Educated, Urbanized – and Narrow-minded?
A Comparative Analysis of Political Ideology in Local Parties
Zürich, May 2009
This paper aims to explore the strength and causal determinants of ideological thinking within political parties. The degree of “ideologization” includes two dimensions: “horizontal couplings”, as they are manifested in intercorrelations between different beliefs or opinions; and “vertical couplings” as they appear in the degree to which specific beliefs are related to more abstract concepts like “left” and “right” The results presented show shockingly high trends of ideologization at least on the left-center section of the spectrum, especially in the vertical dimension. Urbanization stands out as a major causal factor: On both sides of the spectrum, ideological constraints are significantly higher in larger than in smaller communities, and more pronounced in suburban settings than in villages far from the next city. Only in rather small communities, it is found that ideologization correlates positively with the educational level and the modern occupational background of party members, with the exposition of the local sections to the programs of supralocal mother parties, and with the number of other local parties with which they have to compete. In communities of given site, ideological thinking is more pronounced when parties possess a small share of political power (or none at all). Finally, there is evidence that ideological constraints in party policy has increased since the late eighties of the last century, and that it may increase further because younger age cohorts are more prone to clustering issue positions tightly and for relating them closely to abstract notions of “left” and “right”. In contradiction to most conventional wisdom, we may tentatively conclude that political thinking is more “rational” (or at least more flexible and pragmatic) at the peripheries than in the more central spheres of modern societies.
1. „Ideology”: an ambiguous concept
In everyday talk as well as in political discourses and scientific theory building, the term "ideology" has many different, even contradictory, connotations.
On the one hand, "ideological thinking" denotes a tendency
Therefore, there is always a tendency to attribute ideological thinking to adversaries, while exempting oneself from this stigma.
In Marxism, there is an equally negative tone that stresses the causation of ideologies by underlying economic conditions and their function to support reigning power systems by propagating systematically distorted views of society and social relations.
On the other hand, there is also a long tradition in social science to see ideology in a much more positive way: as a correlate of higher intellectual sophistication. In this view, ideological thinking is characterized by high stability, internal logical coherence and consistency, based on the capacity to relate specific issues to more abstract principles and to organize different attitudes to logically consistent wholes (Allport 1962; Converse 1964; Gerring 1997). This view gave rise to the notion that people with high political interest and expertise are most likely to structure attitudes toward political issues in ideologically consistent ways (Converse 1964; Zaller 1992; Federico/ Schneider 2007), while most ordinary citizens (especially the less educated) lack ideological thinking because their political views are found to be incoherent, volatile and logically flawed. (Bishop 2005; Converse 1964; McGuire 1985/1999).
"....the fact remains that many citizens are unable to understand and structure their preferences in terms of the abstract ideological concepts that organize elite political discussion and activity, perhaps compromising their capacity to effectively engage the political system." (Federico/Schneider 2007:222).
Following Federico and Schneider (2007), two dimensions of ideological attitude-structuring may be distinguished:
"Horizontal constraints": or "political belief consistency" (Converse 1964).
This concept relates to the degree to which issue opinions among are interrelated among each other: so that to know one attitude helps much to infer all the others. Scaling methods as well as factor-analytic and cluster-analytic approaches are useful to detect such patterns of mutual interrelation. In this perspective, left and right (or liberalism vs. conservatism) are not ideological poles endowed with an ex ante determined content, but just clusters of issue opinions that appear inductively when intercorrelation matrices are analyzed and dimensionally reduced. This accords with Gabel and Huber who assert that methodologies aiming to place reliably political parties on the left-right dimension in a consistent manner impose no a priori meanings on the concepts of left and right. Consequently the most successful methods are those that locate parties with the "vanilla" method, which reduces ideology to a consistency constraint on parties' issue positions (Gabel and Huber, 2000: 96).
"Vertical constraints" (or "Level of conceptualization" (Converse 1964):
This concept denotes the tendency to relate specific issue opinions hierarchically to higher order constructs of ideology, e. g. one's placement on the left-right continuum (Federico/ Schneider 2007: 232 passim). Such vertical patterns are manifested by measuring differences between issue-attitudes and ideological scale values, or by calculating the correlations between these two. By applying these concepts, experimental research studies have found for instance that survey respondents who are asked to think carefully about a variety of issues show greater constraint among attitudes toward these different issues (Judd and Downing 1990; Lavine et al. 1997). However, this "rationalistic" perspective has been heavily attacked by psychologists who again stress the importance of non-rational factors in the genesis of ideological thought. For instance, Federico and Schneider criticize such views as unduly "cognitivistic" and stress the importance of motivational factors in processes of ideological structuring: e. g. the (situationally conditioned) subjective needs, wants and goals that promote tendencies to aggregate and constrain perceptions, opinions and judgments (Federico/Schneider 2007).
In a similar vein, Jost et al. have argued that ideological thinking may well derive its internal coherence not from logics, but "acquires coherence and structure from psychological needs, motives, and constraints that vary both situationally and dispositionally." (Jost et. al. 2008).
Especially "conservatism" has been widely interpreted as a highly synthetic world view that encompasses
Such views lend themselves easily to the negative evaluation of ideological thinking as a weakness rather than strength: a disposition to let political attitudes and decisions be governed by emotional factors, character traits or habitualizations - factors related to the evaluating subject rather than to the evaluated cause.
While psychologists are professionally disposed to focus on intrapersonal motivations and dispositions, sociologists - even when they are not Marxists - are more inclined to include factors related to social interaction, collective group formation, society and culture. Adopting this perspective, it is fascinating to learn that, there are pronounced differences between various segments of societal elites. First of all, it has been found that highly consistent belief systems are highly pronounced among cultural elites, while they less prevalent among elites highly involved in (political or economic) decisions.
This regularity may indicate that high involvement in powerful roles may be incompatible with the maintenance of ideological consistency because there is too much need to do justice to each particular problem and to adapt to issue-specific and situational conditions. On the other hand, ideologies may well flourish within powerless groups who can easily cultivate consistency because they are not forced to test out whether such views could be upheld "in practice". Secondly, it has been found that ideological thinking is more ubiquitous among human science academics than among those with a background in natural or applied sciences (Lerner et. al. 1990). This may indicate that ideological patterns are heavily conditioned by factors not related to politics at all: in this case: variables related to professional socialization.
2. Left-right ideology: conceptual and theoretical issues
While originating from mutually very distant disciplines (like sociology, psychology, sociology and political science), most studies on ideology converge in the point that ideological thinking in politics is mainly organized along a single dimension spanning between "conservative vs. "liberal" or "left" vs. "right".
A wealth of empirical evidence shows that the left-right axis is almost ubiquitously salient in two ways in a wide range of different countries: as a polarity that structures bipolar political conflicts and as a continuum that allows differentiated comparisons on an ordinal or interval scale (Castles & Mair 1984; Gross & Sigelman 1984; Janda 1980; Laver & Budge 1993; Warwick 1992). Thus, most citizens in most developed democratic countries (including the United States) are willing and able to place themselves on the left-right dimension: a scale usually ranging from 1-10 or from 0-10. (Inglehart/Klingemann 1976; Colomer/ Escatel 2004: 3). Similarly, voters as well as political elites use the LR scale for characterizing social movements, political parties, candidates, news media, issue positions, political programs and governing regimes.
It is well known that the political left-right dimension has originated in France at the time before the revolution, some weeks after the convention of the Estates General in June 1789 (Laponce 1981:47). As an intuitive variable derived from the seat order in parliament, it filled the vacuum created by the demise of estate membership as a dominant criterion of classification. Like the top-bottom model of social stratification, the LR scheme is a simple spatial metaphor that has the characteristic of being translatable in all languages and being potentially adopted by all human cultures (Laponce 1981: 27).
The ubiquity of such one-dimensional schemes is best explained by the "functionalist" theory which assumes that the salience of the LR continuum is particularly high under conditions of high political complexity and low political information (Fuchs/Klingemann 1990). As a starting premise, the theory assumes that most people spend little efforts for acquiring and synthesizing political information, because they have little skills to do that or no available time. However, under conditions of political democracy, all citizens are called to make decisions despite the fact that most of them are unable or unwilling to collect much detailed information on the political sphere. Given these conditions, they have a great need for simplifying stereotypes helping them to decide which parties, leaders or political programs they should support.
Especially when political conditions are highly complex and non-transparent, citizens tend to rely on rather simple heuristic shortcuts in order to gain orientation and to come to non-ambiguous voting decisions with a minimum of personal efforts (Berelson, Lazarsfeld, and McPhee 1954; McKelvey and Ordeshook 1986; Neuman 1986; Popkin 1994; Hinich and Munger 1994). Such labels facilitate also political communication, because they create a common background of meaning that can be presupposed ex ante, so that it has not to be created on every single occasion. The salience of the left-right dimension can be minimal in two party systems, because party preferences can easily be built up without referring to ideological notions. In more complex and competitive multiparty systems, however, many voters will feel a need to locate themselves on the left-right scale in order to identify the parties and candidates most akin to his own views (Fuchs/Klingemann 1990; Knutsen 1998), and parties will use ideology as an indispensable tool for "branding their products" in political campaigns (Downs 1957; Colomer/Escatel 2004). In conformity with Converse's observation that highly educated and politically interested individuals were more inclined to think in ideological terms (Converse 1964), various more recent studies have shown that politically sophisticated individuals are better able to make use of political labels like “left” and “right” because they are better informed about their meaning (Sniderman et.al. 1991; Kitschelt/Hellemans 1990; Klingemann 1979 etc.), and that issue positions and left-right self placements are more tightly correlated when the level of political cognition is high (e. g. Inglehart/Klingemann 1976).
This level of political cognition has been shown to be critically dependent on political interests and activities on the one hand and the educational level on the other (Inglehart/Klingemann 1976).
However, this of course presupposes that there exists a relatively consensual, culturally anchored conception of “left” and “right” in the given national society: so that it is just a question of cognitive capacity whether it is perceived adequately or not. Nie and Andersen have found high consistency and intercorrelations among elites’ responses on a wide range of issues concerning the scope of government activity enabling them to identify clear “liberals” and “conservatives.” On the level of uneducated strata, however, “there is little or no interdependence … in mass attitudes, because mass publics have neither the educational background, the contextual knowledge nor the capacity to deal with abstract concepts that sustain an organized set of beliefs over a wide range of political issues” (Nie/Anderson 1974). Given the high causal relevance of political interest and political communication, it is to be expected that political elites are most likely to maintain tight relationships between left-right self placements and issue positions. In fact, Eurobarometer studies have shown that within a sample of political candidates, the statistical variance explained was four times as high than within a sample of ordinary voters (Inglehart 1989: 368).
The astonishing permanency of the spatial left-right metaphor contrasts sharply with extreme variations of the meanings associated with these two terms during history and across different cultures. Whoever identifies leftism with socialism should well remember that in the beginning 19th century, it was mainly associated with individualism, free enterprise, national independence and - following Rousseau - an endeavour to restore a more perfect form of human society as it had presumably existed in the past (Laponce 1981: 118ff.). Between about 1850 and 1960, leftism was almost exclusively amalgamated to socialist and communist ideologies associated with the various labour movements - thus giving priority to questions of economic organization, class relations and social welfare. Given the long duration of this phase, several scholars have tried to identify the “invariant core” of substantive values and political issues that define difference between left and right.
While S M. Lipset (1954) defined leftism as the fight for more equality, Downs has tied it down to governmental economic interventions (Downs 1957). A comprehensive attempt to identity the transnational semantics of left-right ideologies was made by Laver and Budge on the level of Paneuropean Manifesto data. Applying factor analysis, they have extracted 26 items defining leftist and rightist party positions. On the left side, the list includes issues like governmental control of capitalism, nationalization of enterprises, internationalism and the expansion of social services, while rightist parties were consistently characterized by an emphasis on traditional morality, law and order, free enterprise and national autonomy (Laver/Budge 1993).
In the late sixties, the civil rights movements and the student revolts gave rise to a new, more encompassing understanding of leftism (or: "radicalism"): including the perspective of extending basic standards of human rights and welfare to all kinds of discriminated population segments (like non-whites, females, gays, starving poor in underdeveloped nations etc.). In a broader sense, there emerged potent movements of "social liberalism" directed against all traditional values and habits that sought to prevent full individual self-determination (e.g. the feminist and "pro-choice"-movement). Within the same time period, ecologist issues gained increased prominence on political agendas and were incorporated in programs aiming to catalyze societal and economic change. Finally, processes of globalization and regional supranational integration (within Europe) have given rise to new controversies where basic strategies of foreign policy are at stake.
However, there is no agreement whether all these "postmaterialist" issues become increasingly assimilated to "leftism", or whether they are the nucleus of a second (orthogonal) ideological dimension.
On the one hand, there is considerable evidence that no amalgamation is taking place. Thus, Inglehart presented convincing empirical data which showed that the attitudes of the general electorates of seven European countries toward "materialist" issues and toward "postmaterialist" issues are rather weakly correlated with each other. As correlations are particularly low among younger age cohorts, it is concluded that intergenerational change is contributing to a growing independence of these two dimensions (Inglehart 1984, 25ff; Inglehart 1989: 372ff.).
Following a similar argument, Clark and Lipset suppose that rising affluence and intergenerational change contribute to a growing salience of "social value issues" which constitute "a distinct political dimension from more traditional economic or fiscal issues." (Clark/Lipset/Rempel 1993: 305). According to Clark, a "new political culture"(NPC) is emerging, particularly among the young and better educated strata, in which "leftism" becomes
In contrast to socialist collectivism, this new leftism is more associated with individualism: implying an affirmative position toward markets and private actions, and a growing skepticism toward governmental bureaucracy. All these ideological changes are facilitated in wealthier environments (Clark/Inglehart1998). In a similar vein, Kitschelt has argued that in addition to the traditional LR dimension, a second axis extending between an "authoritarian" and a "libertarian" pole is gaining weight (Kitschelt 1994). Besides social liberalism, ecologism is also said to constitute a new ideological dimension less and less associated with conventional radicalism. Thus, a cross-national study (comparing the populations of England, Germany and the United States) has found that ecologist attitudes are not correlated with standings on traditional left-right issues (like governmental control) (Kessel/Tischler 1984).
Finally, Simon Hix and Fritz Scharpf claim that processes of European integration have given rise to a new dimension that divides those who favor this process from those who want to preserve national autonomy (Hix 1999; Scharpf 1996). They both assert that the public attitudes toward issues of European integration tend to be correlated with Left-Right positions, but constitute a distinct dimension of political conflict in Western Europe. On the other hand, there is also evidence that attitudes toward "old" and "new" political issues are significantly interrelated, and that this linkage has not been eroded by either socio-economic development or intergenerational change.
In a sophisticated empirical study, Weakliem (1991) has shown that the "materialist" and the "postmaterialist" dimension of political ideology show considerable values of interfactor correlations (between .40 and .60). This is quite consistent with Inglehart's findings that, at least for political elites, materialistic and postmaterialistic issue positions appear to remain more closely linked and to be components of an overarching left-right dimension encompassing economic as well as non-economic issues. (Inglehart 1984: 33). According to Inglehart, the functional needs for organizing politics along a single "left-right"-dimension are so imperative that - in the long run at least - postmaterialistic issues will become assimilated to this dimension rather than evolving to an independent second ideological axis (Inglehart 1984: 37).
All this conforms to Sani and Sartoris view that Left and Right are just "empty containers" that can be filled with various meanings by anybody (Sani/Sartori 1983).
In her extensive study based on Eurobarometer data from 1999, Corrie Potter concludes that
Evidently, the LR-dimension has an astounding capacity to absorb new political values, issues and strategic goals (e. g. of social movements) (Mair 1997: 26; Inglehart 1984; Knutsen 1995). However, these new aspects seem to complement and overlay the old ones without replacing them.
Some researchers dealing with the semantics associated with the LR scale have also indicated that basic asymmetries exist between the two poles. Thus, it has been found that post-materialist values are strong predictors of left placement, while opinions on materialist are more determinative for rightist positions (Potter 2001). In several other empirical studies, it was found that rather tight correlations between issue positions and left-right self ratings exist for the left half of the continuum, while on the right side, the explanatory power of political attitudes is much reduced. As Laponce concludes from a metaanalysis of such studies, this is true for most issues conventionally related to the LR-continuum: e.g. attitudes toward economic regulation, nationalism or gender equality (Laponce 1981: 158ff.). As the author himself has verified in a study on local parties in Switzerland, the same is also true for items related to financial policy, immigration policy or environmental protection (Geser 1992). 
In fact, there is wide agreement that the general decline of class voting has not at all contributed to a decline of such ideological factors. To the contrary, it has caused political thinking to be more determined by subjective attitudes, among which ideological concepts of leftism and rightism are still of paramount importance (Knutsen 1988; Kim/Fording 1998; Freire 2006: 367 ). As a consequence, the correlations between LR-self-placements and political attitudes may be become even more conditioned by individual education and individual political exposure: because individuals have to learn the exact meanings of left and right in their respective society, and they have to synthesize different values and issue positions in order to get an overall value on the one-dimensional LR scale. While it is universally acknowledged that ideological contents vary according to the flow and ebb of different political issues, it is sometimes still upheld that such variabilities are just peripheral changes that don’t affect an invariant conservative core:
In the era of conventional class politics, ideological cores were sufficiently defined by stable interests deriving from the position of collectivities within the socio-economic system. In post-materialist ages where particular interests are substituted by generalized "values", this specificity and stability has evidently been lost. Instead, ideologies become more dependent
Consequently, we may well see a higher variability of "leftisms" and "rightisms" in the future, while the degree of ideological cohesiveness may well remain the same, because the same needs for simplification are persisting on the individual as well as on the political level.
However, it cannot be denied that all these different issues were too heterogeneous (and too variable) to be part of a logically consistent ideological belief system.  Thus, the notion of "postmaterialism" (Inglehart) reflects just the fact that the new issues are not primarily related to vital interests of specific societal groups, while the concept of "new liberalism" focuses on a highly generalized attitude of "cosmopolitanism" and "anti-institutionalism" particularly prominent among highly educated younger cohorts (Bell 1979, Brint 1984). Likewise, it is not evident why these new issues can easily amalgamate wit the old "materialist" contents of left and right, as it is almost impossible to find the commonalities that make them parts of a coherent ideological whole. For instance: if it's true that conservatives have an increased need for security: why are they more prone to carry the risks connected with nuclear power plants? And why are leftist parties fighting so vigorously for the abandonment of atomic energy, when this results in such cost increases for electricity that they risk to lose their traditional electorate (lower social strata).
Similarly, it is not evident that attitudes toward EU membership and European integration have become a left-right issue in many countries. Why are the leftist (e. g. in Switzerland) in favor of such a project which is associated with so many developments they don't appreciate: the liberalization of trade and labour markets, the intensified fight against immigration from Southern countries; the demise of national worker protection?  If highly educated strata are more prone to maintain tight constraints among opinions to all these issues: why does this indicate that their thinking is in any way more sophisticated than the less interrelated attitudes of less educated citizens? Doesn't this manifest just the contrary: that educated people are more disposed to take over ready-made collective stereotypes that to rely on autonomous individual reflection, more eager to create in-group conformity than to generate guidelines for pragmatic judgments and decisions?
3. Local parties as a field for studying "ideological cultures"
The following empirical study intends to shed a light on the actual ideological culture in Swiss politics and on its relationship to the status-characteristics of politically active populations. In contrast to the "political culture" which encompasses more basic "rules of the game", ideological cultures may be seen as more variable pattern co-varying with long-term changes of societal value systems as well as more short-term developments on the level of social movements, salient issues or attitudinal "fashions". Nevertheless, the concept of "ideological culture" implies the existence of collective political perspectives, values and goals governing the behavior of individuals and organizations that are transmitted by regular processes of socialization.
In general, average individual citizens are rather poor informants of "culture" because their thinking is heavily shaped by psychological idiosyncrasies and because they often lack the cognitive and intellectual capacities needed to perceive and interpret these collective patterns adequately. Studying politically active elites may be a better approach, but it still assumes that cultural patterns are adequately mirrored in the subjective consciousness of individual minds. By studying groups and organizations, more justice can be done to the basic fact that culture is primarily expressed in the outcomes of collective communications and activities: e.g. in the explicit results of discussions, negotiations and deliberative procedures, in formally stated decisions and action programs or at least in mutually recognized "majority opinions". Political parties have particularly strong links to ideological culture because it is their job to contribute to the aggregation and articulation of collectively shared opinions, values and goals.
It is evident that in general, party groupings will tend to maintain more consistent and integrative ideological standings than individuals, because
This implies that high intercorrelations between different issue opinions (like "socialism" and "ecologism") among parties may not at all reflect a similar covariance on the level of their individual members: but rather their particular success in aggregating "socialist factions" and "ecological factions" into an overarching party platform (e.g. for simple tactical reasons of winning elections). In an even wider and longer perspective, political parties can be seen as active agents that define for everybody else in society what is the (current) meaning of "left" and "right".
Local party sections have the additional virtues of being so numerous that rigorous multivariate methods of comparative analysis can be applied, and of being so low in organizational complexity that a single central member is capable of delivering all the relevant information. Switzerland is outstanding for the fact that formalized part groupings are astonishingly wide-spread even among very tiny municipalities (of 500 or less inhabitants), and that they control to a high degree all major political processes on the communal level (Geser et. al 1994: passim).
Given the mix between direct and representative democracy typical for Swiss politics, most of these local parties have the dual function of influencing elections on the one hand and decisions about specific political issues on the other. With a total number of about 200000 participative adherents (= ca. five percent of Swiss voters), these groupings encompass the major part of all politically active citizen in the country.
4. Reseach questions and Hypotheses
In the following, we want to explore some factors that determine to what extent the issue positions, goals and action courses of political parties are structured by ideological constraints.
Two aspects of "ideologization" have to be distinguished:
1) "Horizontal constraints" consist in the mutual interrelationships between different issue positions. In operational terms, they reach their maximum when knowing the opinion on issue A allows me to predict with certainty the opinions on issues B, C, D... to Z, and they are minimal when no predictions are possible because intercorrelations are zero. Evidently, horizontal constraints can effectively be assessed by factor-analytic methods that explore to what degree intercorrelation matrices between issue opinions can be reduced to very few common dimensions.
2) "Vertical constraints" are defined by the degree to which attitudes to specific issues are related to higher-order ideological concepts like "liberalism" vs. "conservatism" or "left vs. right". In operational terms: they can be assessed by the certainty and precision with which the placement on such ideological scales can be predicted when opinions on all particular issues (A,B,C.... to Z) are known. Evidently, such assessments call for multivariate regression methods that allow to measure the cumulative explanatory power of all the issues in questions.
Minimal ideologization would be defined as a totally unconstrained political standing where the opinion about each issue is generated independently and without guidance by superordinate "Weltanschauungen" or programmatic structures: on the basis of its own intrinsic merits and by adapting to particular needs and problems, current trends in the media or demoscopic surveys, and specific situational conditions. Evidently, such openness would have to be payed with heavy loads of permanent information gathering, communication and consensus-building procedures, and it will make it difficult for a party to formulate programs and to establish and maintain a consistent and clear-cut public identity.
Theoretically, these two aspects could vary independently of each other. Thus, dense horizontal clusterings could be found without any relation to more abstract ideological notions, or issue positions may be tightly coupled to ideological concepts despite the fact that they are mutually unrelated. (This second case is easily identified by multivariate regression equations where different issue positions maintain high explanatory power despite the fact that their mutual interrelationships have been eliminated (by means of statistical controls). However, notions like "leftism" or "rightism" always imply a combination (and positive covariation) of horizontal clusterings and vertical couplings: especially among more educated and politically sophisticated respondents who have enough knowledge about the current meaning of ideological labels and skilled enough in processes of mental abstraction.
On a general level, it has to be expected that in contrast to supralocal (especially national) party organizations, local parties may well maintain a much lower level of ideology, because in the realm of community politics, issues are often defined as non-political problems to be solved by mere common sense or technical expertise (Vidich/Bensman 1968, Geser 2003). Nevertheless, we expect that local parties are basically also subject to the same patterns of one-dimensional left-right polarity as it is reigning in the encompassing political system.
1) We expect that a certain degree of horizontal as well as vertical ideologization is present among all political parties, and that these clusterings and constraints can be unambiguously identified in conventional terms of "left" vs. right".
2) It is to be expected that this main ideological dimension encompasses a broad range of "materialist" as well as "postmaterialist" issue positions. Other ideological dimensions (e. g. related to social liberalism, ecologism, international openness etc.) may also be found, but they are likely to be of secondary importance.
3) We acknowledge the possibility that there are asymmetries between "leftism" and "rightism" in the degree of ideologization. On the one hand, we are open for the notion that "conservatives are the real dogmatists" because they have a heightened need for cognitive structures and a decreased capacity to tolerate ambiguity and change. We recognize that since Adornos seminal work on the “authoritarian personality" (Adorno et. al. 1950) until today, a horde of liberal psychologists have taken great efforts to denounce conservatives as "closed minds" - while asserting at least implicitly that liberals (= they themselves) are exempt from such restrictions by being more open, rational - and better human beings in most relevant cognitive and moral aspects (e. g. Tomkins 1963; Wilson 1973; Altemeyer 1998; Jost et. al. 2003; 2008). On the other hand, we take into account the thorough empirical evidence that falsifies such claims: e. g. the impressionistic historical regularity that leftist (e. g. socialist and communist) movements and parties have been forerunners in political ideologization since the middle of the 19th century, while ideologies on the rightist side (including national socialism) were mainly reactive and have remained on lower levels of coherence and consistency. In more rigorous quantitative terms, this regularity has been corroborated by several studies that have found vertical couplings between issue positions and ideological self-placement to be much stronger on the left-to-center than on the center-to-right section of the Left-Right scale (see. Laponce 1981:158ff; Geser 1992; Potter 2001).
4) It is expected that ideologization is conditioned by various characteristics of the social and cultural setting within which parties operate. Given the notoriously low salience of ideological thinking in (especially rural) communal politics (Vidich/Bensman 1968; Holler 1981), we will expect ideology to become more prominent with increasing city size (or proximity to bigger cities) and increasing degree of "communal politicization".
On a more general level, we follow Converse, Inglehart/Klingemann and many others in assuming that the meaning of left and right is an ingredient of political culture in which not all citizens are equally involved. Therefore, take into consideration the possibility that political ideology is predominantly articulated in the very centers of society and by the more educated strata (Converse 1964; Laponce 1981: 158ff.;Gerring 1997; Bishop 2005). Consequently, especially vertical couplings should be highest in parties that operate in an urban (or suburban) municipality with a highly educated population.
5) For the same reasons, ideological constraints are likely to covary with party membership composition. In particular, we expect higher ideology when larger percentages of party activists stem from modern sectors of the economy and have higher educational degrees.
6) In addition to socio-economic conditions, cultural factors are also likely to affect the tendencies toward ideological constraints. Considering the results of an earlier comparative study conducted by this author, it might be expected that at least vertical constraints are more pronounced in the German-speaking parts of Switzerland than in the French and Italian region (Geser 1992). Such divergences may well be caused by the higher political autonomy of communities and the more pronounced traditions of direct democracy in the German speaking cantons.
7) We remember Lerners's finding that ideological thinking is most pronounced within marginal (e.. g. cultural) elites not involved in far-reaching (political or economic) decisions (see above). This could result from the fact that high policy involvement creates pressures to adapt pragmatically to particular problems and situational conditions, while powerlessness offers better opportunities to preserve "ideological purity" because no full "reality tests" have to be faced. By following this argumentation, it can be assumed that conditions for maintaining coherent ideological beliefs may be better when a party doesn't actively participate in executive political power, or when its share of formal power is rather small. Very high political involvement (e. g. when a party reigns with absolute majority) will be incompatible with high ideological consistency for several reasons: e. g. because such parties have to be responsive to the total population (not only to narrow electoral clienteles) and because taking concrete decisions usually imply that parties adapt to pragmatic conditions hic et nunc, irrespective of precedences, other political issue positions and more abstract ideological stances.
8) We expect that in municipalities of given size, ideological constraints will rise with the number of competing political parties for two reasons. First, every single party has more leeway to appeal to special electorate groups and give priority to internal purity and coherence, because it has not to take the overall perspective of the total community. Secondly - following the functional theory of Fuchs and Klingemann (1990) - parties in complex political systems have to streamline and simplify their positions and programs, so that citizens are better able to make electoral decisions without having to gather and synthesize much information.
9) Ideological constraints should increase to the degree that local parties are penetrated by supralocal mother parties: e. g. by taking over and following their political programs. Especially in smaller communities where endogenous ideology pressures are low, such exogenous influences are likely to be pronounced.
10) Impressed by the widely articulated mantra of "ideological revivals" (Hinich/Munger 1994; McCarthy/Poole/Rosenthal 2006; Baldassari/Bearman 2007 etc.), we ask how degrees of horizontal and vertical ideological constraints change over time. Fortunately, our data sets allow detailed assessments of changes and stabilities between 1989 and 2002, because survey waves with partially identical questions were conducted at these two points of time. Additional insights in current trends of intergenerational ideological change are gained by comparing local party sections with older and younger members
Auch ein hoher Frauenanteil im Vorstand scheint sich eher negativ als positiv auf die Einflussstellung von Ortsparteien auszuwirken, die im Gemeinderat über keine dominierende formelle Machtbasis verfügen (Tab. 8).
5. Data and variables
The following empirical study is based on two mailed-out surveys (conducted in 1989 and 2002) that have included all (about 5000) local party sections in all (about 2800) Swiss communities of all three linguistic regions. The questionnaires were sent by mail to the current heads of these sections. They were asked to provide information about the political goals and values of their grouping as well as on its membership composition, internal organizational structure, political activities and relationship to the supralocal party levels. As the return was about 50% in both waves, a rather large sample of more than 2600 units was achieved: providing the basis for testing rigorously a manifold of hypotheses with multivariate statistical procedures (Table 1).
Given the pronounced legal autonomy of Swiss municipalities (especially in the German speaking cantons), it is not astonishing that local parties also have much leeway in their relation to the supralocal mother parties to which they formally belong. As a consequence, they are rather free to define their political standings by adapting to characteristics of their membership base and various conditions of their local setting. Among many other questions, respondents were asked to place their party section on a left-right scale ranging from 1 (extreme left) to 10 (extreme right).
As seen from Table 2, more than 95% of all participants were ready and able to provide such a judgment, and the whole range of values was actually used. Comparing the two waves, there can be concluded that the notions of "left vs. right" seems to have increased somewhat in salience, because the share on non-respondents has considerably declined (from 4.3 to 2.9 percent); and that only minor changes have occurred in the overall distribution. In particular, extremely leftist and moderately rightist groupings have slightly increased, while sections on the extreme right have lost ground.
Looking at Table 3, it is remarkable that the frequency distributions on most items are highly skewed. Social desirability factors may well have effected that informants were inclined to "agree" to any items proposed: so that results may have been somewhat different when alternative (negative) formulations would have been used. As in Table 2, a closer inspection shows that no clear overall shifts to the left or the right have occurred in the critical period, While support for some leftist demands (budget expansion, closing of nuclear power plants) has increased, welfare commitments to immigrants have declined and critics of the Swiss financial sector (under heavy international attack because of its banking secrecy practices) have lost ground.
Table 3: Distribution of party opinions on 15 salient political isses 1989 and 2002 (percentage values).
Considering the semi-nominal character of these opinion scales, nonparametric methods of analysis have to be applied. For assessing the horizontal constraints (=interrelations between variables), Multiple Correspondence Analysis (MCA) seems an adequate procedure because it parallels factor analysis in reducing intercorrelation matrices to a minimum number of dimensions: with factor scores that indicate the relationship of each variable vis-à-vis these extracted factors, and inertia values that measure the percentage of total variance explained by each factor. The vertical constraints are assessed by correlating issue opinions with the values of the left-right scale by using Spearman's rank correlation coefficients rho, and by calculating a multiple linear regression equation with the LR scale as the dependent variable. Using regression methods seems unproblematic because the (approximately normally distributed) Left-Right scale may well be treated as an interval variable, and because the method is known to be quite robust when nonparametric predictors are included (Labowitz 1970; Anderson 1984). Additional data used in this study are taken from Federal census data of 1990 and 2000, and from several Swiss community surveys conducted in 1988, 1994 and 2005. In these mailed-out surveys, the central civil service officials ("Gemeindeschreiber”) were asked to provide information about the formal political and administrative structure as well a about informal political processes in their community and its relation to the supralocal political levels.
6. Empirical results
6.1 Basic ideological constraints in the total sample
By tackling our first, most basic hypothesis, we explore to what extent Swiss local party sections operate under conditions of horizontal and vertical constraints, whether such constraints are coinciding with notions of "left" and "right", and whether they are equally or unequally pronounced on the left and the right half of the LR scale.
As the most conspicuous result, Table 4 shows that very pronounced constraints exist on the left side of the ideological spectrum. The dense interrelationships between the 15 issue opinions (horizontal constraints) is manifested by a very strong first factor that absorbs 46.7% of the total variance and shows positive factor loadings by all variables included. Such clusterings go along with strong vertical constraints as they are visible in the high correlation of each issue variable with the self placement value on the LR scale and in the high cumulative statistical explanation achieved by the multivariate regression in which all issue variables are included as (mutually independent) predictors (64.5%).
Table 4: Horizontal and vertical ideological constraints in the left-center and the center-right section of the LR scale (total sample of Swiss local parties) (wave 2002).1)
1) All data have been arranged in way that more leftist positions score higher in positive values and positive correlations indicate positive relationships with leftist positions. Thus, the values of the four items that tap rightist issues have been inverted.
On the other hand, no equivalent patterns of rightist ideology can be found. While the MCA procedure produces also a predominant first factor, its explanatory power (25.6%) as well as its scope are on a much lower level. Similarly, vertical couplings to the left-right dimension are so loose that by including all predictors in a linear regression, just 18% of the total variance in LR placement can be explained. Evidently, our data are quite efficient to tap the ideological polarization between leftism and centrist political positions, while they are inadequate to account for differences that set centrist parties apart from the extreme right. As these results conforms with several other studies (mentioned above), they invite the question whether are caused just by a biased selection of political issues, or whether they reflect the fact that "rightism" is more defined by elements of political "style" and behavior rather than by substantive stances toward political issues and goals. However, all speculations that the traditional LR dimensions may be challenged by other (e. g. postmaterialist) dimensions are refuted. While a second sector has been extracted in both scale sections, its profile is very weak and it is neither related to ecologism nor to social liberalism or to nationalist-internationalist aspects as they have been postulated by different scholars (see above section 2).
6.2 The impact of community size and (sub-)urbanization
If it is true that urbanized settings promote higher degrees of ideologization, we should find that horizontal as well as vertical couplings are more dense when a community is either itself urbanized (by having a large number of inhabitants), or when it is located in an urbanized agglomeration rather than in rural areas. This second aspect has been measured by asking respondents how much time is needed to travel from their community to the next larger city (with more than 100 000 inhabitants) by means of public transportation.
Looking first at the impact of community size in the left-center part of the LR scale (Table 5a), it is evident that horizontal as well as vertical couplings are less pronounced in smaller than in middlesized municipalities, and reach maximal values in cities with more than15000 inhabitants. In addition, some shifts in the content of leftist ideology can be observed insofar as "urban leftism" is more tightly associated with the demand for expanding the budget for culture and with negative attitudes toward tax cuts and the expansion of public security measures (while the classical socialist issue of "worker empowerment" loses ground). However, a rather generalized tendency of leftist ideologization seems to go along with increasing city size: encompassing 12 out of 15 items in the case of horizontal coupling and 13 out of 15 issues in the realm of vertical constraints. Inspecting the center-right part of the LR scale (Table 5b), it is evident that rightist ideologies are rather weak in smaller as well as middle-sized communities, while winning considerable ground above the level of 15000. In these larger settings, being "rightist" means particularly: favoring nuclear energy and tax cuts, and opposing public budget expansion, help for asylum seekers as well as subsidized kid daycare facilities and Swiss membership in the EU. As a summary, we may conclude that larger city size promotes ideological structuring on both sides of the LR scale, but somewhat more so in the vertical than in the horizontal dimension.
Table 5: Horizontal and vertical ideological constraints: according to size of community (wave 2002).
a) Left- center section of the LR scale
b) Center-right section of the LR scale
As a next step, we want to explore whether irrespective of community size, ideologization is also affected by the degree to which a community is located in an "urbanized" or "suburbanized" setting. The travelling time needed to reach the next larger city (with more than 100 000 inhabitants) seems to be a fruitful indicator, because it taps not so much the physical distance, but the "social distance" (e. g. relevant for daily work commuting as well as leisure time inter action). In order to exclude artIfact effects arising from the positive correlation between city size and "metropolitan proximity", only middle-sized communities (with populations ranging between 2000 and 5000) are included in the analysis.
As seen from Table 6, vertical ideological constraints are consistently decreasing with increasing metropolitan distance in the left as well as the right half of the ideological spectrum, while horizontal issue clusterings are only affected in the center-right section of the scale. Evidently, effects stemming from urbanization are not restricted to endogenous factors resulting from higher intracommunal politicization, but also from exogenous factors associated with the encompassing urbanized, sub-, peri- or nonurbanized setting in which the community is located.
Table 6: Horizontal and vertical ideological constraints in the center-right section of the LR scale: according travelling distance to the next larger city (over 100 000 inhabitants) (only communities between 2000 and 10000 inhabitants)
6.3 Divergences between cultural regions
Switzerland is a genuine "multicultural" country composed of three regions that don't just differ in language (German, French and Italian), but in a large number of cultural characteristics that also extend to ideological levels. In a previous study (also using data from the nationwide party survey), the author has found that vertical constraints (between specific issue opinions and generalized self placements on the Left-Right dimension) were considerably stronger in German-speaking parts of country than in the two remaining ("Romanic") regions (Geser 1992). Within the theoretical framework presented here, this could well be explained by the fact that German speaking cantons provide their communities with larger autonomy, so that more
Table 7: Horizontal and vertical ideological constraints in the three linguistic regions of Switzerland (only communities between 2000 and 10000 inhabitants).
a) left-center scale section
b) center-right scale section
* Insignificant correlation (p > .10)
genuine "community politics" (instead of mere administrative activities) take place. However, part of the divergences may also be explained by the smaller community size in certain (especially Italian speaking) regions. By focusing on middle-sized communities, it is evident that leftist ideology is most pronounced in the German region and least articulated in the Italian speaking canton (Ticino), with the "Romandie" (= French speaking Western Switzerland) falling inbetween (Table 7a). At the same time, there are considerable divergences on the level of content. In Germanic regions, leftism is predominantly related to issues of fiscal policy and public expenses, while in the Western French cantons, questions related to worker empowerment and nuclear energy (and in Southern parts governmental regulation issues and public security concerns) are of primary importance. Looking at the right scale section (Table 7b), it is evident that while horizontal clustering is not affected by regional culture, vertical constraints are highest in the Romandie and fully inexistent within the Italian speaking population, with German Switzerland occupying an intermediary position. Again, attitudes toward fiscal and welfare policies are most tightly related to rightism in German cantons, while in the French regions, traditional class related issues (worker empowerment and tax-based redistribution policies) are still occupying a crucial place.
6.4 Membership composition
The well-documented empirical regularity that educated people tend to more strict ideological thinking is only partially reproduced. In small communities (with less than 2000 inhabitants), horizontal as well as vertical constraints increase dramatically with rising percentages of highly educated active members, when the total sample is considered. While this regularity may at least partially be caused by the higher educational level in leftist groupings, it does not completely disappear when the ideological direction of the party is controlled. Thus, leftist sections show much higher vertical ideologization when many members possess higher educational degrees. All this contrasts with the conditions in middle-sized communities where overall impacts of education are weaker and more concentrated to the center-right section of the scale (Table 8b).
Apart from formal education, we may speculate that insofar high as political ideologization is a correlate of modern urban society, we will find that it is less pronounced in parties where a large percentage of members stem from “traditional” occupations. For testing this hypothesis, we calculate the percentage of active members who are farmers or self-employed (excluding free professionals who are of course more numerous in modernized settings).
Similar to Table 8, Table 9 shows that such occupational impacts are most pronounced in the smallest communities. Here, trends toward tight ideology are highest in leftist as well as rightist party sections in which the share of traditional occupational strata is insignificant or nil (Table 9a). In middle-sized settings, however, such impacts can only be found in groupings on the center-right section of the LR scale – similar to Table 8a which shows the same asymmetry in the case of education. These findings conform well with the argument that leftist ideologies have always been conceived as a ready-made, highly explicit constructions easy to grasp by low-educated people (e. g. unionized workers), while rightist, conservative ideologies were less formal, so that their grasping and precise definition depends more on factors of individual motivation and skills.
Table 8: Horizontal and vertical ideological constraints in party sections with different percentages of highly educated active members
a) Communities below 2000 inhabitants
b) Communities between 2000 and 10000 inhabitants
Table 9: Horizontal and vertical ideological constraints in party sections with different percent-ages of active members who stem from traditional occupational strata (farmers and self-employed).
a) Communities with less than 2000 inhabitants
a) Communities with 2000-10000 inhabitants
* free professionals excluded
6.5 Complexity of the communal party system
Functionalist theories predict that the salience of the left-right dimension increases with the complexity of the political system, because it provides an easy categorical scheme for comparing and evaluating large numbers of political positions, politicians, regimes or political parties. By synthesizing their issue positions to highly structured bundles labeled as “leftist”, centrist” or “rightist”, even less interested voters can easily make up their mind about their preferences without engaging in cumbersome information gathering and multidimensional evaluations. While the high correlations of ideological constraints with city size may well be partially explained by the fact that larger communities tend to have more local parties, the number of parties may well be a determinative factor independent of size.
As seen from Table 10a and 10b, this hypothesis is only borne out for leftist ideology in smaller communities between 2000 and 5000 inhabitants, while no effects are visible in larger communities and in the right section of the scale. We may speculate that in more sizable communities, ideological tendencies are sufficiently supported by other factors (e. g. the high politicization of community affairs (see 6.6) or the higher impact of supralocal party politics (see 6.8), so that intracommunal factors like the number of local party sections is no longer decisive.
Table 10: Horizontal and vertical ideological constraints of local parties according to the total number of competing local parties in the community
a) Communities with 2000-5000 inhabitants
b) communities with 5000-10000 inhabitants
6.6 Political and apolitical views of community affairs
Especially in smaller settings, community affairs are widely considered as nonpolitical issues to be settled by common sense, technical rationality or by applying supralocal legal rules and administrative procedures: so that there is no place for power play and controversies on the level of values and goals (Geser 2003). In the United States, even larger cities have been affected by the "progressive reform movement" which reinforced nonpartisan managerial conceptions of city government dedicated general "community welfare", not to the interest of particular groupings and electoral clienteles (Kemp 1999). Under such conditions, we expect that the need for ideological structuring is much less than in settings where "real politics" like on cantonal or the national level takes place. For tapping this variable, party heads were asked whether according to their own judgment, community issues were (overall) "nonpolitical questions". Not unexpectedly, respondents who disagreed were more likely to stem from leftist and urban than from rightist and nonurban parties, so that it is indispensable to control these two variables in order to find out whether "community politicization" is an independent causal factor.
Table 11: Horizontal and vertical ideological constraints: according to (non)political interpretation of community issues and LR scale section (only communities between 2000 and 10000 inhabitants)
a) Communities with less than 2000 inhabitants
b) Communities with 2000-5000 inhabitants
c) Communities with 5000-10000 inhabitants
Inspecting Table 11, it is evident that in municipalities of various size, the degree of horizontal ideological structuring tends to be higher when community affairs are defined in political (rather than technical, legal or administrative) terms. These trends are similarly strong in the left and the right section of the LR scale. In the vertical dimension, however, the results are much less consistent, except in the largest size category (5-10000 inhabitants) where the salience of “left” and “right” reaches maximum levels when a highly politicized interpretation of community matters prevails. In addition, Table 11 makes clear that the higher levels of ideology found in leftist parties and sections in larger communities are partially caused by the higher politicization levels of these same groupings, because divergences shrink considerably (especially on the right side), when this variable is controlled.
6.7 Power position of the party within the community
In well established democracies like Switzerland, cynical sayings like "Power corrupts and absolute power corrupts absolutely" (Baron Acton 1887) may only have limited relevance, but they may not be completely mistaken. As in all political systems, parties that without any share in political power have least difficulty in maintaining "ideological purity", because they are not involved in "compromising" decisions and because their responsibility extends just to their members and voter clienteles, not to the community as a whole. By contrast, parties reigning with absolute majority have to shift ideological points of views into the background, because they have to adapt to all sorts of pragmatic considerations (e. g. caused by financial scarcities or organizational deficits) and because they or obliged to focus on the general welfare of the whole population. Therefore, we expect a negative correlation between a party's share of formal power (operationalized as the percentage of seats it holds in the communal executive board) and its degree of ideologization.
Table 12: Horizontal and vertical ideological constraints: according percentage of seats in the communal executive and LR scale section (only communities between 2000 and 10000 inhabitants) 
As seen in Table 12, this prediction is borne out very strongly in the total sample, but much less in the subsamples representing the left and the right section of the scale. Evidently, the total sample effect is mainly caused by the regularity that powerless parties are more likely to be leftist parties - which maintain higher degrees of ideology irrespective of any other conditions. Among rightist parties however, the expected trend seems to hold: sections have lowest levels of horizontal clustering and vertical couplings when they control over 50% (= the absolute majority) of the executive seats.
6.8 Programs of supralocal parties
Swiss local party sections enjoy a rather high autonomy vis-à-vis their supralocal mother parties which is based on formal statutes as well as on the fact that they have to rely basically on their own financial means and organizational capacities. Therefore, it is up to them to what degree they give much weight to the values, goals and programs of the cantonal and national party. Starting again with the premise that political ideologies are mainly generated and maintained on the level of supralocal politics, we expect higher ideological structuring among parties that open themselves up to such influences, instead of cultivating their own local perspectives. As in the case of politicization, it has to be considered that supralocal orientation is generally stronger in the case of leftist and urban parties, so that ideological orientation as well as community size have to be controlled.
Table 13: Horizontal and vertical ideological constraints: according to the importance given to nation al party programs and LR scale section (wave 2002).
Table 14: Horizontal and vertical ideological constraints: according to the importance given to nation al party programs and community size (wave 2002)