Full Terms & Conditions of access and use can be found athttp://www.tandfonline.com/action/journalInformation?journalCode=nses20

Download by: [UVA Universiteitsbibliotheek SZ] Date: 17 September 2015, At: 23:04

School Effectiveness and School ImprovementAn International Journal of Research, Policy and Practice

ISSN: 0924-3453 (Print) 1744-5124 (Online) Journal homepage: http://www.tandfonline.com/loi/nses20

Standardization of lower secondary civiceducation and inequality of the civic and politicalengagement of students

Jacqueline Witschge & Herman G. van de Werfhorst

To cite this article: Jacqueline Witschge & Herman G. van de Werfhorst (2015):Standardization of lower secondary civic education and inequality of the civic andpolitical engagement of students, School Effectiveness and School Improvement, DOI:10.1080/09243453.2015.1068817

To link to this article: http://dx.doi.org/10.1080/09243453.2015.1068817

Published online: 27 Jul 2015.

Submit your article to this journal

Article views: 18

View related articles

View Crossmark data

Standardization of lower secondary civic education andinequality of the civic and political engagement of studentsJacqueline Witschge and Herman G. van de Werfhorst

Department of Sociology, University of Amsterdam, Amsterdam, The Netherlands

ABSTRACTIn this paper, the relation between the standardization of civic educationand the inequality of civic engagement is examined. Using data from theInternational Civic and Citizenship Education Study 2009 among earlyadolescents and Eurydice country-level data, three-level analysis andvariance function regression are applied to examine whether there is arelation between measures of civic education standardization andinequality in three dimensions of civic engagement: civic knowledge,interest in social and political issues, and participation in the community.Inequality is conceptualized as differences in students’ civic engagementbetween schools and the association between social origin and civicengagement. The results demonstrate that accountability is associatedwith more inequality in civic knowledge, whereas centralization is asso-ciated with less inequality in non-cognitive civic engagement.

ARTICLE HISTORYReceived 20 June 2014Final version received29 June 2015

KEYWORDSCivic education;standardization; educationalsystems; social inequality;multilevel analysis; variancefunction regression

Introduction

In the last two decades, interest in citizenship outcomes of education has increased (Eurydice,2012; Van der Wal & Waslander, 2007). For a healthy democracy, it is essential that citizens areinformed and engaged. Concerns regarding social cohesion in society – because of increaseddiversity and individualization – and the allegedly declining civic engagement of young peoplehave made citizenship education a policy priority in many countries (Dalton, 2007; Putnam, 2000;Schulz, Ainley, Fraillon, Kerr, & Cosito, 2010). Education is a particularly powerful institution forpreparing people to participate in democracy because all citizens can be addressed; moreover,social values stabilize when young adulthood is reached (Newcomb, Koenig, Flacks, & Warwick,1967). The contribution of education to various dimensions of civic and political engagement hasbeen demonstrated empirically (Campbell, 2006; Dee, 2004; Hauser, 2000; Milligan, Moretti, &Oreopoulos, 2004; Nie, Junn, & Stehlik-Barry, 1996; Torney-Purta, Lehmann, Oswald, & Schulz, 2001).

A “civic engagement gap” between more and less socioeconomically advantaged people is welldocumented (Dalton, 2007; Levinson, 2012; Verba, Schlozman, & Brady, 1995). This gap poses achallenge to the democratic ideal of political equality, to a representative democracy, and to socialcohesion (Dahl, 2007; Lijphart, 1997; Lipset, 1959; Putnam, 2000). If education provides differentlearning opportunities to different socioeconomic groups, socioeconomic inequality in civicengagement can be magnified. Because formal education is generally considered an importantinstitution that could improve civic awareness and engagement among youth, it can be consideredone of the most important vehicles to address equality in civic engagement.

Comparative studies concerning the cognitive outcomes of education have demonstrated thatfeatures of educational systems relate to academic achievement and how this achievement is

CONTACT: Jacqueline Witschge j.h.m.witschge@uva.nl

SCHOOL EFFECTIVENESS AND SCHOOL IMPROVEMENT, 2015http://dx.doi.org/10.1080/09243453.2015.1068817

© 2015 Taylor & Francis

Dow

nloa

ded

by [

UV

A U

nive

rsite

itsbi

blio

thee

k SZ

] at

23:

04 1

7 Se

ptem

ber

2015

distributed across socioeconomic groups (Fuchs & Wössmann, 2007; Hanushek & Wössmann, 2010;Horn, 2009; Van de Werfhorst & Mijs, 2010). A small body of literature is also emerging thatindicates that features of an educational system may affect the distribution of civic outcomes ofschool (Eckstein, Noack, & Gniewosz, 2012; Janmaat, 2011; Janmaat & Mons, 2011; Janmaat,Mostafa, & Hoskins, 2014; Van de Werfhorst, 2014). This paper examines the relation betweenone institutional feature of education, standardization, and the civic engagement of lower second-ary school students. Standardization concerns the establishment of national or regional standardsfor educational inputs and outputs, and varies substantially between countries. Because standardi-zation can be expected to affect the variation between schools, as will be explained below, it is aparticularly relevant feature of an educational system to study in relation to the distribution of civicengagement between schools and social groups. The research question of this paper is: To whatextent does standardization of civic education in secondary education relate to equality of civic andpolitical engagement of students?

We investigate the civic engagement of pupils using cross-nationally comparative data collectedin the International Civic and Citizenship Education Study (ICCS) 2009 among early adolescents. Wecombine these individual-level student assessment data with country-level indicators of thestandardization of civic education derived from Eurydice data and aggregated ICCS teacher data.This combined dataset enables us to examine whether standardization is related to inequalitiesbetween schools and between students of different social backgrounds. Our results demonstratethat accountability is associated with more inequality in civic knowledge, whereas centralization isassociated with less inequality in non-cognitive civic engagement, notably participation in thecommunity.

Theoretical framework

Education and inequality of civic and political engagement

It is well established that regarding socioeconomic circumstances, more advantaged people aremore civically and politically engaged (Dalton, 2007; Levinson, 2012; Verba et al., 1995). Educationand income are positively associated with different forms of participation, such as voter turnout,volunteering, and community work, and non-institutionalized forms of political participation, suchas signing petitions and boycotts. Education and income are also positively associated with civicknowledge and values and attitudes regarding civic and political issues (Dalton, 2007; Kalmijn &Kraaykamp, 2007; Lijphart, 1997; Marien, Hooghe, & Quintelier, 2010).

Equal opportunities for acquiring civic skills, knowledge, and opportunities for participation areimportant at an individual and societal level (Verba, Schlozman, & Brady, 1997). The civic engage-ment gap presents a challenge to the democratic ideal of political equality, to a representativedemocracy, and to social cohesion (Dahl, 2007; Lijphart, 1997; Lipset, 1959; Putnam, 2000). Fromthis perspective, it is important to understand both the potential and the disadvantages ofeducational systems in their contribution to equalizing civic learning opportunities.

Many mechanisms that cause this “civic engagement gap” have been identified. Socioeconomicbackground determines access to resources such as skills, knowledge, and material and socialcapital (Flanagan & Levine, 2010; Hauser, 2000; Verba et al., 1995). Socioeconomic background alsoaffects political socialization and, therefore, civic attitudes and normative pressures (Flanagan &Levine, 2010; Hauser, 2000). Moreover, these mechanisms are cumulative in nature; civic engage-ment affects the capacity and motivation to become civically engaged (Verba et al., 1995).

The acquisition of skills, knowledge, and social capital as well as political socialization partlyoccurs in schools, for example, through the curriculum, pedagogical method, class climate, andextracurricular activities. In addition to that, the socioeconomic composition of schools and classesalso affects learning opportunities. Therefore, it is plausible that schools affect the distribution ofcivic engagement. Moreover, this distribution may be biased by socioeconomic background. Kahne

2 J. WITSCHGE AND H. G. VAN DE WERFHORST

Dow

nloa

ded

by [

UV

A U

nive

rsite

itsbi

blio

thee

k SZ

] at

23:

04 1

7 Se

ptem

ber

2015

and Middaugh (2009) established some evidence that in California civic learning opportunities insecondary education concerning the available curriculum are distributed unequally to the benefitof socioeconomically advantaged students.

The civic and political dimensions of engagement combined constitute engagement with issuesof public concern.1 The civic and political dimensions of engagement differ because the politicaldimension involves the objective to influence public policy, whereas with civic engagement this isnot necessarily the aim (Zukin, Keeter, Andolina, Jenkins, & Carpini, 2006). To facilitate theexplorative purpose of the present study, three dimensions of civic and political engagement areexamined. The three dimensions are selected based on their relevance concerning an erosion ofyouth civic engagement, the civic engagement gap, and the expected relation to civic education.The first dimension is civic knowledge. In the context of civic education, the cognitive dimension ofcivic engagement is especially relevant to consider. Because civic knowledge is a resource forpolitical voice, a lack of civic knowledge or inequalities in civic knowledge are problematic to thelegitimacy of democracies (Verba et al., 1995). Moreover, civic knowledge is positively related topolitical participation (Galston, 2001). A lack of political participation and social backgroundinequalities in political participation are a significant concern (Dalton, 2007; Levinson, 2012;Putnam, 2000). Second, we examine an affective dimension: interest in social issues and politics.Youth apathy regarding civic issues and politics is also a significant concern (Dalton, 2007).Adolescents are not beyond the “critical period” when social and political attitudes form.Therefore, schools are considered a logical place to develop these attitudes (Flanagan & Levine,2010). Finally, we investigate a participatory dimension of civic engagement: participation in thecommunity. Research shows that this type of participation by youth is associated with adultpolitical participation (McFarland & Thomas, 2006).

Standardization of civic education

In comparative education research, it has been verified that various institutional features ofeducation affect “efficiency” (i.e., maximizing student performance), and “equality” (concerningdispersions in student outcomes or their associations with social origin) (Fuchs & Wössmann, 2007;Hanushek & Wössmann, 2010; Van de Werfhorst & Mijs, 2010). We follow the terminology of thisliterature and conceptualize inequality in civic engagement in the following two ways: (a) differ-ences in students’ civic engagement between schools and (b) the association between social originand civic engagement.

In this paper, the relation of one specific form of institutionalization of an educational systemon the three dimensions of civic engagement of students is examined, namely, standardization.Standardization in education has been referred to as the degree to which there is a “provisionof equal educational standards nationwide” (Allmendinger, 1989, p. 231). We focus in particularon the standardization of civic education. Among different educational systems, there issignificant variety in the degree to which civic education is standardized (Eurydice, 2012;Hahn, 1999, 2010; Hooghe & Claes, 2009; Torney-Purta, Schwille, & Amadeo, 1999). In thispaper, we distinguish between two dimensions of standardization: centralization andaccountability.

In the current debate, criticisms regarding standardization tend to overshadow its benefits.Standardization can undermine teacher autonomy and the possibility to adapt to specific context-driven needs (Levinson, 2011). Standardization can have undesirable side effects, such as teachingto the rating, investing more in students who are performing just below a satisfactory level – theso-called “quick gains” – at the cost of the best and worst performers, increased grade retention,increased learning pressure, and less positive attitudes of students towards learning (Amrein &Berliner, 2003; Carnoy & Loeb, 2002; Jürges & Schneider, 2010; Reback, 2008). Moreover, increasedstudent performance because of high-stakes testing may simply be driven by increased test-specific skills and effort (Jacob, 2005).

SCHOOL EFFECTIVENESS AND SCHOOL IMPROVEMENT 3

Dow

nloa

ded

by [

UV

A U

nive

rsite

itsbi

blio

thee

k SZ

] at

23:

04 1

7 Se

ptem

ber

2015

However, setting standards and holding educational stakeholders accountable for their perfor-mance can also improve efficiency, equity, and transparency (Jürges & Schneider, 2010; Levinson,2011). Setting standards can equalize learning opportunities. Monitoring educational outputsprovides information that allows the identification of problems that should be addressed andpossible measures for improvement.

CentralizationCentralization captures the scope and strength of central – national or subnational – regulationsand recommendations for formal education. Centralization is the antonym of autonomy, which ismore common in the literature because of a trend of decentralizing education across countries(Honig & Rainey, 2012).

Civic education can involve the centralization of educational inputs, such as budget allocation,the civic education curriculum (both approach and content), civic education student assessments,extracurricular civic activities organized by the school, school governance, school and class climate,civic education teacher training requirements, and continued professional development. Theextent of centralization of civic education can also differ to a significant degree between educa-tional systems. In The Netherlands, for example, schools are required to incorporate civic educationin the curriculum, but schools are free to decide how they put this requirement into practice. InFrance, in contrast, the ministry of education decides in great detail on the content of civiceducation (Hooghe & Claes, 2009). Finally, it should be noted that centralization does not equalimplementation. This difference is particularly relevant when assessing the consequences ofcentralization.

AccountabilityAccountability concerns the degree to which actors in education are held accountable for theirperformance (Wössmann, 2003). Across European countries, little standardized testing of civiceducation can be found (Eurydice, 2012). However, accountability of civic education occurs inother ways. The performance in civic education by schools can be evaluated through externalschool evaluations that cover civic education. Moreover, national or international initiatives existfor monitoring the educational systems’ performance in civic education (Eurydice, 2012).

For all civic education accountability systems, the level of accountability depends on its con-sequences concerning sanctions or rewards (Bishop, 1997; Bol, Witschge, & Van de Werfhorst, 2014;Fuchs & Wössmann, 2007). The higher the sanctions for performing poorly and rewards forperforming well, the higher the incentive is to perform better. Moreover, standardization of civiceducation output can be applied to different dimensions of civic education, such as teaching andlearning knowledge, skills, and participation. Actors in education will have the incentive to performbetter on the aspects for which they are held accountable. These aspects and the benchmarks thatare used can influence performance. Finally, the “models of assessment” that are used to ensureaccountability differ between countries. Assessment can be based on the process, improvement,and outcomes of civic education (Dijkstra & De la Motte, 2014). Some examples of information usedfor assessment are: schools’ mission statements, educational development plans, direct observa-tions of learners, and the information provided by parents (Eurydice, 2012).

Standardization of civic education and inequality of civic and political engagement

CentralizationHooghe and Claes (2009) point out that if there are no centrally administrated clear requirementsfor civic education, there can be enormous variation in it across schools, such as in Belgium. Ifeducation (including civic education) is centralized, however, students in the same country orregion will receive similar opportunities for civic learning. Logically, this centralization may causestudents to learn more homogeneous knowledge and skills. It has been demonstrated that

4 J. WITSCHGE AND H. G. VAN DE WERFHORST

Dow

nloa

ded

by [

UV

A U

nive

rsite

itsbi

blio

thee

k SZ

] at

23:

04 1

7 Se

ptem

ber

2015

centralization in an educational system decreases the impact of the particular school on how muchis learned by students, leading to more equal educational outcomes (Horn, 2009; Van de Werfhorst& Mijs, 2010; Wössmann, Luedemann, Schütz, & West, 2009). It is expected that this outcome alsoapplies to civic education. Our first hypothesis is therefore: The centralization of civic education isassociated with less dispersion in the civic engagement among students from different schools(Hypothesis 1).

Given that students from better socioeconomic backgrounds have more learning opportunitiesfor civic and political knowledge and skills at home, socioeconomically disadvantaged students relymore on learning opportunities at school. Several studies have shown that civic education has acompensatory effect for civic learning for students with less civic learning opportunities at home(Campbell, 2008; Langton & Jennings, 1968). Through equalizing all civic learning opportunities,centralized civic education may therefore compensate for the disadvantage of students with arelatively unfavorable social background. Janmaat and Mons (2011) demonstrate that in morecentralized educational systems, there is less inequality in the two civic values of ethnic toleranceand patriotism across ethnic and social groups.2 Therefore, it is expected that the centralization ofcivic education is related to a weaker association between student background and civic engagement(Hypothesis 2).

AccountabilityRegarding the relation between accountability and dispersion of school performance, the literatureprogresses in two opposite directions. Both directions begin from the mechanism of increasedincentives to perform. This mechanism implies that when students, teachers, and schools are heldaccountable for their performance and are subsequently sanctioned and rewarded according to it,their incentives to perform well will be higher. This mechanism is used to explain the finding thataccountability increases average general school performance (Bishop, 1997; Bol et al., 2014; Fuchs &Wössmann, 2007; Jürges & Schneider, 2010; Wössmann et al., 2009). Although this mechanism ofincentives to perform has not been tested in relation to civic education, one may expect that it alsoapplies to this domain; when agents in education are held accountable for civic outcomes, they willhave the incentive to perform better. These incentives to perform may have two divergingimplications for the relation between accountability measures and the distribution of civicengagement.

On the one hand, school accountability may especially cause the schools scoring below thestandard civic outcomes to improve their performance. These schools will have more to gain inimproving their civic outcomes. Moreover, if a school is aware of their relative poor performance incivic outcomes, the involved actors will feel more pressure to improve their performance to avoidthe negative repercussions of performing poorly. Narrowing dispersions between effectively andpoorly performing schools may ultimately be the result of accountability. This assumption leads toour between-school compensation hypothesis of accountability: The accountability of civic educa-tion is associated with less dispersion in civic engagement between schools (Hypothesis 3a).

Alternatively, better performing schools may be concerned with their competitive position andmake efforts to increase the quality of their civic education. Moreover, better performing schoolsmay be better able to respond to school evaluations so that their performance according to theseaccountability measures will be better. These assumptions lead to our between-school competitionhypothesis of accountability: The accountability of civic education is associated with more dispersionin civic engagement between schools (Hypothesis 3b).

Regarding social background inequality in civic engagement, we also include two contrastinghypotheses of compensation and competition. First, accountability measures can be used to holdschools accountable for the performance of students in relation to their social background. Thesemeasures provide incentives to decrease achievement gaps based on social background. Moreover,students with less civic learning opportunities at home may have greater marginal gains fromresources that facilitate civic engagement in school. This discussion leads to our social background

SCHOOL EFFECTIVENESS AND SCHOOL IMPROVEMENT 5

Dow

nloa

ded

by [

UV

A U

nive

rsite

itsbi

blio

thee

k SZ

] at

23:

04 1

7 Se

ptem

ber

2015

compensation hypothesis of accountability: The accountability of civic education is related to aweaker association between student background and civic engagement (Hypothesis 4a).

If, on the other hand, school evaluation of civic education increases the differences in outcomesbetween schools, students with an advantageous social background are expected to benefit morefrom this because they are typically clustered in relatively well-performing schools (Jenkins,Micklewright, & Schnepf, 2008). This conclusion leads to our social background competitionhypothesis of accountability: The accountability of civic education is related to a stronger associationbetween student background and civic engagement (Hypothesis 4b).

Data and methods

Data

For this study, two sources of data were used. First, the International Civic and Citizenship EducationStudy (ICCS) 2009 was used, which was conducted by the International Association for the Evaluationof Educational Achievement (IEA). The ICCS contains the results of a civic and political knowledge testand social background data from approximately 141,000 eighth-grade students from 38 countries aswell as data from teachers, school principals, and national ICCS coordinators. The sample was reducedto students aged 12 to 16. Approximately 98% of the sample remained available for analysis after thisintervention. Second, additional macro-data were obtained from the 2012 Eurydice report CitizenshipEducation in Europe. These data were merged with the ICCS data at the country level. The requiredICCS and Eurydice data were available for 24 European countries.

Dependent variablesThe following three variables were chosen for the operationalization of civic and political engage-ment: civic knowledge, interest in social and political issues, and participation in the community.This choice of variables allows the examination of the cognitive, affective, and participationdimensions of civic and political engagement.

Civic knowledge was measured by a factor score of the five plausible values for the ICCS knowledgetest. This standardized cognitive test comprises 80 questions testing students’ knowledge and skills ofreasoning and analysis in the following four “content domains”: civic society and systems, civic principles,civic participation, and civic identities (Schulz, Ainley, & Fraillon, 2009, pp. 21–31). Through principal factoranalysis of the five plausible test score values, one factor for civic knowledge was retained with aneigenvalue of 4.46 and factor loadings of .94. Subsequently, this factor was z standardized.

Interest in social and political issues was measured by a factor from six items concerning howinterested the student is in local, national, European, and international political issues and national andinternational social issues. Before performing factor analysis, the scoring was recoded so that a higherscore indicates a student is more interested. Once again, principal factor analysis was applied to createan index (eigenvalue of 3.44 and factor loadings ranging from .69 to .82) that was then standardized.

Finally, participation in the community was derived from six items in which the student is askedif he or she has engaged in different forms of civic participation (see Appendix 1 for these items).The answer categories were recoded into a dummy, where 1 indicates a student has engaged withthe particular type of participation in the past, and 0 indicates a student has never engaged with it.Subsequently, factor analysis was performed leading to an index with an eigenvalue of 1.40 andfactor loadings of .38 to .53.

The four dependent variables are mostly moderately positively correlated, with the exception ofthe correlation between participation in the community and civic knowledge.

Independent variablesA brief description of all independent macro-variables can be found in Table 1.

6 J. WITSCHGE AND H. G. VAN DE WERFHORST

Dow

nloa

ded

by [

UV

A U

nive

rsite

itsbi

blio

thee

k SZ

] at

23:

04 1

7 Se

ptem

ber

2015

To measure centralization of civic education, two macro-level variables were introduced. Thefirst variable concerns a dummy variable for whether there were central guidelines on theassessment of lower secondary education students’ active participation in school or in the com-munity in a country in 2010–2011 (yes/no, values 1/0). These data originate from the Eurydicereport Citizenship Education in Europe and are provided by the National Units of the Eurydicenetwork (Eurydice, 2012, p. 10). See Tables 2 and 3 for an overview of the distribution by countryand the basic descriptive statistics.3

Second, a variable was created to measure the centralization of civic education content. Twoitems were taken from the ICCS teacher questionnaire: In planning <civic and citizenshipeducation> or citizenship-related topics for your <target grade> classes, to what extent do youdraw on the following sources? (a) Official curricula, curricular guidelines or frameworks and (b)Official requirements (standards) in the area of <civic and citizenship education>. These two itemswere recoded so that there are the following four categories: 1 = not at all/logically not applicable;2 = to a small extent; 3 = to a moderate extent, and 4 = to a large extent. Subsequently, a factor wascreated with an eigenvalue of 1.83 and factor loadings of .96. This factor was then aggregated tothe country level, standardized, and merged with the individual data. The aggregated means bycountry are presented in Table 2. The basic descriptive statistics are presented in Table 3.

Table 1. Independent variables.

CentralizationGuidelines assessment Central guidelines on assessing lower secondary education students’ active participation in

school or in the community in a country in 2010–2011Centralized content Degree to which teachers draw on official guidelines and requirements in planning civic

education for Grade 8 (factor of two items aggregated to country level)AccountabilityExternal evaluation Evaluation of civic education teaching and learning in a school conducted by evaluators who

report to a local, regional, or central education authority but are not members of staff at theschool concerned

Table 2. Country scores on independent variables.

External evaluation Guidelines assessment Centralized content

Accountability Centralization

n = 23 countries n = 20 countries

Austria 1 0 ‒.86Belgium (Flanders) 1 0 1.39Bulgaria 0 1 ‒.44Cyprus 0 0 ‒.92Czech Republic 1 0 ‒.68Denmark 1 0 2.28England 1 0 .62Estonia 0 0 NAFinland 0 0 .78Greece 0 0 NAIreland 1 1 ‒.23Italy 1 0 .06Latvia 1 1 ‒.99Lithuania 1 0 ‒.55Luxembourg 0 0 NAMalta 0 0 .19Netherlands 0 0 NANorway 1 0 1.72Poland 1 1 ‒1.25Slovakia 0 1 ‒.17Slovenia 1 1 .89Spain 1 1 ‒1.31Sweden 1 0 ‒.08Switzerland NA NA ‒.46

SCHOOL EFFECTIVENESS AND SCHOOL IMPROVEMENT 7

Dow

nloa

ded

by [

UV

A U

nive

rsite

itsbi

blio

thee

k SZ

] at

23:

04 1

7 Se

ptem

ber

2015

Regarding accountability, this paper is concerned with the accountability of schools. This account-ability was operationalized by an external school evaluation of civic education teaching and learning inlower secondary education (ISCED 2). These data are also provided by the National Units of the Eurydicenetwork (2012). The evaluation of civic education teaching and learning is focused on “the quality ofthe teaching and learning process, pupil and student learning outcomes, as well as adherence toofficial curriculum content and to recommended teaching methods”4 (Eurydice, 2012, p. 79). Theexternal school evaluation was conducted by evaluators who report to a local, regional, or centraleducation authority but are not members of staff at the school concerned. A binary variable wascreated indicating whether the type of evaluation is applied in a country (yes/no, values 1/0). In Table 2,an overview of the score by country can be found. The basic descriptive statistics are presented inTable 3.

Control variablesThe individual-level control variables that were included are gender, age, immigration status,religion, and two measures of social background: (a) index of students’ socioeconomic back-ground (“socioeconomic status” [SES]) and (b) parents’ interest in social and political issues(“parents’ interest”). The SES index was composed of the following three socioeconomic vari-ables: parents’ higher occupational status, parents’ highest educational level in approximateyears of education, and the approximate number of books at home. Following the sameprocedure5 by which the national index of socioeconomic background was created by theIEA (Schulz et al., 2009, p. 193), these variables were standardized for the countries in thesample. After this standardization, principal component analysis was applied to create aninternational index of SES. The factor that was retained has an eigenvalue of 1.52 and factorloadings of .84 for parents’ higher occupational status and parents’ highest educational level inapproximate years of education and a factor loading of .29 for the approximate number ofbooks at home. Finally, a z standardized variable was created with a mean of 0 and a standarddeviation of 1.

Parents’ interest in social and political issues concerns a variable for which the categories andtheir corresponding meanings can be found in Appendix 1. The two social background measurescorrelate only modestly with r = .20 overall and with correlations between r = .09 (Malta) andr = .28 (Luxembourg) within countries. All individual-level data were derived from the studentbackground questionnaires completed by the students. The mean, standard deviation, and rangeare presented for every control variable in Table 3. To account for the socioeconomic compositionof the students’ class, we included a variable for the mean level of parents’ interest and SES in themodels.6

Table 3. Descriptive statistics.

Mean Standard deviation Range

Dependent variablesCivic knowledge (67054 obs) 0 1 ‒4.08–3.60Interest in social and political issues (65577) 0 1 ‒1.86–2.45Participation in community (65799) 0 1 ‒.85–3.47

Control variables (67054)Female (1 = yes) .51 .5 0–1Student age 14.30 .5 13–16Religion (1 = religious) .56 .5 0–1Immigrant status (1 = non-native or first generation) .1 .29 0–1Students’ socioeconomic background 0 1 ‒3.7–5.7Interest in social and political issues parents 1.93 .73 0–3

Independent variablesExternal evaluation (n student = 67054; n country = 23) .63 .48 0–1Guidelines assessment (n student = 67054; n country = 23) .30 .46 0–1Centralized content (n student = 42020; n country = 20) 0 1 ‒1.31–2.28

8 J. WITSCHGE AND H. G. VAN DE WERFHORST

Dow

nloa

ded

by [

UV

A U

nive

rsite

itsbi

blio

thee

k SZ

] at

23:

04 1

7 Se

ptem

ber

2015

Methods

Multilevel analysisThe hypotheses were tested with three-level random intercept regression models that account forthe nested structure of the data, with individuals nested in schools and in countries.7 To analyzethe impact of standardization on the equality of civic engagement, cross-level interaction variableswere created with the three institutional feature variables described above and the two individual-level social background variables SES and parents’ interest.

For all three dependent variables, two models were run; first, a model in which the main effect ofexternal evaluation and guidelines for assessmentwas estimated, and in addition, an interaction betweenthese institutional variables with the two micro-social background variables was estimated (Table 4).

The “centralization of civic education content” was available for a different group of countriesthan the other three indicators of standardization. Therefore, a separate model was run in whichthe coefficients of the control variables, centralized content, and a cross-level interaction ofcentralized content with both social background variables are estimated (Table 5).

Because the interest in this paper concerns the impact of institutional features, differences in thesample size for countries were accounted for by including sampling weights so that all countrieshave an equal weight.

Variance function regressionThe hypotheses regarding between-school dispersion of students’ civic and political engagement aremodeled through a variance function regression. In variance function regressions, the variance of anoutcome variable is modeled as functions of covariates (Western & Bloome, 2009; Zeng, Yang Yang, &Land, 2011). For our hypothesis involving between-school inequality, we are interested in between-school within-country variance of the dependent variables. Two steps were taken to model this focus.

First, a random intercept three-level regression model with only the control variables was run forthe three dependent variables. The between-school within-country residuals were stored.Subsequently, the log of the squared between-school within-country residuals were estimatedfor every dependent variable.

In the second step, the log of the squared residuals was taken as the dependent variable in amodel including only the independent country-level variables of interest. To account forheteroskedasticity, a general linear model (GLM) gamma function with a log link was used.However, this method prevents a multilevel approach. To obtain unbiased standard errors, thedata were clustered by country.

The estimated covariates of the GLM model can be interpreted as the effect of the measures ofstandardization (independent variables) on the magnitude of the between-school variability in acountry.

Similar to the multilevel models, the variance regressions were performed with two models forevery dependent variable because of the available sample for the macro-variables. One modelincluded the external evaluation and guidelines for assessment, and a second model included thecentralization of civic education content.

Reliability and validity

Some caveats should be made regarding several issues concerning the data and methods. First, weare using cross-sectional data. This fact means we cannot make any causal claims based on theresults. Findings involving any relations between civic education standardization and civic engage-ment inequality may be a result of preexisting differences.

Second, ICCS 2009 does not contain individual-level data on general cognitive ability. Toaccount for this deficiency as much as possible, we repeated our analysis including the variablecivic knowledge as an independent variable. Logically, this approach could not be implemented in

SCHOOL EFFECTIVENESS AND SCHOOL IMPROVEMENT 9

Dow

nloa

ded

by [

UV

A U

nive

rsite

itsbi

blio

thee

k SZ

] at

23:

04 1

7 Se

ptem

ber

2015

the models with civic knowledge as a dependent variable. A small number of findings did not passthis robustness check, and they are mentioned in the results section. The main effects demonstratethat civic knowledge is positively related to interest in social and political issues and negativelyrelated to participation in the community.

Another possibly influential element is the particular civic or political culture and influential eventsin countries (Almond & Verba, 1989; Isac, Maslowski, & Van der Werf, 2011), especially becausepolitical culture is known to affect the development of educational systems (Green, 1990). To accountfor this element, we re-estimated the models accounting for the control of corruption in 2009 asoperationalized by the World Bank. This variable reflects “perceptions of the extent to which publicpower is exercised for private gain, including both petty and grand forms of corruption, as well as‘capture’ of the state by elites and private interests” (Kaufmann, Kraay, & Mastruzzi, 2009, p. 6).Including this measure did not alter our findings. Other macro-variables such as the human devel-opment index (HDI), the World Bank’s governmental efficacy index, and the Economist’s democracyindex were also considered. However, these variables did not meaningfully contribute to the modelsbecause of the high correlation with the control of corruption variable. The control of corruptionmeasure was chosen because, in our opinion, it measures political culture better than the HDI and thegovernmental efficacy index. In addition, the democracy index includes public-opinion measures ofparticipation in political culture, which could lead to tautological findings.

Two of the three indicators of standardization used in this paper, an external evaluation and guide-lines for assessment, are based on dichotomous measures. This basis allows for comparing standardiza-tion somewhat between countries but does not explain the degree to which a form of standardizationoccurs. The actual difference in the level of standardization between two countries where one scoreszero and the other scores one may in some cases be relatively small. In this way, binary measures mayamplify differences in standardization between countries, thus leading to bias. Furthermore, our binarymeasures of standardization do not comprise qualitative differences of standardization between coun-tries. Forms of standardization are manifested in many different ways across countries. Because ourindicators are blind to these differences, this leaves the risk of finding relations that are actually spurious.

To uniformly measure the standardization of civic education across countries, we use two typesof data: (a) data provided by country experts assigned by Eurydice concerning whether there areofficial standardization measures and (b) teacher data regarding whether teachers are usingcentralized regulations. Both types of data have their advantages and drawbacks regardingreliability and validity. The expert data that are used are valuable because they offer standardizedmeasures of accountability and centralization. These data, however, are sensitive to measurementerror, and, therefore, the results may be biased. The teacher data do not actually concerncentralization, but the perceived implementation of centralization by teachers. This type of datacan simultaneously be considered a drawback and an advantage. The experience of teachers is lessobjective than the actual existence of centralized guidelines. At the same time, this type of dataundermines the risk of influential measurement error that occurs when we use expert data.Moreover, the implementation of centralized regulations drives our hypothesis. When interpretingthe results, we must consider these characteristics that make the data imperfect. We think it isworthwhile to analyze the data because they allow us to study standardization across countriesuniformly.

Results

Results of the multilevel analysis

The results of the three-level models are presented in Tables 4 and 5. Table 4 presents the modelincluding the external evaluation and guidelines for assessment and across-level interaction ofthese two measures of standardization with SES and parents’ interest for all three dependent

10 J. WITSCHGE AND H. G. VAN DE WERFHORST

Dow

nloa

ded

by [

UV

A U

nive

rsite

itsbi

blio

thee

k SZ

] at

23:

04 1

7 Se

ptem

ber

2015

variables. Table 5 presents the models including centralized content with cross-level interactionsfor both social background variables.

The variance components for the three different levels are included in the model. It is clear thatthe vast majority of variance lies at the student level (62 to 94% of the total variance) followed bythe country level (3 to 20%), and the least variance is found at the school level (3 to 18%). Out ofthe four dependent variables, contextual factors – notably, the school – are most important for thecognitive dimension of civic engagement.

Girls score higher than boys on civic knowledge and participation in the community but loweron interest in social and political issues. Age is negatively associated with civic knowledge andpositively associated with the two remaining dependent variables. Identifying oneself as beingreligious is positively related with interest in social and political issues and participation in thecommunity and is not significantly related to civic knowledge. Finally, being an immigrant isnegatively associated with civic knowledge and positively associated with interest in social andpolitical issues. In the model with centralized content (Table 5), being an immigrant is alsopositively related to participation in the community.8

The findings regarding the relation between social background and civic engagement aregenerally consistent with the consensus in the literature. Both students’ SES and parents’ interest

Table 4. Results multilevel models external evaluation and guidelines for assessment.

Civic Knowledge Interest in social and political issues Participation in community

Female .194*** ‒.080*** .139***(.022) (.021) (.020)

Age ‒.081*** .043** .044***(.023) (.015) (.010)

Religious .009 .145*** .078***(.027) (.016) (.014)

Immigrant ‒.247*** .227*** .044(.031) (.031) (.038)

Parents’ interest .085*** .431*** .140***(.015) (.016) (.021)

SES .171*** .027*** .013(.027) (.007) (.020)

Class mean parents’ interest .166* .197*** .094*(.065) (.038) (.014)

Class mean SES .309*** ‒.024† .031(.052) (.014) (.019)

External evaluation .168 ‒.009 ‒.072(.204) (.094) (.086)

Guidelines for assessment .012 .068 .130†(.148) (.093) (.075)

Cross-level interactionsExternal evaluation*parents’ interest .040† .022 ‒.012

(.020) (.020) (.021)External evaluation*SES .053† .017 ‒.005

(.030) (.015) (.020)Guidelines*parents’ interest ‒.023 ‒.029 .017

(.024) (.021) (.015)Guidelines*SES ‒.019 ‒.055*** ‒.023†

(.029) (.015) (.013)Constant .523 ‒1.951*** ‒1.189***

(.412) (.199) (.200)Variance i (student) .578 (61.62%) .809 (93.42%) .876 (90.59%)Variance j (school) .17 (18.12%) .025 (2.89%) .047 (4.86%)Variance k (country) .19 (20.26%) .032 (3.7%) .044 (4.55%)Ni 67054 65577 65799Nj 3152 3150 3151Nk 23 23 23

Note: Standard error between brackets.***p < .001.**p < .01. *p < .05. †p < .1.

SCHOOL EFFECTIVENESS AND SCHOOL IMPROVEMENT 11

Dow

nloa

ded

by [

UV

A U

nive

rsite

itsbi

blio

thee

k SZ

] at

23:

04 1

7 Se

ptem

ber

2015

are positively associated with all the dependent variables with the exception of the relationbetween SES and participation in the community. The aggregate measures of social background(class mean of parents’ interest and SES) are also mostly positively related to the three civicoutcomes. However, the class SES mean is negatively associated with interest in social and politicalissues in the model with external evaluation and guidelines (Table 4) and with participation in thecommunity in the model with centralized content (Table 5). The class SES mean is not significantlyrelated in the other model to these two dependent variables. Mean parents’ interest is notsignificantly related to knowledge in the model with centralized content.

In Table 4, external evaluations and guidelines for assessment do not significantly affect the civicoutcomes. The only significant main effect is found with the positive relation between guidelinesfor assessment and participation in the community with .13 at a significance level of α < .1. InTable 5, we find that centralized content is negatively associated with interest in social and politicalissues and participation in the community with ‒.128 at α < .001 and ‒.086 at α < .05, respectively.

Our main interest in Tables 4 and 5, however, concerns the cross-level interactions among thethree indicators of standardization and the two social background variables.

In Table 4, we see that an external evaluation of civic education significantly interacts positivelywith both parents’ interest (.04 at α < .1) and SES (.053 at α < .1) for civic knowledge. Externalevaluations of civic education are thus associated with more social background inequality in civicknowledge. This result supports Hypothesis 4b: The accountability of civic education is positivelyrelated to a stronger association between student background and civic engagement. For the othertwo dependent variables, no significant interactions between external evaluations and the twosocial background variables are found, falsifying Hypotheses 4a and 4b.

Table 5. Results multilevel models centralized content.

Civic Knowledge Interest in social and political issues Participation in community

Female .196*** ‒.081*** .151***(.022) (.024) (.021)

Age ‒.080** .042** .044***(.025) (.016) (.011)

Religious ‒.004 .144*** .083***(.031) (.017) (.017)

Immigrant ‒.267*** .237*** .053†(.031) (.032) (.032)

Parents’ interest .102*** .437*** .136***(.011) (.011) (.010)

SES .203*** .016† .005(.016) (.009) (.008)

Class mean parents’ interest .136 .191*** .080†(.088) (.045) (.043)

Class mean SES .330*** ‒.027 ‒.054*(.054) (.019) (.023)

Centralized content .088 ‒.128*** ‒.086*(.087) (.028) (.041)

Cross-level interactions

Centralized content*parents’ interest .023* .015 ‒.018*(.011) (.009) (.009)

Centralized content*SES .018 .017 .002(.015) (.011) (.006)

Constant .678† ‒1.916*** ‒1.175***(.405) (.228) (.171)

Variance i (student) .586 (62.94%) .808 (93.95%) .884 (90.76%)Variance j (school) .162 (17.4%) .027 (3.14%) .049 (5.31%)Variance k (country) .183 (19.66%) .025 (2.91%) .041 (4.21%)Ni 58486 57225 57362Nj 2919 2917 2918Nk 20 20 20

Note: Standard error between brackets.***p < .001. **p < .01. *p < .05. †p < .1.

12 J. WITSCHGE AND H. G. VAN DE WERFHORST

Dow

nloa

ded

by [

UV

A U

nive

rsite

itsbi

blio

thee

k SZ

] at

23:

04 1

7 Se

ptem

ber

2015

The existence of guidelines for assessing students’ participation in school and in the community isassociated with less social background inequality in two instances: The positive association betweenSES and students’ interest in social and political issues and participation in the community are smallerwhen there are guidelines for assessment (effect size respectively ‒.055 at α < .001 and ‒.023 at α < .1).These findings support Hypothesis 2: The centralization of civic education is related to a weakerassociation between student background and civic engagement.

In Table 5, we see that the centralized content of civic education is associated with socialbackground inequality of both the cognitive and the participation components of civic engage-ment. The positive association between parents’ interest and the score on the civic knowledge testis magnified by .023 at the α < .05 level for an increase of one on the civic education contentcentralization index. Concerning the range of the content centralization scale of 3.59, the interac-tion effect will be a maximum of ‒.023*3.59. This enhanced inequality with centralization is theopposite of what was expected in Hypothesis 2. The positive association between parents’ interestand participation in the community, in contrast, is diminished by ‒.018 at α < .05 for an increase of1 on the civic education content centralization index. Hypothesis 2 is thus confirmed for participa-tion in the community.9

Both the impact of parents’ interest and SES are associated with standardization variables threetimes. We find indications that guidelines for assessment are associated with SES inequality,whereas centralized content is associated with inequality by parents’ interest. Interestingly, thetwo measures of centralization are associated with another dimension of social backgroundinequality. Civic education content is associated with inequality in civic engagement by parents’interest in social and political issues, whereas central guidelines for assessment of participation inthe community are associated with inequality in civic engagement by SES.

Results in between-school variance

In Table 6, the covariates that estimate dispersion in civic engagement between schools withincountries are presented. Again, because of data availability, the results were derived through twomodels: one model with external evaluation and guidelines for assessment and a second modelwith civic education content centralization. The betas presented in Table 6 are the estimatedeffects of their corresponding independent variable of standardization on the squared residuals ofschools within countries.

Between-school variance of the dependent outcomes is not significantly related to our account-ability measure of external school evaluation. Hypotheses 3a and 3b are thus rejected.

Centralization of civic education is associated with between-school variance of civicengagement in three instances. Civic education content centralization is associated with lessbetween-school variance for civic knowledge and participation in the community. These findingsconfirm Hypothesis 1: The centralization of civic education is associated with less dispersion in civicengagement between students from different schools. Guidelines for assessment, on the other hand,are associated with more between-school inequality for participation in the community. Thisfinding is the opposite of what was expected in Hypothesis 1.

Table 6. Results variance function regression.

External evaluation23 countries

Guidelines assessment23 countries

Centralized content23 countries

Civic knowledge ‒.28 [.21] .23 [.16] ‒.13* [.07]Interest in social and political issues .01 [.12] ‒.01 [.08] .02 [.05]Participation in community ‒.24 [.19] .32* [.14] ‒.28* [.11]

Note: Standard error between brackets.***p < .001. **p < .01. *p < .05. †p < .1.

SCHOOL EFFECTIVENESS AND SCHOOL IMPROVEMENT 13

Dow

nloa

ded

by [

UV

A U

nive

rsite

itsbi

blio

thee

k SZ

] at

23:

04 1

7 Se

ptem

ber

2015

Discussion and conclusion

In this paper, we performed a cross-country comparative study involving 24 European countries toexamine whether standardization of civic education is associated with inequality of civic engagement.Inequality of civic engagement was examined in two ways: (a) between-school within-countryvariance and (b) the association between social background and student civic engagement.

The findings demonstrate that schools and the educational system explain only a modest part ofdifferences in civic engagement; the four dimensions of civic engagement examined in this paperare mostly explained by characteristics at the individual level. This result is consistent with othermultilevel studies of civic outcomes in education (Dijkstra, Geijsel, Ledoux, Van der Veen, & TenDam, 2014; Isac et al., 2011; Isac, Maslowski, Creemers, & Van der Werf, 2014; Quintelier, 2010). Notsurprisingly, schools are the most important for the cognitive dimension.

The results confirm the existence of a civic engagement gap between lower secondary schoolstudents. Both individual social background characteristics and class composition matter. However,as demonstrated in this paper, civic engagement inequality is altered by the standardization ofcivic education.

A distinction was made between two dimensions of standardization: centralization and account-ability. Regarding accountability, we found indications that external school evaluations are associatedwith more social background inequality in the cognitive dimension of civic engagement. This findingsupports the idea that the sanctions and rewards following external school evaluations encouragecompetition between schools to the benefit of students with a more advantageous social background.

Regarding centralization, the findings are more mixed. The first measure of centralization,centralization of civic education content, is associated with less inequality in participation in thecommunity both between schools and regarding parents’ social background. Centralization of civiceducation content is also associated with less background inequality in the civic knowledgebetween schools, but at the same time inequality in civic knowledge by social background islarger in countries with more centralized civic education content. The second measure of centra-lization, central guidelines for assessment of participation in the community, is associated with lesssocial background inequality in interest in social and political issues and participation in thecommunity. Between-school inequality in participation in the community, in contrast, is larger incountries where central guidelines for assessment exist.

These findings indicate that, although it is important to distinguish between different types ofcentralization, centralization is mostly associated with less civic engagement inequality. This isconsistent with our expectation. It can be explained by an equalizing of civic learning opportunitiesbetween schools and the compensatory effect of civic education for parental civic socialization athome. The small number of findings that show that centralization is, in fact, also associated withmore civic engagement can perhaps be explained by the fact that better performing schools maybe better able to implement centralized rules and regulations in an effective manner than schoolsperforming less well.

The centralization of content aspect of centralization only measures centralization if it isimplemented. In practice, centralized guidelines and regulations are not always implemented.Moreover, there can be great variety in how centralized guidelines and regulations are interpretedand implemented. This disparity may have led to bias in the results. In future studies, it may bemore informative to focus exclusively on measures of implementation of centralization to avoidthis bias.

A final note on centralization is that it can refer to both national and regional guidelines andregulations. Building on the theory in this paper, national centralization may imply lower variationin guidelines and regulations than regional centralization. Consequently, national centralizationmay have more of an equalizing effect on civic engagement.

Because of the cross-sectional nature of the data and the limited comparative data on standar-dization of civic education that are reliable and have good validity, the findings of this study should

14 J. WITSCHGE AND H. G. VAN DE WERFHORST

Dow

nloa

ded

by [

UV

A U

nive

rsite

itsbi

blio

thee

k SZ

] at

23:

04 1

7 Se

ptem

ber

2015

be considered with caution. One important improvement that could be made in future research isthe use of longitudinal data to examine whether we are dealing with a causal relation.

Despite these limitations, this paper provides the new insight that standardization of civiceducation is associated with the distribution of civic engagement among secondary schoolstudents. This result strengthens the case that if governments are interested in the civic engage-ment gap, the context of the educational system should be considered, especially becausemeasures of standardization may lead to an undesirable larger gap in civic knowledge. Thepaper suggests that standardization measures might be an effective tool to address the civicengagement gap and warns that well-intended policy measures involving standardization can beharmful to democratic equality.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Funding

This work was supported by the Programme Council for Fundamental Research of the Netherlands Initiative forEducation Research (PROO), Netherlands Organization for Scientific Research (NWO) [grant number 411-10-920].

Notes

1. The social and religious dimensions of engagement with society are not included in the dependent variable ofinterest because they are not necessarily concerned with the common good (Campbell, 2006; Putnam, 2000).

2. Janmaat and Mons (2011) use a different concept than centralization, namely, “territorial differentiation” or the“division of powers among the national, local, and school levels regarding curriculum, textbooks, and assess-ment” (p. 57).

3. From the report, it is not clear if in Latvia these guidelines are available for ISCED2. Latvia’s National EurydiceUnit confirmed to the authors of this article that this is the case.

4. Other areas of civic education on which schools can be both internally and externally evaluated distinguished byEurydice are school culture, school governance, and relationships with thewider community (Eurydice, 2012, p. 79).

5. We followed the same procedure with the exception that we did not impute values for missing data. Instead,we excluded respondents who had a missing value for one of the three variables on which the index is based.

6. At the school level, variables for private/public education and the size of the community in which the school islocated were also considered but did not contribute to the models substantially. A variable for the religiousdenomination of the school was not available.

7. A four-level analysis including the class level was not suitable since in the majority of the participating schoolsstudents from only one class were tested. The school level was chosen in order to increase the comparabilitybetween the multilevel models and the variance function models which concern between-school residuals(see above). The models were re-run using class instead of school as the second level in order to check therobustness of the findings. This did not substantially change our findings.

8. In the robustness checks, we included, in separate models, two additional control variables: (a) the World Bank2009 control of corruption index and (b) civic knowledge (as an independent variable). The findings show thata higher score on the corruption index is significantly associated with higher civic knowledge and lowerinterest in social and political issues as well as participation in the community. Civic knowledge is positivelyrelated to interest in politics and negatively to participation in the community.

9. When we include civic knowledge in the model as a control variable, this coefficient is slightly smaller and lessefficient with ‒.016 at α < .1.

Notes on contributors

Jacqueline Witschge is a PhD candidate in sociology at the Amsterdam Institute for Social Science Research (AISSR)and the Amsterdam Centre for Inequality Studies (AMCIS). Her research deals with the question to what degreeeducational systems, including the organization of civic education, affect the distribution of civic competences acrosssocioeconomic groups. Her main research interests are as follows: social stratification and mobility, educationalinequality, political socialization, and civic engagement. She has published a paper in Social Forces.

SCHOOL EFFECTIVENESS AND SCHOOL IMPROVEMENT 15

Dow

nloa

ded

by [

UV

A U

nive

rsite

itsbi

blio

thee

k SZ

] at

23:

04 1

7 Se

ptem

ber

2015

Herman G. van de Werfhorst is a professor of sociology at the University of Amsterdam and director of the AmsterdamCentre for Inequality Studies (AMCIS). He studies the impact of educational institutions on various inequalities. Hisrecent publications on this topic appeared in, among others, the Annual Review of Sociology, Social Science Research,Social Forces, Comparative Education Review, Demography, Acta Sociologica, European Sociological Review, Research inSocial Stratification and Mobility, and Oxford University Press.

References

Allmendinger, J. (1989). Educational systems and labor market outcomes. European Sociological Review, 5, 231–250.Almond, G., & Verba, S. (1989). The civic culture: Political attitudes and democracy in five nations. London: Sage.Amrein, A. L., & Berliner, D. C. (2003). The effects of high-stakes testing on student motivation and learning. A research

report. Educational Leadership, 60(5), 32–38.Bishop, J. (1997). The effect of national standards and curriculum-based exams on achievement. The American

Economic Review, 87, 260–264.Bol, T., Witschge, J. H. M., Van de Werfhorst, H. G., & Dronkers, J. (2014). Curricular tracking and central examinations:

Counterbalancing the impact of social background on student achievement in 36 countries. Social Forces, 92, 1545–1572.

Campbell, D. E. (2006). What is education’s impact on civic and social engagement? In R. Desjardins & T. Schuller (Eds.),Measuring the effects of education on health and civic engagement: Proceedings of the Copenhagen symposium (pp.25–169). Paris: CERI, OECD.

Campbell, D. E. (2008). Voice in the classroom: How an open classroom climate fosters political engagement amongadolescents. Political Behavior, 30, 437–454.

Carnoy, M., & Loeb, S. (2002). Does external accountability improve student outcomes? A cross-state analysis.Educational Evaluation and Policy Analysis, 24, 305–331.

Dahl, R. A. (2007). On political equality. New Haven, CT: Yale University Press.Dalton, R. J. (2007). The good citizen: How a younger generation is reshaping American politics. Washington, DC: CQ

Press.Dee, T. S. (2004). Are there civic returns to education? Journal of Public Economics, 88, 1697–1720.Dijkstra, A. B., & De la Motte, P. I. (Eds.). (2014). Social outcomes of education: The assessment of social outcomes and

school improvement through school inspections. Amsterdam: Amsterdam University PressDijkstra, A. B., Geijsel, F., Ledoux, G., Van der Veen. I., & Ten Dam, G. (2014). Effects of school quality, school citizenship

policy, and student body composition on the acquisition of citizenship competences in the final year of primaryeducation. School Effectiveness and School Improvement. Advance online publication. doi:10.1080/09243453.2014.969282

Eckstein, K., Noack, P., & Gniewosz, B. (2012). Attitudes toward political engagement and willingness to participate inpolitics: Trajectories throughout adolescence. Journal of Adolescence, 35, 485–495.

Eurydice. (2012). Citizenship education in Europe. Brussels: Education, Audiovisual and Culture Executive Agency.Flanagan, C., & Levine, P. (2010). Civic engagement and the transition to. The Future of Children, 20(1), 159–179.Fuchs, T., & Wössmann, L. (2007). What accounts for international differences in student performance? A re-examina-

tion of using PISA data. Empirical Economics, 32, 433–464.Galston, W. A. (2001). Political knowledge, political engagement and civic education. Annual Review of Political Science,

4, 217–234.Green, A. (1990). Education and state formation: The rise of the education systems in England, France and the USA.

Basingstoke: Macmillan.Hahn, C. L. (1999). Citizenship education: An empirical study of policy, practices and outcomes. Oxford Review of

Education, 25, 231–250.Hahn, C. L. (2010). Comparative civic education research: What we know and what we need to know. Citizenship

Teaching and Learning, 6, 5–23.Hanushek, E. A., & Wössmann, L. (2010). The economics of international differences in educational achievement (NBER

Working Paper No. 15949). Cambridge, MA: National Bureau of Economic Research.Hauser, S. M. (2000). Education, ability, and civic engagement in the contemporary United States. Social Science

Research, 29, 556–582.Honig, M. I., & Rainey, L. R. (2012). Autonomy and school improvement: What do we know and where do we go from

here? Educational policy, 26, 465–495.Hooghe, M., & Claes, E. (2009). Civic education in Europe: Comparative policy perspectives from the Netherlands,

Belgium, and France. In J. Younnis & P. Levine (Eds.), Engaging young people in civic life (pp. 235–272). Nashville, TN:VanderBilt University Press.

Horn, D. (2009). Age of selection counts: A cross-country analysis of educational institutions. Educational Research andEvaluation, 15, 343–366.

16 J. WITSCHGE AND H. G. VAN DE WERFHORST

Dow

nloa

ded

by [

UV

A U

nive

rsite

itsbi

blio

thee

k SZ

] at

23:

04 1

7 Se

ptem

ber

2015

Isac, M. M., Maslowski, R., Creemers, B., & Van der Werf, G. (2014). The contribution of schooling to secondary-schoolstudents’ citizenship outcomes across countries. School Effectiveness and School Improvement, 25, 29–63.

Isac, M. M., Maslowski, R., & Van der Werf, G. (2011). Effective civic education: An educational effectiveness model forexplaining students’ civic knowledge. School Effectiveness and School Improvement, 22, 313–333.

Jacob, A. (2005). Accountability, incentives and behavior: The impact of high-stakes testing in the Chicago PublicSchools. Journal of Public Economics, 89, 761–796.

Janmaat, J. G. (2011). Ability grouping, segregation and civic competences among adolescents. International Sociology,26, 455–482.

Janmaat, J. G., & Mons, N. (2011). Promoting ethnic tolerance and patriotism: The role of education systemcharacteristics. Comparative Education Review, 55, 56–81.

Janmaat, J. G., Mostafa, T., & Hoskins, B. (2014). Widening the participation gap: The effect of educational track onreported voting in England. Journal of Adolescence, 36, 473–482.

Jenkins, S. P., Micklewright, J., & V. Schnepf, S. (2008). Social segregation in secondary schools: How does Englandcompare with other countries? Oxford Review of Education, 34, 21–37.

Jürges, H., & Schneider, K. (2010). Central exit examinations increases performance . . . but take the fun out ofmathematics. Journal of Popular Economics, 23, 497–517.

Kahne, J., & Middaugh, E. (2009). Democracy for some: The civic opportunity gap in high school. In J. Younnis & P.Levine (Eds.), Engaging young people in civic life (pp. 29–58). Nashville, TN: VanderBilt University Press.

Kalmijn, M., & Kraaykamp, G. (2007). Social stratification and attitudes: A comparative analysis of the effects of classand education in Europe. The British Journal of Sociology, 58, 547–576.

Kaufmann, D., Kraay, A., & Mastruzzi, M. (2009). Governance matters VIII: Aggregate and individual governance indicators1996–2008 (Policy Research Working Paper No. 4978). Washington, DC: The World Bank.

Langton, K. P., & Jennings, M. K. (1968). Political socialization and the high school civics curriculum in the UnitedStates. American Political Science Review, 62, 559–567.

Levinson, M. (2011). Democracy, accountability, and education. Theory and Research in Education, 9, 125–144.Levinson, M. (2012). No Citizen Left Behind. Cambridge, MA: Harvard University Press.Lijphart, A. (1997). Unequal participation: Democracy’s unresolved dilemma. American Political Science Review, 91, 1–14.Lipset, S. M. (1959). Some social requisites of democracy: Economic development and political legitimacy. American

Political Science Review, 53, 69–105.Marien, S., Hooghe, M., & Quintelier, E. (2010). Inequalities in non-institutionalised forms of political participation: A

multi-level analysis of 25 countries. Political Studies, 58, 187–213.McFarland, D. A., & Thomas, R. J. (2006). Bowling young: How youth voluntary associations influence adult political

participation. American Sociological Review, 71, 401–425.Milligan, K., Moretti, E., & Oreopoulos, P. (2004). Does education improve citizenship? Evidence from the United States

and the United Kingdom. Journal of Public Economics, 88, 1667–1695.Newcomb, T. M., Koenig, K. E., Flacks, R., & Warwick, D. P. (1967). Persistence and change: Bennington college and its

students after 25 years. New York, NY: John Wiley.Nie, N., Junn, J., & Stehlik-Barry, K. (1996). Education and democratic citizenship in America. Chicago, IL: University of

Chicago Press.Putnam, R. (2000). Bowling alone: The collapse and revival of American community. New York, NY: Simon & Schuster.Quintelier, E. (2010). The effect of schools on political participation: A multilevel logistic analysis. Research Papers in

Education, 25, 137–154.Reback, R. (2008). Teaching to the rating: School accountability and the distribution of student achievement. Journal

of Public Economics, 92, 1394–1415.Schulz, W., Ainley, J., & Fraillon, J. (Eds.). (2009). ICCS 2009 technical report. Amsterdam: International Association for

the Evaluation of Educational Achievement.Schulz, W., Ainley, J., Fraillon, J. Kerr, D., & Cosito, B. (2010). ICCS 2009 international report: Civic knowledge, attitudes,

and civic engagement among lower secondary school students in 38 countries. Amsterdam: International Associationfor the Evaluation of Educational Achievement.

Torney-Purta, J., Lehmann, R., Oswald, H., & Schulz, W. (2001). Citizenship and education in twenty-eight countries: Civicknowledge and engagement at age 14. Amsterdam: International Association for the Evaluation of EducationalAchievement.

Torney-Purta, J., Schwille, J., & Amadeo, J. (1999). Civic education across countries: Twenty-four national case studiesfrom the IEA Civic Education Project. Amsterdam: International Association for the Evaluation of EducationalAchievement.

Van der Wal, M., & Waslander, S. (2007). Traditional and non-traditional educational outcomes: Trade-off or comple-mentarity? School Effectiveness and School Improvement, 18, 409–428.

Van de Werfhorst, H. G. (2014). Changing societies and four tasks of schooling: Challenges for strongly differentiatededucational systems. International Review of Education, 60, 123–144.

Van de Werfhorst, H. G., & Mijs, J. (2010). Achievement inequality and the institutional structure of educationalsystems: A comparative perspective. Annual Review of Sociology, 36, 407–428.

SCHOOL EFFECTIVENESS AND SCHOOL IMPROVEMENT 17

Dow

nloa

ded

by [

UV

A U

nive

rsite

itsbi

blio

thee

k SZ

] at

23:

04 1

7 Se

ptem

ber

2015

Verba, S., Schlozman, K. L., & Brady, H. E. (1995). Voice and equality: Civic voluntarism in American politics. Cambridge,MA: Harvard University Press.

Verba, S., Schlozman, K. L., & Brady, H. E. (1997). Reply to reviews. Voice and equality: Civic voluntarism in Americanpolitics. American Political Science Review, 91, 427–430.

Western, B., & Bloome, D. (2009), Variance function regression for studying inequality. Sociological Methodology, 39,293–326.

Westheimer, J., & Kahne, J. (2004). What kind of citizen? The politics of educating for democracy. American EducationalResearch Journal, 41, 237–269.

Wössmann, L. (2003). Schooling resources, educational institutions and student performance: The internationalevidence. Oxford Bulletin of Economics and Statistics, 65, 117–170.

Wössmann, L., Luedemann, E., Schütz, G., & West, M. (2009). School accountability, autonomy, and choice around theworld. Cheltenham: Edward Elgar.

Zeng, H., Yang Yang, Z. H., & Land, K. C. (2011). Variance function regression in hierarchical age-period-cohort models.American Sociological Review, 76, 955–983.

Zukin, C., Keeter, C., Andolina, M., Jenkins, M., & Carpini, M. X. D. (2006). A new engagement? Political participation, civiclife, and the changing American citizen. New York, NY: Oxford University Press.

Appendix 1. Underlying items

Interest in politics/social issues

How much are you interested in the following issues?

(1) Political issues within your <local community>?(2) Pol issues in your country?(3) International politics?(4) European politics?(5) Social issues in your country?(6) Social issues in other countries?

Civic participation in the community

Have you ever been involved in activities of any of the following organizations, clubs, orgroups?

(1) Youth organization affiliated with a political party or union.(2) Environmental organization.(3) Human rights organization.(4) A voluntary group doing something to help the community.(5) An organization collecting money for a social cause.(6) A group of young people campaigning for an issue.

Categories parents’ interest in social and political issues

0 = not interested at all1 = not very interested2 = quite interested3 = very interested

18 J. WITSCHGE AND H. G. VAN DE WERFHORST

Dow

nloa

ded

by [

UV

A U

nive

rsite

itsbi

blio

thee

k SZ

] at

23:

04 1

7 Se

ptem

ber

2015