low school connectedness,
truancy and suspension were associated with higher
rates of drug use 2-4 years later.
Fill an important research gap
by systematically reviewing
high-quality quantitative studies
examining school effects on young
people’s drug use as well as those
examining how individuals’ drug
use relates to their experiences of,
and attitudes to, school.
Suldo S., Riley K., Shaffer E.
(2006). Academic Correlates
of Children and Adolescents
Life Satisfaction
School Psychology
Literature review to provide a
comprehensive summary of the
multiple school-related correlates
of life satisfaction. Also intended
to identify limitations in existing
knowledge for future studies
Not stated,
many American
(more than 150
schools), UK,
Norway, Finland,
Latvia, China,
Being treated fairly, supportive teachers and feelings of
safety are important components of school climate
No relationship intelligence – global happiness
Students’ perception of academic
ability, teacher support and overall
satisfaction with school strongly
related to children’s global life
Feldman, A., Matjasko, L.
(2005). The Role of School-
Based Extracurricular
Activities in Adolescent
Development: A
Comprehensive Review and
Future Directions
Review of Educational
Review literature on school-based
activity participation patterns
focusing on academic achievement,
substance use, sexual activity,
psychological adjustment,
delinquency and young adult
USA, Sweden (1)
Structured school-based extracurricular activity
participation is associated with positive adolescent
developmental outcomes
There is evidence that activity
participation has positive influence
on development and outcome.
Extent of impact needs refined
inquiry. Suggest a comprehensive
theoretical framework to include
social networks and supportive
adult relationships.
Maddox S. & Prinz R.
(2003). School bonding in
children and adolescents:
Assessment, and Associated
Clinical child and family
psychology review
To review conceptualization,
measurements, theories of school
Not stated.
Mainly American,
one from New
School-bonding is a multidimensional concept. The
social development model establishes school bonding
as a protective factor promoting positive life outcomes
through its effects on opportunities for pro-social
External constraints, demographic factors and
individual factors affect the youth’s opportunities
and skills for involvement. This affects amount of
reinforcement, which promotes attachment. This in
turn affects attitudes and subsequent behaviors e.g.
academic performance
In prevention programs
conceptualization of school
bonding should include the
domains attachment to school,
attachment to personnel,
school commitment, and school
Table 2.6 Characteristics and results of the 6 reviews that focus on the indirect relations between school factors, and mental health and learning
24 25
Dewey, J.(1999). Reviewing
the Relationship Between
School Factors and
Substance Use for
Elementary, Middle and High
School Students
The Journal of Primary
Review school-related variables
(such as academic performance,
absenteeism, and noncompletion)
related to substance use.
Mostly the USA,
1 West Germany
(Berlin), 1 Canada
Greater number of risk factors connected to increased
substance use.
Higher grades – lower drug use and vice versa
Several studies indicate causal relationship
between GPA, absenteeism, educational aspirations,
noncompletion and substance use, others don’t.
School climate (teachers, peer students, environment,
undisciplined atmosphere) plays an important role in
substance use, but considering neighborhoods, peer
and family variables, school functions as a mediator.
The statistical relationship can
establish an antecedent, but not a
causal connection
La Paro K., Pianta R. (2000).
Predicting Children’s
Competence in the Early
School Years: A Meta-
Analytic Review
Review of Educational
Estimate cross-time relations
between children’s early school
outcomes and measures of their
skills and abilities in preschool or
Early assessments make small to moderate effects to
the predictability of children’s early school success.
Effect size moderate for predicting academic/cognitive
assessment (explains 25% of the variance)
Provides empirical support that
defining and assessing “readiness”
in other terms than skills and
abilities would add important
information to current practices
They describe the effect of forms of summative assessment on motivation for learning.
In the reviewed articles it is difcult to discriminate descriptions of motivation from
descriptions of achievement, because changes in achievement are seen as indirect
indications of changes in motivation for learning. Thus, the review provides indirect
indicators of mental health factors that are important for achievement. The authors use
terms such as self-esteem, locus of control and self-efcacy in their results and discussion
sections, which makes it possible to relate the results to the reviews focused on self-
beliefs. Concerning personality precise/sequential learning disposition in students is
related to high self-esteem and internal locus of control. Concerning classroom factors,
feedback is a signicant factor inuencing students’ willingness to invest effort in a
task. In addition, greater effort is associated more with learning goals than performance
goals. The review focusing on frequency of participation in extra-curricular school
activities (Feldman & Matjasko) tries to relate participation in extracurricular activities
both to achievement and psychological well-being. One problem is that participation in
activities can be seen as a sign of well-being. Accordingly, investigating the relationship
between participation and well-being is partly circular. However, a relatively strong
relation between participation in extracurricular activities and achievement is also
reported. This relationship is mediated by peer networks and adult support.
The review results displayed in table 2.5. indicate that several types of problems with
mental health are related to achievement and learning. Whether these relations are
causal and the direction of such cause-effect relations are less evident. However, several
of the reviews (e.g DeSocio et al, 2004; Dewey, 1999; Hembree, 1988; Ma, 1999, and
Trout et al. 2003) conclude that there are causal relations between mental health
problems and achievement and that these relations are bidirectional. The same patterns
is seen independent of type of mental health problem (mental health overall, anxiety,
depression, conduct problems) This indicates that it may be more a matter of mental
health problems interfering with learning and achievement and vice versa, than that
there exist specic relations between mental health and achievement and learning. An
alternative explanation is that studies of the relations between mental health problems
and achievement require more concrete and specic measures of how aspects of mental
health problems functionally interfere with learning and achievement (Dewey, 1999;
Ma, 1999) as well as a broader span of achievement measures (Trout et al, 2003). Most
mental health measures are related to diagnoses, e.g. ADHD, depression and conduct
disorder, rather than how symptoms related to these diagnoses functionally interfere
with learning. Most achievement outcomes employed in the studies are represented by
result on tests and/or reading, writing and mathematics. Few achievement outcomes
concern other subjects, e.g. science, wood craft and social sciences, and relatively few
concern actual grades.
In several of the reviews in table 2.6. that focus on the indirect relations between school
factors and mental health, both classroom factors and other school factors are included.
Concerning classroom factors the two most frequently mentioned are teacher-student
relationship and student-peer relations. Three reviews (Dewey, 1999; Fletcher et al, 2008;
Suldo et al, 2006) report that teacher-student relations are related to low drug use and
global life satisfaction, and one study reports that the quality of peer relations is related
26 27
to low drug use. It is however from these studies difcult to draw any conclusions about
causality and the direction of cause-effect relations. The relatively few review results
related to classroom factors as well as the lack of studies allowing causal conclusions to
be made indicate that future research need to focus on longitudinal studies investigating
the relations between class room factors and mental health. Five of the six reviews
displayed in table 15 (except for LaParo et al.) explicitly discuss the relation between
school factors outside the classroom and mental health. They use partly different terms
such as ethos, participation, school climate, school connectedness, school attachment
and school bonding. These terms seem to be partially overlapping conceptually and are
difcult to discriminate. It might be, as for self-evaluations, that it is more important
to dene the ecological level at which they are measured than to dene what kind of
student school perception they stand for. Presuming that the concepts are all aiming
at approximately the same environmental level, the discussed terms are describing a
student’s general perception of school, in one way or another. Common to the reviews
are that they report a negative relationship between these school factors and problems
with mental health, and a positive relationship between these factors and positive life-
outcomes. It is however difcult to draw any conclusions about causality. There are
at least two reasons for these difculties, the rst is that few of the reviews makes a
clear distinction between longitudinal and cross sectional studies. The second is that the
positive life outcomes are described in general terms. There seem to be few, if any, good
alternatives to achievement in describing functioning in school that does not overlap
with terms classied as positive aspects of mental health.
2.6 Mapping
We now go into the mapping of empirical studies identied in the full text review. Of
the 1033 references that were full-text screened, 471 were deemed by the reviewers as
falling within the scope of this review (see Figure 2.1) and potentially relevant for data-
extraction. A description of the characteristics of the studies that resulted from the
analysis of the completed protocols is presented in this section (the protocol employed
for the full text screening is available in the Appendix, section 8.1).
The purposes of the papers were classied by the reviewers according to nine dened
purposes and an open alternative to be dened. The same paper could be classied with
as many purposes that were relevant.
The most frequent purpose identied by the reviewers was “Effects of mental health
on achievement and learning” (P2 in the protocol) which was found in more than half
of the papers (N=255). Many papers, about a third, were concerned with the effects of
background factors on mental health and achievement (P9) (N=160). The purpose of
investigating the effects of achievement and learning on mental health and well being is
present in about a third of the papers (N=153). The other purposes were all represented
by a number of references in decreasing order from peer effects on mental health (P3,
about a quarter), to effects of school climate (P7, present in 16 percent of the papers),
teacher effects on mental health and well being, and antecedents and consequences of
victimisation (P4 and P5, identied in 12 percent of the papers).
Figure 2.3. Reports included in the mapping phase (N=471) and their distribution
according to kind of purpose (more than one purpose is allowed for the same report)
28 29
The least represented purposes were the effects of educational aspects on the mental
health and well being of risk groups (P8) and the effects of various organisational
aspects (P6). These were identied in about 10 percent of the included papers.
For the intent of this review it is a positive result that all the purposes are represented
by a number of papers and that the two purposes that correspond to the main theme of
the review are well represented in the included papers. The papers with the purpose of
studying the effects of background factors are concerned with the study of confounders,
from the point of view of the aims of this review.
We looked at the combinations of purposes with cross tabulations. About 35 % of the
papers having the purpose P2 (effects of mental health on achievement) have also been
identied with the purpose P9 (effects of background factors). About 30% of the reports
having the purpose P1 (effects of achievement on mental health) also have P9. More
than 50% of the papers having the purpose P1 (effects of achievement on mental health)
were also identied with the purpose P2 (effects of mental health on achievement),
meaning that these two purposes are relatively often treated in the same studies. P3 and
P5 proved to be purposes that are partially but not completely overlapping. In fact, just
about 20% of the papers having P3 (effects of peers) was identied also with P5 (effects
of victimisation), while the overlap between P4 (teacher effects) and P7 (climate effects)
was somewhat higher, about 50%. In summary, the analysis of the review shows that
the categories of purposes that we identied in the protocol were relevant for this set of
referencesand that there was a certain overlap between the categories of purposes, but
they apparently also served to identify distinct types of studies.
2.6.2 Mental health and well being aspects
The aspects of mental health and well being were categorised in the protocol with 23
codes. The reviewers were instructed to identify all the possible relevant themes that
were presented in the reviewed study. The occurrence of these categories in the included
references is presented in Figure 2.4.
The most frequently represented themes are the positive aspects of mental health
dened as self-perception, self-esteem, and self-concept (C4b). These aspects were
identied by the reviewers in more than 35 % of the references. Among the frequently
identied aspects we nd, in descending order, internalising symptoms C1a, (mood,
depression), externalising symptoms C2f, (antisocial disorder, violence) C2d (conduct
disorder) and internalising C1b (anxiety). These aspects were identied in about 25 %
to 35 % of the references.
A second group of mental health aspects was identied in less than 20% and more than
10% of the references. These are the other externalising aspects dened in the protocol
as attention (C2e), hyperactivity (C2c) and impulsivity (C2b) and also a positive aspect:
well being (C4a).
Figure 2.4. Occurrence of each mental health aspect in the papers included in the mapping (N=471), in percent (more than one aspect was allowed)
A third group of mental aspects was identied in just more than 5 % of the references:
coping (C4b) and the internalising aspects psychosomatic disorder (C1f) and stress
The categories in the group Other psychiatric symptoms (C3) were not well represented
among the references. One of these, tics, was not represented at all. An explanation of
the low number of references coded with C3 is that during the abstract review and the
full text review we required a normal population or at least a comparison group, if the
study had a special population.
Table 2.7 Rotated component matrix of the mental health aspects identified in the included articles (N=471).
The mental health aspects with means below 0.04 are excluded from the analysis. Extraction Method: Principal
Component Analysis. Rotation Method: Varimax with Kaiser Normalization (only loadings >0.40 are showed)
In order to describe how the mental health aspects co-occur in the included articles, we
performed factor analyses of the codes (see Figure 2.4) from the full text protocols. In
the factor analysis we exclude the mental health aspects with means below 0.04. We
retained in the factor analysis 17 of 23 mental health aspects.
The total variance explained is about 44 per cent. The rst factor explains about 16%
of the variance and all the externalising aspects except C2f (antisocial, violence) load
on this factor. However, anxiety also has small loadings on the rst factor. The second
factor explains about 10% of the variance and all the internalising mental health aspects
load on this factor. The third factor explains about 8 % of the variance and all the
positive mental health aspects load on this factor. The fourth factor explains about 8%
of the variance and antisocial disorders load on this factor, along with suicidal and self-
harm behaviour (C2d, C2e). The fourth factor is also negatively related to aspects of self-
Rotated component matrix
1 2 3 4
C1a internalising: mood, depression .56
C1b internalising: anxiety .43 .53
C1d internalising: sleep disorders .53
C1e internalising: stress .55
C1f internalising: psychosomatic symptoms .65
C1h internalising: suicidal- and self harm .40 .41
C2a externalising: hyperactivity .79
C2b externalising: impulsivity .74
C2c externalising: attention .73
C2d externalising: conduct disorder .62
C2e externalising: oppositional behaviour .61 .
C2f externalising: antisocial disorders, violence .64
C4a positive aspects: psychological well-being .61
C4b positive aspects: self-esteem, etc. . -.58
C4b positive aspects: coping .61
C4b positive aspects: resilience .41
C4b positive aspects: mastery .57
concept and self-esteem (C4b). This mean that the papers that are related to violence
and antisocial disorder are not concerned with self-concept and self-esteem.
It is quite interesting to see that the rst three factors quite closely correspond to the
three major aspects of mental health: externalising, internalising and positive aspects.
There is also some overlap between these major categories. Anxiety can be identied
in references that focus on externalising mental health aspects, but apparently not on
those on antisocial disorders. Suicidal and self-harm behaviour can be related also to the
externalising theme antisocial disorders and violence. The fourth factor is characterised
mainly by externalising mental health aspects, but also of an internalising suicidal and
a positive aspect. Nevertheless, this pattern suggests that this type of research tends to
represent each of the major areas of mental health with multiple indicators, but that there
is not so much overlap between the aspects. However, to investigate how the different
aspects of mental health combine in different studies an analytical technique which uses
the individual paper as a unit, such as cluster analysis, would be more appropriate.
The educational characteristics were categorised in the full text protocol with 17
categories sorted into seven main groups (D1-D7). The denition of the categories build
on the identication of the educational process and its outcomes as complex phenomena
that can be investigated at the individual level, at the level of the interaction between
teacher and student, at the level of didactic activities, at the level of interaction in the
group of students in the class and in the school, at the level of the school organisation
and management, and at the level of the national regulations (curriculum, grades,
modality of selection).
Furthermore, a range of combinations of approaches in the study of educational
processes are possible: for instance the phenomenon of student failure can be considered
as an outcome at the individual level of an interaction between individual factors and
the educational system. However, some studies falling in this category may have a
main interest in the identication of risk factors at the individual level, while others are
interested in which features of the educational system may contribute to an undesirable
increase of school failure at the national level. Other studies may have an interest in the
consequences of the experiences of school failure for the individual.
Group D1 was related to individual aspects and related concepts of the educational
process that are used in the research in this eld and that the studies are indexed with
(achievement, teaching, instruction, teaching behaviour, goals, homework, independent
work, test, grades, evaluations). D2 is concerned with issues of selection and ability
grouping; D3 with special educational issues as referral, labelling, individual plans,
inclusive education; D4 with school failure, drop out, truancy, perceived stress, school
adjustment; D5 with the relationship in the educational environment (teacher-students,
peers, school climate, harassment); D6 treats organisational aspects as leadership,
management, funding, accountability systems, educational reforms, reward and
discipline, grade transitions. D7 is concerned with the educational system in a broad
sense at the national level; international comparison also t in here.
32 33
The instructions asked the reviewers to code each study with all the appropriate codes.
The content of the categories was discussed at the group meetings.
The most common educational theme in the studies reviewed was Achievement (D1a).
This may mean that some measure of student achievement is employed in the study,
not necessarily that the main focus of the study is on the learning process as such. The
second most common theme was Relations with peers (D5b) that was identied in more
than 30 % of the references.
A third group is constituted by themes that were identied in about between 20% and
15% of the references. They are the following, in descending order: School adjustment,
Tests and assessment, Relations teacher-student, School failure, Relations and climate,
Bullying and victimisation. A fourth group is represented by themes that appear in more
than 5% of the references: Perceived stress, Grade transition, Teaching and instruction.
The last group includes six themes that were identied in less than 5% of the references:
Special education, Selection and ability grouping, School organisation, Reward and
discipline, Educational reforms, National educational systems.
Table 2.8 Rotated component matrix of the Educational aspects identified in the included articles (N=471). The
mental health aspects with means below 0.04 are excluded from the analysis. Extraction Method: Principal
Component Analysis. Rotation Method: Varimax with Kaiser Normalization (only loadings >0.40 are showed).
Table 2.8. presents the results of a factor analysis conducted in order to investigate co-
occurrence of different educational aspects in the studies. The total variance explained
by this analysis is 63 per cent, and each of the six factors explain about 10 per cent of
the total variance.
Figure 2.5 Occurrence of each educational themes in the studies included in the mapping (N=471) in percent.
1 2 3 4 5 6
D1a Educational aspects: Achievement -0.77
D1b Educational aspects:
Teaching, instruction, methods 0.80
D1c Educational aspects:
Test, grades assessment 0.56
D3 Educational aspects:
Special education, individual plans 0.83
D4a Educational aspects:
School failure and dropout 0.81
D4c Educational aspects: Perceived stress 0.82
D4d Educational aspects: School adjustment 0.78
D5a Educational aspects:
Relations teacher-student 0.78
D5b Educational aspects: Relations with peers 0.73
D5c Educational aspects:
Relations climate, connectedness 0.52
D5d Educational aspects:
Bullying, victimization, violence, 0.70
D6d Educational aspects: Grade transition 0.65
34 35
Teaching, instruction &methods and Relations teacher-student have high loadings on
the rst factor, so this may be interpreted as a classroom teaching factor. The second
factor contrasts Achievement and Bullying, victimization, violence, which indicates
that these aspects are not included in the same paper. On the third factor Special
education, Tests, grades and assessment load highly. School adjustment and Relations
with peers load highly on the fourth factor, which could represent the broad theme of
social achievement. The fth factor loads on Stress and Grade or school transitions
while School failure and dropout and Relations climate &connectedness load highly
on the sixth factor. This factor analysis gives indications about how the educational
themes that are present in the references are combined together. It seems therefore that
the following themes could be identied: Teachers and teaching; Peer victimisation;
Assessments and provisions; Social adjustment; Changes and transitions; and Failure
and participation.
2.6.4 Background and context
Section G in the full text protocol was introduced in order to assess if the studies
reviewed made comparisons between groups or took into account individual,
background and context characteristics that could be moderators of the relationship
between mental health and educational variables. Out of the papers reviewed in full
text, 295 are concerned with differences among students, such as differences in ability
and background, gender etc. There are 106 papers that are concerned with individual
risk factors, pre-term born, malnutrition, abuse and neglect, and violence. There are
also 66 papers that are identied as investigating other differences (for example: twin
studies, urban and rural schools, race and ethnicity).
2.6.5 Method
Figure 2.6. Occurrence of each type of methodology in the studies included in the mapping (N= 471) in percent.
The literature searches included longitudinal studies as a search term, even though this
was not a requirement. The high presence of longitudinal studies is therefore expected.
In this section the papers could also be classied into more than one category. Out of
43 papers identied as using a multilevel approach, 34 used a longitudinal approach.
A large part (41 of 48) of the studies with a comparative approach is also longitudinal.
Figure 2.7 Occurrence of papers with participants from various ages and type of schools (N= 471) in percent.
The largest age group/type of school was students aged 10-15 (B1c) and attending middle
school and junior high school. This age group of students was focussed upon in 72 % of
the papers. The age groups 6-9 and 16-19 were represented in about 40% of the papers.
The age group 2-5 was represented in about 20% of the papers.
2.6.7 Relevance
The reviewers were asked to evaluate the relevance of the report examined for the specic
purposes of this review with two questions. The reviewers identied 284 studies with
a specic focus on themes that are highly relevant for the systematic review and 362 as
having only a broad relevance.
2.6.8 Context of the study
The major part of the included papers describes studies from USA (63%). The remaining
37% describe studies from 30 different countries. Canada is the second largest
contributor of included studies (5.5 %). Great Britain, Australia, Finland, China and
Sweden contribute with a relatively great amount of studies among the other countries.
The Netherlands, Norway, Germany, Belgium, New Zealand and Greece contribute
between 10 and 3 papers, while the other 16 countries contribute one or two papers.
2.6.9 Publication Year
The papers included were distributed over the ten years 1999-2009, but with a clear
pattern of increase over time. During the years 1999-2002, between 20 and 30 papers
were published each year that met the search terms and the inclusion criteria of this
review. From the years 2006-2008 there were 70 and 80 references each year. This
trend may be due to both a general increase in the number of articles published each
year, and to an increase in the interest in the issues that are the object of this review.
2.6.10 Clusters
The combinations of contents of the studies included in the mapping have been analysed
by means of cluster analysis. A solution with 12 clusters, computed with the k-means
program in SPSS, is presented here to give a picture of how the mental health factors
and the educational factors combine in the literature included in this review.
36 37
The themes of academic achievement and self-concept were common in the most of
studies and accordingly they are present in almost all the clusters to somewhat varying
Cluster Description N=
1 Mental health and academic achievement 45
2 Anti-social behaviour and social achievements in educational settings 68
3 Assessment, self-concept and depression 15
4 Transitions, social relationships and depression 12
5 School failure, social relationships and mental health 30
6 Teaching, relationships with teachers and self-concept 43
7 Achievements in several domains and well being 115
8 Stress, anxiety and school failure 34
9 School climate, adjustment and anti-social behaviour 42
10 Educational factors (teacher, climate, methods) school failure, stress
& well being 11
11 Self-concept and academic achievements 48
12 Mental health and social relationships in educational settings 5
Valid 468
Missing 3
In the rst cluster of references, we nd studies that are concerned with a wide range
Table 2.9 References
included (N=471) sorted
by country of provenience
of internalizing and externalizing symptoms and also with
academic achievement, school failure and adjustment.
About 10 per cent of the references were grouped into this
In the second cluster the studies consider the symptoms of
anxiety, hyperactivity, attention, and above all anti-sociality
in combination with social relationships and achievements,
in particular with peers (about 15 %).
The studies in cluster 3 form a small group (3 %) concerned
with above all depression and self-concept, but also
anxiety and to a certain degree other symptoms; among
the educational factors the most prevalent is test and
assessment and to a lesser degree stress, peer relationships
and victimization.
Cluster 4 is also a small group, characterised by studies
concerned with depression, anxiety and other symptoms
- but not with self-concept - and with relationships with
peers, assessments and grade transitions among the
educational factors.
Cluster 5 includes studies focussing on school failure and
adjustment, together with relationships with teachers and
peers including victimisation, in combination with several
mental health factors, except for the positive mental health
indicators. This cluster comprises 6 % of the references.
Cluster 6 is concerned with achievement, teaching and
teachers factors in combination with self-concept, coping
and depression among the mental health factors (less than
Country N
USA 300
Canada 26
United Kingdom 17
Australia 15
Finland 14
China 12
Sweden 12
Netherlands 10
Norway 9
Germany 8
Belgium 7
New Zealand 7
Italy 5
Multiple 4
Greece 3
Chile 2
France 2
Slovenia 2
Korea 2
Israel 2
Spain 2
Denmark 1
Austria 1
W. Indies 1
Croatia 1
Turkey 1
Singapore 1
Yemen 1
Mexico 1
India 1
Taiwan 1
Total 471
Figure 2.8 Distribution of
the references included per
publication year (N= 461, data
from 2009 omitted because
limited to two months, N=10)
Table 2.10 Clusters of the
references included in the data-
38 39
or national level (D6, various characteristics of the educational system, D7, reforms,
etc.) are present in the material, but only to a very limited extent. We can conclude that
these aspects are not sufciently well represented to allow for this review to answer
questions concerning educational aspects at the system level.
Another eld that is not well represented is the educational organisational level. There
are many studies that are considering individual factors, and there is also a certain
amount of studies that are considering the relationships at the classroom level, but
studies that examine the higher organizational levels are very scarce. Since it is well
known from organisational research in other sectors that different aspects of the work
environment (management, values, goals) inuence the well-being and health of the
employees, such research is needed in the educational area as well.
It may be noted, however, that quite a few of the references are concerned with relations
among different aspects of mental health and academic achievement, which should
provide an adequate material of studies to be discussed in the narrative synthesis.
We now turn to the second purpose which is according to conduct an in-depth review to
answer to the following questions:
- Which are the causal relationships between mental health and academic achievement?
- How are these relationships inuenced by other factors, both related to the individuals
and their social background, and to factors in the educational environment (evaluation
system, tests, grades, selection procedures, special educational system, teaching
methods, social climate).
First of all, it can be concluded that the mapping has shown that a substantial number of
references investigate effects of academic achievement on mental health and/or effects
of mental health on academic achievement, which indicates that an empirical basis is
available in the form of published papers which satisfy the inclusion criteria that we have
applied so far. However, before we go any further into this literature we need to discuss
the nature of the research problem and the methodological requirements it poses.
3.1 Methodological issues in the study of relations
between academic achievement and mental health
For a long time, research has demonstrated that a relationship exists between mental
health and academic achievement. Thus, problem behavior is associated with low
academic achievement (Hinshaw, 1992), as is depression (Asarnow et al., 2005;
Chen et al., 1995; Marmorstein & Iacono, 2001) and anxiety disorders (Stein & Kean,
2000; Woodward & Fergusson, 2001; Kessler, 2003). Mental illness is also related to
premature termination of education (Kessler et al., 1995), and educational disparities
Cluster 7 is a quite large group of references (about 25 per cent) that treat social and
academic achievements and assessments, and among the mental health aspects well
being and self concept are the most frequent of all.
Cluster 8 has a main focus on educational aspects as school failure, grade transition,
peer-relationships and on mental health aspects as anxiety, depression and stress.
Cluster 9 has a focus on school climate and adjustment among the educational factors
and on self-concept and anti-social behaviour among the mental health aspects (8 %)
The quite few studies in cluster 10 consider several aspects of educational settings (related
in particular to relationships with teachers, to teaching methods, failure, and school
climate) and just some aspects of mental health: in particular stress, psychosomatic
symptoms, well being and self-concept.
Cluster 11 is characterized by studies that focus only on the themes of anti-social
behaviour, self-concept and academic achievement (more than 10 % of the studies).
Finally, the few studies in cluster 12 consider several internalising and externalising
mental health symptoms and several educational factors but mainly those concerned
with relationships, except those related to academic achievement.
2.7 Discussion and Conclusions
From the mapping of the literature we can draw the conclusion that there seems to be a
large enough amount of literature to review that is concerned with what can be dened
as the individual level of the educational factors, namely, the students’ academic and
social achievements and failures.
The different mental health factors seem to be well represented among the references,
with the exception for the group that was dened as other psychiatric symptoms
(autism spectrum, psychosis, etc.). One reason for this may be found in the inclusion
criteria adopted in the review, requiring the presence of a comparison group in the
case of studies with a special population. It is possible that studies investigating the
links between school and mental health factors that are concerned with these other
psychiatric symptoms do not include comparisons groups. Another possible explanation
is that the longitudinal studies do not consider this group of symptoms. This may in turn
be because children with these symptoms were not attending school in regular settings
in the past. This situation has changed during the last years and in many educational
systems, students with autism spectrum disorders now attend regular schools and
classes. Future longitudinal studies therefore need to consider this group.
References that treat mental health factors and educational factors at the organisational
40 41
due to mental illness persist through life (Miech et al., 2005).
Four different causal models may explain these relations (Hinshaw, 1992). According to
the rst model one or more factors (e.g., poor attention or a troubled family environment)
inuence both mental health and academic achievement. This model implies that the
relation between mental health and achievement is spurious, because the correlation is
due to a common cause, such as cognitive ability or environmental inuences. Second,
it may be that achievement-related problems cause mental health problems. For
example, school failures may lead to loss of self-esteem and depression. According to
the third model, mental health problems inuence academic achievement. For example,
disruptive behaviours might decrease focus on learning activities, thereby worsening
school performance. The fourth model states that academic achievement and mental
health cause each other in a reciprocal relation. For example, school failure may cause
problem behaviour, which in turn causes further deterioration of school performance in
a negative feedback loop.
These four models are not easy to distinguish from one another in empirical research,
and there is considerable controversy about which of these four models provide valid
explanations of observed correlations between mental health and academic achievement.
The fact that the distinction between these four models is a considerable simplication
does not make things any easier. There is thus evidence that a causal effect is often due to
indirect chain effects (Rutter, 1994) and complex mediating mechanisms. Thus, different
forms of mental health problems may cause one another, and academic achievement
may be causally involved in these relations. For example, conduct problems may cause
achievement problems; these may in turn affect self-esteem negatively, which may cause
depression. This sequence of causal relations implies that we will observe a correlation
between conduct problems and depression. Furthermore, Rutter (2009) identied ten
different ways in which risk mechanisms for mental disorders may be conceptualized,
and simple additive effect models fail to capture many of these mechanisms.
As has been observed many times before (e.g., Hinshaw, 1992) great methodological
challenges thus are encountered in research aiming to make causal inferences about
relations between mental health and academic achievement. It is generally agreed that
randomized experimental designs provide the strongest basis for causal inference, but
we do not have any such studies in our set of references. One reason for this is that
intervention studies are not within the scope of this review because another review has
that focus. Another reason is that many of the questions we ask have not been studied
experimentally, and cannot be studied experimentally for ethical reasons.
Natural experiments and estimations with instrumental variables provide the next best
basis for credible causal inference (Feinstein et al., 2006). However, there are very few
such studies available in this area of research.
Longitudinal designs also offer possibilities, under certain conditions, to make inferences
about causality (e.g., Hinshaw, 1992; Loeber & Farrington, 1994), even though it must
be realized that the complex nature of the phenomena investigated makes it necessary
to be careful in design, analysis and interpretation of longitudinal research. Indeed,
Hinshaw (1992) concluded a review of longitudinal studies on relations between
externalizing problem behaviour and achievement with the statement:
the material covered herein underscores the potential complexity of causal mechanisms
underlying the association of interest. Given the interactions and transactions among
social, familial, linguistic, and neurobehavioral variables that may culminate in the
overlap between underachievement and externalizing behavior, teasing apart the
effects of any single background factor is likely to be quite difcult or even misguided.
The challenge for the eld is to derive explanatory models with sufcient rigor and
complexity to handle the diversity of causal factors. (Hinshaw, 1992, p. 151)
It is, however, our expectation that there has been a development both when it comes
to the conduct of longitudinal studies, and when it comes to explanatory models since
the Hinshaw (1992) review was published. The number of longitudinal studies has
increased, and suggestions for design improvements by Hinshaw (1992), Rutter (1994)
and others should have increased the quality of the studies. Great progress has also been
made in development of analytical techniques appropriate for longitudinal designs,
such as latent variable growth curve modeling techniques (e.g., Muthén, 2004).
Because of the progress that we expect has been made, the in-depth review will take
42 43
advantage of the large number of longitudinal studies that resulted from the search.
It should be noted, however, that not all studies based on longitudinal data necessarily
support causal inference. Hinshaw (1992) emphasized that a good longitudinal study
of relations between mental health and achievement should include assessments of
achievement and mental health at initial as well as follow up periods. Such designs
allow analyses that focus on change in both independent and dependent variables, and
such analyses have the advantage that time-invariant omitted variables do not bias the
results. Hinshaw (1992) also emphasized that an adequate longitudinal design should
include measures of relevant antecedent variables, in order to be able to control for
common causes. It also is necessary to use analytical techniques that take advantage of
the longitudinal design in statistically appropriate ways.
However, even longitudinal studies that satisfy these requirements may fail to yield
correct inferences. One reason for this is that there may be problems of measurement
due to scale problems and to random and systematic errors. There also are problems
of construct validity and of the degree of specicity of measures and effects (Hinshaw,
1992). Measures of positive aspects of mental health, in particular, are aficted by such
conceptual problems.
The mapping reported in chapter 2 illustrates the great heterogeneity within the domains
of mental health and academic achievement, and the studies also are heterogeneous
in many other respects, such as age groups investigated, language and culture. The
relatively limited number of studies in combination with this heterogeneity precludes
use of quantitative techniques of meta-analysis for the in-depth review. Instead a
narrative synthesis has been attempted. Even though this syntheses has not managed
to follow the recommendations for the conduct of narrative syntheses made by Popay et
al. (2005), we have tried to relate theoretical notions in the literature to the empirical
ndings in the reviewed studies.
3.2 Studies included in the data extraction
From the 471 articles that were included in the mapping we selected a subset of studies
to be included in the data extraction. One criterion for inclusion was that the study
should have as an aim to study either effects of academic achievement on mental health,
or effects of mental health on achievement or both (i.e., P1 and/or P2 according to the
mapping). A second criterion for inclusion was that the study according to the judgement
of relevance should have a specic focus on the issues of the review. The third criterion
for inclusion was that the study should have a longitudinal design.
When these three criteria were applied 180 references were included.
3.2.1 Description of the studies
In the following section, some descriptions of these 180 selected studies are presented.
The 180 references included in the data-extraction, were published in 95 different
journals, representing the following disciplinary areas: developmental psychology, child
and adolescent psychiatry, educational and social psychology, educational research,
youth and adolescent research, sociology, public and community health, pediatrics,
criminology, and health economics (see Table 3.1)
Table 3.1 Number of references for different journals among the references included in the data-extraction;
only journals with more than one contribution are presented in the table.
The data sources of the references included in the data extraction are presented in Table
3.2 below. However, the distribution of these references among the databases has to
been regarded as indicative, since a large amount of duplicates were removed from the
retrieved references.
Table 3.2 Number of references from different databases
The large number of journals and the variety of disciplines that they represent,
show that the issues that are considered in this review are focussed upon in various
elds of knowledge and research. However, a core area can be identied, where the
density of references is higher, consisting essentially of developmental psychology and
psychopathology, together with educational psychology.
However, it is also important to emphasize the large number of disciplines contributing
research, such as sociology, epidemiology, educational research, youth studies,
criminology, health economics, paediatrics and learning disability studies. This result
conrms that the broad literature search strategies adopted were necessary to cover the
whole eld of relevance.
3.2.2 Data-extraction procedures
Each reviewer was assigned 15-20 references to extract data from, a few from each
publication year. The reviewer completed for each reference a protocol that was
developed for this purpose (see the data-extraction protocol in section 5.5. in the
The data-extraction protocol is structured with the following sections:
Section A: Administrative details
Section R: Relevance
Section B: Study aims and rationale
Section C: Study mental health focus
Section D: School and achievement
Section F: Sample and context
Section H: Study Method
Section I: Methods, groups
Section J: Sampling strategy
44 45
Section K: Methods – Data Collection
Section L: Methods – data analysis
Section G: Results and conclusions
Section M: Quality of study: description
Section N: Overall quality of study
The rst step of the data-extraction was to evaluate if the study met the additional
inclusion criteria specied for the data-extraction. The rst criterion was that the study
should have a “high” relevance with respect to the review question specied for the
narrative synthesis. Studies that were not rated high were not included in any further
The second criterion was that there had to be at least two measurements of the outcome
Periodical N
Child development 14
Developmental Psychology 12
Journal of educational psychology 9
Development and psychopathology 7
Journal of Youth & Adolescence 6
Journal of Child Psychology & Psychiatry & Allied Disciplines 4
Journal of Clinical Child & Adolescent Psychology 4
Criminology 3
Health economics 3
International Journal of Behavioral Development 3
Journal of abnormal child psychology 3
Journal of Adolescent Health 3
Journal of Adolescent Research 3
Journal of Emotional & Behavioral Disorders 3
Journal of School Psychology 3
Prevention Science 3
Social Development 3
Social Forces 2
American Educational Research Journal 2
Archives of Paediatrics & Adolescent Medicine 2
British Journal of Educational Psychology 2
Cognitive Therapy and Research 2
European Psychologist 2
Journal of Adolescence 2
Journal of Clinical Psychology 2
Journal of Educational Research 2
Journal of Learning Disabilities 2
Journal of Personality 2
School Psychology Quarterly 2
Social Development 2
Social problems 2
Social science & medicine 2
Social Science Research 2
Sociology of Education 2
variable, an initial measurement and follow-up measurements. If the study did not have
at least two measurements of the dependent variable, the relevance of the design of
the study for this review was evaluated as “low”. It should be pointed out that it was
not necessary that the variable was measured with exactly the same instruments at the
different occasions, but if academic achievement was the outcome considered in the
study an initial measure for instance of cognitive ability should be present.
Of the 180 studies considered, 60 was
evaluated as having low relevance for the
purposes of this review, either because of
their focus or because an initial measure of
the outcome variable was lacking, 51 were
judged as having medium relevance and 69
high relevance (see Figure 3.1).
The data-extraction protocol asked a series
of questions about the methodology of the
study and asked for a further evaluation of the quality of the study based on these
methodological evaluations. Out of the studies that went through a full data-extraction,
44 were evaluated as having medium quality and three low quality.
The following chapter presents a narrative synthesis of the studies included in the data-
extraction that got the evaluation “high” with respect to relevance and quality (N= 52).
We expect that these references have the characteristics (content, focus, design, method,
quality) that are required to answer the questions of the present study. However, even
though the studies satisfy strict methodological requirements, it is not possible to claim
that all of them allow causal inferences for reasons discussed above.
As is clear from previous chapters, mental health is a complex phenomenon dened by
several different aspects, and it is reasonable to expect that the relations to academic
achievement are quite different for different aspects. The narrative synthesis is
organized in such a manner that a basic division is made between studies that investigate
academic achievement as a determinant of mental health on the one hand, and studies
that investigate mental health as a determinant of academic achievement on the other
hand. The rst category of studies has been grouped according to different aspects of
schooling, and the second category has been group according to which aspect of mental
health (internalizing problems, externalizing problems, and positive aspects of mental
health) is studied.
4.1 Schooling and academic achievement as determinants of
mental health
Database N
PsycInfo 58
Medline 44
CSA* 32
Academic Search Premier 25
EJS E-Journals 10
Psycarticles 8
Business Source Premier 1
*CSA includes: ASSIA, Applied Social Sciences
Index, Abstracts Social Services Abstracts,
Sociological Abstracts
46 47
There are numerous theories about how schooling and academic achievement can
inuence mental health. In this context it is not possible to discuss all these, but we will
briey refer some important theories and models.
Frydenberg (2008) summarized previous research on the main sources of stress in the
life of adolescents by identifying three main categories of concerns: things that relate
to success and failure at school, peer and family relationships, and social issues such as
the environment, poverty and unemployment. Frydenberg also emphasized that these
stresses and strains vary according to age, context and culture. It is, nevertheless, of
great signicance as a starting point for this review that things that relate to academic
achievement is identied as one of the main sources of stress, and also that the
importance of this source varies as a function of individual and contextual factors.
Negative effects of school failure on mental health may be accounted for in terms of
theories of stress. The concept of stress seems to have been borrowed from the eld of
physics, and represents the situation when an object breaks or is deformed by forces of
outside pressure. One example of an inuential model is the transactional view of stress
(e. g., Lazarus, 1991), which regards stress as occurring when there is an imbalance
between the perceived demands of the environment and the perceived resources of the
A more recent conceptualization is the Conservation of Resources (COR) theory
developed by Hobfoll (1989, 2001). According to this theory individuals strive to retain,
protect and build resources in order to maximize coping capacity and limit psychological
distress. Stress occurs when resources are threatened or lost. Hobfoll (1989) thus denes
psychological stress as a reaction to the environment in which there is loss of resources,
or a lack of resource gain following investment of resources. Both perceived and actual
loss of resources, or lack of gain of resources, is sufcient for producing stress.
Resources are dened as objects (e.g., possessions), personal characteristics (e.g.,
problem solving skills, mastery, self-esteem, learned resourcefulness), conditions
Figure 3.1 Literature flow from mapping to narrative synthesis
48 49
(e.g., being a student), and energies (e.g., social networks, time). Different kinds of
environmental circumstances may threaten people’s resources, and cause loss. Loss
is important for two reasons. The rst is that resources have instrumental value in
themselves, and the second that they have symbolic value in helping to dene for people
who they are.
According to COR theory effects of resource loss and resource gains are asymmetrical.
Thus, one rst principle of the theory is that resource loss is disproportionally more
salient than is resource gain, so given equal amounts of loss and gain, loss will have
more impact. Furthermore, resource gain is more important in the context of resource
loss than in non-threatening situations. A second principle of COR theory is that people
must invest resources in order to protect against resource loss and to recover from
resource loss. One implication of this principle is that those with lower resources are
more vulnerable to resource loss and less capable of resource gain. They are likely to
take a defensive position to guard their resources, and they are least successful under
high stress conditions. Another implication is that initial loss increases the likelihood
of future loss, and that initial resource gain increases the likelihood of future resource
gain. This causes loss cycles and gain cycles, but because loss is more potent than gain,
loss cycles will have greater impact than gain cycles.
While there is a considerable amount of supportive evidence for the principles of COR
theory in research on adults (e.g., Hobfoll, 1989, 2001), the amount of research conducted
in school contexts is limited. However, the COR theory is a simple, comprehensive and
powerful theory which has several attractive features when accounting for effects of
schooling and academic achievement on mental health. The theory thus emphasizes
the dynamic nature of development of resources and stress, and it integrates notions of
positive aspects of mental health with notions of mental health problems. Furthermore,
as was emphasized by Frydenberg (2008, p. 8), the theory makes it possible to identify
resources which are malleable and which have broad application, and which therefore
are interesting targets for intervention. High self-esteem, optimism, and good problem-
solving skills are examples of such resources mentioned by Frydenberg (2008, p 8).
There is also a tradition of research which has viewed academic and social difculties
as stressors that cause depression, anxiety and dysphoric mood states. Cole (1991)
has proposed a competency-based model of depression. The fundamental idea is that
depressive symptoms in childhood reect underlying views of low self-competence. These
views are based on the acquisition of positive or negative feedback about performance in
a particular domain. Such domains include, for example, academic performance, social
acceptance, athletic competence, personal conduct, and physical appearance. According
to the Cole model, repeated exposure to negative feedback in a wide variety of domains
inhibits the emergence and differentiation of positive self-schemata.
Some of this negative feedback comes from self-evaluations based on social comparisons.
Another source of information consists of the responses from others. Teachers, parents,
peers, and others develop expectations about a child, which they directly and indirectly
convey to the child. Such feedback increases the likelihood that the child will act in a
50 51
manner that conrms these expectations, as has been observed in the research on self-
fullling prophecies. Cole (1991) also referred to theories within symbolic interactionism,
which claim that what children believe about themselves results from their perception
and subsequent internalization of others’ appraisals. A personal evaluation by peers,
teachers, parents, and other signicant persons are conveyed to the child and shapes
the child’s self-image. Frequent positive feedback promotes the development of positive
self-schemata that inhibit the emergence of depression. Frequent negative feedback
inhibits the development of such schemata, which potentially predisposes the child for
In the somewhat similar dual-failure model, Patterson and Stoolmiller (1998) proposed
that development of depressive symptoms among young schoolchildren occurs as a
function of both academic failures and rejection among peers. In particular, they argue
that children with externalizing behaviour problems run a risk of being rejected by their
peers, which in combination with academic failure increases the risk for development of
depressive symptoms. This model thus posits a mechanism through which externalizing
behaviour problems cause development of internalizing problems.
Relations between externalizing behaviour and academic underachievement have been
in focus in much research. Reviews by Rutter (1974) and Hinshaw (1992) concluded
that children with externalizing problems are over-represented among children with
learning disabilities, poor grades, and grade retention. Hinshaw (1992) also concluded
that in early and middle childhood, there is a link between hyperactivity-inattention and
underachievement, while by adolescence there are links between antisocial behaviour
and underachievement.
Hinshaw (1992) observed that alternative causal models to explain these links have
rarely been rigorously tested. He gave several reasons for this, such as the difcult
methodological and conceptual issues involved in linking the domains of externalizing
behaviour and academic achievement, inadequate assessment of antecedent variables in
the designs, insufcient control of correlated predictors and of antecedents, and the lack
of prospective longitudinal studies. He also noted that measurement issues concerning
both domains offer great challenges.
However, in the review Hinshaw (1992) observed that there are some suggestions
of causation from one domain to the other. For example, early reading failure may
predispose to subsequent increases in externalizing behaviour. There also are
suggestions that a subgroup of children with reading failure without early externalizing
behaviour problems may develop delinquency by late adolescence. Hinshaw (1992)
emphasized that that there may be distinct causal mechanisms for different subgroups of
children, and research strategies are needed that can uncover subgroups with different
developmental trajectories.
We now turn to a presentation and discussion of longitudinal studies satisfying the
inclusion criteria, and which focus on effects of schooling and academic achievement
on mental health.
4.1.1 Trajectories of development of academic achievement
and mental health
One group of studies investigates the development of mental health and academic
achievement from early years throughout the school system (see Table 4.1). These studies
often aim to identify relations between development in the two areas, and to determine
the actual direction of relationship between mental health and academic achievement.
The broad, open and complex character of these studies makes it appropriate to present
them rst.
Table 4.1 Studies of trajectories of development of achievement and mental health
These studies are discussed in detail below. Studies
Roeser et al. (1999; ID3031) investigated children’s academic motivation and mental
health as a way of describing different patterns of adjustment during adolescence. The
original sample consisted of 875 elementary school-aged children in three cohorts, their
parents, and their teachers. The children and their families were contacted 6 years later
with a follow up the next year. The children were then in grades 8, 9, and 11. A total of
491 children participated in all three waves of measurement.
Cluster analysis was used to delineate patterns of academic functioning and mental
health during eighth grade. Four equal-sized clusters were identied. Some children
were doing consistently well or consistently poorly across the academic and emotional
domains of functioning (well-adjusted and multiple problems groups, respectively).
Other children could be characterized as having mild difculties in only one or the other
of these domains of functioning during eighth grade (the poor motivation and poor
mental health groups). The children in the four clusters were followed up in high school
and the concurrent and predictive validity of the clusters was investigated.
Children in the multiple problems group showed long-term continuity in terms of their
poor academic motivation, low grades, and low self-esteem during the years between
elementary and high school. One interesting nding was that their mental health
actually improved across the transition to high school and there was no decline in self-
esteem or achievement during this transition.
The poor school motivation group without accompanying psychological distress showed
declining perceptions of their academic competence and value of the school between
elementary school and the later grades of middle school. Roeser et al. (1999) observed
that declines in these children’s self-perceptions of competence occurred before declines
in their academic values, and suggested that children who increasingly experience
difculty in school over time start to devalue school as a way of protecting their self-
52 53
For the children in the poor mental health group who showed continued school
motivation to learn there were no long-term trends in academic motivation or self-
esteem, and the level of emotional distress in this group of adolescents was stable over
time. Roeser et al. (1999) suggested that these adolescents’ difculties arose in the early
adolescent period, during which time data was not collected. However, they observed
that these adolescents, along with the poor motivation group, are likely to go unnoticed
their emotional difculties, because they are not failing academically and they are not
acting out.
It must be observed that even with the wealth of information collected for this study it
still is limited by the fact that the clustering of students was based on the information
that was available in grade eight. It thus is reasonable to expect that further relations
and patterns could have been identied if an even better coverage of the school years
had been possible.
Dodge et al. (2008; ID8441) investigated factors inuencing the development of
antisocial behaviour from age 5 to 18. They identied several domains of predictors
(early adverse context, harsh and inconsistent discipline, school social and cognitive
readiness, early externalizing behaviour problems, school social and cognitive failure,
parental monitoring and communication, and deviant peer associations), which were
hypothesized to operate on antisocial outcomes by directly inuencing the next domain
in the hypothesized temporal sequence. This next domain both mediates the impact of
the prior domain and causes further antisocial development in an incremental manner.
The study included 387 children, who were selected to represent the population of a
set of schools, along with 367 high-risk children who were oversampled. Weighting
was used in the analyses to reect the oversampling of high-risk children. Beginning
in kindergarten and lasting through grade 11, annual measurements were collected
from multiple sources that included the teacher, the peer group, administrative school
records, the mother, the child, and interviewer ratings. Data was analyzed with path
modelling techniques.
Each of the seven predictor domains was signicantly related to adolescent violence,
and all temporally adjacent domains were signicantly related to each other. Each
succeeding domain not only mediated the preceding domain, but it also provided
a signicant increment beyond the previous domain in enhancing the prediction
of violence outcomes. In particular it was found that school failure was a signicant
predictor of adolescent violence and lack of parental monitoring. School failure also
partially mediated the prediction of adolescent violence from early externalizing
problems, and it signicantly incrementally predicted adolescent violence, controlling
for childhood externalizing behaviour problems.
Masten, Hubbard et al. (1999; ID5217) investigated the development of resilience,
focusing on how resilient adolescents differ from maladaptive peers. A sample of 205
children (114 females, 91 males; 27% minority) was recruited in elementary school
and was followed over ten years. Three major domains of competence (academic
achievement, conduct, and peer social competence) were assessed from childhood
through adolescence, along with aspects of adversity and major psychosocial resources.
Both variable-centred and person-centred analyses were conducted to test the
hypothesized signicance of resources for resilience.
The two analytical approaches gave similar results and supported the importance of IQ
and parenting as resources for the development of competence and as protective factors
with respect to the development of prosocial behaviour in a high adversity context. IQ
was a signicant predictor of social competence in childhood but not so in adolescence.
The combination of low IQ and high adversity was related to conduct problems that
were evident in childhood and worsened over time. The analyses also identied a group
of maladaptive youth which differed from resilient youth in current and past resources,
and in psychological well-being. The maladaptive adolescents tended to have a history
of adversity, low resources, and broad-based competence problems.
Study Study purpose Main conclusion
Roeser, R. W., Eccles, J. S., &
Freedman-Doan, C. (1999).
Academic functioning and mental
health in adolescence: Patterns,
progressions, and routes from
To describe different patterns of
adjustment during adolescence
with respect to academic
motivation and mental health
Four main patterns of school
adjustment and mental health
could be discerned
Dodge, Greenberg et al.
(2008). Testing an idealized
dynamic cascade model of the
development of serious violence
in adolescence
To test a dynamic cascade model
of the development of violence in
There are relations between
early externalizing problems,
school failure, anti-social
behaviour, in particular regarding
the influence of academic and
social failures.
Masten, A.S., Hubbard, J.J. et
al. (1999). Competence in the
context of adversity: Pathways
to resilience and maladaptation
from childhood to late
To study how intellectual
functioning and parenting quality
relate to multiple dimensions
of competence over time from
childhood to late adolescence,
particularly in the context
of adversity. To study how
resilient adolescents differ from
maladaptive peers who have
not succeeded in the context of
adversity and from competent
peers who are also successful
but have not experienced serious
Four main conclusions were
drawn: (a) The development
of competence is related to
psychosocial resources, (b) good
resources are less common
among children growing up in
the context of adversity, (c) if
reasonably good resources are
present, competence outcomes
are generally good, even in
the context of chronic, severe
stressors, and (d) maladaptive
adolescents tend to be stress-
reactive and have a history
of adversity, low resources,
and broad-based competence
Masten, A.S., Burt, K.B. et al.
(2004). Resources and resilience
in the transition to adulthood:
Continuity and change
To study if resilience in early
adolescence endures to
adulthood and if it is possible
to predict resilience that
emerge in young adulthood,
despite maladaptation in early
Success in developmental tasks
over time was associated with
not only a history of success in
earlier tasks but also a set of
resources in childhood and early
adolescence. These resources
were related to young adulthood
competence in developmental
tasks, but these effects were
largely mediated by competence
along the way.
54 55
These ndings suggest that good parents and cognitive skills are general advantages
for development that may be particularly important for overcoming serious chronic
Masten, Burt et al. (2004; ID4848) examined patterns of continuity and change in adaptive
behaviour over the transition to adulthood with a focus on the adaptive resources. This
study extended by ten years the examination of competence, adversity, and resilience
in the ten-year longitudinal study by Masten,Hubbard et al. (1999) described above.
The assessments were focused particularly on the quality of competence in multiple
domains; lifetime adversity, disadvantage, or stressful life events; and differences in the
children or their families that might account for good versus poor adaptation.
From the study four main conclusions were drawn: (a) the development of competence
is related to psychosocial resources; (b) good resources are less common among
children growing up in the context of adversity; (c) if reasonably good resources are
present, competence outcomes are generally good, even in the context of chronic, severe
stressors; and (d) maladaptive adolescents tend to be stress-reactive and have a history
of adversity, low resources, and broad-based competence problems. Discussion
The four studies presented above all demonstrate that there is considerable consistency
from childhood into young adulthood in the presence of problems both in the area of
mental health problems and in the area of academic achievement. However, the studies
investigate somewhat different aspects of the process of development, and they propose
somewhat different mechanisms to account for the relationship between mental health
and academic achievement.
According to Roeser et al. (1999) an important mediator of the continuity of problems
for the multiple problems group is the child’s internalization of negative self-
perceptions of academic competence, negative feelings of self-worth, and a sense of
school as uninteresting and unimportant. Children in the multiple problems group did
not necessarily show any cognitive decits compared to children in the other groups
during rst grade, which suggests that it was not low intelligence that accounts for the
long-term difculties. Roeser et al. (1999) suggested that it was inappropriate skills and
behaviour rather than general cognitive incompetence that set in motion the maladaptive
trajectory of development. Poor skills could lead to failure, frustration, poor conduct,
teacher disapproval, and so on. It also is interesting to note that the self-esteem of the
multiple problems group improved in the transition to high school. Roeser et al. (1999)
argued that as a means of protecting their self-esteem and emotional well-being, they
may have turned to their peers and to engagement in antisocial activity in forming a
view of their worth and self esteem.
Dodge et al. (2008) relied on a dynamic cascade model to account for the development
of chronic violence in adolescence. The model posits incremental roles of numerous
life experience factors as causes of development, and they showed that each succeeding
domain not only mediated the preceding domain, but that it also provided a signicant
increment beyond the previous domain in enhancing the prediction of violence outcomes.
Thus, an early social context of disadvantage predicts harsh and inconsistent parenting.
This predicts social and cognitive decits, which predicts conduct problem behaviour,
which predicts elementary school social and academic failure, which predicts deviant
peer associations, which ultimately predicts adolescent violence. They interpreted this
patterned sequencing as being consistent with a dynamic model specifying reciprocal
impact of parenting on peer relations and peer relations on subsequent parenting, as
well as reciprocal inuences between the self and the environment. The model thus
explains how early risk is realized across time and also how each new developmental era
affords new risk.
In the two studies by Masten et al. (1999, 2004) the concept of resilience plays a central
role. The term resilience is used to refer to the nding that some individuals have a
relatively good psychological outcome despite suffering risk experiences that would be
expected to have disastrous effects.
The Masten, Hubbardet al. (1999) study demonstrates that resources in the form of
general intellectual skills and good parenting protect against negative effects of adverse
conditions on conduct and prosocial behaviour. This result is in agreement with Hobfolls
theory that resources in the form of problem-solving skills and a social network provide
protection against consequences of loss. Similarly, Masten, Burt et al. (2004) concluded
that psychosocial resources are important for resilience in young adulthood.
4.1.2 Effects of early academic achievement
The search resulted in several studies dealing with early school failure. Several of these
studies investigated effects of different kinds of reading problems, as is clear from Table
4.2. Much previous research has shown early reading instruction to be crucial for school
outcomes (e. g., Maughan, Hagell, Rutter and Yule, 1994). It has also been demonstrated
that there is an overlap between reading and behaviour problems, which will tend to
decrease achievements further for children with reading problems (Rutter et al., 1970).
The studies in this group are presented and discussed below. Studies
Halonen et al. (2006; ID4700) investigated three main research questions: (1) to what
extent do children’s problems in the acquisition of pre-reading and reading skills
predict their subsequent externalizing and internalizing problem behaviour; (2) to what
extent does children’s externalizing and internalizing problem behaviour predict the
emergence of reading difculties; and (3) to what extent does children’s internalizing
problem behaviour predict their externalizing problem behaviour; or is it rather that the
externalizing problem behaviour predicts internalizing behaviour?
The sample included a total of 196 (104 boys, 92 girls) 5- to 6-year-old children
56 57
participating in the Jyväskylä Entrance into Primary School study in Finland. There
were six measurement points and at each point children’s reading skills and problem
behaviours were assessed using identical measurements. At Time 1, children’s pre-
academic skills were also investigated. The research questions were examined using
path modelling with latent variables.
Slow progress in learning to read was found to predict an increase in internalizing
problem behaviour both during the preschool year and in rst grade. However, children’s
internalizing problem behaviour also predicted subsequent problems in reading
performance both during the transition from preschool to primary school and in primary
school. Another main nding was that internalizing problem behaviour predicted an
increase in externalizing problem behaviour during the transition from preschool to
primary school. After children entered primary school externalizing problem behaviour
started to predict internalizing problem behaviour rather than vice versa.
The results of this study suggest that difculties in learning to read, and internalizing and
externalizing problem behaviours are developmentally linked in a cumulative manner.
Trzesniewski et al. (2006) used a longitudinal twin design to evaluate three competing
hypotheses about the source of the association between reading achievement and
antisocial behaviour. One hypothesis was that the relation is explained by genetic
inuences that are common to both reading and antisocial behaviour; the second
hypothesis that the relation between reading achievement and antisocial behaviour
can be explained by common environmental antecedents; and the third hypothesis
that reading achievement and antisocial behaviour are causally related. They also
investigated the hypothesis that reading achievement and antisocial behaviour are
related because ADHD children tend to display antisocial behaviour.
The study was based on two consecutive birth cohorts in a birth register of twins born
in England and Wales, and comprised about 1000 same-sex twin pairs, about half of
which were monozygotic and half of which were dizygotic. There were two waves of
measurement, at ages 5 and 7.
Most of the analyses were based on data for boys, because the association between
achievement and antisocial behaviour was stronger for boys. Modelling of the genetic
sources of variance showed that both reading and antisocial behaviour were inuenced
Study Study purpose Main conclusion
Halonen, A., Aunola, K., Ahonen, T.,
& Nurmi, J. E. (2006). The role of
learning to read in the development
of problem behaviour: A cross-lagged
longitudinal study
To study whether children’s
problems in the acquisition of pre-
reading and reading skills predict
their subsequent externalizing and
internalizing problem behaviour,
and whether children’s externalizing
and internalizing problem behaviour
predict the emergence of reading
Slow progress in learning to read
predicted an increase in internalizing
problem behaviour both during the
preschool year and in first grade.
Children’s internalizing problem
behaviour also predicted subsequent
problems in reading performance
both during the transition from
preschool to primary school and in
primary school
Morgan, P. L., Farkas, G., Tufis, P. A.,
& Sperling, R. A. (2008). Are reading
and behaviour problems risk factors
for each other?
To examine if reading problem in
first grade was related to behaviour
problems in third grade; and if
behaviour problems in first grade
related to reading problems in third
Children with reading problems
in first grade were more likely
to display poor self-control,
externalizing behaviour problems,
and internalizing behaviour problems
in third grade.
Ackerman, B. P., Izard, C. E., Kobak,
R., Brown, E. D., & Smith, C. (2007).
Relation between reading problems
and internalizing behaviour in school
for preadolescent children from
economically disadvantaged families.
What are the relations between
reading problems and internalizing
behaviour reported by teachers
The results tie reading problems
to emotional distress in school
and support conclusions about
the direction of effects and