POLITICS IN SERBIA 1990-2002 : A CLEAVAGE OF WORLD VIEWS

The paper analyzes socio-psychological sources of political divisions in post-communist Serbia. Following the argument that authoritarianism is intrinsically associated with the opposition to pro-democratic political change, it is hypothesized that authoritarianism is associated with the support for the former communists, and increasingly over time for radical nationalists. The data analysis utilizes three data sets, from 1990, 1996 and 2002, that is from periods that represent three crucial stages in the development of the Serbian post-communist politics. Discriminant analysis of party preferences showed that preferences for authoritarian political options and ideological orientations were tied to authoritarianism as an individual difference variable and specific socio-structural characteristics. The paper offers an interpretation of the Serbian politics throughout 1990s in terms of a cleavage of world views.

Political alignments, according to the dominant perspective in social sciences, tend to reflect stable social divisions or cleavages (Lipset and Rokkan, 1967, Bartolini and Mair, 1990, Antonić, 1998).The often presumed intricate link between the so conceived cleavage politics and political stability (Whitefield, 2002) suggests that in a country that experienced a great degree of instability such as Serbia, the politics of social cleavages can play only a marginal role.Yet as will be argued in this paper, while the Serbian society has certainly not enjoyed stability, the political system did exhibit certain stable features.The facts that ten years were needed for the first crucial political change to take place, and that very few political parties have been electorally relevant, are distinctive signs of this political 'stability'.Such political intransigence, it will be shown, was a manifestation of a deep political divide which had its roots in social structure and associated values and world-views, a divide that goes beyond the narrow sociological definition of a social cleavage.
At the end of the XX century, the emerging post-Communist political parties possessed alternative methods for mobilizing potential voters than parties that provided ground for the development of the cleavage-politics model.Mass media in particular, provide an efficient access to a much wider audience, thus making most of the organizational work of the earlier times superfluous.Today, party elites might find it much more efficient, especially in the shorter run, to appeal to voters' sociopsychological characteristics, and not necessarily or directly to the interests associated with their socio-structural location.
In this way, distinct world-views accompanied by negative stereotypes about political enemies, can become the central dividing lines of the national politics.The depth of such divisions can be large enough to prevent cooperation across the campborders, as happens in cases where socio-economic groups are pitted against each other.The general argument of this paper is that political attitudes, values, or ideologies, with only marginal or indirect links with social structure, but heavily rooted in distinct world-views, can provide ground for exceptionally stable political orientations.
The specific contribution this paper seeks to make is through the elaboration of a specific individual-dispositional trait that connects social structure with ideological orientations and further with political preferences, namely authoritarianism.This concept, conceived as a personality or attitudinal dimension has frequently been used for explaining anti-democratic political preferences (Adorno et al., 1950, Altemeyer, 1988).The perspective adopted presently, however, is in line with Gabennesch's (1972) conception of "authoritarianism as world view" that emphasizes socialization experiences rather than intra-family processes (cf.Bojanović 2004).Authoritarianism on the individual level is conceived as a world view rooted in reified perception of the social reality.Particular positions in social structure, such as low education or cultural deprivation, constrain development of broader social perspectives.If an individual's "social perspectives are narrow, the social world will tend to appear to him as a reified entity" (Gabennesch, 1972, p. 869, emphasis in original).Those coming from authoritarian micro-world tend to see the entire social world as 123 organized in authoritarian fashion, and the hypothesis proposed here is that they tend to support authoritarian political options.As Enyedi aptly argues, "authoritarianism is important because it unites political goals with psychological dispositions, providing political orientation with nonpolitical anchors" (2005, p. 6).
The present research is specifically intended to explain structuration of political oppositions in Serbia, but could be applicable to other cases in that have experienced difficulties in establishing a consensus about transition towards the democratic regime (e.g., Krause 2004).There are, however, multiple reasons to expect not only exceptionally strong political role of authoritarianism in Serbia in general (e.g., Mihailović, 1991, Kuzmanović, 1995), but to expect very specific associations.First of all, authoritarianism should be strongly related with the support for the former socialist regime.In the early 1990s former communist authorities represented the established authorities -to whom the authoritarians are submissive by definition.
With the advent of multipartism, these authorities appealed to social strata commonly associated with higher levels of authoritarianism -those characterized by older age, lower education, rural residence.However, the loyalty that these strata showed towards the Socialist Party was not based only on the contemporaneous party propaganda, but also on the fact that the post-World-War-II period under the communist rule represented for them the time of unprecedented growth in material standard of living, and, for that matter, in the degree of freedom and respect, if not 'democracy', they enjoyed comparing to the pre-communist period.Thus, the disintegration of the socialist system meant not only a change of government or of the ruling party but the collapse of the entire world-view cultivated for nearly half a century.In addition, we should expect clear evidence of the well-established link between authoritarianism and nationalist attitudes, given the degree of nationalist mobilization in Serbia.
Break-up of a political system, drastic economic decline, disputes over the borders of the national community, more or less open armed conflicts with neighboring republics and ethnic groups, and finally a war with the most powerful military alliance the world has ever witnessed, are certainly events that can be associated with the feeling of threat.Under such condition, we should expect especially strong political role of authoritarianism (Feldman and Stenner, 1997).
The theoretical relevance of the outlined research problem has relevance broader than the research tradition on psychological authoritarianism.The argument has a number of parallels with the theory of post-materialist value change (Inglehart, 1990, Dalton 1988).It is related to the research on the structure of political attitudes (Eysenck, 1975, Middendorp, 1978, Evans, Heath and Lalljee 1996).Obviously, the problem is also linked to the research on socio-political cleavages (Lipset andRokkan, 1967, Bartolini andMair, 1990), and to a number of specific models intended to explain structure of political opposition in the post-communist context (e.g., Kitschelt, 1992, 1995, Mair 1997).

REVIEW OF THE POLITICAL ELITE IN SERBIA
The empirical analysis is preceded by a brief qualitative review of the political processes on the level of political elites in Serbia in order to provide a background for interpretation of the empirical analysis.It is possible to differentiate three stages in the development of the party system in Serbia.Political pluralism started with the opposition between the former communists and anticommunist nationalists, just as in a number of other countries in the region.The Serbian peculiarity was that the anticommunist side was far too weak to seriously challenge the semi-reformed communists, partly because communists themselves earned a nationalist reputation.Socialist Party of Serbia (SPS), a direct heir of the former ruling League of Communists of Serbia, won the first pluralist elections in December 1990 easily, obtaining 194 (77.6%) out of 250 seats with just 46.1 % of the votes cast (thanks to the majoritarian two-round electoral system).The rest of the 56 seats were divided among 13 parties and coalitions and 8 independent candidates.With 15.8% of the votes, the strongest opposition party was the Serbian Renewal Movement (SPO), distinguished by the passionately nationalist rhetoric of its leader, a well-known writer Vuk Drašković.
Liberal-democratic perspective was the third major political option, represented by the Democratic Party (DS) who obtained about 7% of the votes.The Serbian Presidential elections, held at the same time, clearly show the balance of power between the Socialists and the opposition: Slobodan Milošević won approximately four times more votes than the main challenger, Vuk Drašković.
In 1990, supporters of the nationalist opposition were more nationalist than those who voted for Milošević.However, it was difficult to beat Milošević on the nationalist ground, since he was already perceived as a statesman defending the Serbian state and the national interests (Goati, 1998).2Consistently non-nationalist were only voters supporting Ivan Đurić, then a candidate supported by several democratically oriented parties, but they still lacked wider popular support.
As soon as the interethnic conflicts broke out, the main opposition party, the SPO, quieted down its nationalist rhetoric.In several attempts to unite forces with the rest of the democratic forces, the opposition to the socialists was recast in terms of the conflict between authoritarianism and democracy, if not explicitly in terms of the struggle between the 'good and evil'.This marks the second phase, which lasted throughout most of the 1990s (Antonić, 1998(Antonić, , 2000)), characterized by increasing polarization and confrontation.The authoritarian side has been inhabited by the so-cialists and radical nationalists, while the self-proclaimed 'democratic opposition' consisted of ideologically rather heterogeneous members united by the desire to remove Milošević from power, if not always by particularly deep commitment to democracy.
The election of September 2000 and the so-called "bulldozer revolution" (Birch, 2002) of October 5, 2000, mark the end of the rule of the SPS and Milošević and beginning of the third phase in the Serbian political development.Coalition of democratic-oriented parties (DOS3 ), headed by Vojislav Koštunica of DSS (Democratic Party of Serbia), won a persuasive victory (53% of the seats) over SPS and JUL (Yugoslav United Left) who ended up with about 41% of the seats.For the Radicals this was their worst result ever (approximately 8% of the votes cast).Public support for the Socialists seems to be lastingly diminished: the socialist candidate in the first round of the 2002 presidential elections won barely above 100.000votes.Serbian Radical Party (SRS) seemed to be weakened initially too, but not nearly so much as the SPS.Their recovery was fast: Vojislav Šešelj was able to win about one million votes in the unsuccessful December 2002 presidential elections, attracting many of the former SPS voters. 4In addition, the extreme nationalist bloc obtained another party in the parliament, namely Party of Serbian Unity (SSJ). 5PO and Vuk Drašković were among the losers of this transformation.Drašković fared rather poorly as a presidential candidate in October 2002, and the SPO remained without a single representative in the Serbian parliament.Although it seemed that once the main leader of the democratic opposition was on the way out of the political stage in Serbia, his party successfully returned to the parliament by the end of 2003.
This period has also witnessed an increasingly important division within the democratic camp between the more traditionalist and conservative perspective represented by the DSS, against the economically liberal and culturally libertarian DS.December 2003 parliamentary elections did not bring clear victory to any of the three political camps.Although SRS appeared as the strongest single party, none of the parties from the 'democratic bloc' seem to be willing to enter such coalition and socialists have too few representatives.On the other side, DSS rejects forming a government with DS, but the support it receives from SPO and newly formed G17 is insufficient to form a majority government.It remains to be seen whether this postelectoral bargaining will result in a new political constellation, or in new elections.
In the empirical analysis I focus on three key periods of the pluralist politics in Serbia.I examine data from 1990 when the first pluralist elections were held in Yugoslavia, and which show the relative strength and profiles of the main competing forces.Authoritarianism versus democracy divide is specially relevant for the second examined period, i.e., year 1996.This was the year of heightened fight between the opposed political blocs, when the democratic opposition was gaining strength, as evidenced by its victory on the local level elections in that year.Finally, relatively contemporary situation is assessed on the basis of survey conducted in 2002, when a new division was opening on the Serbian political scene after the first regime change, but now within the democratic bloc.

HYPOTHESES
The principal hypothesis starts from the assumption that reified world-view implies resistance to change, preference for order, obedience to rules and established authorities, and more general psychological conservatism, and at the same time, distrust of foreign influences, and negative attitude towards outgroups.It follows, then, that that authoritarianism should particularly facilitate support for two ideological dimensions: socialism of the old regime and nationalism.The former reflects the preference for the established order, for the known and familiar (authoritarian conventionalism).The later draws its appeal from the need for rejected outgroups (authoritarian aggression), and to follow the lead of the authorities (authoritarian submission) (e.g., Altemeyer 1981Altemeyer , 1988)).On the other side, variables facilitating development of the reified view of the social world, such as older age, low education, rural residence, should be accompanied by higher level of authoritarianism (Gabennesch 1972).For this perspective, it is irrelevant whether these variables imply social closure as required by the classical notion of social cleavages.
Thus, if the hypothesis about the mediating role of the authoritarian world view is plausible, the most relevant discriminating variables should include structural variables predictive of authoritarianism, such as lower education and age, then authoritarianism itself, and ideological orientations expressing the authoritarian worldview, that is preference for the old socialist regime and ideas, and nationalism.Finally, such dimensions should discriminate parties of democratic orientation from those endorsing socialist and extreme nationalist ideology.
Given the element of conventionalism in the definition of authoritarianism, nationalism in the beginning should be less authoritarian than later, since it was ideology that challenged the established authorities, and especially because of radically different political style pursued by the then nationalist leaders and parties (secondary hypothesis).
Since the analysis aims to establish ideological and other differences between supporters of different parties, not what is best able to predict support for particular parties, the method of choice is discriminant analysis.The focus is on dimension of competition, rather than on ideological differentiation.To sum up, it is expected that the main discriminant functions will join together in the hypothesized manner demographic, dispositional, and ideological dimensions.Purely structural, or ideological, or dispositional discriminant functions would run contrary to this hypothesis, as well as functions that combine variables in an unexpected way (e.g., higher education with authoritarianism).

METHOD
This section presents empirical analyses of the connection between social structure, ideology and party preferences.Data for years 1990 and 1996 come from Serbian and Yugoslav election studies, provided by ZENTRALARCHIV FUER EMPIRISCHE SOZIALFORSCHUNG, KOELN, in further text referred to as ZA Studies (ZA Study 2903(3rd wave), 1990;andZA Study 2911, 1996). 6These studies are based on national random quota samples, covering the entire territory of Serbia, excluding Kosovo.The sampling method is generally multi-stage stratified sampling, designed to be representative according to major demographic variables (e.g., age, sex, residence, education).The field research is conducted as face-to-face interviews in respondents' homes, and organized by established local polling agencies.
The second main source of data is an independent survey of a random sample of Belgrade residents (N=502), conducted in February 2002.Polling was conducted by the Center for the Advancement of Empirical Research, Belgrade, Serbia, (Udruženje za unapređenje empirijskih istraživanja) under the guidance of Professor Srećko Mihailović.Sampling was done according to the random route method, where the random selection of streets was done in 25 randomly chosen localities in 13 Belgrade municipalities (with three stages in street selection: municipality, local community, and street).Face-to-face interviews with adult respondents (18 years and over) were conducted in respondents' homes and completion of each interview lasted approximately 90 minutes.The entire questionnaire contained 210 questions.
The aim of the analysis is to empirically establish the most important dimensions that divided the relevant Serbian parties.The analysis is mainly focused on the five most influential parties, with strongest electoral base and continuity over time: SPS, SRS, DS, DSS and SPO.Discriminant Analyses are performed on the data from each three examined time periods.Variables entered into analysis include (1) the key socio-structural variables, such as age, rural or urban residence, level of education, church attendance, (2) authoritarianism scales to capture the authoritarian world view, (3) left-right self-identification scale as an indicator of broader ideological or value orientation, and (4) scales measuring nationalism and socialist or pro-Communist orientation, to measure the two most salient specific ideological orientations.The dependent variable is the expressed vote choice for presidential candidates in 1990, and for parties in the other two years.
Instead of analyzing single items, the approach adopted here is to construct relatively reliable multi-item measures of the key constructs: authoritarianism, nationalism, and socialist orientation.Data sets from 1990, 1996 and 2002 contain sufficient number of items for accomplishing this goal.Although the employed scales differ in each year, I believe there is enough commonality to treat them as comparable operationalization of the same constructs. 7Items composing these scales are given in Appendix.
One way of establishing functional equivalency of the applied operationalizations of the same constructs in different surveys, is to examine their relationships with relevant 'other' variables.The problem in applying this strategy when data come from different time points is that possible differences might be caused both by differences in construct operationalizations and changes in the relationships with the analyzed variables.It helps if there are theoretical grounds for expecting different degrees of stability over time for different variables and their relationships.Age and education are two stable socio-demographic characteristics which often have predictable attitudinal correlates.They are well-known correlates of authoritariannism, and the relationship between them is rather stable over time and cross-culturally, and should therefore change little from 1990 to 2002.As Table 1 clearly shows, correlation between age and authoritarianism is virtually identical in the three surveys, despite the differences in the operationalization of both constructs (in the 1990 survey, age is measured by 5 age-group categories).The association between authoritarianism and education decreased somewhat from 1990 to 2002, but the differences are minimal, again supporting the functional equivalency of the authoritarianism measures.The relationships of the other two scales, measuring socialist and nationalist orientations, with age and education differ more significantly in the three surveys.These differences, however, are more likely to result from the changes in substantive relationships than to be measurement artifacts.As will be elaborated in more details in the main analysis, nationalist attitudes over time became more prevalent among older and less educated respondents (it is worth noting a small but significant tendency of younger respondents to be more nationalistic in 1990).The association of the pro-socialist orientation and education does not exhibit clear trend -it is moderately negative in all three surveys.There is also a small tendency of the socialist attitudes to be increasingly associated with older age, perhaps reflecting the importance of early socialization for this orientation.Thus, as expected the documented stability over time supports the functional equivalency of three different measures of authoritarianism.The same largely applies to the measures of pro-socialist orientation.The modest trend of increasingly older respondents favoring socialist views most likely reflects substantive changes rather than bias in operationalization.One of the main argument of this paper is that the political role of nationalist attitudes changed over time in Serbia.These preliminary analysis shows that changes over time are likely to reflect substantive changes rather than simply methodological differences.

SOCIAL, SOCIO-PSYCHOLOGICAL AND ATTITUDINAL FOUNDATION OF POLITICAL PREFERENCES Discriminant Analysis, 1990
Seven variables were entered into Discriminant Analysis in order to construct functions that are best able to differentiate supporters of four main candidates participating in presidential election in 1990 (Table 2).The first of the two stati-stically significant Discriminant Functions (Rc1=.55,p<.001) is defined primarily by left wing identification (correlation with the function is .74),high scores on socialist egalitarianism dimension, and with older age, but also with significant projections of nationalism (in negative direction!), authoritarianism, and low education.This dimen-sion, therefore, unites pro-socialist orientation with the opposition to nation-alism, and is, as expected, positively related with authoritarianism.Socio-structural variables related to this dimension are those expected to be predictive of the authoritarian worldview: older age and lower education.
This function strongly divides voters of Slobodan Milošević from the supporters of the opposition candidates, but especially from the nationalist opposition (SPO and SRS), as shown by group centroids in Figure 1.Comparing to the socialist supporters, nationalists were less authoritarian when this worldview was associated with the pro-socialist orientation, which is explainable if we take into account that their distinct ideological marker in this period was anticommunism.
The second function is less powerful discriminator (Rc2=.36),but it shows specific differences between the nationalist voters and supporters of the democratic candidate Ivan Đurić.This dimension is primarily defined by nationalism (correlation with the function is .81),and to a lesser degree by authoritarianism and lower education.It seems that authoritarianism of the nationalist orientation is more psychological in nature, since age is unrelated to this Discriminant Function.
Positive coefficient for socialist egalitarianism perhaps shows that supporters of Ivan Đurić are those the most opposed to the rigid egalitarianism of the old system.Despite this, nationalism remains unrelated with left-wing identification.
According to the expectation, authoritarianism was related to both pro-socialist and nationalist ideological orientations.Mean voters' scores on ideological dimensions presented in Figure 2 show that the non-authoritarian perspective is associated primarily with supporters of the democratic opposition, and, more interestingly, that the nationalist candidates attracted considerably less authoritarian voters than the socialists (supporting the secondary hypothesis).Thus, strong parallelism between average scores for socialism and authoritarianism should be noted, as well as the sharp rise in nationalism scores on the right wing.

Figure 2. Ideological differences between supporters of different presidential candidates in 1990
(Note: For Left-Right scale, higher score means right-wing identification)

Discriminant Analysis, 1996
Two Discriminant Functions proved significant in 1996 data as well, but the relationships are stronger than in 1990 (Rc1=.77 and Rc2=.50, both p<.0001).Their structure (correlation between variables and functions) is presented in Table 3.On the ideological level, the first function is defined by high scores of left wing identification, socialist regime support, and a zero coefficient for nationalism.This nonnationalist pro-socialist orientation is strongly authoritarian (the highest coefficient), and is related with older age, lower education, and rural residence.It strongly divides socialist voters from all the rest opposition parties, including the radicals (Figure 3). 8 The second Discriminant Function is, similarly as in 1990, strongly determined by nationalism (the highest coefficient).This time, however, it is related with explicitly right-wing identification, while the correlation with pro-socialist orientation approached zero level.Authoritarianism is again predictive of the nationalist Discriminant Function, as is lower education, but it remained unrelated to age.The second function shows specific appeal of the Radical party, but especially against the parties of democratic i.e., non-nationalist opposition including previously strongly nationalist SPO (Figure 3), while the socialists are in-between the two camps. 8The latter were in heightened, at least rhetorical, opposition to the SPS after the fall of Krajina in mid-1995.It again seems that the first function, indicating pro-socialist orientation, is rooted in the authoritarian world-view given its strong basis in structural variables, especially age and education.We can also observe that in mid 1990s, nationalist ideology already attracted increasingly authoritarian population.In 1990, Šešelj's voters were rather non-authoritarian, while in 1996 their authoritarianism scores are higher than among the rest of the opposition (Figure 4).

Discriminant Analysis, 2002
Six years later, one Discriminant Function proved highly significant in ideologically differentiating the five main political actors 9 (Rc=.62,p<.001), while the second function approached the border of statistical insignificance (Rc=.35,p<.05) (Table 4).
The main change, however, is in the content of the first function.While in 1990 socialism and nationalism loaded the first function in opposite direction, and in 1996 nationalism was unrelated to the socialism dimension, here we see that nationalism has the highest loading on the first function, followed by socialism, and is related with the left wing identification.Unsurprisingly, this function also strongly correlates with authoritarianism, and accompanying demographic variables -in the first place low education, but also church attendance, lower economic position, and age.
Supporters of DOS/DS obtained the lowest scores on this dimension (Figure 5), while both socialists and extreme nationalists are far on the opposite side.DSS is in-between, corresponding to its recent middle-of-the-road ideological positioning.
The second function, that could be termed right-wing nationalism, is correlated primarily with the right-wing identification, somewhat less with the nationalism scale, and still less with church attendance.The function is unrelated to socialism scale and authoritarianism, as well as to the background variables.At the low end of this dimension we find SPS voters, followed by the DOS/DS voters, while Koštunica and Šešelj supporters are among the high scorers.This could be interpreted as purely ideological dimensions, showing classical elements of the conservative-traditionalist ideology -right wing identification, nationalism and religiosity.
A possible reason why this dimension is not related to authoritarianism could be that it partly captures non-conventional nationalism of the radicals, and more democratically oriented nationalism of the Koštunica's DSS.The main reason, however, is that the low end of this dimension is occupied by authoritarian voters of the SPS (see Figure 6).We can also observe that the extreme rightists became more sympathetic to the left whether in terms of self-identification or in terms of attitudes. 9DS and DOS voters who would not vote for DSS were joined.The sample contained too few SPO voters, so the party is omitted from this part of the analysis.A new parliamentary party SSJ is included.

DISCUSSION
Comparison of findings for the three time periods reveals certain continuities as well as changes in the relationship between authoritarianism and the two key ideological dimensions.Authoritarianism, together with the relevant structural variables, was consistently related with the strongest discriminant functions.Thus, the main political divides included different worldviews, in addition to ideological and structural layers.Political divides structured in this way maybe do not qualify for the label of political cleavages, but they certainly do not lack width and depth.
When nationalist and socialist orientations were opposed, as proven by the first discriminant function in 1990, authoritarianism was related with the former orientation and negatively with nationalism, in accordance with the secondary hypothesis.At the time, nationalism was perhaps too an unconventional ideology and general outlook to be related with authoritarianism.At the same time, when nationalism was pitted against the liberal opposition, as documented by the second discriminant function in 1990, authoritarianism accompanied nationalist attitudes.In 1996, nationalism and pro-socialist orientation were loading orthogonal dimensions, and authoritarianism was related with both dimensions.In 2002, nationalism and socialism converged into a common discriminant dimension together with authoritarianism.There, however, remained a trace of a right-wing nationalist orientation, without significant loading of authoritarianism, since it was contrasted not with the liberal opposition but with the few remaining socialist supporters.It suggests that nationalism of the DSS supporters might be more traditionalist, conservative and clerical, but not particularly authoritarian.Hence, authoritarianism participated in discriminating the main Serbian political actors in accordance with the main hypothesis: it went along with nationalism and with the pro-socialist orientation.
The second aspect of the main hypothesis demands that authoritarianism be rooted in specific social condition.The findings strongly support this hypothesis: in all cases high loadings of authoritarianism on a particular discriminant function were followed by significant loadings of variables such as older age and lower education, but occasionally also by rural residence and poorer economic conditions.Education seems especially relevant in this regard.Authoritarianism had significant projection on five out of six discriminant functions, and in all cases education loaded significantly and in the opposite direction.This corroborates Gabennesch's ideas about social roots of the authoritarian world-view.
Concerning the evidenced changes over time, the particular trajectory of the relationship between nationalist and socialist ideological dimensions deserves a particular attention.In 1990, they were opposed -they loaded the first function in the opposite directions.In 1996, they loaded different orthogonal functions, where nonauthoritarian democratic opposition was facing two authoritarian opponents -the nationalists and socialists.In 2002, the socialist and nationalist orientations finally converged into a single discriminant function, together with their common denominator authoritarianism.It is a rather unique feature of the Serbian delayed postcommunist transition that nationalism and socialism converged so strongly into a single political dimension.In 2002 this combination was particularly understandable after years of SPS and SRS collaboration, and their joint opposition to the DOS government.
Correspondence of the ideological oppositions on the mass level and ideological conflicts on the level of party elites, outlined in the qualitative analysis, raises the question to what extent the observed changes can be accounted for simply by agency, i.e., strategic maneuvers of party elites engaged in the power struggle.The presented findings cannot give complete answer to this dilemma, but it is certain that these explanations are not alternative to each other.Without negating the influence of party elites, it seems clear that there is more stability in the structure of the obtained discriminant functions than strictly in association of these dimensions with particular parties.For example, when SPO abandoned extreme nationalist rhetoric it lost some voters who rather opted for parties more in line with their sociopsychological profile.It seems that common socio-psychological ground made particular maneuvers more or less appealing to specific sections of the electorate.

139
Hence the apparent ease with which Serbian voters shift from left (support for SPS) to right (support for Radicals).
The pervasive influence of age suggests the relevance of socialization experience, just as in Inglehart's theory of value change.The difference is, however, that the present model operationalizes the dispositional variable that transfers the socialization experiences into the present, while in Inglehart's research program the accessed level in the need hierarchy remains hypothetical and operationalized is only the presumed effect -the materialist or post-materialist value orientation.
The obtained findings pose a challenge to speculate about the explanation of the prolonged weakness of the non-authoritarian democratic bloc in Serbia.According to Kitschelt's (1992) model, the strength of the libertarian-pro-market camp is dependent on the economic development since resource endowment is crucial for future market competition.And indeed, the loadings of structural variables, such as age and education, clearly show that potential "losers of transition" have been more ready to vote for the Radicals and the Socialists.On the other side, those with greater social and cultural capital seem to be more open towards the economic reform, and to support reform-oriented parties.
However, Serbia's relative economic backwardness can only be a part of the story.On the one side, it is difficult to overemphasize the importance of the then contemporary political processes and the feeling of threat they induced.The dissolution of the socialist Yugoslavia and secession of the other Yugoslav republics forced the national issue to the foreground.But this process was certainly helped by the ability of the leaders of the authoritarian bloc to appeal to specific sociopsychological features of the electorate and secure stubborn support despite obvious economic and other hardships.At the same time, it also shows the inability of the more democratically oriented leaders and parties adjust their rhetoric to the sensibility of large segments of the electorate.
The relevance of psychology-based world views is clearly seen in crisis situations.During the 1990's all social strata suffered, yet the popularity of the SPS remained high.Only when the zigzags of the party elite and the military defeat undermined the image of the SPS as either credible or effective representative of any orientation started the voters to desert.The implication is that leaders can secure stubborn popular support even in the face of economic hardships, as long as they manage to appeal to the psychological tendencies or world-views of significant portions of the public.
The final question is to what extent the political divisions described in the paper can be conceived in terms of the cleavage theory.The major structural variables such as rural-urban division, age, religiosity, are relevant, but their influence is relatively small comparing to the role of specific ideological orientations, such as nationalism or socialism.However, even to the degree that these variables work, they do not work because of organizational encapsulation or closure of relevant social groups.Given that the main discriminant dimensions contained ideological, dispositional, and socio-structural variables -such divisions could perhaps be labeled as cleavages of world-views.

CONCLUSIONS
Multivariate analysis of party preferences in 1996 and 2002 showed that preference for authoritarian political options was tied to authoritarianism as an individual difference variable.Supporters of different parties tend to come from somewhat different socio-economic and socio-cultural groups, and to hold different political attitudes.The central political divide throughout the 1990s, that between authoritarianism and democracy, is well reflected on the level of voters' characteristics.The less educated, older, and rural population proved to be more authoritarian, nationalist, and nostalgic about the communist times, and tended to support the SPS and SRS.One the other side, have been those with greater social and cultural capital, those with better chances for developing broader social perspectives.The divide between the authoritarian and democratic blocs was thus not only political and ideological, but also psychological and cultural.Political immobilism in Serbia during the 1990s can at least partly be explained by the depth of this division and by the impossibility -both for voters and politicians -to cross the border between the two positions.
From the theoretical point of view, the most important result is that socialstructure consistently covaries with a more general world-view.As Gabennesch (1972) argued, authoritarianism is a psychological reflection of particular social conditions marked by material and -especially -educational deprivation.The political role of such individual-dispositional variables is perhaps particularly strong in contexts when social structure is still not crystallized, and when it is not easy to establish clear connections between particular economic interests, political actors, and policies.In the context of "flattened societies" (Wessels and Klingemann, 1994), the formation of politically relevant groups may be based more on values and world-views than on social categories, although there is a strong connection between them.Serbs should be proud of their people.

REFERENCES
There are few nations that contributed to the world's culture and science as ours.
No nation has such a glorious and at the same time tragic history as the Serbs.
Our country should follow its own way, not caring about the expectations of the West.
It is true that Albanians in Kosovo were victims of the persecution by the Serbian state.*Nationalism endangers the development of our country.* It is more important that a politician be a strong patriot than that he/she be an expert.
Our country should seek a peaceful reunification of those parts of the neighboring countries that are inhabited by ethnic Serbs with Serbia (e.g., Republika Srpska).
Schools should pay more attention to the patriotic education of young people The Serbian people is a victim of an international conspiracy.
The Serbian people often suffered because it was too good towards the others.Socialism 6 .60For workers it is better to be employed in state-owned firms, than in private of privatized ones.
The state should provide job to everybody who wants to work.
Trade unions should have more say in government business.
The transfer of state-owned companies to private hands will help very much in solving the economic problems of our country.*Unprofitable factories and mines should be closed down immediately even if this leads to unemployment.*Education should be accessible to everyone, therefore it should be free.
Authoritarianism 10 .84 The most important virtues a child has to learn are obedience and respect for authority.
All true patriots are obliged to take measures against those condemned by the leaders of the country.
Young people need strict regulations and determination to fight for their families and their country.
Nowadays in our country most of the damage is done by those who do not respect our leaders and the order of the society.
Young people sometimes have rebellious thoughts, but as they grow up, they should condemn these and adapt.
It is nothing wrong to have nudist beaches in our country.*Immoral conditions in our country are partly due to the fact that both teachers and parents forgot that physical punishment is still the best way of upbringing.
It would be better for everyone if the authorities would censor the newspapers and films so that rubbish be kept away from the youth.
Most of our social problems would be solved if we get rid of the immoral and pervert persons.

Note:POLITIKA
All scales constructed as the first principal component of the included items.* Direction reversed for inclusion in the respective scale.REZIME Politics in Serbia 1990-2002: A Cleavage of World Views 147

Table 2 . Structure matrix of two significant functions that discriminate supporters of four main presidential candidates in 1990
Figure 1.Group centroids, 1990.(Data source: ZA Study 2901, 1990.)

Scales in 1996 survey (ZA Study 2911)
All scales constructed as the first principal component of the included items.