Civic Knowledge, Civic Skills and Civic Engagement

European Educational Research Journal, Volume 2, Number 3, 2003 Civic Knowledge, Civic Skills and Civic Engagement CARMINE MAIELLO, FRITZ OSER & HORST BIEDERMANN University of Fribourg, Switzerland ABSTRACT In this article the authors suggest that civic skills and civic knowledge are key components of the political information perception process and try to determine the differential effects of these variables on civic engagement. Starting from a model proposed by Torney-Purta, Lehmann, Oswald & Schulz, they developed an alternative model to explain the likelihood to vote in the International Association for the Evaluation of Educational Achievement Civic Education Study Standard Population of 14 year-olds tested in 1999. Results indicate that home environment and school-related factors predict civic knowledge and skills equally well. However, the direct effects of civic knowledge on likelihood to vote appear to be stronger than the corresponding effects of civic skills in 23 out of 28 countries. The evaluation of total effects highlights the meaning of an open classroom climate for discussion with reference to the likelihood to vote. Introduction Civic engagement is one of the most important instruments in democracy (Hauser, 2000). The destiny of countries lacking political participation is in the hands of a reduced number of people (oligarchy) who hold the power and deny it to anybody else. In the long term, democracy cannot survive without civic engagement. To deeply understand the concept of civic engagement, consider the case in which you decide to commission a constructor to build a house. Imagine that you wanted brown window frames and that the constructor is painting the window frames blue. Without any objection from your side you will get blue window frames although you wanted brown ones. The only way to get brown window frames is to remind the constructor of your instruction and 384

CIVIC KNOWLEDGE, SKILLS, AND ENGAGEMENT advise him to do as you said. Without your intervention the constructor like any government will do whatever he or she decides to do despite your wishes. Therefore, civic engagement can be considered as a control mechanism within democracy. It is indispensable and assures that a country stays democratic in the long term. Unfortunately, most of the European countries and the USA have experienced a general decline in community and civic engagement since the 1960s (Weiser, 2001). Among other variables, educational factors have been identified as key factors with reference to civic engagement and political participation (Hauser, 2000). Therefore, participation wishes and participation behaviour can be partly considered to be a result of knowledge about the importance of civic engagement. But, civic engagement leads to positive results only if it is driven by correct interpretation of political issues. Appropriateness of political judgement implies that people understand politics to a certain extent. This means that people must necessarily obtain basic civic knowledge and enough civic skills to correctly understand political information in order to work out suitable political judgements and, consequently, positively contribute to decisions on public issues. In this contest it is important to note that, ideally, information should be unbiased to guarantee accuracy of judgement. Manipulation of information and information sources represents a threat because it leads to assumptions that are based on incorrect information. Since deductions are only true if the primary premises are true, freedom and quality of information are necessary prerequisites for unbiased judgements and effective participation. In this regard, participation itself depends on the quality of information, which in turn depends on freedom of information media and on the individual ability to recognise true facts within the available information, which normally contains interpretative elements and may omit important details. It seems reasonable to see civic knowledge and skills as acting to filter interpretative and suggestive components of information in order to distinguish between facts and nonfacts. Figure 1 illustrates the theory outlined above. In this theoretical model information is considered as made up of facts and manipulative components. Some of these components are omission of important details, interpretative elements and masked or unmasked suggestions. Civic knowledge and skills help to filter the manipulative elements to identify the facts behind them (perception process). The filtered information leads to more or less unbiased deductions and judgements depending on the quality of the filtering process (political judgements). Afterwards the filtered information can be used to effectively participate in political discussions and processes (civic engagement). Finally, active participation leads in the long term to an effective control of the actions of political institutions (political decisions). Besides their role as information filters, civic knowledge and skills are thought to affect participation motivation, which is itself considered as a determinant factor with reference to political activity (Figure 1). 385

Carmine Maiello et al Facts Information Manipulative Components Knowledge Perception Process Skills Political Judgment Participation Motivation Civic Engagement Political Decisions Figure 1. Hypothesised role of civic knowledge and civic skills with reference to information processing and civic engagement. The crucial difference between civic knowledge and civic skills is linked to the correlation with cognitive abilities and intelligence. While civic knowledge does not depend on cognitive abilities and merely refers to knowledge about politics and political institutions, civic skills are positively correlated with both cognitive abilities and intelligence. It has been hypothesised that cognitive abilities may fill the role generally attributed to civic knowledge with reference to civic engagement (Hauser, 2000). However, the literature does not provide evidence of the supremacy of skills over knowledge. In fact, there is a lack of studies on the differential effects of these variables on civic engagement, and even primary analyses (Torney-Purta et al, 2001) of the International Association for the Evaluation of Educational Achievement (IEA) data have failed to evaluate the effects of these variables on civic engagement. Torney- Purta et al (2001) introduced a model to explain civic engagement in which no distinction was made between civic knowledge and civic skills. As shown in our replication of this model (Figure 2), another weakness is that no covariance between the predictor variables is assumed. Further, no attempt was made to explain the student s report of whether he or she has learned in school about the importance of voting. The objective of the present secondary analysis of the IEA data was to fill these gaps by developing an alternative and more sophisticated model to 386

CIVIC KNOWLEDGE, SKILLS, AND ENGAGEMENT explain civic engagement, distinguishing between civic knowledge and civic skills. Gender (male) Home Literacy Resources.18.27.04 e1 Civic Knowledge.15 Expected Years of Further Education.13.09 Open Classroom Climate Participation in School Council -.10.07 -.03.03.08.23 Evenings Spent Outside Home.10.06 Watching TV News Having Learned about Voting -.02.14.20 Standardized estimates chi-square=6964.962 df=29 p-value=.000 gfi=.946 agfi=.898 rmsea=.098 pclose=.000 Likelihood to Vote e2.16 Figure 2. Replication of the model proposed by Torney-Purta et al (2001) to explain civic engagement. Sample Students data from the IEA Civic Education Study were used to develop the model. The desired population included all students from 28 countries enrolled in schools on a full-time basis in that grade in which most students are aged 14. Participating countries were Australia, Belgium, Bulgaria, Colombia, Chile, Cyprus, the Czech Republic, Denmark, England, Estonia, Finland, 387

Carmine Maiello et al Germany, Greece, Hong Kong (SAR), Hungary, Italy, Latvia, Lithuania, Norway, Poland, Portugal, Romania, Russia, the Slovakian Republic, Slovenia, Sweden, Switzerland, and the USA. Testing took place in most countries between March and June 1999. More details about sampling are given in Torney-Purta et al (2001). The general model reported in this article refers to the entire population from 28 countries. Multi-sample analyses were used to integrate and confirm the findings. The Input Model Starting from the model described above we developed an alternative model with the same variables used by Torney-Purta et al (2001). The dependent variable in this model is the likelihood to vote in national elections, which is considered as an indicator of civic engagement and reflects the motivation of students to vote when they become adults. This variable refers to conventional civic engagement only. Since it is rank scaled, a logistic model would be more appropriate from a statistical point of view. However, using a logistic approach would make the comparison with the model of Torney-Purta et al (2001) more difficult. Civic knowledge, civic skills and the student s report of whether he or she has learned in school about the importance of voting are endogenous intervening variables. The error terms associated with these variables are assumed to be correlated to each other. With reference to measurement, civic knowledge and civic skills are Item Response Theory (IRT) scores derived from a one-parameter Rasch scaling of multiple-choice items. The student s report of whether he or she has learned in school about the importance of voting is a rank scaled multiple-choice item. Exogenous variables include individual characteristics (gender), home environment (home literacy resources), school-related factors (expected years of further education, open classroom climate, participation in school council), out-of-school activities (evenings spent outside home) and perception of political information (frequency of watching television news). Gender is a dummy variable where male students are coded 1, female students 0. It is the only individual characteristic included in the model because others would not have been comparable across countries. Home literacy resources gives the student s report of the number of books at home and is a 5-level rank scaled variable with 1 meaning less than 11 books, 2 11 to 50 books, 3 51 to 100 books, 4 101 to 200 books and 5 more than 200 books. This variable has proven to be an important predictor of educational achievement in previous studies (Beaton et al, 1996) and may be seen as an indicator of the emphasis placed on education, the resources available to acquire literacy, and the academic support a student finds in his or her family (Schulz, 2002). The students reports on how many years of further education they expect to complete after the current year are thought to reflect individuals 388

CIVIC KNOWLEDGE, SKILLS, AND ENGAGEMENT aspirations as well as parental and peer influences. A similar variable has proven to be a good predictor of knowledge in the 1971 IEA civic education study. The Rasch estimates of classroom climate indicate individual students perceptions of the atmosphere for expressing opinions and discussion in class, and involve students relations with peers in the school setting as well as with teachers. This variable was a strong predictor of knowledge, attitudes and participation in the 1971 IEA Civic Education Study. The reported participation in a school council is a dummy variable coded 1 for students who reported participation and 0 for students who did not participate. This variable is related to the school environment but may also reflect individual activity (Schulz, 2002). The students reports on the evenings spent outside home are said to be a predictor of anti-social behaviour (Currie et al, 2000) and negatively predicted civic knowledge in previous studies (Torney-Purta et al, 2001; Schulz, 2002). Students reports about how often they watch news on television were coded 1 for never, 2 for rarely, 3 for sometimes and 4 for often. Television was reported in most countries as the most important source of news among 14 year-old students. The input model is shown in Figure 3. Formally, it is a multiple indicators, multiple causes (mimic), recursive model with 11 observed variables and 4 error terms.[1] Senate weighted population data Gender (male) Civic Knowledge e1 1 Home Literacy Resources Expect. Further Education 1 e4 Likelihood to Vote Open Classroom Climate Participation in School Council Freq. Watching TV News 0 Having Learned about Voting 1 e3 Evenings Spent Outside Home Civic Skills 1 e2 Figure 3. Mimic model of civic education (input model). Data source: IEA Civic Education Study, Standard Population of 14 year-olds tested in 1999. 389

Carmine Maiello et al Results The model allows the estimation of a larger number of indirect effects than the model proposed by Torney-Purta et al (2001), takes the covariances between the predictor variables into account, fits the data better and explains 20% of the variance of the likelihood to vote (Figure 4). Senate weighted population data Gender (male) Home Literacy Resources Expect. Further Education Open Classroom Climate Participation in School Council Freq. Watching TV News Evenings Spent Outside Home.16.06.26.12.08.08.18 Civic Knowledge -.10.07.04.12.22.19.00 -.08 Civic Skills e1.15 -.03.02.08.10.05 -.11.03.24.04.08.00 e2.17.14 -.02.06.08.08 Having Learned about Voting -.08 Standardized estimates discrepancy/df=3.012 / 3 gfi=1.000 agfi=1.000 pclose=1.000 e4.20 Likelihood to Vote.19 -.05 e3.65 Figure 4: Mimic model of civic education (standardised solution); estimation ML; covariances between exogenous variables not shown. Data source: IEA Civic Education Study, Standard Population of 14 year-olds tested in 1999. Table I reports all standardised direct effects along with their statistical significance. The best predictor of likelihood to vote is having learned about the importance of voting, which is itself best predicted by an open classroom climate for discussion. Both civic skills and knowledge are best predicted by expected further education, home literacy resources and open classroom climate, but civic knowledge predicts likelihood to vote much better than civic skills. In fact, civic knowledge and the students reports of whether they have learned in school about the importance of voting seem to be major predictors of likelihood to vote. Meaningful, though less important, are the direct contributions of open classroom climate and frequency of watching television news. 390

CIVIC KNOWLEDGE, SKILLS, AND ENGAGEMENT Table I. Standardised direct effects. However, an inspection of the total effects on likelihood to vote obtained by summing all direct and indirect effects (Table II) gives a quite different picture of the predictor significances. The outcomes demonstrate that open classroom climate and frequency of watching television news have stronger total effects on likelihood to vote than civic knowledge. Table II. Standardised total effects on likelihood to vote. Subsequently performed multi-sample analyses support these findings. Results reveal that the model structure holds in each of the 28 countries (χ2/df = 1.711; GFI = 0.999; AGFI = 0.977) and that predictions of the likelihood to vote are better made with civic knowledge than with civic skills in 23 out of 28 countries. However, in the remaining five countries (England, Sweden, USA, Bulgaria and Latvia) the regression weights associated with civic skills are higher than those associated with civic knowledge. Multi-sample analyses demonstrate further that an open classroom climate for discussion has considerable total effects on the likelihood to vote within each participating country with the exception of Romania and the Czech Republic (Table III). 391

Carmine Maiello et al Australia 0.194 Belgium 0.214 Bulgaria 0.168 Chile 0.211 Colombia 0.206 Cyprus 0.133 Czech Republic 0.079 Denmark 0.192 England 0.171 Estonia 0.172 Finland 0.144 Germany 0.157 Greece 0.236 Hong Kong (SAR) 0.198 Hungary 0.144 Italy 0.163 Latvia 0.178 Lithuania 0.139 Norway 0.155 Poland 0.142 Portugal 0.151 Romania 0.080 Russia 0.218 Slovakian Republic 0.143 Slovenia 0.122 Sweden 0.223 Switzerland 0.097 USA 0.129 Table III. Multi-sample analyses: standardised total effects of open climate on likelihood to vote in 28 countries. Discussion The model proposed in this article has been used to provide an answer to relevant theoretical questions on the meaning of civic knowledge and civic skills for the prediction of likelihood to vote. Results indicate that civic knowledge seems to be a better predictor of likelihood to vote than civic skills in the majority of the countries involved in the analyses. From a theoretical point of view this outcome indicates that civic education may be an effective means to enhance likelihood to vote and that civic skills are not necessarily required to improve this motivation for engagement. However, there are five countries in which civic skills predict likelihood to vote better than civic knowledge, but it is difficult to explain this result. Further, likelihood to vote represents only a single and very conventional aspect of civic engagement. Therefore, it is quite possible that other aspects of engagement may require more skill than this kind of participation. 392

CIVIC KNOWLEDGE, SKILLS, AND ENGAGEMENT Results also indicate that civic knowledge and civic skills are best predicted by expected further education, home literacy resources, and an open climate for discussions. Interestingly, civic knowledge and civic skills are equally well predicted by these same indicators although their effects on the likelihood to vote are unequal. Reflections on this outcome may lead to a deeper understanding of the meaning of these variables. In detail, these findings suggest that expected further education, home literacy resources, and an open climate for discussions are in fact major determinants of civic knowledge and skills and have indirect as well as direct effects on civic engagement. However, we speculate that the full degree of these effects on civic engagement cannot be completely estimated in this study because voting is an easily understandable behaviour and represents only a limited facet of engagement. As a consequence, civic knowledge appears to be a more important predictor than civic skills, but the real reason for that is related to the evidence that such a simple behaviour as voting is most likely essentially dependent only on knowledge and does not require a skills component in most countries. We hypothesise, therefore, that future consideration of other important engagement variables could fully clarify which aspects of civic engagement are primarily affected by civic skills and which aspects such as voting simply require civic knowledge. This point is certainly of major theoretical relevance for education because this distinction is crucial in order to recognise differential patterns of educational influence on civic engagement. Another emerging variable in this context is an open classroom climate for discussion, which seems to play a major role as best indicator of the student s report of whether he or she has learned in school about the importance of voting, which is in turn the best predictor of likelihood to vote. Consideration of the total effects reveals further that open classroom climate for discussions affects civic engagement even more than civic knowledge. This outcome, which was additionally confirmed by results of multi-sample analyses in 26 out of 28 countries, implies that courses on civic education should be less teacher centred and should include group work and open discussions. Certainly, the findings reported in this article do not provide responsible leaders in school affairs and civic education with a ready-to-use recipe for direct application in educational practice. Nevertheless, it helps to understand which role is played by crucial educational variables with reference to civic engagement. For example, an open classroom climate for discussion influences likelihood to vote as well as civic knowledge. This is contrary to expectation since classroom discussion is itself practice for advanced civic skills. This finding may be an artefact of the age at which the students were sampled since 14 year-olds are not likely to have consistently achieved formal operational thinking. Civic knowledge represents a concrete operational measure, as does likelihood to vote. This suggests that an open classroom climate for discussion may benefit even younger children as preparation for civic skills later in adolescence. The observation that home literacy opportunities and watching 393

Carmine Maiello et al television news contribute directly and indirectly to civic knowledge and to likelihood to vote also has educational implications, particularly for home school collaboration in preparation for civic education. These suggestions, of course, would require further research both with older youth and perhaps longitudinally. Broad understanding of the role played by crucial educational variables is a goal that is largely emphasised and demanded by political institutions. It can be achieved through building and evaluation of contrasting alternative statistical models. The alternative model proposed in this article represents a useful improvement of the model proposed by Torney-Purta et al (2001) in the IEA report on civic education. It accounts for more sources of variance, fits the data better and contributes to answer important theoretical questions. As it is particularly evident in the case of an open classroom climate for discussions, our model demonstrates that taking indirect effects into account may lead to a more complete picture of the meaning of certain variables with reference to the explanation of civic engagement. The methodological conclusion is that concentration on direct effects leads to incomplete results and future work must include more theoretically relevant intervening variables to recognise additional and probably still unknown indirect effects of the considered predictors. However, the contribution of this article is fundamentally limited by the fact that likelihood to vote is only a single and restricted motivational aspect of civic engagement. Further, the explained variance of this dependent variable still remains too low. Future work must therefore look for more powerful indicators, taking into account intervening variables and interaction terms. Correspondence Carmine Maiello, Department of Education, University of Fribourg, Rue Faucigny 2, CH 1700, Fribourg, Switzerland (carmine.maiello@unifr.ch). Note [1] One parameter has been constrained to zero because it was not significant in the international sample as well as in each of the country samples. References Beaton, A.E., Mullis, I.V.S., Martin, M.O., Gonzalez, E.J., Kelly, D.L. & Smith, T.A. (1996) Mathematics Achievement in the Middle School Years: IEA s Third International Mathematics and Science Study (TIMSS). Chestnut Hill, MA: TIMSS International Study Center, Boston College. Currie, C., Hurrelmann, K., Setterobutte, W., Smith, R. & Todd, J. (2000) Health Behavior of School-aged Children: A WHO cross-national study (HBSC) international report. Copenhagen: World Health OrganiSation Regional Office for Europe. 394

CIVIC KNOWLEDGE, SKILLS, AND ENGAGEMENT Hauser, S-M. (2000) Education, Ability, and Civic Engagement in the Contemporary United States, Social Science Research, 29, pp. 556-582. Schulz, W. (2002) Explaining Differences in Civic Knowledge: multilevel regression analysis of student data from 27 countries, paper presented at the annual meeting of the American Educational Research Association, New Orleans. Torney-Purta, J., Lehmann, R., Oswald, H. & Schulz, W. (2001) Citizenship and Education in Twenty-eight Countries. Amsterdam: International Association for the Evaluation of Educational Achievement. Weiser, E.B. (2001) The Functions of Internet Use and their Social, Psychological, and Interpersonal Consequences, Dissertation Abstracts International, 61(7-B), 3906. 395