Predictors and correlates of youth political knowledge in Serbia 1

It is considered that political knowledge is one of the most prominent features of democratic citizenship, which makes its analysis especially relevant in a democratising society. This paper explores the predictors and correlates of youth political knowledge in Serbia. Political knowledge was measured by 12 multiple-choice questions pertaining to three areas: everyday politics, the rules of political game and foreign politics. Research participants were 788 fourth-grade students from 25 Belgrade secondary schools. The results show that the level of youth political knowledge is generally low and unevenly distributed among students. Its main predictors are interest in politics, academic achievement, gender and father’s interest in politics. Other predictors that are usually regarded as very important (e.g. socioeconomic status, civic education) proved to be non-significant. Although political knowledge is positively correlated with political tolerance, political activism and diffuse political support, the coefficients are very low. The concluding part discusses possible causes of the obtained results.

As in any decision-making process, information and knowledge are of crucial importance in making meaningful political decisions.The level of citizens' awareness about the rules of political game, current public affairs and government positions, their own rights and obligations -all have a profound impact on their ability to act and think in accordance with their deeper beliefs and best interests.In that sense, political knowledge is a critical component of citizenship and democracy in general.Citizens' political knowledge strengthens their ability "to link their policy views and choices to their evaluations of politicians and parties.Better-informed citizens can more easily identify with the parties and politicians whose policy positions are more consistent with their own views" (Delli Carpini & Keeter, 1996, p. 251).Without relevant information there is no solid foundation either for weighing the promises of political authorities or for fair judgment of their performance.Hence, "political knowledge is to democratic politics what money is to economics: it is the currency of citizenship" (Delli Carpini & Keeter, 1996, p. 8).When only the minority of citizens are able to effectively use their democratic rights due to the low political knowledge levels, democracy becomes not effective, but rather managerial (Delli Carpini & Keeter, 1996).
The more equitably political knowledge is distributed among citizens, the more likely it is that politics will reflect public interests, and that, consequently, the public will be more supportive of it.Political knowledge is, in other words, closely related to political power.Numerous studies have indeed shown that increased political knowledge is related to more active participation in various political activities (Delli Carpini & Keeter, 1996;Inglehart, 1979;Klingemann, 1979;Krampen, 2000) and increased political tolerance (Delli Carpini & Keeter, 1996), which figure as two fundamental features of a democratic political outlook.However, a manifold of previous studies conducted worldwide in the past few decades reached a uniform conclusion -the level of political knowledge of an ordinary citizen is (very) low (Delli Carpini & Keeter, 1991;1996;Gronlund & Milner, 2006;Kuklinski et al., 2000;Prior, 2002;Rhine, Bennett, & Flickinger, 2001;Vettehen, Hagemann, & Van Snippenburg, 2004).Poor, shameful, pathologically low, miserable -those are just some of the adjectives used to describe the level of political knowledge of ordinary citizens around the globe.Some authors argue that this is the main proof for the adequacy of the elitebased model of democracy (e.g.Neuman, 1986), which justifies the passivity and exclusion of the politically ignorant citizens from politics.Others claim that political ignorance does not prevent citizens from making meaningful political decisions using heuristics (Mondak, 1993;Lupia, 1994) or messages that come from political elites (Zaller, 1992).Page and Shapiro (1992) even argue that voting "errors" due to the lack of political knowledge tend to cancel each other out, so that these random variations can do little harm to the aggregate or overall coherence.
If political knowledge is a necessary precondition for effective democracy, analysis of its predictors should be a very useful tool for finding a way to promote it.Revealing the features of youth political knowledge is therefore especially relevant.First of all, youth are the population stratum whose support and patterns of political participation will be crucial for the future development of democracy in a society.On the other hand, numerous surveys have shown that young people are the least knowledgeable age group (Jennings, 1996;Delli Carpini & Keeter, 1991;1996;Rhine et al., 2001;Gronlund & Milner, 2006).Furthermore, the "profile" of a more politically knowledgeable young person is, in some sense, a mirror image of a more knowledgeable adult.Some of the main predictors of youth political knowledge levels also predict political knowledge in adult population.Gender (Jennings & Niemi, 1981;Niemi & Junn, 1998), socioeconomic status of the family (Niemi & Chapman, 1998;Torney-Purta et al., 2001), academic achievement (Jennings & Niemi, 1981;Niemi & Chapman, 1998;Niemi & Junn, 1998), attendance of civic education courses (Denver & Hands, 1990;Finkel & Ernst, 2005;Ichilov, 2007) or family "politisation" (Niemi & Chapman, 1998) are some of the main predictors of youth political knowledge.Males, students from families with higher socioeconomic status, better academic achievers, those who attend civic education or have more politically interested parents are more politically knowledgeable youth groups.
Effects of this kind of political socialisation in a specific social context such as Serbian are virtually unknown.Various analyses of youth political orientations in Serbia thus far have neglected the topic of political knowledge.There have been some attempts to measure political knowledge, but those studies used either students' own estimations of how much they were informed about politics (and only one specific group of students) (Baucal et al., 2009) or specific questions constructed for other purposes in public opinion polls (Pantić & Pavlović, 2009).Both have some methodological shortcomings.The analysis of political knowledge of the citizens of Serbia based on a knowledge test is not to be found in the last two or three decades.
Since political knowledge figures as one of the most prominent features of a democratic citizen (at least in theory), its analysis in Serbia has special relevance.In a society where democracy has not yet consolidated and even started to lose positive connotation among citizens (Pavlović, 2010), the development of political culture which would be compatible with the recently established democratic political system is of utmost importance.The first step towards making the political elite more responsive, the political process more transparent and democracy more effective is facilitating the development of a more attentive and critical public.An important part of this process is to increase political knowledge.Therefore, this paper is aimed at analysing some of the main characteristics of youth political knowledge in Serbia.What and how much young people in Serbia know about politics, which are the main predictors of youth political knowledge and what are the relationships between political knowledge and other relevant political orientations -these are just some of the questions left unanswered so far.Accordingly, the primary goals of this paper are: 1) to analyse youth political knowledge level, 2) to analyse the main sources of variation in political knowledge, and 3) to analyse the relationship between political knowledge and other relevant political orientations.

Method
Sample.Research participants were fourth-grade students from four different types of secondary schools at the territory of the city of Belgrade (N=788): grammar (n=202), medical (n=168), economic (n=211) and technical (n=207).Research was conducted in 25 randomly chosen secondary schools.The population of Belgrade secondary schools was first stratified by type of school (grammar, medical, economic and technical) and type of settlement (city centre and suburbs).Within each combination of stratification criteria, a separate list of schools was formed.Sample size was arbitrarily set to approximately 30 classes in total, i.e. 800 students (200 students per four types of schools), with the ratio of city centre versus suburban students of 3:1.The number of schools necessary to accomplish the planned sample size was chosen from the list (using randomly generated numbers).In every chosen school, the first available class of students was taken.In the majority of the included schools, one class of students filled out the questionnaire; in five cases (some of the secondary schools in the suburbs) two classes of students in one school were chosen.There were 58% female and 42% male students in the sample.Students were 17 to 20 years old (M=18.15).One quarter of students in the sample (25%) attended secondary schools in the Belgrade suburbs; the remainder (75%) attended secondary schools in the central parts of Belgrade.
Procedure.Data were collected during school classes, from February to April 2010, as a part of a more extensive study.It took approximately 35 minutes for a student to fill out the whole questionnaire.The study was conducted under the official approval of the Ministry of Education of the Republic of Serbia, as well as school management in each secondary school.Before the questionnaire was administered, the students were presented with the basic pieces of information regarding the research and were asked for their consent for participation.

Variables and instruments
Questionnaire was used for data collection.Numerous relevant variables were included: Socio-demographic characteristics.This group of variables included gender, parents' education level (primary and less, secondary, university and higher), type of school (general or vocational), academic achievement (average grade level at the end of the third grade) and elective course (civic or religious education).Students evaluated their family financial situation choosing the most appropriate answer from the following list: (1) we do not have enough money, not even for food (2) we have enough money for food, but we can hardly buy any new clothes (3) we have enough money for food and clothes, but not for bigger purchases (like TV or refrigerator), (4) we have enough money even for bigger purchases (5) we can buy almost everything we want.

Interest in politics.
Students' interest in politics was measured by one question: "Generally speaking, how much are you interested in politics?",followed by a four-point scale (1.not at all, 2. somewhat, 3. a lot, 4. very much) (M=1.92,SE=.03, SD=.89).
Parents ' and peers' interest in politics.Students were also asked to estimate their mother's, father's and best friend's interest in politics, using a similar question and a fourpoint scale SE=.03,SD=.89,mothers' political interest M=2.11,SE=.03,SD=.81,best friends' political interest M=1.72,SE=.03,SD=.90).
Political discussion.Answers to three questions pertaining to frequency of political discussions with father, mother and best friend expressed on a four-point scale (1.never, 2. rarely, 3. sometimes, 4. often) were summed and treated as a measure of the frequency of students' political discussion (α = .72;M=5.46,SE=.08,SD=2.11).
Media exposure.Students were asked to estimate how often they followed politics in four different types of media (television, radio, newspapers and Internet) on a five-point scale (1.never, 2. rarely 3. 1-2 times a week, 4. 3-5 times a week, 5. daily).Students' scores were summed and treated as a measure of media exposure (α = .71;M=12.56,SE=.14, SD=4.16).
Political tolerance.Political tolerance was measured by one form of the least-liked group method (Sullivan, Piereson, & Marcus, 1979).Based on the findings from the pilot survey which showed that homosexuals were one of the least-liked groups among youth, students were asked whether a homosexual person should be allowed: (1) to give a speech on homosexuality, (2) to work as a teacher at school/university and (3) to run for the President of Serbia.The number of approved rights, ranging from 0 (does not approve of any right) to 3 (approves of every right), was treated as a measure of political tolerance (M=.97,SE=.04, SD=.38).
Political activism.Based on previous differentiations of forms of political behaviour (Clagget & Pollock, 2006;Conway, 1990;Sigel & Hoskin, 1981;Verba, Nie, & Kim, 1979) students were asked whether they participated in any of the following 12 types of political behaviour: (1) persuading others to vote, (2) writing a protest letter to the media editor, (3) contacting local or national officials, (4) signing a petition, (5) engaging in a political campaign of a party or a candidate, (6) writing graffiti with political content, (7) wearing a bumper sticker or a T-shirt with political message, (8) joining the political groups on the Internet, (9) visiting websites of a party or a politician, (10) joining political meeting, (11) peaceful or (12) violent demonstrations.The number of types of political behaviour in which the student participated was treated as a measure of political activism (M=2.14, SE=.07, SD=.84).
Diffuse political support.Diffuse political support is an expression of general political (dis)satisfaction with the current political authorities.It is based on the belief that political authorities are legitimate, their actions good and just, and their decisions respectful ergo obligatory.It was measured by Muller et al. (1982) scale which was shortened to four items and adapted (α = .69;M=9.17, SE=.10,SD=2.85;range=13,min=4,max=17).Political efficacy.Political efficacy is " the feeling that individual political action does have, or can have, an impact upon the political process " (Campbell, Gurin, & Miller, 1954, p. 187).It was measured by the original four-item scale (Campbell et al., 1954) (α = .70;M=8.00, SE=.11,SD=3.09;range=15,min=4,max=19(Delli Carpini & Keeter, 1996, p. 65), numerous researchers distinguish between two main types of political knowledge (Delli Carpini & Keeter, 1991;1993;1996;Gronlund, 2007;Gronlund & Milner, 2006;Jennings, 1996).The first is knowledge about everyday politics -"factual" or "surveillance" knowledge; awareness of the current political situation (parties, people, issues etc.).The second is knowledge about the rules of the game -"taught" or "citizenship" knowledge, awareness of general and relatively stable rules and procedures in the functioning of political system.Some other themes are also considered as relevant: knowledge of foreign affairs (Delli Carpini & Keeter, 1996;Maghami, 1974), political geography (Delli Carpini & Keeter, 1996;Prior, 2005) or political history (Delli Carpini & Keeter, 1996;Jennings, 1996).Based on these empirical analyses as well as specific recommendations about what citizens should know about politics and how political knowledge should be tested (Delli Carpini & Keeter, 1993;1996), political knowledge of students was measured by twelve multiple-choice questions, covering three areas of political knowledge: everyday politics, the rules of the game and foreign politics.Questions were first tested in a short pilot survey2 to check their difficulty and item-total correlations which served as criteria for item selection.Item-total correlations of the final test are shown in Appendix 1.
Four answers plus "does not know" (DK) option were offered for every question.There are some methodological issues about the inclusion of DK option and the widespread practice of encouraging respondents to choose it (by formal instructions, telling interviewers not to insist on obtaining answers etc.).Some authors argue that when DKs are encouraged, test scores reflect two systematic factors: knowledge and propensity to guess (Mondak, 2001).In line with this, there is some evidence that, when DKs are not encouraged, there is a 15% increase in political knowledge scores (Mondak & Davis, 2001) or that nearly 50% of the well-known gender differences in political knowledge can be explained by differences in response set -males are more prompt to guess (Mondak & Anderson, 2004).Finally, this implies that incorrect and DK answers should not be treated as a single category (Mondak, 1999).However, others suggest that we should be careful about accepting Mondak's opinion that personality related propensity to guess affects knowledge scores and invalidates group differences.When subsequently asked to give the best guess, people who initially chose a DK alternative fare statistically no better than chance (Sturgis, Allum, & Smith, 2007).Bearing all this in mind, DK alternative was offered but not encouraged.The following instruction was given to students: "Here are several questions related to some political issues.Answer them by choosing one of the offered alternatives.Read them carefully because only one of them is correct." Structural equation modelling showed that the three-factor solution (RMSEA = .05,χ² [51] = 170.80,p<.01 AGFI = .95)was a better fit than one-factor solution (RMSEA = .06,χ² [54] = 230.23,p<.01, AGFI = .93).However, high interfactor correlations among three factors (ranging from .70 to .94)supported the idea that political knowledge in this research can be viewed as a unidimensional phenomenon, which is in line with some previous analyses (Delli Carpini & Keeter, 1993;1996).The number of correct responses was used as a measure of the level of political knowledge.

The level of students' political knowledge
Generally speaking, the level of youth political knowledge is low.Although the questions tap some of the basic and most fundamental pieces of political information, the majority of students were able to provide a correct answer to only three out of twelve questions.The data are presented in Table 1.Note: Everyday politics questions refer to the Serbian political situation in the period when the research was conducted (February-April 2010).
On the other hand, the minority of students had knowledge of some of the most basic features of the Serbian political system.The number of seats in the Serbian Parliament or the percentage of election votes that makes election threshold, for instance, is familiar to less than one half of students.Fifty-eight students (7%) did not answer any question correctly, while only five (less than 1%) answered every question in the right way.The mean of correct responses was 4.34 (SD=2.66,SE=.09).
When students were not able to provide a correct answer, they either gave the wrong one or stated that they did not know.These two indicators of the absence of knowledge call for some further reconsideration, in line with the above-mentioned issues.Mondak (1999) argues that before aggregating these two types of answers, some attempt to assess the basic pattern of results is necessary.Simply put, upon disaggregating DK and incorrect answers we should find that some important predictors of political knowledge produce comparable effects on both indicators.This would mean that political knowledge is discrete and that these two types of answers can be treated as a single category.More detailed elaboration of this topic is beyond the scope of this paper, but, for the sake of illustration, proportions of correct, incorrect and DK answers were analysed by gender and parents' level of education, i.e. some of the most prominent predictors of political knowledge (figures are shown in Appendix 2).The results show that males, compared to females, gave fewer incorrect (t [786]=-2.09,p<.05) as well as DK answers (t [786]=-2.36,p<.05).Although there are some similar trends, with the increase of mother's education level, the proportions of incorrect (F [2, 785]=.26,p=.76) or DK answers (F [2, 785]=.67,p=.58) do not change significantly.Similar holds for father's level of education -there are no significant differences in mean proportions of incorrect (F [2, 785]=2.01,p=.13) or DK answers (F [2, 785]=1.98,p=.13).In other words, some of the most important predictors of political knowledge have similar or comparable effects on both indicators of the absence of knowledge, which corresponds to what Mondak (1999) describes as an acceptable pattern of results.Such limited evidence does not rule out possible differences in response sets but at least provides some justification for treating incorrect and DK answers as one category for the present purposes.

The predictors of students' political knowledge
Multivariate regression model which includes several predictors is significant (F [13, 628]) = 15.18,p<.01) and explains 22% of variance in political knowledge scores (Table 2)3 .The strongest predictor of students' level of political knowledge is the level of their political interest.Those more interested in political affairs are at the same time more knowledgeable of politics and vice versa.Other relevant predictors of the level of students' political knowledge are gender, academic achievement and level of father's political interest.Male students are more knowledgeable of politics than female.The same applies to students with higher academic achievement, as well as to those whose father was, in their own estimation, more interested in politics.None of the other variables included in the model proved to be significant predictors of the level of students' political knowledge.

The correlates of students' political knowledge
Data in Table 3 illustrate the relationship between the level of students' political knowledge and four other types of political orientations.Three out of four correlation coefficients are significant and positive, but generally (very) low.The positive correlation coefficient between political knowledge and political tolerance indicates that the higher student's knowledge about politics, the more tolerant he/she is towards civil rights of a marginalised social group.In addition, higher level of political knowledge indicates increased support for some of the basic elements of the Serbian political system, i.e. diffuse political support is more prominent.Finally, political knowledge is positively correlated with political activism, which suggests that students who are more knowledgeable of politics are also more politically active and vice versa.Non-significant correlation between political knowledge and sense of political efficacy implies, surprisingly, that the belief in one's own possibility of political influence has nothing to do with what or how much one knows about politics.

DISCUSSION
Several conclusions can be drawn based on these results.First of all, the level of political knowledge of secondary school students in Serbia is low.The majority of students were able to give correct answers to only three out of twelve questions.Bearing in mind an abundance of previous studies (Delli Carpini & Keeter, 1991;1996;Gronlund & Milner, 2006;Prior, 2002;Vettehen et al., 2004), this finding is somewhat expected.A young person in late adolescence or early adulthood is in a very specific life position.Struggling with his/her own personal "crisis", choosing the future profession, finding a mate or starting the family -in sum, all the so-called "start up" problems (Jankowski & Strate, 1995;Nie, Verba, & Kim, 1974) keep a young citizen away from politics.Still, it seems that the majority of Serbian new voters lack some of the basic political facts, which may not prevent them from voting, but still calls into question their ability to act rationally and in their best interest.
Political knowledge/ignorance is not evenly distributed among Serbian secondary school students.Previous studies show that some of the most important predictors of youth political knowledge are gender (Jennings & Niemi, 1981;Niemi & Junn, 1998;Niemi & Chapman, 1998), academic achievement (Jennings & Niemi, 1981;Niemi & Junn, 1998) or the family "politisation" (Niemi & Chapman, 1998).Results of this research support such claims, as well as the general finding that interest in politics is one of the most important predictors of political knowledge (Delli Carpini & Keeter, 1996).
Acquisition of political information is a demanding activity, which depends not only on personal ability and motivation, but also on certain social factors beyond and over individual control.Students who know more about politics are not only those who are more interested in politics or have higher academic achievement, but also those who come from families where politics is more present and prominent.The level of father's political interest is a significant predictor of student's political knowledge; having a more politically interested father implies that there are more political cues at home and more opportunities to learn politics.
Finally, it seems that Serbian politics, as in many other countries worldwide, is male business.Male students are more politically knowledgeable than female.This fact is usually explained by gender specific patterns of socialisation, different opportunities for males engage in political world, political limits placed on females in contemporary politics etc. Roughly one-quarter to one-half of the relations between gender and political knowledge can be explained by structural, situational and attitudinal variables (Delli Carpini & Keeter, 1996).Possible explanations for gender differences in political knowledge among Serbian youth can be found in the reality of political life in Serbia which is dominated by males, the traditional views that females have no place in politics, as well as in more profound interest of females in personally related topics at the expense of politics (Delli Carpini & Keeter, 1996).The influence of gender differences in response sets (Mondak & Anderson, 2004) cannot be ruled out either.It is important to note that gender cleavage in Serbia is established as early as in adolescence.
On the other hand, some other variables did not prove to be relevant, which is inconsistent with previous research.Civic education, by definition, should be one of the most important instruments of developing citizenship virtuespolitical knowledge among other things -especially in "new" democracies.It should, at least to some extent, alleviate the pre-existing differences between students and serve as a mechanism of decreasing the prevalence of specific social groups among the more knowledgeable citizens.Unlike numerous previous studies (Denver & Hands, 1990;Finkel & Ernst, 2005;Ichilov, 2007;Niemi & Junn, 1998), this research has shown that civic education attendance is not a significant predictor of the political knowledge level.This may stem from the fact that such topics are absent from civic education curricula, while there are also some general problems with these courses in Serbia: their sudden inclusion in educational system, no promotion and system support, no textbooks or evaluation standards, questionable teachers competencies etc. (Baucal et al., 2009).
Indicators of student's socio-economic status (e.g.parents' level of education) are not significant predictors of political knowledge levels, which is contrary to results of previous studies.It is possible that the role of SES is limited to increasing (political) motivation by socialising students to the political world and stimulating their interest in it, as well as facilitating the development of cognitive abilities necessary for effective learning in general.These are some of the most important reasons why formal education is significant for political knowledge (Delli Carpini & Keeter, 1996).There is evidence that after controlling for intelligence formal schooling makes no significant contribution to the level of political sophistication (Luskin, 1990) or becomes a much weaker predictor of political knowledge (Delli Carpini & Keeter, 1996).It is possible to interpret the data presented in this paper in this way -after controlling for political interest and academic achievement, SES no longer makes a significant contribution.This may be less a demonstration that SES is not important than a specific way of how it does matter -by promoting development of some other relevant dispositions that are more directly related to political knowledge.
Finally, it can be argued that political knowledge is indeed a currency of citizenship in Serbia, although its value is very low.The presented data confirm previous findings about the positive relationship between political knowledge and political tolerance (Delli Karpini & Keeter, 1996), political activism (Delli Carpini & Keeter, 1996;Inglehart, 1979;Klingemann, 1979) or political support (Gronlund & Milner, 2006).Although direct comparisons are not possible, while previous research shows that political knowledge is, for example, the most important predictor of political tolerance (Delli Karpini & Keeter, 1996) or political activism (Krampen, 2000), the relationships between these and political knowledge in this study are very weak.Further research is necessary to draw more definite conclusions, but for now it seems that in the case of Serbian youth the relationship of political knowledge with other types of political orientations is far from those registered in other socio-political contexts.

CONCLUSIONS
Results of this study show that the ability to acquire political knowledge depends not only on personal motivation and capacities, but also on the characteristics of informational environment and opportunity to learn politics.The family plays a certain part by providing political cues and helping to establish initial interest in politics as a wide and strong motivational basis for learning politics and probably some other intellectual tools important in the process of political learning.The school also plays its part.Although civic education is not a significant predictor of political knowledge, relevance of academic achievement may indicate the importance of some other type of contextual knowledge acquired at school (history, geography, sociology etc.) that makes learning about politics easier.Youth political knowledge is, in that sense, unevenly distributed among the new voters in Serbia, and generally low.It is only weakly correlated with several relevant political orientations often mentioned in literature.
Previous findings show the stability of the level of individual's political knowledge across time (Delli Carpini & Keeter, 1991;1996;Jennings, 1996).Therefore, the prevailing political ignorance among youth calls into question not only the quality of students' political decisions or of precursors of political participation in general, but also the quality of functioning of the recently established democratic political system which is firmly rooted in the idea of an informed and knowledgeable citizen.
Since there are limitations to the present study, further research on this and related topics would be fairly useful.Exploring other areas of political knowledge that this analysis did not cover (e.g.political history), youth understanding of basic political concepts and ideas (i.e.political sophistication) and its relation with political knowledge, the relationship between political knowledge and political beliefs and values, the relevance of some other variables as possible predictors of political knowledge (e.g.party identification) -to name just a few, would help clarify this phenomenon in more detail and draw more precise conclusions.

Figure 1 .Figure 2 .
Figure 1.Proportions of correct, incorrect and DK answers by gender ).Based on the widely accepted assumption that what citizens should know about politics is, simply put, "what government is and what it does"

Table 1 .
Percentage of correct, incorrect and DK answers to twelve political knowledge questions

Table 2 .
Relative effects of predictors on students' political knowledge

Table 3 .
Correlation between political knowledge and political efficacy, political tolerance, diffuse political support and political activism