Parenting practices and school dropout: a longitudinal study.

Abstract

Adolescents' perceptions of parenting style and parental involvement in their education were examined longitudinally and related to school dropout among Icelandic youth (N = 427). Results indicated that adolescents who, at age 14, characterized their parents as authoritative (showing acceptance and supervision) were more likely to have completed upper secondary school by age 22 than adolescents from non-authoritative families, controlling for adolescents' gender, socioeconomic status (SES), temperament, and parental involvement. Parenting style seems to more strongly predict school dropout than parental involvement. Further, parenting style may moderate the relationship between parental involvement and dropout, but not in all groups; only in authoritative families does parental involvement decrease the likelihood of school dropout. Furthermore, even after controlling for previous academic achievement, adolescents from authoritative families were less likely to drop out than adolescents from authoritarian and neglectful families. These findings emphasize the importance of encouraging quality parent-child relationships in order to reduce the likelihood of school dropout.

Full text

Page 1

PARENTING PRACTICES AND SCHOOL DROPOUT:
A LONGITUDINAL STUDY
Kris^ana S. Blondal and Sigrun Adalbjamardottir
ABSTRACT
Adolescents' perceptions of parenting style and parental involvement in their
education were examined longitudinally and related to school dropout among
Icelandic youth (N = 427). Results indicated that adolescents who, at age 14,
characterized their parents as authoritative (showing acceptance and supervi-
sion) were more likely to have completed upper secondary school by age 22
than adolescents from non-authoritative families, controlling for adolescents'
gender, socioeconomic status (SES), temperament, and parental involvement.
Parenting style seems to more strongly predict school dropout than parental
involvement. Further, parenting style may moderate the relationship between
parental involvement and dropout, but not in all groups; only in authoritative
families does parental involvement decrease the likelihood of school dropout.
Furthermore, even after controlling for previous academic achievement, ado-
lescents from authoritative families were less likely to drop out than adoles-
cents from authoritarian and neglectful families. These findings emphasize
the importance of encouraging quality parent-child relationships in order to
reduce the likelihood of school dropout.
When adolescents drop out of school, the results are psychosocially
and economically costly, for both the individual and society. As our
modem knowledge-based societies increasingly rely on a highly skilled
labor force, young people without upper secondary education are more
The ongoing longitudinal research project reported in this study was sup-
ported by grants to the second author from the Icelandic Council of Science
and the Research Foundation of the University of Iceland. In addition, this
part of the study was supported by a grant to the authors from the Eimskip
University Fund at the University of Iceland. We would like to thank Pamela
Davis-Kean, Jon Torfi Jonasson, and Sif Einarsdóttir for their insighitftil con^-
ments. The young people, parents, teachers, and principals who kindly con-
sented to participate in this project are also gratefully thanked. In addition,
thanks are due to the Educational Testing Institution of Iceland and the Na-
tional Statistical Institute of Iceland which provided educational status data.
Sigrun Adalbjamardottir, Faculty of Social and Human Sciences, University
of Iceland, Reykjavik, Iceland
Requests for reprints should be sent to Sigrun Adalbjamardottir, Faculty of
Social and Human Sciences, School of Social Sciences, Oddi, University of
Iceland, 101 Reykjavik, Iceland. E-mail: sa@hi.is
ADOLESCENCE, Vol. 44, No. 176, Winter 2009
Libra Publishers, Inc., 3089C Clairemont Dr., PMB 383, San Diego, CA 92117

Page 2

vulnerable than ever before. They have fewer work opportunities, and
are less likely to return to education and training later in life, com-
pared to those who finish school (e.g., Rumberger & Lamb, 2003). They
also face a higher risk of various negative outcomes; they may be unem-
ployed, live in poverty, have health problems, and engage in antisocial
behavior (see Rumberger & Thomas, 2000).
In recent years the problem of school dropout has received increased
attention. The European Union has proposed a common benchmark
for the member states: by the year 2010, the early school-leaving rate
should be no more than 10% in any given country (Council of the
European Union, 2004). In the U.S. this problem has also been ad-
dressed nationally as one of the National Educational Goals adopted
in 1990 (U. S. Department of Education, 1990). Moreover, in the federal
reform plan, the No Child Left Behind Act of 2001, all states are re-
quired to incorporate graduation rates into their accountability sys-
tems for high schools (U.S. Department of Eiducation, 2002), In Iceland,
where this study was conducted, the dropout problem is also of concern;
currently the Icelandic government is presenting educational reforms
that aim to reduce dropout (Upper Secondary School Act No. 92/2008).
The family has been recognized as one of the primary contributors to
children's success at school (Rumberger, 1995). Studies of the famil/s
influence on school dropout, however, have at least four important
shortcomings. First, such studies tend to focus too strongly on struc-
tural characteristics, such as parents' socioeconomic status (SES). Sec-
ond, studies examining parental influence on school success have
mainly focused on the relationship between parental involvement in
their child's education and academic achievement but seldom on school
dropout. Third, findings about the relationship between parental
involvement and school success have been inconsistent, and fourth,
most studies on school dropout are cross-sectional. The purpose of this
study was to explore more general aspects of parenting in relation to
school dropout; we examine the relationship of both parental involve-
ment and parenting style with school dropout. Moreover, we use a
longitudinal design.
THE FAMILY AND EDUCATIONAL OUTCOMES
Research on family influences has been criticized for focusing on
such structural characteristics as parents' socioeconomic status to ex-
plain children's school success and failure (e.g., Alexander, Entwisle, &
Horsey, 1997). Findings consistently show that students in higher SES
730

Page 3

groups are academically more successful and less likely to drop out of
school than students in lower SES groups (see McNeal, 1999; Rosen-
thai, 1998). These studies, however, provide little insight into what is
occurring in family life that helps the students succeed at school
(Davis-Kean, 2005). Studies in this area have also been criticized for
using a narrow definition of parental support (see Jeynes, 2007).
Parental Involvement
Studies on the influence of parenting on school outcomes have
mainly focused on specific parental practices such as involvement in
their child's education, mostly in relation to academic achievement and
rarely in relation to school dropout (McNeal, 1999; Rumherger, 1995).
Common indicators of parental involvement include contacts between
parents and school, parental involvement in school activities, parent-
child communication about school, parental supervision involving
homework, and parents' educational aspirations for their child (Fan &
Chen, 2001; McNeal, 1999). Despite the many studies on parents'
involvement and children's, academic achievement, the nature of the
relationship remains unclear (Jeynes, 2007; McNeal, 1999). In their
meta-analysis. Fan and Chen (2001) concluded that the association
between parental involvement and students' academic achievement
was weaker than expected. Moreover, the findings of these studies
have been inconsistent. In some, parental involvement seems to relate
positively to children's achievement (e.g., Hoge, Smit, & Crist, 1997);
other studies indicate no association, or even a negative one. For exam-
ple, McNeal (1999) found that adolescents whose parents participated
in the parent-teacher association got lower grades than their peers.
In one of the few studies in the area of school dropout, Alexander
and his colleagues (1997) fovmd that yovmg people whose parents had
low expectations for their educationsil attainment at the beginning of
their schooling were more likely to drop out. Other studies indicate
that parental involvement in parent-teacher organization activities
and parental supervision of adolescents' homework also reduces the
risk of dropping out (McNeal, 1999; Rumberger, 1995). However,
McNeal's findings (1999) suggest that the association between parental
involvement and school success varies depending on the indicator used
for success. For example, parent-child communication about school was
positively associated with higher grades but did not seem to reduce
the risk of dropout.
Parenting Style
The literature outlined above suggests that to better understand
parenting in relation to school dropout it might be more useful to look
731

Page 4

at parenting style as an indicator of the general quality of the parent-
adolescent relationship, instead of focusing solely on specific parental
behaviors such as participation in their children's education. Accord-
ingly, such an approach might help parents better understand how to
motivate and encourage their children's educational aspirations and
support their success at school.
Parenting style is most consistently associated with Baumrind's
(1971) pioneering work. In their well-known approach Steinberg and
his colleagues (Lambom, Mounts, Steinberg, & Dornbusch, 1991;
Steinberg, Lambom, Darling, Mounts, & Dornbusch, 1994) assessed
parenting styles using the interactive effects of the parenting dimen-
sions of support and control. In their measure they used a fourfold
typology of parenting style based on Baumrind's work (1971) and on
the Maccoby and Martin (1983) revision of her conceptual work.
The prototypes of the parenting styles are authoritative, authoritar-
ian, indulgent, and neglectful. Authoritative parents are accepting,
warm, and encouraging toward their children but at the same time
firm; they impart clear standards for their children's behavior, enforc-
ing developmentally appropriate expectations without being intrusive
or restrictive. Authoritarian parents are demanding and controlling,
but not responsive or warm. They have clear rules that their children
are not supposed to question. Indulgent parents are responsive and
warm. They allow considerable self-regulation, but are lenient and
avoid confrontation. Neglectful parents are neither responsive nor de-
manding. They do not monitor or guide their children and do not sup-
port them or relate to them with warmth.
This typology of parenting style has been considered promising since
it provides an opportunity to explore relationships between the multi-
dimensional characteristics of parenting and the adjustment of youth
(Adalbjamardottir & Hafsteinsson, 2001; Lambom et al., 1991;
Steinberg et al., 1994; Türkei & Tezer, 2008). Compared to their peers
raised in non-authoritative families, children from authoritative fami-
lies have been shown to score higher on a wide variety of measures of
psychosocial development and competence (see overview by Steinberg,
2001). Regarding education, adolescents who perceive their parents as
authoritative are more likely to earn higher grades than adolescents
who perceive their parents as non-authoritative (e.g., Adalbjamardot-
tir & Blondal, 2004; Baumrind, 1991).
In addition, Steinberg and his colleagues (Steinberg, Lambom, Dorn-
busch, & Darling, 1992) argue that the association between parental
involvement and academic achievement may vary across parenting
style. Their findings indicate that the emotional context of authorita-
732

Page 5

tive parenting increased the effectiveness of parental involvement for
adolescents' academic achievement. Adolescents whose parents were
involved in their child's education did better in school only if their
parents were authoritative in their parenting style. These findings
suggest that in understanding the processes through which parenting
style relates to child development, we must separate different aspects
of parenting, namely (a) the specific practices parents use in order to
help their children attain particular goals such as heing successful
at school and (b) the parenting style or emotional context where the
socialization takes place (Steinberg, 2001).
Researchers have rarely examined the relationship between a
broader conceptualization of parenting style and school dropout. An
exception is the cross-sectional study by Rumberger and his colleagues
(Rumberger, Ghatak, Poulos, Ritter, & Dombusch, 1990) based on
Baumrind's work. Their study provides some indication that adoles-
cents of permissive parents are more likely to drop out of school than
adolescents raised in authoritative and authoritarian families. In their
study no distinction was made between parents who were caring but
permissive and those who were uncaring and neglectful. Moreover,
the group of school dropouts was too small to take into account other
important factors hke parents' SES.
Research on parenting has been criticized for ignoring the bidirec-
tional interactive relationship between parent and child by not taking
into account the possible connection between characteristics of the
child, such as temperament, and the parent-child relationship (see
Collins, Maccoby, Steinberg, Hetherington, & Bomstein, 2000). Tem-
perament is considered to be constitutionally based individual differ-
ences in self-regulation and reactivity that are fairly stable from early
on, but can also be modified by experience (Collins et al., 2000; Windle,
1992). For example, some researchers (Crockenberg, 1986) have found
that individual differences in temperament may moderate the quality
of interactions with parents and therefore relate to children's and ado-
lescents' psychosocial and behavioral adjustment.
School Dropout
Definition. Definitions of school dropout can be divided into two cate-
gories. The inclusive definition depends on the norm of education in
the society, while the exclusive definition defines dropouts as those
who do not complete the education they have started (see Jonasson &
Blondal, 2004). In Iceland the inclusive definition is used more often,
as upper secondary education is presently the norm for the society and
therefore those who do not complete upper secondary education are
733

Page 6

defined as school dropouts independent of whether or not they start
upper secondary school. In Iceland compulsory education lasts 10
years, from age 6 to 15, and over 90% of those in each age cohort start
upper secondary education at age 16 even though it is not compulsory.
Upper secondary school is normally a four-year system (including both
academic programs which qualify students for higher education and
programs for industrial arts and other work-oriented courses—^VET).
This means that most students are expected to complete their studies
at age 20. Still the system is flexible as everyone has the legal right
to upper secondary education regardless of age and many students use
that flexibility (Ministry of Education, Science and Culture, 2002). In
this study we therefore deflne school dropouts as those who have not
completed, and are not enrolled in, upper secondary education as of
age 22.
Academic achievement and school dropout. Students who show low
academic achievement have consistently been shown to be at higher
risk of dropping out of school than are other students. Academic
achievement is one of the strongest predictors of school dropout (see
Battin-Pearson et al., 2000; Jonasson & Blondal, 2002; Rumberger,
1995). Moreover, it may mediate, at; least partly, the relationship be-
tween parenting style and school dropout as adolescents' achievement
varies by parenting style (Adalbjamardottir & Blondal, 2004). There-
fore parenting style may influence the likelihood of dropping out of
school partly because of its influence on adolescents' academic
achievement.
THE PRESENT STUDY
The literature on the association between parenting practices and
school performance is limited in at least three important ways. First,
few studies focus on the relationship between parenting practices and
school dropout. Second, they have provided little insight into what it
is in family life that enhances the probability of a student finishing
upper secondary school. The purpose of this study is to contribute to
this understanding. Third, studies of dropout have been criticized for
using cross-sectional rather than longitudinal data (Doll & Hess,
2001). Since school dropout often seems to result from a long-term
process of withdrawal from school (Finn, 1989), it may be especially
important to analyze the predictors of school dropout longitudinally.
In this study we use a longitudinal design.
The major aim of this study is to explore and compare the relation-
ship between piirenting style and parental involvement on the one
734

Page 7

hand and school dropout on the other. We use the well-known typology
of parenting style in Baumrind's (1971, 1999) tradition to explore how
various parenting styles (authoritative, indulgent, authoritarian, and
neglectful) relate to school dropout. Further, we use a longitudinal
design that covers an unusually long period in studies of socialization
(Collins et al., 2000); the adolescents were followed from age 14 until
age 22. This long period makes it possible to assess relationships over
time bet\yeen parenting practices and school dropout.
Our main hypothesis is that parenting style predicts school dropout.
We expect that adolescents who at age 14 characterize their parents
as authoritative are more likely to finish their upper secondary educa-
tion by age 22, compared to adolescents who perceive their parents as
non-authoritative. We expect this relationship to persist even after
taking into account adolescents' background (gender and SES) and
temperament, as well as their previous academic achievement at the
end of compulsory school (age 150[^ Furthermore, we explore possible
mediation effects of achievement on the relationship between parent-
ing style and school dropout.
Second, we hypothesize that pairental involvement at age 14 predicts
school dropout. We expect that the adolescents who at age 14 perceive
their parents as willing to participate in their education are more likely
to complete upper secondary school. However, compared to parental
involvement, we expect parenting style to relate more strongly to
school dropout. Third, we expect parenting style to moderate the rela-
tionship between parental involvement and school dropout.
METHOD
Participants
This study is part of a larger ongoing, longitudinal study: the Reyk-
javik Adolescent Risk-Taking Longitudinal Study (RAR-LS; Adalbjam-
ardottir, 1994) with a focus on 427 adolescents who were classified
at baseline (age 14) into one of four parenting styles: authoritative,
authoritarian, indulgent and neglectful. These 427 were a subgroup of
the initial sample of 1,010 14-year-old students (51% female) drawn
from the population of students attending the 9th grade of compulsory
school in Reykjavik, the capital city of Iceland. Approximately 90% of
Reykjavik's 9th-grade public school population participated in the
study at baseline. The case loss was the result of absenteeism (9%)
and parental exclusion (8 cases, less than 1%). The sample was homo-
geneous with respect to culture (native Icelanders), religion (Lutheran)
and language (Icelandic).
735

Page 8

Of the initial sample, 474 subjects could be categorized into the four
parenting styles; 47 were missing information on one or more of the
research variables, which is not surprising given the longitudinal de-
sign of the study. Table 1 presents descriptive data of the study vari-
ables for the total sample (N = 1,010), the sample for those who were
classified into the four parenting styles at baseline (iV = 474), and the
final sample used in this study (N = 427).
Table 1
Proportions (%) or S4ean and Range Values for the Study Variables in the Total
Sample, Total Sample within Parenting Styles, and Study Sample
Variable
Gender
Males
Females
Socioeconomic status
Higher status
Lx)wer status
Temperament
Range
Parenting style
Authoritative
Neglectful
Authoritarian
Indulgent
Parental involvement
Range
Academic achievement
Range
School dropouts
Completers
Total sample
(^= 1,010)
n %/M
478
532
529
467
999
184
184
56
50
995
932
319
659
47.3
52.7
53.1
46.9
2.57
0-9
38.8
38.8
11.8
10.5
3.74
1-4
6.71
1.33-9.67
32.6
67.4
Parenting style sample
(A/=474)
n %/ M
225
249
244
223
453
184
184
56
50
453
442
164
294
47.5
52:5
52.2
47.8
2.54
0-9
38.8
38.8
11.8
10.5
3.69
1-4
6.64
2.33-9.67
35.8
64.2
Study sample
n %/M
196
231
230
197
427
170
158
51
48
427
427
143
284
45.9
54.1
53.9
46.1
2.49
0-8
39.8
37.0
11.9
11.2
3.70
1-4
6.69
2.33-9.67
33.5
66.5
736

Page 9

A series of analyses was conducted to determine whether the attri-
tion in sample size from age 14 to 22 was due to systematic effects.
The 47 subjects who were missing one or more study variables were
compared to the 427 for whom we had complete data. Males were more
likely to be missing one or more variable(s) x^(l> ^ ~ ^474) = 4.24, p =
.04 as were those from the lower SES groups, x^d. N = 467) = 5.22,
p = .02. Moreover, on average, those with missing data had more
temperament difficulties, i(467) = -3.48, p = .00, and lower grades at
the end of compulsory school, i(440) = 3.54, p = .00, and were more
likely not to have completed upper secondary school by age 22, x^d, N
= 458) = 14.75, p = .00. No differences were observed for parenting
style and parental involvement.
Thus, relative to those who were missing one or more of the study
variables, the final sample was biased in favor of students with more
school success, fewer temperament difficulties, and higher SES, but it
was not biased with, regard to parenting variables. These attrition
biases suggest that generalization of results needs to be considered
cautiously. One implication might be that our estimates of effects are
restricted as the variability of risk factors for dropping out is con-
strained.
Procedure
Permission for the study was granted by the Icelandic Data Protec-
tion Commission, the Ministry of Education, and the Educational Test-
ing Institution of Iceland. All of the principals at the 19 compulsory
schools in Reykjavik provided written permission to collect data for
their schools. Letters describing the study were sent to the adolescents
and their parents. The parents were asked to contact the research
project if they or their adolescent did not want to participate in the
study. Researchers have found that in studies reqiairing active paren-
tal consent for an adolescent to participate, well-functioning families
tend to be overrepresented (see Lambom et al., 1991). The self-report
questionnaire was administered during school hours with the help of
trained data collectors. The adolescents were informed that they could
refuse or discontinue participation at any time and were assured that
their answers were strictly confidential.
In addition to the survey data, information from the Educational
Testing Institution of Iceland was obtained: the students' performance
on the standardized national tests given at the end of compulsory
school (10th grade). Moreover, Statistics Iceland, the national statisti-
cal institute, provided information on the upper secondary education
of the participants at age 22.
737

Page 10

Measures
School dropout. The participants were considered to have dropped
out of school if they had not completed, and were not registered in, an
upper secondary school at age 22.
Academic achivement. A composite of grades on standardized na-
tional tests in Icelandic, mathematics, and English at the end of com-
pulsory school (10th grade, age 15) was used to measure previous
academic performance. The grades on the three subjects were highly
correlated (ranging from r = .60 to r = .75) and were thus averaged
to provide an index of academic performance.
Socioeconomic status. Socioeconomic status was assessed using the
HoUingshead (1975) Index which links parents' SES with their educa-
tion and occupation. Those considered to have lower status were un-
skilled and skilled manual workers and workers in service occupations.
In the higher status category were executives, teachers, university-
educated specialists, professionals, and owners of businesses.
Temperament. The Revised Dimensions of Temperament Survey
(DOTS-R) developed by Windle and Lemer (Windle, 1992; Windle &
Lemer, 1986) was used. This scale consists of 54 items assessing 10
temperament attributes defined as: (1) Activity level—general (7
items, a = .80), (2) activity level—sleep (4 items, a = .82), (3) approach-
withdrawal (7 items, a = .56), (4) fiexibility-rigidity (5 items, a = .77),
(5) mood quality (7 items, a = .86), (6) rhythmicity-sleep (6 items, a =
.71), (7) rhythmicity-eating (5 items, a = .70), (8) rhythmicity-daily
habits (5 items, a = .48), (9) distractibility (5 items, a = .70), and (10)
persistence (3 items, a = .65). In line with Windle (1992) we created
one temperament construct; for each of the attributes, a dichotomous
score was derived (0, 1) with 1 indicating that the adolescents were in
the 30th percentile along with those who had the most difficulties on
a given temperament attribute (see Windle, 1992). The possible range
was 0-10 with higher scores indicating more temperament difficulties.
Parental involvement. Three dimensions commonly used in studies
on parental involvement were used: adolescents' perception of parents'
assistance with homework, parent-child communication about school,
and educational aspirations (Fan & Chen, 2001; McNeal, 1999). The
adolescents were asked about perceived but not actual parental sup-
port—the factor usually evaluated in studies on parental involvement
(e.g., McNeal, 1999; Rumberger, 1995). The adolescents were asked
how willing their mother/father was to assist them with homework,
show interest in their studies, and encourage them to get further edu-
cation. The Cronbach's alpha for the six items (three items for each
parent) was .82. Higher scores refiect more parental involvement.
738

Page 11

Parenting style. Adolescents' perceptions at age 14 of parenting
styles were measured using the Acceptance/Involvement and Strict-
ness/Supervision scales developed by Lambom and her colleagues
(1991) based on Baumrind's (1971) work. The Acceptance/Involvement
Scale assesses the perceptions adolescents hold of their parents' af-
fection, responsiveness, and involvement (10 items, a = .75). Typical
statements included "When (he/she) wants me to do something, (he/
she) explains whjr" and "I can count on (him/her) to help me out if I
have some kind of problem." On the Strictness/Supervision scale ado-
lescents are asked about their parents' limit setting, monitoring, and
supervision (8 items, a = .77).Typical statements to be rated include
"In a typical week, what is the latest you can stay out on Friday or
Saturday night?" and "How much do your parents REALLY know
where you go at night?"
Of the 1,010 adolescents participating in the study at age 14, 474
were classified according to one of four parenting styles: authoritative,
authoritarian, indulgent, and neglectful. Following the methodology of
Lambom and her colleagues (1991) the sample was trichotomized on
the two scales of Acceptance/Involvement and Strictness/Supervision,
and those results were combined to yield four different parenting
styles. Authoritative parents were defined as those who scored in the
upper third of both scales. Authoritarian parents were those who
scored in the lowest third on the Acceptance/Involvement scale, but in
the highest third on the Strictness/Supervision scale; indulgent par-
ents were those who scored in the highest third on the Acceptance/
Involvement scale, but in the lowest third on the Strictness/Supervi-
sion scale. Finally, neglectful parents were those who scored in the
lowest third on both scales. To distinguish more clearly between the
four styles, we omitted from our analysis those adolescents whose per-
ceptions of their parents placed them in the middle third on either
scale.
Analysis
Logistic regression analyses were performed to determine whether
the six independent variables—SES, gender, temperament, parental
involvement, parenting style, and previous school achievement—pre-
dicted the likelihood of school dropout. The analysis was conducted in
four steps. First, we explored the association between parenting style
and parental involvement at age 14 on the one hand, and school drop-
out on the other, controlling for gender, SES, and temperament. Next,
we examined whether the relationship between parental involvement
and school dropout varied across parenting style. Finally, we comnared
739

Page 12

two logistic models, with and without academic achievement, to deter-
mine whether achievement partly mediates the relationship between
parenting style and school dropout (e.g.. Baron & Kenny, 1986).
Students who had completed upper secondary school by age 22 were
coded 0 and those who had dropped out were coded 1. In a logistic
regression, given the increased value of the independent variables, an
odds ratio greater than 1 indicates an increased risk of dropping out
and an odds ratio below 1 indicates reduced risk. The independent
variables were adolescents' (1) gender (males coded as 0 and females
as 1); (2) SES (lower status coded as 0 and higher status as 1); (3)
temperament; (4) parenting style (which was quadrivalent, and repre-
sented by three dummy coded variables with authoritative serving
as a comparison group); (5) parental involvement; and (6) academic
achievement. The continuous variables—^temperament, parental
involvement, and academic achievement—^were centered at the sample
mean in the regression models to account for problems associated with
multicoUinearity and interaction terms (Aiken & West, 1991). More-
over, this method allows for easier interpretation as the regression
coefficient for a given variable represents its effect when all other con-
tinuous variables in the model are at their mean.
RESULTS
All six of the independent variables correlated with school dropout
as follows: gender r=-.15,p = .00; SES r = -.28, p = .00; tempera-
ment r = .11, p = .03; parenting style Cramer's V = .30, p = .00;
pairental involvement r = — .16, p = .03; and academic achivement r
= -.58,p = .00. Furthermore, the correlations indicated a low to mod-
erate significant intercorrelation between the independent variables;
gender was an exception as it correlated only with parenting style.
The correlation ranged from .12 to .41. The strongest correlation was
between the two parenting variables: parenting style and involvement.
Adolescents from authoritative and indulgent families were most likely
to perceive their parents as involved in their education, while adoles-
cents from neglectful families perceived the least parental involve-
ment, F(3, 423) = 29, p = .00, Tukey p < .05.
The first step of the logistic regression analysis was conducted to
explore the relationship between parenting style and parental involve-
ment on the one hand and school dropout on the other, controlling for
gender, SES, and temperament. Table 2 illustrates the results. First,
as Table 2, Model 1 indicates, the school dropout rate was lower among
740

Page 13

Table 2
Multiple Logistic Regression for Relationship of Adolescents ' Perceived Parenting
Practices with later School Dropout, Controlling for Students ' Background,
Temperament and Academic Achievement
Variable
Females
Higher SES
Temperament
B
-0.70
-1.12
0.03
Parental involvement -0.33
Parenting style
Neglectful
Authoritarian
Indulgent
1.04
1.44
0.78
Academic achievement
Constant
Nagelkerke R^
-0.50
.22
Modell
SE
0.24
0.23
0.07
0.24
0.30
0.38
0.38
0.29
a
OR
0.49**
0.33***
1.03
0.72
2.83***
4.22***
2.19*
0.61+
B
-0.78
-0.45
0.63
1.41
0.65
-0.97
-0.84
.48
Model 2'
SE
0.27
0.27
0.32
0.43
0.45
0.11
0.32
OR
0.46**
0.64+
1.87*
4.11**
1.92
0.38***
0.43**
Note. OR = Odds ratio.
• Model 1, likelihood ratio x\l, N = 427) = 74.19; p = .000.
* • Model 2, likelihood ratio x\6, N = 427) = 180.70; p = .000.
+ p<.01. *p<.Q5. **p<.01. ***p<.QO\.
femades than among males and those of higher SES. Second, pai-enting
style at age 14 predicted school dropout, controlling for gender, SES,
temperament, and parental involvement in the child's education. Ado-
lescents who at age 14 characterized their parents as authoritative
were more likely to have completed upper secondary school by age 22
than adolescents who perceived their parents as neglectful, authoritar-
ian or indulgent. As also shown in Table 2, Model 1, neither tempera-
ment nor parental involvement at age 14 significantly predicted school
dropout, controlling for parenting style, gender, and SES. These find-
ings suggest that parenting style relates more strongly to school drop-
out than does parental involvement.
741

Page 14

The second step of the logistic regression was conducted to explore
possible interaction effects between parenting style and parental
involvement on school dropout, controlling for SES, gender, and tem-
perament. A significant interaction term emerged in relation to paren-
tal involvement, between the authoritative and neglectful parenting
styles (6 = 1.34, SE = 0.65, OR = 3.82, p = .,04). Among adolescents
who perceived their parents as authoritative, parental involvement
decreased the risk of dropping out (r = -.20, p = .01) but this was not
true among adolescents who perceived their parents as neglectful (r =
.01, p = .87). This interaction term, however, did not significantly
improve the main effect of Model lpresented in Table 2 (likelihood
ratio x'(3, N = 427) = 5.16; p = .16).
The third step of the multiple logistic regression analysis was con-
ducted to explore whether the parenting style at age 14 predicted
school dropout at age 22, controlling for SES and gender as well as
previous academic achievementat age 15. Table 2, Model 2 shows the
results. Parenting style predicted school dropout even when the strong
predictors of previous academic achievement, as well as SES and gen-
der, were taken into account. As shown in Table 2, Model 2, adolescents
who perceived their parents as authoritative at age 14 were more likely
to have completed upper secondary school at age 22 compared to those
who perceived their parents as either authoritarian or neglectful.
Taken together, parenting style, SES, gender, and academic achieve-
ment explEiined 48% of the variation in school dropout.
Finally, to assess a possible mediation effect of achievement on the
relationship between parenting style and school dropout, we compared
the effects of parenting style in Model 2 using the same model without
achievement. Including achievement in the analysis produced a drop
in the effects of parenting style on school dropout observed in Model
2 (see Thorlindsson & Bemburg, in press). Results showed that the
coefficients for comparing the authoritative to the other parenting
styles dropped substantially when achievement was included; for ne-
glectful parenting they dropped by 48% (from 1.22 to 0.63), for authori-
tarian parenting by 8% (from 1.54 to 1.41), and for indulgent parenting
by 18% (from 0.79 to 0.65). These findings indicate that parenting style
has an effect on the odds of dropping out, in part because it influences
adolescents' academic achievement.
DISCUSSION
One of the main findings of this study is that perceived parenting
style at age 14 predicts dropout from upper secondary school. Adoles-
742

Page 15

cents who at age 14 perceived their parents as authoritative were more
likely to have completed upper secondary school by age 22 compared
to those who perceived their parents as authoritarian, neglectful or
indulgent. Our findings are particularly noteworthy, as they remained
pronounced even after we took into account the influential factors of
parents' SES (e.g., Rosenthal, 1998), adolescents' gender (OECD,
2007), parents' involvement in their education (McNeal, 1999; Rum-
berger et al., 1990), and adolescents' temperament (see Collins et al.,
2000). ^
A second major finding is that a broader conceptuahzation of parent-
ing seemed to be a stronger predictor of school dropout, rather than
the specific behavior depicted in parental involvement in their child's
education. Once we took into account adolescents' gender, SES, and
temperament, as well as parenting styles, parental involvement no
longer related significantly to school dropout. This finding supports
the suggestion that using broader characteristics of the parent-child
relationship—like parenting style rather than merely parental
involveinent in the child's education—might be a more meaningful way
to detect the essential role the family plays in the children's education
(Jeynes, 2007; Steinberg, 2001).
Also, interestingly, we found some indication that parental involve-
ment decreased the likelihood of dropping out within authoritative
families, but not among neglectful families. This finding is in Une with
those of Steinberg and his colleagues (1992) who found, in their cross-
sectional study, that adolescents who experienced high parental
involvement showed better academic achievement than their peers in
authoritative families but not in non-authoritative families. Accord-
ingly, our findings in this longitudinal study may further support the
importance of separately assessing specific parenting practices and
broader characteristics of parenting style to understand the processes
involved in parental practices in relation to adolescents' educational
outcomes (see Steinberg, 2001).
One explanation for the weaker relationship between dropout and
parental involvement compared with dropout and parenting style
might be that parents become more involved if their child needs educa-
tional support (see McNeal, 1999). It should be noted, however, that
in this study, rather than asking the adolescents about parental
involvement in action, we asked them how willing their parents were
to support them whether or not they needed it. In other words, adoles-
cents who do well academically and need less actual parental involve-
ment, compared to those who struggle in their studies, may still
perceive their parents as highly involved, knowing that the support is
there if needed.
743