A Critical Review of the Literature on School Dropout

Abstract

This paper reviews the growing literature on early school leaving. We clarify what is at stake with early school leaving, and touch upon underlying problems and methodological issues raised in the literature. The paper investigates the levels, the methods and models with which the topic has been studied, and discusses potential (dis)advantages of each of those. We focus on early school leaving in all its complexity, and on the interplay of relevant (levels of) factors, rather than on just certain factors, typically located in individual students, schools or families. The findings in the literature are discussed and placed into perspective. Finally, a wide set of policy measures are discussed.

Full text

A Critical Review of the Literature on School
Dropout
Kristof De Witte, Sofie Cabus, Geert Thyssen,
Wim Groot, Henriette Maassen van den Brink
TIER WORKING PAPER SERIES
TIER WP 14/14

1
A Critical Review of the Literature on School Dropout
Kristof De Witte
ᵼ
**
, Sofie Cabus
ᵼ
, Geert Thyssen
ᵼ Ŧ
, Wim Groot
ᵼ Ɣ
, Henriëtte Maassen van den Brink
ᵼƔ
ᵼ
Top Institute for Evidence Based Education Research, Maastricht University, Kapoenstraat 2, 6200 ML
Maastricht
Ŧ
Faculty of Language and Literature, Humanities, Arts and Education, Université deLuxembourg, Route de
Diekirch B.P. 2, 7220 Walferdange, Luxembourg
**
Faculty of Economics and Business, KU Leuven, Naamsestraat 69, 3000 Leuven, Belgium
Ɣ
Amsterdam School of Economics, University of Amsterdam, Roeterstraat 11 , 1017 LW Amsterdam
April 2013
Abstract
This paper reviews the growing literature on early school leaving. We clarify what is at stake
with early school leaving, and touch upon underlying problems and methodological issues
raised in the literature. The paper investigates the levels, the methods and models with
which the topic has been studied, and discusses potential (dis)advantages of each of those.
We focus on early school leaving in all its complexity, and on the interplay of relevant (levels
of) factors, rather than on just certain factors, typically located in individual students,
schools or families. The findings in the literature are discussed and placed into perspective.
Finally, a wide set of policy measures are discussed.
Keywords: School dropout; Literature review; Determinants; Policy measures
JEL-Classification: I21; I28
1 Introduction and problem statement
The high dropout rates in Western countries sharply contrast with the social and economic objectives
that have been formulated by government officials and policymakers in order to achieve sustainable
economic growth. School dropout has been defined as leaving education without obtaining a minimal
credential (most often a higher secondary education diploma).
1
In the OECD countries, on average 72%
of all 25- to 34-year-olds had completed a year 12 equivalent in 1999 (Business Council of Australia,
2002a). Another report mentions a year 12 equivalent level of education in the European Union of
77.3% of the population in 2005; a level similar to that of the United States, albeit one that has only
slightly improved since 2000 (European Commission, 2006). These rates mask several things: first, the
diversity of standards by which school dropout and completion are measured across various studies
(from “event” and “status dropout rates” to “graduation” and “status completion rates” or even
1
Early school leaving has often been referred to as “dropout”, early “withdrawal”, or “attrition” from high
school, and before the 1960s also “student elimination”. We will use these terms interchangeably throughout
the paper.

2
“averaged freshman graduation rates”; Cataldi et al., 2009); second, the plurality of differential criteria
underlying them (the age, grade and time range: e.g., “permanent” versus “temporary” dropout or
“stopout”, types of credentials: e.g., a regular or adult high school diploma versus a GED or alternative
diploma, grade entrance versus completion, intra- or inter-school enrolment, etc.; Hammack, 1986,
Pittman and Haughwout, 1987; Rumberger and Lamb, 2003; Blue and Cook, 2004; Entwisle et al., 2004
and 2005; Dalton et al., 2009); and third, the interests involved in their measurement (e.g., on the part
of schools receiving funds according to a “capitation” formula; cf. Entwisle et al., 2004).
In order to reduce the dropout rates, the “No Child Left Behind Act” (2001), and the “Lisbon
2000” and the “Europe 2020” goals have been formulated in the United States and Europe,
respectively. The former aimed at an average high school graduation rate of 90 percent, whereas the
latter expressed the desire that at least 85 percent of all 22-year-olds in the European Union complete
upper-secondary education and maximum 10% of all pupils leave school early by 2012 (i.e., an
objective to halve the dropout rate between 2002 and 2012; see: US Department of Education, 1990;
European Commission, 2006).
Despite increasing attention on the part of policy makers, school dropout still is a serious issue.
The growing literature on early school leaving indicates that school dropouts, compared with their
graduated peers, are more frequently associated with long-term unemployment, poverty, bleak health
prospects, sustained dependence on public assistance, single parenthood (in females), political and
social apathy, and (juvenile) crime (Christenson et al., 2000; Business Council of Australia, 2002b;
Rumberger and Lamb, 2003; Kaufman, Alt and Chapman, 2004; Vizcain, 2005 and references therein).
However, as Smith (2003) has argued, there is something naïve about the use of such
associations, as they do not necessarily imply causation. It is indeed increasingly recognized that
caution is required in interpreting such correlations, as the decision to drop out of school may be
driven by exogenous factors, or may even result from systemic flaws, rather than factors intrinsic to
dropouts themselves (Rumberger and Lamb, 2003; Business Council of Australia, 2002). Structural
inequality may not only cause early school leaving, but also for, e.g., health problems or poverty that in
turn may be at the origin of dropping out.
In contrast to previous literature reviews on school dropout (e.g. Rumberger, 1994), this paper
does not aim to fully summarize the dropout literature. Instead, it focuses on hitherto unchallenged
commonplaces, possible underlying problems, methodological issues and research trends. It attempts
to analyse the complex interplay of factors in its entirety rather than to concentrate on certain factors
one-sidedly, as to avoid reproducing stereotypes. The literature is thereby carefully pondered with the
aim of producing an overview of factors that may be most predictive of early school leaving (indicative
of correlation), either by themselves or in interaction with other predictors. From this, and again in
contrast to the previous reviews, we try to highlight aspects found in the literature that unite both
dropouts and graduates, and that have a positive influence on all parties involved. In other words, it is
aimed to pinpoint characteristics susceptible to improvement, from which both potential early school
leavers and their fellow pupils may benefit. This focus significantly distinguishes our literature review
from previous ones on early school leaving, as does its subsequent connection to important policy
measures. By presenting policy measures next to predictive variables of early school leaving, we

3
highlight their close interrelation. Indeed, in line with an evidence-informed paradigm, policy measures
should focus on what research indicates as the most predictive measures.
This paper has benefitted from journal articles, books and reports from the past three decades
(until 1980, with the exception of Reich and Young (1975) which provides a lowly cited yet nonetheless
interesting starting point of this paper). To this end we have used the search engines ERIC (Educational
Resources Information Center) and Google Scholar. As an additional criterion for inclusion, we have
pragmatically restricted the literature search to English language literature. The review’s emphasis is
on early school leaving at the level of secondary (or high school) education.
2
The keywords “school
dropout” or “school leaving”, and “secondary education” or “high school” have been used in search for
abstracts. Using these keywords, Google Scholar yielded the highest number of hits (over thousands),
whereas ERIC only provided us with 12 abstracts. To limit the total number of hits in Google Scholar,
we also have included the keywords “school to work” or “transition”. ERIC then excluded all abstracts
from the hit list, while in Google Scholar, we still retained about 600 abstracts.
The greater part of the existing literature has described only one or some dropout
determinants, has not provided an overview of, or clear connections to, other dropout determinants,
and has only to a limited extend been informative about studies on dropout prevention strategies. This
finding is in line with Wilson et al. (2011), who have found in total 167 experimental or quasi-
experimental studies eligible for inclusion in their systematic review on school dropout and
completion. There are two main reasons why high quality studies of dropout prevention measures or
interventions are lacking. First, as various observed and unobserved factors influence the decision to
leave school early, evaluations may fail to show program effectiveness. This would result in
‘publication bias’ (i.e., negative or insignificant results are not published). Second, there is a general
lack of uniformity and transparency with respect to school attendance and enrollment registration.
Many studies therefore have to rely on surveys/questionnaires or (costly) local experimental settings.
Due to self-reported data on attendance behavior and sample selection, this may lead to difficult
statistical inference.
This paper is organized as follows. The next section explores common stereotypes with regard
to dropout. Section 3 looks at current research approaches to early school leaving. In turn, a
conceptual framework fitting a wide range of potential predictors of early school leaving is presented
in Section 4. Section 5 discusses the predictors of early school leaving at student, family, school and
community level. Using the insights for the predictors, we link them with common policy measures in
Section 6. Finally, Section 7 concludes the paper with policy advice and scope for further research.
2 Stereotypes on school dropout
Over time, potential predictors of non-graduation have generally been looked for, firstly, among
individual students and their families, subsequently in schools, teachers and fellow-pupils, and only at
a later stage in the broader context or environment (neighbourhoods, peers networks and labour
2
Note that, in this literature review, we did not focus on the level of post-secondary (or college) education, so
that it is somewhat underrepresented (exceptions being, Bynum and Thompson, 1983; Smith and Naylor, 2005;
Perna et al., 2008).

4
markets; cf. Rumberger, 2004a). Moreover, attention has first been focused on immutable variables
(demographic and other intractable risk factors like gender, race and ethnicity, parental education,
income, property ownership and place of residence, home language), creating the impression that
early school leaving is in part a natural process – literally “attrition” – largely impervious to change
efforts (e.g. Finn, 1989; Appleton et al., 2008; Christenson et al., 2008).
Perhaps the focus should not so much be on dropping out as a problem of perceived or actual
failures of pupils, schools and the costs associated to it, but on dropout as an indication and origin of
fundamental inequities (Smeyers and Depaepe, 2006, p. 8-9). This perspective shifts the focus towards
school attendance and completion as a right of citizens that is to be safeguarded in any democracy
(Dorn, 1996) and calls for a more nuanced view on the many determinants of dropout (cf. Dorn, 1996).
While, it is increasingly recognized that early school leaving is a complex, multi-dimensional
phenomenon with numerous causes and consequences, it is still sometimes seen as a single symptom
of related problems (Dorn, 1996). And although early school leavers are increasingly considered as a
heterogeneous group (Rumberger, 1987; Jarjoura, 1996), they are still described in broad categorical
terms loaded with negative connotations. Dropout stereotypes thus risk being reproduced, in spite of
overwhelming evidence of their untenability. Two stereotypes often mentioned in policy debates
presume a correlation between dropout on the one hand and delinquency and unemployment on the
other.
Yet, the frequently drawn association between dropout and delinquency is all but univocal. A
study based on a large-scale nationally representative probability sample revealed that the propensity
to engage in delinquency after early school leaving depends on the reason for leaving and the poverty
status of the youth involved (Jarjoura, 1996). The study found that only those who leave education
early for personal reasons were more prone to display offending behaviour; those leaving for
economical reasons in fact appeared less inclined to offend than those who graduate, independent
from their poverty status.
Likewise, the connection between dropout and unemployment is ambiguous. Whereas fewer
employment opportunities for young adolescents helped increasing high school attendance and
graduation rates from the mid-1940s onwards (Dorn, 1996), so have more job market opportunities in
times of economic revival increased dropout rates (Olsen and Farkas, 1989; Marks and Fleming, 1999;
Cabus and De Witte, 2011). In countries like the United States, Australia and some European countries
(e.g. Portugal and Spain), teenagers have been drawn to the labour market in greater numbers (Cabus
and De Witte, 2012). While most of (school-leaving) youth become engaged only in part-time jobs with
short-term employment contracts, this need not imply a break from schooling, as to a certain extent it
mirrors increased enrolment in part-time education provisions. However, a significant number of them
are pulled out of school due to the attractiveness of the labour market (cf. Business Council of
Australia, 2002b). Unfortunately, once excluded from full-time employment, and without minimal
credential, dropouts’ experiences on the job market often do not qualify for an equivalent credential
(Dorn, 1996).
“Dropout discourse” has thus linked early school leavers with unemployment, urban poverty
and juvenile delinquency (often serving as a substitute for race and class) (cf. Dorn, 1996). Observed
determinants have thereby acted as stereotypes. The stereotype, par excellence, of the “culturally

5
deprived”, unintelligent, unskilled, unadjusted, non-white male adolescent, who ends up unemployed
and delinquent, has increasingly been qualified. Nonetheless, youngsters who display at least some of
the characteristics just mentioned are still taken as a starting point in recent studies. This can partly be
related to the way in which early school leaving has long been approached methodologically, that is: as
an aggregate of combined probabilities with regard to an array of separate risk factors, the overall
average of which is thought to represent the typical dropout (Reich and Young, 1975). In this context
Swadener (1995, p. 25), among others, has stressed: ‘what is particularly troubling and problematic is
the degree to which [for instance] children’s race, gender, class, first language, family makeup, and
environment all target them for this at-risk label and associated interventions.’ Many students
encounter circumstances that might place them at risk, and yet all – however hindered by nurture or
nature – are also “at promise” (Swadener, 1995). That this is true, also for “culturally diverse” children
and youth, is illustrated by a study of Herbert and Reis (1999). These authors looked at high-achieving
minority students, and focussed on why they stayed in school and achieved well, in spite of the many
risk factors they faced, rather than the other way around.
In sum, we can conclude that the problem of early school leaving implies more than the notion
of students failing to achieve academically and graduating from school. The issue may then not only be
how to better prepare them for schooling, or even how to attune schools more to their diverse needs.
Some may fail merely within the academic system, but nevertheless be forced to remain therein, as it
is believed that only schools can provide the kind of formal education and credentials needed for
successful transition to work and adulthood in general (Reich and Young, 1975; Swadener, 1995; Dorn,
1996). Whether or not one subscribes to this view and sees dropping out as problematic in itself, or
one views it as part of a broader problem related to questions of inequity, if the aim is to prevent
youngsters from leaving education early, then it seems worthwhile considering how school dropout
determinants are currently studied in the literature.
3 Underlying ‘determinants’ of early school leaving
The complexity of early school leaving is reflected, among other things, in the levels on which it has
been studied, and the kind of models and methods by which it has been investigated. Most studies on
school dropout seem to focus either on the national level, the state level, the level of a district, county
or city, or that of an individual school (see Table 1). The topic, moreover, has been analysed by means
of a large variety of models and methods (see Table 2).
< Table 1 about here >
Studies on the national level often analyse the same data sets. Used most frequently for the
United States are national longitudinal data. The latter, in particular, draw on the four studies thus far
conducted in the frame of the National Education Longitudinal Studies (NELS) programme of the
National Center for Education Statistics (NCES), namely: the National Longitudinal Study of the High
School Class of 1972, concluded in 1986 (NLS-72); the High School and Beyond (HS&B) study, which
started in 1980 and ended in 1993; the National Education Longitudinal Study from in 1988 (NELS:88),

6
with its follow ups until 2000; and the Education Longitudinal Study, initiated more recently in 2002
(ELS:2002). Also studied often, are cross sectional data obtained from the National Center for
Educational Statistics’ (NCES) Common Core of Data (CCD), a primary census database. Commonly
explored as well, are data from the Current Population Survey (CPS), the National Longitudinal Survey
of Youth (NLSY), and the National Longitudinal Surveys of Labour Market Experience of the Bureau of
Labour Statistics (BLS). Among other national data sources encountered frequently in US national level
studies, are the Monitoring the Future Study (MTF) of the National Institute on Drug Abuse, which
commenced in 1975 and is still on-going, and the statistical reports of the GED Testing Service (GEDTS),
a programme of the American Council on Education. Similarly, national level studies on early school
leaving in Australia seem to draw upon common data from the Australian Bureau of Statistics (ABS),
the National Centre for Vocational and Educational Research (NCVER), the Ministerial Council for
Employment, Education, Training and Youth Affairs (MCEETYA) and the Australian Council for
Educational Research (ACER), the latter of which conducted, for instance, national longitudinal surveys
as part of the Youth in Transition (YTT) programme. In the UK data mostly obtained from the
Department of Education and Skills (DfES), or the Higher Education Statistical Agency (HESA).
As has been noted by Vizcain (2005) relying on such broad-scale data sets has both advantages
and disadvantages. While it allows for consistency in patterns across time and space, extrapolating
information on early school leaving from national level data sources could also obfuscate trends on a
more local level. Conversely, it is evident that findings on a local level need not hold true on a wider
scale. However, this does not imply that there is nothing to be learnt from studies on such level. On
the contrary, as Fendler (2006, p. 56 and 61) has contended: ‘when research findings [are] held to be
generalizable from one setting to another, that practice confuses induction with prediction.
[Incidentally,] within statistical modelling, there is no basis for trust or certainty in the generalizability
of findings [as] probability is precisely not certainty. … [Arguably,] generalizability has itself become a
habitual expectation that continues to validate belief in itself.’ Indeed, what may be more problematic
with respect to studies at the level of individual schools and school systems, is that the statistics on
which they generally rely are still based on the grades in which students are, and on administrative
estimations of early school leaving, rather than students’ age and graduation; an important limitation
(cf. Dorn, 1996; and Allensworth, 2005; De Witte and Rogge, 2013).
From a methodological perspective, empirical-analytical or quantitative research predominates
the literature. Studies using more qualitative data are in short supply, which seems surprising, given
the nature of the topic that, in all its complexity, is inextricably bound up with meaning and values,
requiring a great deal of interpretation and judgement. At any rate, it seems an illusion that empirical
identification of all relevant factors and interactions will one day be complete, as has been suggested
by Frank (1990). Methodological pluralism (i.e., the use of mixed methods) is recommendable,
whereby the choice of method should depend on the research question(s) one seeks to answer
(Herbert and Reis, 1999).
< Table 2 about here >

7
Rumberger (2004a) and Plank et al. (2005) have observed that most studies apply standard
logit and multivariate models. Bivariate approaches (i.e., between-group comparisons) have become
less popular, yet to date they are still sometimes adopted to describe early school withdrawal. This is
mainly the case in policy-oriented statistical reports (e.g., Dalton et al., 2009). However, bivariate
analyses do not allow for interaction effects, such that the multiple dimensions of early school leaving
are at risk to being underexposed, and ‘stereotypes’ being sustained. The question should not only be
(whether and) which factors may increase the chance of early school leaving, for whom, why and when
(Willet and Singer, 1991), but also whether, when and why that this may be a problem, and – if
necessary – what could be done about it. In order to be able to find at least some answers to questions
like these, one needs a framework that is able to accommodate for a broad spectrum of relevant
factors. We aim at presenting such a framework in what follows. Thereafter, we explore a number of
often cited, potential “predictors” of early school leaving in Section 5.
4 A “photofit” of those most at risk?
Most often school-related characteristics are revealed as determinants of dropout over and above
family-related, work-related and other motives (Rumberger, 2004; Dalton et al., 2009). However, a
large part of the literature is still focused on factors not related to the school, but to pupils themselves
and their families. And even though many studies at least hint at the importance of both “proximal”
and “distal” factors – that is: aspects related to students, their families, schools and teachers, as well
as the community (from neighbourhoods to labour markets and society at large) – a considerable
number of studies focus only on one or some of these types of aspects (see, e.g., Ekstrom et al., 1986).
Indeed, the majority of research on early school leaving still endeavours to pin-point personal and
social characteristics of potential dropouts that may differentiate them from graduates, so as to create
a kind of “photofit” of those most at risk, for whom targeted intervention measures can then be
devised (Viscain, 2005). We have explicitly chosen not to follow this strategy. Rather, this review
attempts to locate and highlight aspects that unite both early school leavers and graduates alike, and
that may well exert a positive influence on all parties involved.
There exist various theoretical frameworks to model school dropout. The most early
frameworks are those developed by Tinto (1975), Spady (1970, 1971) and Finn (1989). The latter
author considers a lack of self-esteem as an important reason for student withdrawal, whereas the
former authors consider the lack of an optimal match with the school as a critical reason for school
dropout. Bean (1980) and Bean and Metzner (1985) include the labour market as a reason for student
attrition. The most cited theoretical framework, and the one adapted here – as illustrated by Table 3 –
is indebted to the work of Rumberger (1983, 2001 and 2004a), which is regularly cited in the literature
(e.g., Plank et al., 2005 and references therein). The typical distinction between “individual factors”
(student characteristics) and “institutional factors” (family, school and community characteristics), as
made by Rumberger, is however abandoned here. In our opinion, it might give the impression that the
weight to be given to individual student factors equals that of all institutional factors taken together. It
is a divide that could suggest, albeit involuntarily, that there are only two major lines of inquiry to
follow, which seems at odds with the acknowledgement that there are numerous causes of early

8
school leaving (Blue and Cook, 2004). With regard to the latter, Rumberger (2004a) contends that it is
a near hopeless task to prove sustained causal effects of the many factors involved in early school
leaving, the more so because their impact changes over time. In fact, like most scholars, he considers
early school leaving to be merely the last phase of a dynamic, cumulative and multidimensional
process of disengagement.
< Table 3 about here >
In the framework, the various observed predictors of early school leaving on the level of
students, families, schools and the community are explored separately. Nevertheless, they are
inextricably bound up with each other. It makes no sense to view these characteristics isolated from
each other, as they interact in countless ways. Neither student attributes, nor family or school
characteristics can be seen apart from society at large (Reich and Young, 1975). Attempting to
disentangle their effects from each other by means of ever more sophisticated statistical modelling,
may thus not only prove to be a tremendous challenge (cf. Rumberger 2004a), perhaps it is not even
always worth the effort (Smeyers, 2006).
5 Potential predictors of early school leaving
5.1 Student-related factors
One of the student-related factors that have been associated with early school leaving is academic
achievement (sometimes referred to as academic ability). It is most commonly measured using cross-
sectional data via standardized testing (particularly on mathematics and language), by local school
tests and (exit) exams, but also by other indicators, e.g. school retention and enrolment in special
education, remedial or college-preparatory tracks. To an increasing extent this is done longitudinally,
in order to discern the effect of students’ pathways in terms of achievement or skills (Cooper and
Chavira, 2005). Whether measured by exam success (e.g., Dustmann and van Soest, 2007), grade point
average (e.g., Entwisle et al., 2004), test scores (e.g., Ekstrom et al., 1986; Dalton et al., 2009) or
literacy and numeracy skills level (Business Council of Australia, 2002a), most scholars have found that
early academic achievement in elementary and secondary school is predictive of early school leaving
(Rumberger, 2004a). Entwisle et al. (2004), however, found no effect in terms of composite test-score
quartiles. Allensworth (2004 and 2005), moreover, questions whether it can be shown to have a direct
effect. He suggests it may also lead to less retention, and hence a lower chance of dropout. Plank et al.
(2005) further found no effect of academic achievement for older subgroups of students (i.e., those
past the typical grade age); for them grade retention may predict early school leaving more accurately.
The latter authors additionally stress that not only early achievement, but also grade point average
(and course taking) in the most recently completed term could adequately predict non-graduation.
Thus, even very recent negative experiences in terms of achievement can be decisive factors.
An even stronger – if not the strongest – predictor of early school leaving, however, is grade
retention, sometimes bracketed together with an accumulation of credit deficits. Many studies suggest
that being past the typical age in a grade significantly increases the hazard of leaving school early

9
(Rumberger, 2004a). According to Plank et al. (2005) and Entwisle et al. (2004 and 2005) being ‘off-
age’ is a factor that overshadows most other effects, including academic ability or achievement. The
last-mentioned authors furthermore add that grade retention significantly increases the likelihood of
leaving school permanently, rather than just temporarily. They ascribe this effect to the fact that being
retained in the strictly age-based school system is associated with the stigma of being unintelligent,
having failed, and lagging behind. Other scholars, including Allensworth (2004 and 2005), agree that
the strong correlation of pupils’ grade level and early school leaving is entirely explained by age. In
addition, they suggest that several factors “pulling” students away from school, such as teenage
pregnancy and high school employment, have a higher chance of occurring the older students become.
However, they draw attention to the possibility that research on teacher-initiated retention has not
always been successful in accounting for possible intermediary variables, such as overall
disengagement from school that could cause retention and thus dropping out. Retention by a
“promotional gate” (high stakes standardized testing), they show, revealed less univocal effects; it only
significantly increased the likelihood of early school leaving in those students already off-age.
Nevertheless, this kind of retention seems hardly recommendable, not least because it worsens
disparities between students of different gender or skin colour. While schools in practice often still
consider grade retention as necessary, Blue and Cook (2004) conclude that it provides only a short-
term solution at best.
Other predictors of both early school leaving and graduation are academic and professional
aspirations or expectations, even if exogenous factors are taken into consideration (e.g., Rumberger,
1983). Not unrelated to this, is the influence of engagement, typically measured by school attendance
or absenteeism, and (good or problematic) behaviour (Rumberger, 2004a). Quite some scholars found
that a lack of engagement in elementary and middle school predicted early withdrawal from high
school. Appleton et al. (2008) has noted in particular the effect of psychological and cognitive subtypes
of engagement. However, Entwisle et al. (2005) have remarked that even if engagement is a good
estimator of non-graduation, it is not one as powerful as grade retention. In turn closely related to
engagement, and other factors mentioned, are negative attitudes, feelings, perceptions and traits,
which potentially result in problematic comportment and discipline problems. Such attributes may
include: an externalized locus of control (Ekstrom et al., 1986; Blue and Cook, 2004); low motivation
(Adams and Becker, 1990; Herbert and Reis, 1999); a problematic temperament, disposition, or
feelings of inferiority and self-defeat (Entwisle et al., 2004); lack (versus abundance) of sensitivity and
resilience to overcome problems and adversity (Herbert and Reis, 1999); psychological or behavioural
problems like aggression, anxiety, and disciplinary problems, suspensions, cutting classes or trouble
with the police (Ekstrom et al., 1986; Viszcain, 2005).
Apart from that, substance (ab)use is sometimes mentioned as a factor contributing to early
school leaving. Fergusson et al. (2003) found that students who used cannabis had a higher risk of
school leaving, even though nothing in their prior school history indicated that this would become the
case. Prior school leaving – also termed previous withdrawal, temporary dropout or “stopout” – has
furthermore been found to affect the likelihood of non-graduation. With respect to this, DesJardins et
al. (2006) have noted the importance of both its occurrence and duration. In itself stopping out already
has the potential to predict a higher chance of early school leaving, but in case stopouts occur

10
repeatedly, and when the first stopout is rather lengthy, the likelihood of more stopouts and eventual
dropout increases. Similarly, truancy is known to have a deleterious effect (cf. Rumberger, 1983; Olsen
et al., 1987; and Henry, 2007). Finally, teenage pregnancy, marriage and parenthood have been shown
to result in a higher probability of leaving school before graduation (Rumberger, 1983: Kalmijn and
Kraaykamp, 2003).
Even though, as mentioned above, it may not be opportune to focus too much on immutable
variables, we mention some demographic or background factors often cited in the literature. With
respect to gender, many studies indicate that males have a higher propensity to drop out than
females. USNCES (US) suggests that, in general, (event) dropout rates have not tended to differ
significantly across both sexes over the last 30 to 35 years (Kaufman et al., 2004; Cataldi et al., 2009).
As for race and ethnicity, there seems to be much debate and considerable contradiction. Especially US
and Australian studies suggest that being black, Hispanic/Latino or indigenous, rather than Caucasian,
increases the likelihood that one leaves education early. On the other hand, in this context it has been
suggested that being Asian/Pacific descent decreases this probability (e.g., Bynum and Thompson,
1983; Ekstrom et al., 1983; Business Council of Australia, 2003b; Ishitani and Snider, 2006). Over the
past few years the gap between white and non-white youths has closed, albeit slowly and rather more
among females than males (Kaufman et al., 2004; Dalton et al., 2009; Cataldi et al., 2009). However,
other scholars contend that race and/or ethnicity do not have a significant effect once accounted for
factors as family background and student characteristics (e.g., Rumberger, 1983; Balfanz and Legters,
2004; Plank DeLuca and Estacion, 2005; Entwisle et al., 2004 and 2005; DesJardins et al., 2006).
Among minority students, the time since their immigration may play a key role. With regard to
this, contradictory findings emerge from the literature. Based on NELS:88 data, Brisboll (1999) and
Viscain (2005) have suggested that not the recently immigrated (Hispanic/Latino) pupils have a higher
chance of leaving education early, but surprisingly third generation immigrants are more likely to drop
out. In contrast, Blue and Cook (2004) and Cataldi et al. (2009) have mentioned, on the basis of more
recent data, that Hispanic/Latino students born in the US tend to have lower dropout rates than
second or higher generation students. This may raise questions about the adequacy of terms like race
or ethnicity. The connection with one’s ethnic origins or cultural background may well fade over time,
even if one continues to be labelled as belonging to a certain ethnicity. An overview of student related
factors described in earlier literature is provided in Table 4.
< Table 4 about here >
5.2 Family-related factors
Among family-related factors, “social class” or “socioeconomic status“ (SES) is the most contested one.
Often it is measured by parents' (or guardians’) occupational status, education and income, all of
which are sometimes considered influential (e.g., Dalton et al., 2009). More frequently, only some of
these factors are deemed predictive of early school leaving. Thus, for instance, parents’ educational
level, and the educational aspirations for their children, is mentioned by many scholars, among whom
Duchesne et al. (2005), Ishitani and Snider (2006), and Koball (2007). Parental employment is also
believed to be an adequate estimator of the students’ likelihood of leaving education before

11
graduating (see, e.g., Marks and Fleming, 1999; and Business Council of Australia). In addition, families’
“cultural index”, or the extent to which they have reading material available in the household, has
been argued as a more solid predictor of early school leaving across all racial and both sex groups
(Rumberger, 1983).
The school dropout determinant over which most disagreement exists is family income.
Several scholars stress the importance of parental income, either without clear specifications (e.g.
Dorn, 1996; Blue and Cook, 2004; Ishitani and Snider, 2006; Ou and Reynolds, 2006; Cataldi et al.,
2009); or only in case parents’ income is below the poverty line (Orthner et al., 2002); or when low
family income is combined with structural aspects such as family disruption (Suet-Ling, 2000). Others
have stated that its influence holds good only among whites (Rumberger, 1983), while others again
have contended that aspects like “human capital” and parents' acquaintance and comfort with the
school system are of more importance, as is the case for the factor race/ethnicity (Frank, 1990;
Duchesne et al., 2005; Plank et al., 2005).
More unanimity is observed with regard to family structure; students from large families, that
is with five or more siblings, prove to be disadvantaged in terms of graduation prospects (e.g. Kalmijn
and Kraaykamp, 2003; Dustmann and van Soest, 2007); children from single-parent households also
seem to be more likely to dropout (Bridgeland et al., 2006); as do children with step parents (Olsen
and Farkas, 1989; Plank et al., 2005). Parental support or involvement is also known as a predictor of
school dropout, irrespectively of income and ethnicity (Cooper et al., 2005). In fact, it may be the single
most significant family factor scholars have agreed upon (Ishitani and Snider, 2006). Finally, the
emotional climate of the parent-child relationship is an important predictor, often in interaction with
other family aspects (Duchesne et al., 2005). An overview of family related factors is presented in
Table 5.
< Table 5 about here >
5.3 School-related factors
With respect to school-related aspects, the type of school may correlate with students’ educational
outcomes, including eventual graduation. Grammar schools that are more selective tend to have fewer
early school leavers than non-selective, secondary modern technical or vocational schools (Dustmann
and van Soest, 2007). In addition, Balfanz and Legters (2004) have asserted that if a school has more
“promoting power” (that is: an overall higher percentage of pupils passing timely from one grade to
the following) – perhaps evidently – dropout is less. Thus, schools that are attended by minority
students tend to have low promoting power, especially majority minority schools. With regard to
college leaving, it may also matter whether one has been at an independent or state (Local Education
Authority) school – at least in the United Kingdom (Smith and Naylor, 2005). If students first attend a
private independent school, their level of (university degree) performance tends to be lower, which
could be explained by the fact that in college eventual “ability deficits” of these students are no longer
compensated by higher resources available in their previous school. Similarly, students attending a
public/government school, rather than a Catholic or other private high school in the US, generally have
a higher chance of leaving school early (Dalton et al., 2009), as do students frequenting a “poverty

12
school”, that is: a school with a high percentage of students on free or reduced-price lunch programs
(cf. Okpala et al., 2001; and Dalton et al., 2009). As Rumberger (2004a) has argued, such effects may in
part be due to schools’ student composition, an aggregate of students’ individual characteristics on a
social level. From the literature it seems clear that a balanced student composition (contrary to the
one in majority minority schools) is one to be aimed at.
Closely related with the type of schools are schools’ resources, a standard most frequently
defined by class size (e.g., Pittman, 1993) and the teacher-pupil ratio (e.g., Balfanz and Legters, 2004).
In fact, one of the reasons why independent schools may perform better and why parents often
choose independent schools, is because they have small-sized classes. As Smeyers (2006) has
contended, there are a number of reasons why smaller class sizes and lower teacher-pupil ratios may
have a positive effect on school achievement. For one thing, various aspects may differ between
smaller and larger classes, among which teachers’ educational practice (Van Klaveren and De Witte,
2013). Historically, however, the latter has been shown to be resistant to change (e.g., Cuban, 1993;
Tyack and Cuban, 1995; Depaepe et al., 2000), which may explain the small benefits found related to a
smaller class size. Other aspects may reduce the benefits of small-sized classes, such as the age of
students, their well-being, teachers’ workload, etc. – all measures that in Smeyers’ (2006) opinion may
prove to be difficult, if not impossible, to objectify and investigate empirically.
Different and more structural school aspects explored perhaps to excess in the literature are
school size and programme diversity. This is a topic over which there has been considerable debate.
Some scholars have contended that smaller schools (counting, for instance, less than 1,500 students;
cf. Blue and Cook, 2004) are likely to result in lower rates of early school leaving (Pittman, 1993). In
contrast, Pittman and Haughwout (1997), among others, have demonstrated what may seem self-
evident, namely that the effect of school size on dropout is almost entirely related to schools’ social
climate, and more particularly the influence of student participation as well as the amount of problems
in the school environment. In general, larger schools have greater programme or curriculum diversity,
but a less positive social climate. However, Plank et al. (2005) have pleaded to move away from such
general assertions; pupils’ diverse skills, interests, and learning needs have to be taken into account
when varying effects of school size and programme diversity emerge.
More important, perhaps, than the somewhat intractable characteristics of a school is the
latter’s policy and regular practice. A crucial factor seems to be schools’ social and academic climate,
made operational, e.g., through a general sense of cohesion, a high level of participation in school
activities, smooth student-faculty interaction and the extent to which there are problems at school
(Pittman and Haughwout, 1987; Finn, 1989). Also group differences in educational attainment may
play a role (Ou and Reynolds, 2006), as well as academic and social integration (Pitman, 1993), and the
kind of courses available (e.g., academic or college preparatory versus vocational courses) (Viscain,
2005; Business Council of Australia, 2002b). Plank et al. (2005) have noted that with respect to course
taking, there is a clear effect for younger students, but not for older ones. This may be due to the fact
that pupils who made it through the earliest grades, are better situated to make it to the final grades.
Be that as it may, having available appropriately challenging courses in each case appears important
(Herbert and Reiss, 1999), as well as having plenty of opportunities for extracurricular activities, and
after-school, summer or special programs (e.g., Pitman, 1993; Herbert and Reiss, 1999). Moreover,

13
teachers’ experience (Adams and Becker, 1990), expectations (Dalton et al., 2009), support (Herbert
and Reis, 1999), and instruction quality (Blue and Cook, 2004) are all aspects that influence the
propensity to drop out. Crucial thereby seems to be students’ perceptions of teacher (and teaching)
quality, rather than that of school principals (e.g., Bridgeland et al., 2006; Rumberger, 2004a).
With regard to instructional quality, Blue and Cook (2004) also stress the importance of
cultural relevance and student-teacher cultural synchronization; school environments and teacher
attitudes and comportments devaluing and/or negating students’ cultural identity and diversity risk
alienating students and creating resistance to learning, in spite of apparent talent. This issue is
inextricably connected to schools’ social capital, that is: the presence of caring teachers (Blue and
Cook, 2004), an enjoyable school culture (Business Council of Australia, 2002a) and good student-
faculty interaction (Pittman and Haughwout, 1987). A summary of school related factors described in
earlier literature is presented in Table 6.
< Table 6 about here >
5.4 Community-related factors
As has been stressed earlier, student attributes, school characteristics and family background factors
cannot be viewed apart from the broader context in which they are embedded and by which they are
inevitably influenced. Neighbourhood characteristics – the geographical location of families’ residence,
eventual housing problems, lack of playgrounds and green areas (Rumberger, 1983 and 2004a) – may
have detrimental effects on students’ school performance, either directly or indirectly. If youths live in
poor and distressing environments they may be more susceptible to early school leaving (Blue and
Cook, 2004). Just as “urbanicity” may to some correlate heavily to early school leaving, so could a
whole region in which students live be associated with higher dropout rates. This used to be the case,
for instance in the South of the US (Ekstrom et al., 1986), although the latter no longer seems to be the
case (Kaufman et al., 2004).
At least equally important appears to be the presence of a network of high achieving and high
aspiring peers in children’s and youths’ environment. This factor could exert an influence independent
from other variables (Cooper et al., 2005). In addition, employment or apprenticeship opportunities
could act as powerful “pull factors” stimulating students to stop out or drop out (Olsen and Farkas,
1989; Pittman, 1993; Marks and Fleming, 1999). Much depends on the type of employment in which
youth engage, the intensity of the work exercised, the amount of stress associated to the job, whether
or not a stable work pattern is maintained, whether one is male or female, and whether one works in
order to support one’s family or not (Rumberger, 2004a; Entwisle et al., 2004 and 2005; and Dustmann
and van Soest, 2007).
Of course, many other community factors and societal mechanisms could play a crucial role,
like social discrimination and prejudice (Herbert and Reis, 1999). Such processes have caused minority
groups to be “streamed” into special and vocational education tracks for ages. They may moreover still
be responsible for differences in dropout and downward mobility between minority and majority
students (Kalmijn and Kraaykamp, 2003). An overview of community related factors is presented in
Table 7.

14
< Table 7 about here >
5.5 The complex interaction
As Smeyers (2006) has contended, within education it is perhaps not so important to observe that
numerous variables are at work, of which many undoubtedly are relevant but, rather, which of these
factors have a more significant influence on dropout. Not the many separate elements are likely to be
relevant but precisely the complex and dynamic interactions between them (Smeyers, 2006, p. 103-
104 and 107).
For example, the interaction of ethnicity (or race) and sex, respectively, with attitudes,
subjective norms (perceived expectations of teachers), perceived behaviour control, and retention
seems noteworthy. Blue and Cook (2004), for instance, have found for the US that if a student is black
or Hispanic and male, he is more likely to display negative attitudes towards education, perceive his
teachers as having low expectations of him, and situate the locus of control over important things in
his (school) life outside of himself. Thus, at least some minority students evidently risk ending up in a
vicious circle.
Similarly, the interaction between parental involvement, on the one hand, and ethnicity, family
income, and home environment, on the other hand, seems to be of some importance. Okpala et al.
(2001) found in this respect that, although parental involvement matters a great deal, its effectiveness
depends on the kind of involvement parents show, but also, and perhaps equally essential, on their
ethnicity, income and home environment. In other words, cultural and structural barriers may have to
be removed before parental involvement can be successful.
Likewise, employment among high school students, with or without grade retention, does not
by definition result in early school leaving. It has been observed that the job market heavily interacts
with students’ family background. Entwisle et al. (2004 and 2005) observed in particular that students
from less advantaged backgrounds working after school were not more likely to drop out, contrary to
their more well-off counterparts. In fact, the schoolwork of students from families with very low
incomes, did not even deteriorate when the so-called “intensity threshold” of twenty hours of work
per week was surpassed. For them, and for other students from disadvantaged backgrounds,
employment may help them acquire otherwise unobtainable human capital. The authors further found
that among retained students, those who maintained an intensive (adult) but stable work pattern
between the age of fifteen and sixteen had a lower risk of dropping out than those who took on easy
(typical teenagers’) jobs at the age of fifteen and more tough (adult) ones the following year.
In addition to these interactions, “intermediary” factors (that is: factors that cannot easily be
situated on just one of the levels involved) could matter substantially. For instance, “cultural
discontinuities” that originate from frictions or fissures between students’, families’, schools’ and
society’s goals, values, perceptions, activities, styles of communication, etc. (cf. Cooper et al., 2005),
may play a key role.
6 Policy strategies

15
Any policy decision of relevance must necessarily focus on the whole aggregate of factors at the level
of students, families, schools and the broader environment. Whether one views dropping out as a
problem in itself or not, there is neither a single or simple solution to be found. Yet, however many
“support factors” one envisages, they will have to concern more than just students and their families
(Frank, 1990; and Dorn, 1996), contrary to what is still sometimes suggested (cf. WWC Intervention
Report, 2006). Adequate policies will have to address both the “social” and “academic” issues
associated with early school leaving. With regard to this, Rumberger (2004b) has discussed strategies
of a “systemic” nature (involving programmes that try to ameliorate students’ environments by
supporting and/or restructuring them with the help of resources and other forms of assistance) and
strategies of a more “programmatic” nature (attempting, rather, to influence students’ behaviours,
thoughts and feelings, values). A combination of both strategies is probably advisable, even if dropping
out in itself is the sole issue targeted. As many studies have discussed policy recommendations that
may be both effective and meaningful. In accordance with the framework outlined above, hereafter
we will therefore zoom in on several proposed and/or tested policy measures.
6.1 Measures aimed at students
Since research indicates (most often by correlations) that grade retention is the worst culprit among all
student-related risks factors with regard to early school leaving, it is of primary importance to restrict
its use (Dorn, 1996; Entwisle et al., 2005; Vizcain, 2005). As Orthner et al. (2002) have noted, the issue
of grade retention versus promotion is heavily charged; it seems neither wise to delay children’s entry
into high school or transition to a higher grade, nor to advance them without the skills necessary to
succeed in later years. The key is, they state, to identify those at risk of grade retention as soon as
possible, and to provide special care for them, both within and outside school.
Similarly, Adams and Becker (1990) have recommended that teaching support be offered to
first-year students, but insisted on its availability for more experienced students as well. Orthner et al.
(2002), in turn have added that purposive assistance is best arranged even before kindergarten and
should moreover be complemented by extracurricular activities (involving music, dance, drama and
the like) and after-school programmes. Particularly disadvantaged children and youth would benefit
from the latter. Orthner et al. (2002) argue: ‘an integrated strategy with clear objectives is much more
effective than a diverse strategy with multiple objectives. Children [...] need their own integrated,
community-supported strategy with clear direction and mobilized in-school, after-school, and
community-based resources to ensure that they arrive and leave school ready to learn and succeed’ (p.
119).
Promising strategies to enhance academic achievement, even among minority students from
disadvantaged backgrounds, may be found in peer and adult counselling programmes. Teachers,
coaches, peers, family members, and sometimes mentors from community programmes have proved
capable of motivating students to achieve and even strive for academic honours by acting as
supportive role models (Herbert and Reis, 1999). Measures aimed at facilitating social attachments
among all those involved is essential, especially at key moments in pupils’ school live, like the
transition into high school (Blue and Cook, 2004). In addition, it appears worthwhile to devise

16
programmes addressing students’ (culturally diverse) attitudes toward and perceptions of school
responsible for underachievement (Ekstrom et al., 1986; Vizcain, 2005).
Finally, there seems to be agreement among scholars that for disadvantaged students work
during high school needs not to be discouraged (Ekstrom et al., 1986; Entwisle et al., 2005). Yet, at the
same time there is a need for clarification of the circumstances in which work either increases or
decreases students’ propensity to leave school early. Entwisle et al. (2005), therefore, as a measure of
precaution, have advised that students be dissuaded from taking up an adult job before the age of
sixteen.
6.2 Measures aimed at families
In order to be effective, policies should not involve students alone but will have to engage students’
parents (or guardians) as well (Reich and Young, 1975). Since involvement of parents in the academic
achievement of their children has proved to be extremely important, parent engagement strategies
seem a necessary path to follow. If well conceived, these may help parents supervise and regulate
their sons’ and daughters’ activities, discuss with them eventual problems and promote in their
children a certain degree of self-reliance (Bridgeland et al., 2006). There is some evidence that early
childhood (preschool) intervention programmes have positive effects in this regard (cf. Ou and
Reynolds, 2006).
Cooper et al. (2005) have stressed the importance of high and unambiguous expectations on
the part of parents as well as other adults involved in students’ school life, such as counsellors,
teachers, school principals, etc. They have warned, however, against a paternalistic attitude, not least
towards parents from low-income or minority groups.
One way to ensure that parents feel understood is to foster their supportive activities through
parent discussion groups. Herbert and Reis (1999) have recommended that such groups be set up by
school counsellors but run by successful parents in their homes.
More generally, policies have to be focussed on optimizing families’ living conditions in order
to secure an inviting environment for studying and a healthy degree of student responsibility in the
household (Blue and Cook, 2004; Haelermans and De Witte, 2013), and, moreover, on obtaining a safe
emotional climate and parent-child relationship (Duchesne et al., 2005). Finally, welfare programmes
need to offer assistance for single parents who suffer a dramatic income loss after having divorced
(Suet-Ling, 2000).
6.3 Measures aimed at schools
Since the 1980s, it has increasingly been recognized that apart from personal guidance of students,
also strategies have to be developed to influence schools’ organization (Dorn, 1996). The literature
focused on schools’ environment, teacher and teaching characteristics, and schools’ relation to both
families and community.
With regard to the former, Swadener (1995) and te Riele (2006) have stressed that the focus
needs to be on establishing school environments adapted to the needs of diverse students, rather than
the other way around. In a similar vein, Balfanz and Legters (2004) and Bridgeland et al. (2006) have
called for student outreach, especially in case of difficulty, and underlined the value of a school climate

17
that cherishes academics and maintains high standards. Yet the school atmosphere, Blue and Cook
(2004) have stressed, should at the same time be authentic and caring and defer to pupils’ cultural
diverse identities and home languages, while seeing the latter as strengths rather than weaknesses.
Pittman and Haughwout (1987) have advised schools to remain sufficiently small (that is: not to merge
into mega-schools) and to foster a positive social climate through a high degree of pupil participation,
while containing problems as much as possible.
Also in view of this social climate, teaching approaches have been proposed that involve
discussion and conversation, while relating the school to students’ lives (Cooper et al., 2005;
Bridgeland et al., 2006). Other scholars have suggested increased personalization (Balfanz and Legters,
2004; Blue and Cook, 2004; Lee and Burkam, 2003) and technological orientation (Pittman, 1993) in
teaching. With respect to content, some have recommended the development of literacy and
language across various courses, as well as instruction of complex thinking (Cooper et al., 2005). In
general, educational programmes should be intensive and courses challenging (i.e., more academic
and less remedial) in order to close eventual gaps in terms of achievement (Lee and Burkham, 2003).
Finally, in terms of teacher and trainer quality, coherent and long-term professional
development strategies, guidance, care and support for teachers are advocated (Balfanz and Legters,
2004; European Commission, 2006). Some scholars plead for teachers to be allowed to concentrate
their instruction activities in one or two terms, as to increase their teaching quality (Adams and Becker,
1990).
7 Discussion
7.1 Alternative credentials as an answer to school dropout
This literature review has made clear that the role of the economy, politics, and society in general is
often left out of the picture. Moreover, school systems’ organization and its effect on early school
leaving is also still underexplored. As a Dutch case study (Kalmijn and Kraaykamp, 2003) has suggested,
its very conception may sometimes lead to unequal chances of “attrition” between majority students,
on the one hand, and minority students, on the other. Rather than having “dropped out”, the latter
may often have been “facilitated out”, in other words: driven out of the common education system by
teachers’ and other personnel’s low aspirations and incitements to leave (cf. Vizcain, 2005, p. 469).
In this case, there should be an alternative for the minimum credential. In the US, early school
leavers in the past four decades have been encouraged to obtain an alternative credential – a high
school equivalency – by taking a GED (General Educational Development) test, which the American
Council of Education administers. Over the years, however, the number of students doing so has risen
to such a point that the credential’s economic value has been put into question (cf. Rumberger and
Lamb, 2003). Similarly, in Australia many early school leavers, instead of returning to school at a later
age, choose to attend a TAFE (Technical and Further Education) “college”, which is supposed to
provide an equivalent to a senior school certificate, especially given that having attended school up
until year 12 has more and more become a prerequisite for entry. Another frequently chosen pathway
is that of VET (Vocational Education and Training), through which early school leavers can obtain a

18
Certificate II – a year 12 equivalent, according to the OECD. Yet also in Australia, the value of such a
credential has been questioned (Business Council of Australia, 2002b).
So, while alternative diplomas have been, and continue to be, advanced as adequate, if not
ideal answers to early school leaving, they do not put alternative credential holders on the same
footing as high school graduates on the labour market and thus fail to solve problems early school
leaver encounter there. This has also been the conclusion of Dorn (1996), who has criticized the
reliance on high school credentials for adult education. In his view, ‘the dynamics of credentials has
fostered an artificial demand for alternative credentials that has supplanted [...] adult education’ (p.
133), which seems much more important than the credentials in question. In fact, “credentialism” may
one of be greatest problems with regard to dropout.
7.2 Trends and future research
Finally, some last words on current trends within the dropout literature and on viable directions for
future research may not be out of place. One prominent trend in current research on early school
leaving is to move away from investigating whether a certain factor increases the risk of non-
graduation in students in general, and to explore instead when and in whose school careers are more
likely to exert a positive or negative influence. This requires more complex, longitudinal and/or
retrospective studies on dropping out as a long-term process of disengagement (Bridgeland et al.,
2006). More research is also needed on ethnic differences in early school leaving (Kalmijn and
Kraaykamp, 2003). While several scholars have pleaded for more experimental, evidence-based
research, particularly with respect to dropout prevention/intervention programmes (e.g., Plank et al.,
2005; Sinclair et al., 2005), others have denounced the fact that empirical research ‘often plays it too
safe and engenders more of the same [...], more details of what is in the end irrelevant. Instead, to
make real progress, empirical research should take risks and play a more imaginative, possibly
dangerous game’ (Smeyers, 2006).
References
Adams, J. L. and W. E. Becker (1990). Course Withdrawals: A Probit Model and Policy
Recommendations. Research in Higher Education 31(6), 519-538.
Allensworth, E.M. (2004). Ending Social Promotion: Dropout Rates in Chicago after Implementation of
the Eight-Grade Promotion Gate. Consortium on Chicago School Research.
Allensworth, E.M. (2005). Dropout Rates after High-Stakes Testing in Elementary School: A Study of the
Contradictory Effects of Chicago's Efforts to End Social Promotion. Educational Evaluation and
Policy Analysis 27(4), 341-364.
Appleton, J. J., S. L. Christenson, et al. (2008). Student Engagement with School: Critical Conceptual
and Methodological Issues of the Construct. Psychology in the Schools 45(5), 369-386.
Balfanz, R. and N. Legters (2005). Locating the Dropout Crisis. Which High Schools Produce the Nation's
Dropouts? Where Are They Located? Who Attends Them? Center for Research on the Education of
Students Placed at Risk (CRESPAR).

19
Ball, K. and S. Lamb (2001). Participation and Achievement in VET of Non-Completers of School. Report,
ACER.
Bean, J.P. (1980). Dropouts and turnover: the synthesis and test of a causal model of student attrition.
Research in Higher Education 12, 155-187.
Bean, J.P. and Metzner, B.S. (1985). A conceptual model of nontraditional undergraduate student
attrition. Review of Educational Research 55 (4), 485-540.
Blue, D. and J. E. Cook (2004). High School Dropouts: Can we Reverse the Stagnation in School
Graduation? Study of High School Restructuring 1(2), 1-11.
Bridgeland, J.M., J.J. Dilulio and K.B. Morison (2006). The Silent Epidemic. Perspectives of High School
Dropouts. Civic Enterprises.
Business Council of Australia (2002a). Realising Australia’s Commitment to Young People. Scope,
Benefits, Cost, Evaluation and Implementation. Dusseldorp Skills Forum and Applied Economics.
Business Council of Australia (2002b). Young Persons' Education, Training and Employment Outcomes
with Special Reference to Early School Leavers. Dusseldorp Skills Forum and Applied Economics.
Business Council of Australia (2003a). The Cost of Dropping Out: The Economic Impact of Early School
Leaving. Dusseldorp Skills Forum and Applied Economics.
Business Council of Australia (2003b). Overview of Transition Programs: Policies and Programs.
Retrieved 30/03/2010 from http://www.bca.com.au/content.asp?newsID=87400.
Bynum, J. and W. Thompson (1983). Dropouts, Stopouts and Persisters: the Effects of Race and Sex
Composition of College Classes. College and University 59(1), 39-48.
Cabus, S. and De Witte, K. (2012), Naming and shaming in a fair way. On disentangling the influence of
policy in observed outcomes. Journal of Policy Modeling 34, 767-787.
Cabus, S. and De Witte, K. (2011), Does School Time Matter? On the impact of compulsory education
age on school dropout. Economics of Education Review 30, 1384-1398.
Cataldi, E.F., J. Laird and A. KewalRamani (2009). High School Dropout and Completion Rates in the
United States: 2007 (NCES 2009-064). National Center for Education Statistics, Institute of
Education Sciences, US Department of Education. Washington, DC. Retrieved 10/03/2010 from
http://nces.ed.gov/pubsearch/pubsinfo.asp?pubid=2009064.
Christenson, S.L., M.F. Sinclair, C.A. Lehr and C.M. Hurley (2000).”Promoting Successful School
Completion”. In D. Minke and G. Bear (Eds.), Preventing School Problems-Promoting School
Success: Strategies and Programs that Work, pp. 377-420.
Cooper, C. R., G. Chavira, et al. (2005). From Pipelines to Partnerships: A Synthesis of Research On How
Diverse Families, Schools, and Communities Support Children's Pathways Through School. Journal
of Education for Students Placed at Risk 10(4), 407-432.
Christenson, S.L., A.L. Reschly, J.J. Appleton, S. Berman, D. Sprangers, and P. Varro (2008). Best
Practices in Fostering Student Engagement. In: A. Thomas and J. Grimes (Eds.). Best Practices in
School Psychology. Washington, D.S.: National Association of School Psychologists, pp. 1099-1120.
Cuban, L. (1993). How Teachers Taught: Constancy and Change in American Classrooms, 1880-1990.
New York: Teachers College Press.

20
Dalton, B., E. Gennie and S.J. Ingels (2009). Late High School Dropouts: Characteristics, Experiences,
and Changes Across Cohorts (NCES 2009-307). National Center for Education Statistics, Institute of
Education Sciences, US Department of Education. Washington, DC.
Depaepe, M. et al. (2000). Order in Progress: Everyday Educational Practice in Primary Schools,
Belgium, 1880-1970 (Series: Studia Paedagogica, vol. 29). Leuven: Leuven University Press.
DesJardins, S.L., D.A. Ahlburg, et al. (2006). The Effects of Interrupted Enrolment on Graduation from
College: Racial, Income, and Ability Differences. Economics of Education Review 25(6), 575-590.
De Witte, K. and M. Csillag (2013), Does anybody notice? On the impact of improved truancy reporting
on school dropout. Education Economics. In Press.
De Witte, K. and Rogge, N. (2013), Dropout from secondary education: all's well that begins well.
European Journal of Education 47 (4), 1-20.
De Witte, K. and Van Klaveren, C. (2012), Comparing students by a matching analysis – on early school
leaving in Dutch cities. Applied Economics 44 (28), 3679-3690.
Dorn, S. (1996). Creating the Dropout. An Institutional and Social History of School Failure. Westport-
Connecticut/London: Praeger.
Dustmann, C. and A. van Soest. Part-Time Work, School Success and School Leaving. Empirical
Economics 32, 277-299.
Ekstrom, R., M.E. Goertz, J.M. Pollack, and D.A. Rock, Who Drops Out of High School and Why?
Findings from a National Study. Teachers College Record 87(3), 356-373.
Entwisle, D.R., K.L. Alexander and L. Steffel-Olson (2004). Temporary as Compared to Permanent High
School Dropout. Social Forces 82(3), 1181-1205.
Entwisle, D.R., K.L. Alexander and L. Steffel-Olson (2005). Urban Teenagers. Work and Dropout. Youth
Society, 37 (3), 3-31.
European Commission, Education and Culture (2006). Detailed Analysis of progress towards the Lisbon
Objectives in education and training: Analysis of Benchmarks and indicators? Retrieved
10/03/2010 from
http://ec.europa.eu/education/policies/2010/doc/progressreport06annexes.pdf.
Fendler, L. (2006). “Why Generalizability is not Generalizable”. In: P. Smeyers and M. Depaepe (Eds.).
Educational Research: Why 'What Works' Doesn't Work. Dordrecht: Springer, pp. 51-64.
Fergusson et al. (2003). Cannabis and educational achievement. Addiction 98(12), 1681-1692.
Finn (1989). Withdrawing from School. Review of Educational Research (59)2, 117-142.
Frank, J. R. (1990). High School Dropout: A New Look at Family Variables. Social Work in Education,
13(1), 34-47.
Haelermans, C. and K. De Witte (2013). Does residential social mobility improve educational
outcomes? Evidence from the Netherlands. TIER working paper series.
Hammack, F. M. (1986). Large School Systems' Dropout Reports: an Analysis of Definitions, Procedures,
and Findings. Teachers College Records 87(3), 324-341.
Henry, K. L. (2007). Who's Skipping School: Characteristics of Truants in 8th and 10th Grade. Journal of
School Health 77(1), 29-35.
Herbert, T.P., and S.M. Reis (1999). Culturally Diverse High-Achieving Students in an Urban High
School. Urban Education 34(4), 428-457.

21
Ishitani, T. T. and K. G. Snider (2006). Longitudinal Effects of College Preparation Programs on College
Retention. IR Applications 9, 1-10.
Jackson, T. and Armor, D.J. (1992). Carrots or sticks for high school dropouts? The Public Interest 106,
76-90.
Jarjoura, G. R. (1996). The Conditional Effect of Social Class on the Dropout-Delinquency Relationship.
Journal of Research in Crime and Delinquency 33(2), 232-255.
Kalmijn, M. and Kraaykamp, G. (2003). Dropout and Downward Mobility in the Educational Career: An
Event-History Analysis of Ethnic Schooling Differences in the Netherlands. Educational Research
and Evaluation, 9(3), 265-287.
Kaufman, P., M. N. Alt and C.D. Chapman (2004). Dropout Rates in the United States: 2001 (NCES 2005-
046). US Department of Education. Washington, DC: National Center for Education Statistics.
Kemple, J.J. (2004). Career Academies: Impacts on Labor Market Outcomes and Educational
Attainment. New York: MDRC (Manpower Demonstration Research Corporation).
Kingston, P.W. (2000). The Classless Society. [Series: Studies in Social Inequality.] Stanford: Stanford
University Press.
Koball, H. (2007). Living Arrangements and School Dropout Among Minor Mothers Following Welfare
Reform. Social Science Quarterly 88(5), 1374-1391.
Lamb, S. and R. Rumberger (1998). The Early Work and Education Experiences of High School Dropouts:
a Comparative Study of the United States and Australia. Longitudinal Surveys of Australian Youth
Research Report, ACER.
Lee, V. E., and Burkam, D. T. (2003). Dropping out of high school: The role of school organization and
structure. American Educational Research Journal, 40(2), 353-393.
Marks, G. and N. Fleming (1999). Early School Leaving in Australia: Findings from the 1995 Year LSAY
Cohort. Research Report, No. 11, ACER.
Ministerial Council for Employment, Education, Training and Youth Affairs (MCEETYA) (2000). National
Report on Schooling in Australia.
Okpala, C.O., A.O. Okpala, et al. (2001). Parental Involvement, Instructional Expenditures, Family
Socioeconomic Attributes, and Student Achievement. Journal of Educational Research 95(2).
Olsen, R.J. and G. Farkas (1989). Endogenous Covariates in Duration Models and the Effect of
Adolescent Childbirth on Schooling. Journal of Human Resources 24(1), 39-53.
Orthner, D. K., P. G. Cook, et al. (2002). Welfare Reform, Poverty, and Children's Performance in
School: Challenges for the School Community. Children & Schools 24(2), 105-?
Ou, S.-R. and A.J. Reynolds (2006). Early Childhood Intervention and Educational Attainment: Age 22
Findings From the Chicago Longitudinal Study. Journal of Education for Students Placed at Risk
11(2), 175-198.
Pakulski, J. and Waters, M. (1996). The Death Of Class. Sage: London, Thousand Oaks and New Delhi.
Pittman, R. and P. Haughwout (1987). Influence of High School Size on Dropout Rate. Educational
Evaluation and Policy Analysis 9(4), 337-343.
Pittman, R.B. (1993). The 21st Century and Secondary School At-Risk Students: What's Ahead for
Teachers in Rural America? [Conference Proceedings).

22
Perna, L.W. , H. Rowan-Kenyon, A. Bell, S.L. Thomas and C. Li (2008). A Typology of Federal and State
Programs Designed to Promote College Enrolment. Journal of Higher Education 79(3), 244-267.
Plank, S., S. DeLuca and A. Estacion (2005). Dropping Out of High School and the Place of Career and
Technical Education: A Survival Analysis of Surviving High School. National Research Center for
Career and Technical Education, S.P.M.N. & National Dissemination Center for Career and
Technical Education, C.O.H.
Reich, C. and V. Young (1975). Patterns of Dropping out. Interchange 6(4), 6-15.
Robinson, L. (1999). The Effects of Part-Time Work on School Students. Longitudinal Surveys of
Australian Youth Research. Report No. 9, ACER.
Rumberger, R. W. (1983). Dropping Out of High School: The Influence of Race, Sex, and Family
Background. American Educational Research Journal 20(2): 199-220.
Rumberger, R.W. and S.P. Lamb (2003). The Early Employment and Further Education Experiences of
High School Dropouts: a Comparative Study of the United States and Australia. Economics of
Education Review 22, 353-366.
Rumberger, R.W. (2004a). “Why students drop out of school?” In: G. Orfied (Ed.), Dropouts in America:
Confronting the Graduation Rate Crisis, Cambridge. MA: Harvard Education Press, pp. 131-155.
Rumberger, R.W. (2004b). “What Can be Done to Reduce the Dropout Rate?” In: G. Orfied (Ed.),
Dropouts in America: Confronting the Graduation Rate Crisis. Cambridge, MA: Harvard Education
Press, pp. 243-254.
Sinclair, M.F., S.L. Christenson and L. Thurlow (2005). Promoting School Completion of Urban
Secondary Youth with Emotional or Behavioral Disabilities. Exceptional Children, 71, 465-482.
Smeyers, P. and M. Depaepe (2006). “On the Rhetoric of ‘What Works’. Contextualizing Educational
Research and the Picture of Performativity”. In P. Smeyers and M. Depaepe (Eds.).Educational
Research: Why 'What Works' Doesn't Work. Dordrecht: Springer, pp. 1-16.
Smeyers, P. (2006). “The Relevance of Irrelevant Research; the Irrelevance of Relevant Research.” In: P.
Smeyers and M. Depaepe (Eds.). Educational Research: Why 'What Works' Doesn't Work.
Dordrecht: Springer, pp. 95-108.
Smeyers, P. and M. Depaepe (2009). The Educationalization of Social Problems. Dordrecht: Springer.
Smith, R. (2003). Research and Revelation: What Really Works. In: P. Smeyers and M. Depaepe (Eds.)
(2003). Beyond Empiricism: On Criteria for Educational Research. Leuven: Leuven University Press,
pp. 129-140.
Smith, J., and R. Naylor (2005). Schooling Effects on Subsequent University Performance: Evidence for
the UK University Population. Economics of Education Review 24, 549-562.
Spady, W. (1970). Dropouts from higher education: an interdisciplinary review and synthesis.
Interchange 1, 64-85.
Spady, W. (1971). Dropouts from higher education: toward an empirical model. Interchange 2, 38-62.
Suet-Ling, P. (2000). The Effects of Change in Family Structure and Income on Dropping Out of Middle
and High School. Journal of Family Issues 21(2), 147-169.
Swadener, B.B. (1995). “Children and Families ‘At Promise’: Deconstructing the Discourse of Risk”. In:
B.B. Swadener and S. Lubeck (Eds.). Children and Families "At Promise": Deconstructing the
Discourse of Risk. Albany: State University of New York.

23
Swadener, B.B. and Lubeck, S. (1995). “The Social Construction of Children and Families ‘At Risk’: An
Introduction”. In: Beth Blue Swadener and Sally Lubeck (Eds.). Children and Families "At Promise":
Deconstructing the Discourse of Risk. Albany: State University of New York.
Teese, R. and A. Walstab (2002). Early leaving in Victoria: Geographical Patterns, Origins, and Strategic
Issues. Educational Outcomes Research Unit, University of Melbourne.
te Riele, K. (2006). Schooling practices for marginalized students practice‐with‐hope. International
journal of inclusive education, 10(1), 59-74.
Tinto, V. (1975). Dropout from higher education: A theoretical synthesis of recent research. Review of
Educational Research 45, 89-125.
Tyack, D. and Cuban, L. (1995). Tinkering toward Utopia. A Century of Public School Reform. Cambridge
/ London: Harvard University Press.
Van Klaveren, C. and De Witte, K. (2013), How are teachers teaching? A nonparametric approach.
Education Economics. In Press.
Vizcain, D.C. (2005). Investigating the Hispanic/Latino Male Dropout Phenomenon: Using Logistic
Regression and Survival Analysis [Unpublished Doctoral Dissertation: Department of Educational
Measurement and Research, College of Education, University of South Florida].
Willet, J.B., and J.D. Singer (1991). From Whether to When: New Methods for Studying Student
Dropout and Teacher Attrition. Review of Educational Research 61(4), 407-450.
Wilson, S.J. Tanner-Smith, E.E., Lipsey, M.W., Steinka-Fry, K. and Morrison, J. (2011). Dropout
prevention and intervention programs: effects on school completion and dropout among school-
aged children and youth. Campbell Systematic Reviews 8, DOI: 10.4073/csr.2011.8.

24
Tables
Table 1: Level of analysis
National level studies
Observed in:
Rumberger, 1983; Ekstrom, Goertz, Pollack and Rock, 1986; Pittman and
Haughwout, 1987; Olsen and Farkas, 1989; Jarjoura, 1996; Suet-Ling, 2000; Business Council
of Australia, 2002b; Rumberger and Lamb, 2003; Kaufman, Alt and Chapman, 2004; Plank,
DeLuca and Estacion, 2005; Balfanz and Legters, 2005; Smith, J., and R. Naylor, 2005; Ishitani
and Snider, 2006; Henry, 2007; Koball, 2007; Dustmann and van Soest, 2007; Dalton, Gennie,
and Ingels, 2009; Cataldi, Laird and KewalRamani, 2009.
State level studies
Observed in:
Olsen
et al.
, 1987; Entwisle, Alexander and Steffel
-
Olson, 2004; Entwisle,
Alexander and Steffel-Olson, 2005.
County, district or city level studies
Observed in:
Reich and Young, 1975; Okpala
et al.
, 2001; Orthner
et al.
, 2002; Allensworth
,
2004; Allensworth, 2005; Vizcain, 2005; Ou and Reynolds, 2006; De Witte and Van Klaveren,
2012
School level studies
Observed in:
Herbert and Reis, 1999.

25
Table 2: Observed approaches, methods and models
-
Instrumental
-
variables approach
Observed in: Allensworth, 2005
-
Logistic regressions
* including univariate and multinominal regression analyses, multivariate and multi-level
regressions, (multiple- spells) competing risk models
Observed in: Suet-Ling, 2000; Kalmijn and Kraaykamp, 2003; Duchesne et al, 2005;
Allensworth, 2005; Entwisle, Alexander and Steffel-Olson, 2005; Vizcain, 2005; DesJardins
et al, 2006; Henry, 2007
-
Ordinary least squares regressions
Observed in: Okpala et al., 2001
-
Probit models
Observed in: Rumberger, 1983; Adams and Becker, 1990; Jarjoura, 1996; Allensworth,
2005; Ou and Reynold, 2006
-
Survival analyses/event history models/time hazard models/cox regression models
* including discrete-time survival analyses, non-proportional hazards models, trajectory or path
analyses
Observed in: Ekstrom, Goertz, Pollack and Rock, 1986; Kalmijn and Kraaykamp, 2003;
Allensworth, 2005; Vizcain, 2005; Duchesne et al, 2005; DesJardins et al, 2006.
-
Hierarchical generalized linear models (HGLM)
Observed in: Allensworth, 2005
-
Difference
-
in
-
difference (DD) analyses
Observed in: Koball, 2007; Cabus and De Witte, 2011
-
Case study and ethnographic methods
* including interviews and participant observation
Observed in: Herbert and Reis, 1999

26
Table 3: Common predictors of early school leaving in the literature (references in following tables)
DROPOUT PREDICTORS
OBSERVED EFFECT
INTERACTION EFFECT(S)
STUDENT FACTORS
* psychological and behaviour
factors
- academic ability/achievement
- grade retention/repetition
- educational and occupational
aspirations
- engagement
(often made operational by
absenteeism and discipline
problems)
- high school employment
- teenage pregnancy & marriage
* demographic factors:
-
gender
- race
/ethnicity
-
immigration status
-
language background
- disabilities
- if higher, lower dropout risk
- if the case, higher dropout risk
- if higher, lower dropout risk
- if more absenteeism and/or
discipline problems, higher
dropout risk
- if intensive, inadequate, stressful
and unstable, higher dropout risk
- mixed findings
- mixed findings
- mixed findings
- mixed findings
- if native speaker, lower risk
- if the case, higher risk
e.g., with gender, race/ethnicity, and employment
opportunities
e.g., with race/ethnicity
e.g., with family background, perceived behaviour
control, and expectations from teachers
FAMILY FACTORS
* structural characteristics
-
socioeconomic status (parental
-----
education and employment)
- family structure (single-parent, -
----- step- and/or large families)
* underlying processes
- social capital (relationships
between parents, children,
other families and school)
↔
human / cultural capital:
(parental education)
↔ financial capital
(income, ownership)
-
if lower, then higher dropout risk
- no independent effect
-
if more, lower dropout risk
- if higher, lower dropout risk, but
perhaps no independent effect ?
- no independent effect
e.g., with parent-child relationship
with income (they both matter)
SCHOOL FACTORS
- school type
(incl. student composition )
-
school resources (e.g.: class
-l size & teacher-
pupil ratio)
- structural characteristics of
schools (e.g.: school size)
-
school policies and practices
* social and academic climate
(discipline policy considered
fair, high attendance rates,
and advanced course taking)
* teacher & teaching quality
* school social capital
(student-teacher relationship)
- if public & a-selective, higher risk
-
if balanced, lower dropout risk
- no independent effect
- if smaller, lower dropout risk, but
perhaps no independent effect?
-
if stimulating, lower dropout risk
, lower dropout risk
-
if higher, lower dropout risk
-
if better, lower dropout risk
e.g., with teaching quality and practice
e.g., with school social climate

27
COMMUNITY FACTORS
-
neighbourhood characteristics
- high-achieving vs. dropped-out
friends
-
employment opportunities
---
job scarcity & low salaries
--- long working hours
- social discrimination/injustice
-
if detrimental, higher dropout risk
-
lower & higher dropout risk, resp.
-
if job scarcity, lower dropout risk
- if > 20 working hours, higher risk
-
if the case, higher dropout risk
with gender
with student’s SES-background
with race/ethnicity
Table 4: Overview of student factors
A.
Psychological and behavioural factors
1.
academic achievement and ability
:
 if higher, lower dropout risk (-)
Observed in: Rumberger, 1983; Ekstrom, Goertz, Pollack and Rock, 1986; Herbert and Reis,
1999; Lamb and Rumberger, 1998; Ball and Lamb, 2001; Teese and Walstab, 2002; Vizcain,
2005; Dustmann and van Soest, 2007; Entwisle, Alexander, Steffel-Olson, 2004; Dalton,
Gennie and Ingels, 2009; Allensworth, 2004, 2005
↔ interaction with age: Plank, DeLuca and Estacion, 2005
2.
academic and professional aspirations
:
 if higher, lower dropout risk (-)
Observed in: Rumberger, 1983; Entwisle, Alexander, Steffel-Olson, 2004; Dustmann and
van Soest, 2007.
3.
(teacher
-
initiated) grade retention, accumulation of credit deficits
:
 if more, higher dropout risk (+)
Observed in: Rumberger, 1983; Olsen et al., 1987; Robinson, 1999; Blue and Cook, 2004;
Kaufman, Alt and Chapman, 2004; Plank DeLuca and Estacion, 2005; Entwisle, Alexander
and Steffel-Olson, 2004 and 2005; Vizcain, 2005; Dalton, Gennie and Ingels, 2009; Cataldi,
Laird and KewalRamani, 2009.
↔ retention by standardized tests: less univocal effects (Allensworth 2004 and 2005)
4.
previous “stopout”
, truancy
:
 if it occurs, it is repeated and/or sustained, higher dropout risk (+)
Observed in: Rumberger, 1983; Olsen et al., 1987; Adams and Becker, 1990; DesJardins et
al., 2006; Henry, 2007; De Witte and Csillag, 2013
5.
engagement
:
 if stronger, lower dropout risk (-)
Observed in: Finn, 1989; Entwisle, Alexander and Steffel-Olson, 2005; Appleton,
Christenson et al., 2008.

28
6.
attitudes ,
feelings, perceptions, traits and comportment (including discipline)
:
 if positive, lower dropout risk (-)
Observed in: Ekstrom, Goertz, Pollack and Rock, 1986; Adams and Becker, 1990; Herbert
and Reis, 1999; Blue and Cook, 2004; Entwisle, Alexander and Steffel-Olson, 2004;
Duchesne et al., 2005; Viscain, 2005.
7.
early pregnancy (and perhaps marriage)
:
 if the case, higher dropout risk (+)
Observed in: Rumberger, 1983: Kalmijn and Kraaykamp, 2003.
↔ no clear effect if controlled for underlying preferences or opportunities (Olsen and Farkas, 1989)
8.
substance (cannabis) (ab)use
:
 if the case, higher dropout risk (+)
Observed in: Fergusson et al., 2003.
B.
Demographic (background) factors
9.
sex/gender
:
 if male, higher dropout risk (+)
Observed in: Rumberger, 1983; Bynum and Thompson, 1983; Business Council of Australia,
2002b; Duchesne et al., 2005; Ou and Reynolds, 2006.
↔ marginal effect in the long run (e.g., further education) (Business Council of Australia, 2002b)
↔ no significant effect over the past 35 years
Observed in: Kaufman, Alt and Chapman, 2004; Cataldi, Laird and
KewalRamani, 2009)
Where (-) and (+) denote a negative and positive relationship with early school leaving, respectively.
‘’ indicates the main finding, while the symbol ↔ refers to alternative findings.
Table 5: Overview of family factors
1.
family structure
:
 if no biological, two-parent family, higher dropout risk (+)
Observed in: Rumberger, 1983; Olsen and Farkas, 1989; Kalmijn and Kraaykamp, 2003;
Plank, DeLuca and Estacion, 2005; Bridgeland, Dilulio, Morison, 2006; Dustmann and van
Soest, 2007.
2.
family culture/social climate
:
 if free from stressors, warm and supportive, lower dropout risk (-)
Observed in: Frank, 1990; Ou and Reynolds, 2006; Pitman, 1993; Herbert and Reis, 1999;
Kalmijn and Kraaykamp, 2003; Duchesne et al., 2005; Cooper, C. R., G. Chavira et al., 2005;
Ishitani and. Snider, 2006; Bridgeland, Dilulio and Morison, 2006; Dustmann and van Soest,
2007.

29
3.
socioeconomic status
:
 if higher, lower dropout risk (-)
Observed in: Ekstrom, Goertz, Pollack and Rock, 1986; Lamb and Rumberger, 1998;
MCEETYA, 2000; Teese and Walstab, 2002; Business Council of Australia, 2002a; Entwisle,
Alexander and Steffel-Olson, 2005; Dalton, Gennie and Ingels, 2009.
- parental education:
Observed in: Rumberger, 1983; Olsen and Farkas, 1989; Frank, 1990; Business Council of
Australia, 2002a; Kalmijn and Kraaykamp, 2003; Entwisle, Alexander and Steffel-Olson,
2004; Duchesne et al., 2005; Plank, DeLuca and Estacion, 2005; Ishitani and. Snider, 2006;
Dustmann and van Soest, 2007; Koball, 2007; Dalton, Gennie and Ingels, 2009
↔ no independent effect if controlled for child-parent relationship, parental support/involvement
Observed in: Rumberger, 2004a
- parental employment:
Observed in: Marks and Fleming, 1999; Business Council of Australia, 2002a.
- cultural index:
Observed in: Rumberger, 1983.
Where (-) and (+) denote a negative and positive relationship with early school leaving, respectively.
‘’ indicates the main finding, while the symbol ↔ refers to alternative findings.
Table 6: Overview of school factors
1.
school type (incl. student composition)
:
 if selective, independent, and with high promoting power, lower dropout risk (-)
Observed in: Okpala et al., 2001; Balfanz and Legters, 2004; Smith and Naylor, 2005;
Dustmann and van Soest, 2007; Dalton, Gennie and Ingels, 2009.
2.
school resources
:
 if a higher teacher-pupil ratio or larger class size, higher dropout risk (-)
Observed in: Pittman, 1993; Balfanz and Legters, 2004; Rumberger, 2004a.
↔ no effect independent from, e.g., teaching practice and age of students
Observed in: Smeyers, 2006

30
3.
school policies and practices
:
- social and academic climate:
 if challenging, inclusive and problem-free, lower dropout risk (-)
Observed in: Pittman and Haughwout, 1987; Finn, 1989; Pitman, 1993; Herbert and Reiss,
1999; Business Council of Australia, 2002a and 2002b; Ou and Reynolds, 2006; Viscain,
2005.
- teachers’ experience, expectations, support, and teaching quality:
 if higher, lower dropout risk (-)
Observed in: Finn, 1989; Adams and Becker, 1990; Herbert and Reis, 1999; Blue and Cook,
2004; Dalton, Gennie and Ingels, 2009.
- school social capital:
 if positive, with strong cohesion, and care, lower dropout risk (-)
Observed in: Pittman and Haughwout, 1987; Finn, 1989; Herbert and Reis, 1999; Business
Council of Australia, 2002a; Blue and Cook, 2004.
Where (-) and (+) denote a negative and positive relationship with early school leaving, respectively.
‘’ indicates the main finding, while the symbol ↔ refers to alternative findings.
Table 7: Overview of community factors
1.
neighbourhood characteristics
:
 if more distressing, higher dropout risk (-)
Observed in: Rumberger, 1983; Blue and Cook, 2004, Rumberger, 2004a.
2.
friends/peer networks
:
 if positive influence from high-aspiring and achieving peers, lower dropout risk (-)
Observed in: Rumberger, 1983; Herbert and Reiss, 1999; Cooper et al., 2005.
3.
employment conditions
:
 if more jobs available, unstable job pattern, higher stress at work, longer working hours, and
work for family support higher dropout risk (+)
Observed in: Olsen and Farkas, 1989; Marks and Fleming, 1999; Entwisle, Alexander and
Steffel
-
Olson, 2004; Entwisle, Alexander and Steffel
-
Olson, 2005.
4.
social
discrimination and prejudice
:
 if more, higher dropout risk (+)
Observed in: Herbert and Reis, 1999.
Where (-) and (+) denote a negative and positive relationship with early school leaving, respectively.
‘’ indicates the main finding, while the symbol ↔ refers to alternative findings.

TIER WORKING PAPER SERIES
TIER WP 14/14
© TIER 2014
ISBN 978-94-003-0080-4