Political orientation in youth beyond the left-right divide: testing a three-factor model of political orientation and its relatedness to personal values

ABSTRACT The utility of the left-right dimension of political orientation, including among youth, has been questioned recently. Moreover, a sizeable share of young voters in Slovenia has a hard time describing their political orientation. In present research we thus explored whether a multi-dimensional model of social, economic and international dimension, could be better suited for describing young people's political orientations. In addition, as values precede political stances, we explored the relatedness of personal values with proposed dimensions of political orientation and whether the predictive power of regression models containing control background variables could be improved by their inclusion. We used FES Youth Study Southeast Europe 2018/19 data of 1,015 14-29-year-old Slovenians. The three dimensions exhibited good fit to the data and were relatively independent of the left-right placements, offering valuable additional insight into political thinking of youth. Moreover, the values were predictive of scores on all three dimensions over and beyond the included control variables, and shares of explained variances were higher than those with the left-right dimension. While the study has some limitations, findings suggest that a multi-dimensional approach to political orientation could be beneficial in research dealing with the aspects of the ever-changing political and social landscape.


Introduction
The findings of the FES Youth Studies and other studies, exploring experiences and attitudes of youth in countries of Southeast Europe, show that a sizable number of participants has a hard time placing themselves on the traditional spectre of political orientation between left and right (Kirbiš and Flere 2011;Kirbiš and Zagorc 2014;Naterer et al. 2019). Over 42% of Slovenian participants reported in 2018 that they do not know what left-right divide means or represents, with the percentage growing since previous waves of the study. Perhaps, this could be in part explained by the fact that the political landscape today is more complex and issues that are salient in daily politics overreach a one-dimensional continuum. Indeed, extant research on the political landscape in Slovenia, mapping the political agendas of political parties offers strong support for a three-dimensional model of political landscape (Bakker, Jolly, and Polk 2012). Coupled with the fact that political engagement of youth across various countries is in decline (OECD 2016;Zalc, Becuwe, and Buruian 2019), the inability of youth to place themselves on the political spectrum could point to a dissonance between individual and collective conceptualisations of the political landscape. The resulting 'political confusion' could, in turn, lead to lower rates of engagement in conventional political activities by the upcoming generation of voters and potential worsening of the state of democracy (Parvin 2018). In the present research, we tested whether the proposed three-factor model is better suited than a single left-right dimension for describing the individual conceptualisation of the political landscape of the Slovenian youth. To gather more insight into the validity of the proposed factors, we also explored how they relate to basic human values that represent a foundation of political ideologies (Rokeach 1973;Schwartz, Caprara, and Vecchione 2010), but have previously been only weakly related to the left-right placement in general samples (e.g. Piurko, Schwartz, and Davidov 2011).

Political orientation
Regardless of its dimensionality, the central role of political orientation is that of simplifying the political landscape, enabling individuals to make political decisions (e.g. voting) relatively effortlessly by comparing positions of political candidates or parties' positions with their political orientations (Jost, Federico, and Napier 2009). Political orientation thus serves as kind of a roadmap to the type of society one imagines to be optimal, and could, in that sense, also be linked to personal values (e.g . Feldman 2003;Schwartz, Caprara, and Vecchione 2010).
The left-right (liberal-conservative) political divide is most commonly utilised in research across various academic disciplines. It is also used casually in media and everyday conversations around the globe (Jost, Federico, and Napier 2009). Despite stemming from over 200-year-old division of the French assembly during French Revolution between proponents of change vs those standing for stability, its validity in present, more complex society, has been questioned repeatedly over the past decades, with calls for various multi-dimensional conceptions to take its place (e.g. Ashton et al. 2005;Conover and Feldman 1981;Heath, Evans, and Martin, 1994).
The left-right dimension contains two interrelated concepts, namely change vs. stability and rejection vs. acceptance of inequality (see Jost, Federico, and Napier 2009, for a review). While some researchers advocate that left-right dimension is still the most stable across the literature (e.g. Benoit and Laver 2006;Jost, Federico, and Napier 2009), others found it deviates significantly not only from political agendas of the parties but also the perceptions of individuals (e.g. Feldman and Johnston 2014). A range of multi-dimensional models has thus been proposed. Some decompose the leftright dimension into two orthogonal dimensions, namely liberalism and conservatism (e.g. Conover and Feldman 1981), others on the social and economic dimension of political ideology (e.g. Saucier 2000; Duckitt et al. 2002), while some authors propose more than two dimensions of political ideology (e.g. Sprague 1989;Saucier 2000;Marks et al. 2006).
There are, however, notable differences in political landscapes between countries. This may result in varied meaning of left-right dimension (Huber and Inglehart 1995). Various voting systems entail a vastly different number of active political parties, who may in turn focus on wider or narrower scope of social issues. Due to historical and social factors, some topics might be very salient and highly polarising only in some countries. Accounting for the described differences, we focus in the present research on a three-dimensional model, which has been proposed (and tested) based on the actual political agendas of parties in various European countries, including Slovenia (Bakker, Jolly, and Polk 2012). The model proposes three dimensions: social, economic, and the European dimension.
The social dimension includes different non-economic issues, like the questions related to nature, climate, lifestyle, and community. These non-economic issues are also reflected in various namings of this dimension that can be found in the literature (e.g. material/ post-material, libertarian/authoritarian, green-alternative-libertarian/traditional-authoritarian-nationalist or GAL/TAN). Here, the 'left' side of the spectrum emphasises the role of personal freedom, issues like protection of natural resources, equal rights for women and minorities, tolerance of abortion, euthanasia, and same-sex marriages. In contrast, the 'right' side of the spectrum rejects those ideas, often in the name of social order and stability (Marks et al. 2006). The party positions on such issues are diverse in Slovenia (Bakker, Jolly, and Polk 2012); however, not all the topics mentioned are equally prominent. Stemming from the social reforms after World War II, for example, egalitarian gender role attitudes are strongly present in Slovenia (Bakacsi et al. 2002;Kirbiš and Tavčar Krajnc 2013). On the other hand, in the last decade, there have been fierce debates about the rights of national minorities, immigration, same-sex marriages and family code (the latter two also resulting in national referendums; see Freedomhouse 2021 for yearly reviews of the main socio-political issues in Slovenia).
The second, economic dimension, pertains to the role of the state in the economy. The right emphasises a smaller state's role (i.e. less regulation, more privatisation, lower taxes and overall 'leaner' state). In contrast, the left emphasises a more active role of the state in the economy (e.g. higher spending, more regulation; Marks et al. 2006). While the role of the state in the economy has been gradually diminishing with the shift to market economy in Eastern European countries (Bakacsi et al. 2002), including Slovenia, the division between the economic left and right is still influenced by the strong social welfare system, a legacy of the socialistic past, and political parties rarely advocate for positions on the (far) economic right. Consequently, the positions of parties along this dimension in Slovenia is somewhat less dispersed than with the social dimension (Bakker, Jolly, and Polk 2012), although in recent years, party positions have become more dispersed (Norris and Inglehart 2019).
The third, European dimension, identified in various European countries, pertains to the topic of European integration vs. national sovereignty. There is some evidence that the stance of political parties on this dimension is also linked to domestic policies (Marks et al. 2006;Bakker, Jolly, and Polk 2012). As current politics of the European Union continues to focus, for example, on asylum policy, green policy and strengthening of the power of the EU parliament (all those are in turn also present to some degree in the domestic politics of member states; Hagemann, Hobolt, and Wratil 2017;Hobolt and Rodon 2020), national parties take a stance on European integration not in isolation but based on whether their politigal agenda aligns with said topics. We thus expand the content of European dimension slightly in this article to include topics such as immigration policy and strengthening of the national identity, besides the mere support (or a lack thereof) for the European integration, as both are related to the position of country in the international space. To fit its content, we further decided to rename the European to the 'international dimension'. While the importance of European integration (i.e. euroscepticism) is rarely disputed in Slovenia, the issues about the role and strenght of the military (as well as reintroducing military conscription), strengthening of the national identity and immigration policy have been prominent public topics in the last few years. Still, of the three described dimensions, the international one might be the least salient in Slovenian politics (see also Bakker, Jolly, and Polk 2012).
The three proposed dimensions, however, are not completely orthogonal, but the relationship between them differs significantly between countries. In some, the economic left is related to the social left, while in others, the relationship is inversed (Bakker, Jolly, and Polk 2012). Consequently, some researchers argue that regardless of the (sometimes) modest correlations, the distinction between dimensions is not warranted (e.g. Jost, Federico, and Napier 2009). In Slovenia, however, the placement of political parties along those dimensions was found to be the most distinct among the studied countries (Bakker, Jolly, and Polk 2012), and many young people have trouble placing themselves on the single dimension (Naterer et al. 2019), suggesting the individual political space is indeed more complex and warrants more nuanced approach to its measurement. This is also conceivable in practical terms: while a majority of youth in Slovenia consider it is important that the state provides social and economic security (e.g. tuition-free education, jobs; Naterer et al. 2019), which would be considered a position of the economic left, they might differ significantly in their positions on same-sex marriage or abortion, which have little financial implications. Alternatively, individuals might support the European integration regardless of their position on the equality of income between the rich and the poor as support for European integration is not contingent on either of the two positions.
It is essential, however, to point out that the proposed three-dimensional model has been devised using a top-down approach, looking into political programs of active political parties (i.e. political elites), which may not correspond to perceptions of individual voters (Jost, Federico, and Napier 2009). This is especially problematic with young voters, and youth in general, as they exhibit political preferences that differ from the on-going political discourse of older generations (e.g. Büttner et al. 2019;Pew Research Center 2019). To our knowledge, the three-dimensional model has not yet been tested on young people using a bottom-up approach. In the present research, we thus tested the model by grouping the individual ideological stances on particular topics comprising each of the dimensions, rather than looking into correlations between subjective placements on each of the pre-defined dimensions.
Moreover, the left-right dimension in the one-dimensional model represents an entire spectrum of attitudes, expectations and ideologies, making its meaning subject to individual interpretations, and one's individual position hard to interpret in terms of support for specific policy issues. Various researchers thus report of significant differences in its meaning across countries (Huber and Inglehart 1995;Thorisdottir et al. 2007). The proposed three-dimensional model, on the other hand, includes dimensions that are more clearly defined in terms of underlying policy topics. It leaves less space for individual interpretation but is still general enough to make the dimension readily applicable to a range of countries and political systems, even if some topic comprising a dimension is not salient in each of them (Benoit and Laver 2006). Thus, the multi-dimensional models may be more useful in exploring the interplay of specific worldviews with other constructs of interests in international settings.

Values
One of the constructs most closely linked to political orientation are values, as political orientation can be considered an expression of values and individual's placement on the political continuum as ideological self-identification motivated by one's values (Knutsen 1995). Moreover, as orientations in multi-dimensional models are more specific and the topics they pertain are more interrelated compared to a single leftright dimension, they might themselves be considered a form of values, either instrumental (the ways in people may reach what they desire; Rokeach 1973) or terminal (end states of public policy; Tetlock 2000).
More broadly, values are defined as cognitive representations of desirable goals that are independent of a situation. While specific values vary in importance, they serve as guiding principles for an individual (Schwartz 1992). In other words, they are prescriptive beliefs that guide people's identities, attitudes and behavioural norms (Rokeach 1973).
The key characteristics of values are their reference to desirable goals, their ability to motivate people, and their transcendence of specific actions and situations. In addition, they serve as standards for selection and evaluation of actions, people, and policies, and guide one's actions (Schwartz 2012b). In their capacity to guide people's thinking and decision-making in the political sphere, they are the foundation underlying political orientations (Rokeach 1973;Schwartz, Caprara, and Vecchione 2010). For example, voters tend to vote for parties and candidates who promise to shape society in a way that resonates with their expectations and those who offer to strive for strengthening the values important to them.
In his theory, Schwartz (1992;2012b) has identified ten basic individual human valuesself-direction, stimulation, hedonism, achievement, power, security, conformity, tradition, benevolence, and universalism. Those ten basic values are believed to encompass the full range of motivationally distinct values across cultures and could be perceived as universal, making them the value set of choice in various research endeavours. While each value is distinct in its content, some values share common broad motivational goals, and the pursuit of one value (or group of values) can conflict with the pursuit of another (Schwartz, Caprara, and Vecchione 2010). In his circular model, Schwartz (2012a) thus proposed four groups of values aligned along two bipolar dimensions, namely openness to change vs. conservation, and self-enhancement vs. self-transcendence.
In research linking values with political orientation, however, more specific (culturelevel) values have also been identified and measured, deviating from Schwartz's theory of basic human values. Those include, for example, rule-following or acceptance of inequality (Thorisdottir et al. 2007), traditional family values, equal opportunity, moral tolerance, limited government (Goren 2005), law and order, traditional morality, equality, free enterprise, civil liberties, and blind patriotism (Schwartz, Caprara, and Vecchione 2010). As those are less stable over time and more vulnerable to situational variation than basic values defined by Schwartz, there is also no consensus regarding their number or content, making it hard to define and measure the 'entirety' of political values (Schwartz, Caprara, and Vecchione 2010). Due to its theoretic comprehensiveness, we thus decided to use Schwatz's model of basic human values in this study and draw a clear distinction between the basic values and more specific political orientations for legibility, even if those too share some of the characteristics of values.
The role of values in the political sphere is further supported by empirical evidence: they were found to provide a general structure to political attitudes (Feldman 2003;Schwartz, Caprara, and Vecchione 2010) and guide individuals' voting decisions (i.e. voting for a party, whose political goals are congruent with one's value orientations; Caprara et al. 2006;Piurko, Schwartz, and Davidov 2011). Value priorities were also found linked with relevant political issues, such as acceptance of immigrants, joining humanitarian and environmental organisations, and in predicting political activism in various countries (Schwartz 2007). Values might be critical determinants of voting behaviour (Caprara and Zimbardo 2004;Caprara et al. 2006;Schwartz, Caprara, and Vecchione 2010), as they account for more variance in voting intentions than demographics, as is the case, for example, in Italy (Caprara et al. 2006).
The importance of specific values in predicting political orientation varies significantly. Among the values that are most related to voters' preference are universalism, tradition, security, conformity, and openness (Barnea 2003), but also benevolence and self-direction vs. conformity, security, achievement and power (Caprara et al. 2006). However, specific relations could be stronger when looking into support of specific policies or parties, especially when ideological cues of political actors are more salient (e.g. Barnea and Schwartz 1998). While there is ample evidence linking values to decisionmaking in the realms of politics, the relationship between values and general political orientation is less clear. The variance in left-right political orientation explained by the constellation of ten basic values in previous studies was relatively low, especially in post-socialist societies (Caprara et al. 2017), including Slovenia (Piurko, Schwartz, and Davidov 2011). This could partly be accounted for by the fact that the meaning of left-right distinction differs significantly across societies. The latter could be especially unclear when comparing post-socialistic countries to the ones with a longer democratic history. In post-socialist countries, there is a more widespread public support (e.g. Svallfors 2012) for topics that are polarising in Western democracies (e.g. USA) and perceived as ideologically loaded in terms of neoliberalism vs. socialism (e.g. universal public healthcare). This leaves the continuum between left and right to be filled by individuals themselves with other topics (see also Benoit and Laver 2006), likely those most salient to them at the time.
Another explanation for relatively low prediction power of values in general political orientation scores could also lie in the questionable exhaustiveness of the one-dimensional approach to adequately describe the entire range of political orientations. In the studied demographic group, this 'inexhaustiveness' could well be represented in the aforementioned high percentage of unaligned youth. As different opinion polls show that topics youth today perceive as most important are not necessarily aligned with those of the older individuals, the need to explore the dimensionality of the perceived political landscape is even more pertinent.

Research aims
In the present research, we tested whether the proposed three-dimensional model, while suitable for describing agendas of political parties, could also be applied to young people. They are at a critical point within the political socialisation process, form their worldviews and make their first voting decisions (Walker and Iverson 2016). As we described how values should and, at least to an extent, actually do relate to political orientations, we also tested the validity of the three-dimensional model in this demographic. We examined whether the predictive power of values in explaining political orientations is higher when values serve as predictors in the proposed three-dimensional model than in the onedimensional model.
Two main research aims were formed. First, we wanted to explore the structural validity of the proposed three-factor model by grouping participants' attitudes towards various concrete issues and topics, salient in daily politics. Secondly, we wanted to compare the relationship between values and both traditional left-right and proposed a three-factor model of political orientation. We expected the latter to represent the political perceptions of the participants better. Moreover, we expected that the variance of participants' political orientation measured with the three-factor model, explained by personal values, will be higher, compared to the share of explained variance on the traditional left-right dimension, reported by Piurko, Schwartz, and Davidov (2011).
Moreover, as the proposed three dimensions (economic, social and international) are more specific than the left-right continuum, we expect the correlations between values and each dimension would generally be higher compared with the left-right continuum correlations. Specifically, we expect that the correlations between basic values and leftright placement will be low or insignificant for all values. On the other hand, we expect that both social and economic left will correlate positively with self-transcendence (caring for others by supporting both social rights and economic equality), while support for ideas of the economic right (smaller role of the state in the economy, support for individual initiative) would correlate positively with self-direction values. Finally, we expected that higher scores on the international dimension (favouring the international agenda and immigration while caring less about national identity and the strength of the military) would be negatively related with conservation values (i.e. security).
The article extends the literature in two ways: by testing an alternative model of political orientation that might better explain the worldviews of youth today and by exploring the relations between values and political orientations in a representative sample of Slovenian youth. As previous research on post-socialist societies suggests that values in those societies are relatively independent of the political orientation along the leftright divide in adult samples, we were interested whether the two constructs are also unrelated in the younger generation that will shape the political landscape in the near future, and whether their values could better explain participants' placements in alternative models of political orientation. With values being linked with political behaviour, their relatedness to political orientation could serve as another indicator of the model's validity. At the same time, a better understanding of political orientation in youth could be utilised in promoting their political engagement, which is crucial for their positive development, physical and mental health (Lerner et al. 2005;Evans and Prilleltensky 2007;Ballard, Hoyt, and Pachucki 2019), as well as the general condition of democracy in the society (Parvin 2018).

Materials and methods
Participants and procedure Data used in this article was collected as part of the FES Youth Study Southeast Europe 2018/19, and is publically available for research purposes (Friedrich-Ebert-Stiftung 2019). The stratified quota sample for Slovenian participants, collected in the study, is representative of youth between 14 and 29 years old living in Slovenia (population was stratified by 12 statistical regions and five types of communities). The whole sample was also balanced by gender, age group and attained educational level (see Naterer et al. 2019 for detailed description). Data was collected in 2018 using computer-assisted personal interviews, consisting of oral and written part for questions tapping to more personal topics. In sum, 1201 person was contacted, with a relatively high response rate (84,4%).
The sample used in this article included 1015 participants (51% male), aged between 14 and 29 (Mage = 21,68, SD = 4,47). Regarding their educational level, 28,6% of participants have completed their primary education, 26,4% vocational or technical secondary school, 23,9% high school and 14,6% held a bachelor degree. The percentages of participants with unfinished primary school, masters or doctoral degree were small. Almost a half of participants lived in settlements with less than 2000 inhabitants (48,2%), 17,0% lived in settlements with between two and five thousand inhabitants, and 16,7% in one of the two cities in Slovenia with more than 100 thousand participants.

Instruments and measures
Political orientations General political orientation along the traditional left-right divide was assessed using one question (When people talk about their political beliefs, they mostly speak about left-wing and right-wing. How would you place your views on this scale, generally speaking?). Participants were asked to mark their political beliefs on a semantic differential between 1left and 11right. They were also given an option to choose that they do not know how to place themselves on the given differential.
The participants further rated their agreement with a set of 13 items, describing various governmental priorities (To what extend should the national government focus on the realisation of the following goals?) on a 5-point scale (1not at all, 5very much), and their agreement with three additional items about different economic policies, again stating their agreement on a 5-point scale (1completely disagree, 5completely agree). We assigned each of the 16 items to one of the three proposed dimensions of political orientations (see Table 1), based on dimensions' theoretical descriptions in the literature (Marks et al. 2006;Bakker, Jolly, and Polk 2012). Three of the items (improving the position of young people, fostering population growth, fight against crime and corruption) did not fit within the definition of any proposed dimension and were omitted from further analyses. Two of the items on economic policies were also omitted as one (development of private entrepreneurship) is merely opposite of another item (government ownership of business and industry should be increased). The other (economic growth and development), while pertaining to the economy, is overreaching in terms of tapping into the role of the government in the economy (i.e. people on both sides of the economic dimension spectrum could advocate for economic growth, but the way how to reach it may differ). On the other hand, the item dealing with the reduction of unemployment was retained as it suggests an active role of the state in providing jobs, thus interfering to an extent into the economy. The validation process of the proposed factorial structure of political orientation is further described in Statistical analyses.

Value orientations
Within the FES Youth study, values were measured in several different ways. First, the participants rated their agreement with 20 items, which were then grouped into five basic value orientations (values of political and civic engagement, consumerist values, values of personal success, family values, and values of autonomy and responsibility; Naterer et al. 2019), and with seven items regarding their priorities when choosing a job, which measure values implicitly. However, as the items used were not constructed as a scale from the onset, we decided to also look into individual items that roughly correspond to basic human values as proposed by Schwartz Theory of Basic Values (Schwartz 1992;2012b). While this approach of measuring values deviates from the common practice of using established questionnaires, the use of alternative measures of values it is not unprecedented (e.g. Schwartz 2012b). Specific items that corresponded to basic human values and value dimensions were in our analysis grouped into 4 main motivational clusters of values, as some values were only measured using a single item. Moreover, previous research found that some values are highly correlated and could not be fitted into a single regression model. The items used to measure each value dimension are presented below. Notes: * The scores for international dimension items were reversed in subsequent analyses so higher scores mean more international orientation.
Self-enhancement (power and achievement). According to Schwartz (2012a), power and achievement values share a common motivational emphasis, and were previously found to be highly correlated (e.g. Davidov, Schmidt, and Schwartz 2008). Thus, we grouped them into a single score with five indicators (How important to you is: graduating from university, having a successful career, getting/being wealthy, wearing branded clothes, and How important to you personally is having the feeling of achieving something when it comes to choosing a job).
Self-direction. Self-direction was measured using two items (How important to you is: taking responsibility, being independent). While the value of self-direction fits into the motivational emphasis of openness to change, we present it separately here, as hedonism and stimulation were not measured by any of the items within the FES Youth study framework.
Self-transcendence (universalism and benevolence). As with self-enhancement, universalism and benevolence were grouped into a single score, measured by four items (How important to you is: participating in civic actions, being faithful to partner, being faithful to friends, and How important to you personally is the possibility to do something valuable for society when it comes to choosing a job).
Conservation (tradition, conformity, and safety). The two values were again combined and measured using five items (What young people need most of all is strict discipline by their parents, It is acceptable to cheat on taxes if you have a chance, It is acceptable to use connections to 'get things done', How important to you personally is job security when it comes to choosing a job, and How important for a happy life is living in a good country).

Control variables
In line with previous studies looking into relations between values and political orientation (e.g. Thorisdottir et al. 2007;Piurko, Schwartz, and Davidov 2011;Schwartz 2012b;Caprara et al. 2017), we also collected data on participants' age, gender, attained educational level, socioeconomic status, and type of settlement they live in. The settlements in Slovenia were divided into five groups based on the number of inhabitants: under 2,000 inhabitants, 2,000-5,000, 10,000-20,000, 50,000-100,000 and more than 100,000 inhabitants. The socioeconomic status was measured using a single item (Which of the following descriptions most adequately describes the financial situation in your household), where participants chose between five options (1 -We don't have enough money for basic bills, 5 -We can afford to buy whatever we need for a good living standard). They also had an option of choosing the option 'I don't know' (3,4% of participants chose that option).

Statistical analyses
First, the items on political worldviews and importance of various socio-political issues were divided into groups representing three dimensions of political orientation described in extant literature (Marks et al. 2006;Bakker, Jolly, and Polk 2012). Confirmatory factor analysis was conducted using R package lavaan (Rosseel 2012) to test whether the tree factor structure fits the data. As data distribution deviated significantly from the normal distribution with most of the items, and some missing values were present, robust maximum likelihood estimator (MLR) was used to assess the model fit without imputing the missing values. Further, correlations between the proposed factors and traditional left-right orientation were computed. While the confirmatory factor analyses were run using the whole sample, the correlations between factors and traditional left-right dimension (as well as subsequent analyses including values) were computed using a reduced sample of participants who reported on their left-right orientation (N = 584), which allowed comparisons of different models. To test whether the proposed three-factor model is suitable for both those who reported their left-right position and those who were unable to do so, we also tested for its measurement invariance across those two groups. In evaluating the fit of the models, we considered a decrease in CFI equal to or greater than .01 and an increase in RMSEA of .15 as indicative of model noninvariance (Chen 2007).
After the three-factor model of political orientations was confirmed, we tested the correlation between values and both traditional left-right political orientation and the three factors. Correlations were calculated on both item and scale levels to offer a more comprehensive insight into relatedness of specific value orientations with political orientations. In contrast, only the scales (motivational dimensions) were entered into regression models. We fitted two multiple regression models (one for the traditional left-right dimension, and another for proposed three-factor structure of political orientations), testing their predictive power in explaining the variance of political orientations above and beyond the control variables.

Dimensionality of political orientations
Confirmatory factor analysis of the initial model (Model 1) exhibited a poor fit to the data (see Table 2). One of the items (Government ownership of business and industry should be increased.) had weak loading on its factor (b = .26), and social and economic factors were highly correlated (r = .86). Modification indices further suggested that one of the items from economic factor (reduction of unemployment) loads higher on social factor, so we tested two alternative models (Model 2.1 and 2.2). In Model 2.1, the item on the reduction of unemployment was added to the social factor, and in Model 2.2 it was dropped from the model.
Both alternative models exhibited good fit to the data (see Table 2); however, we decided that Model 2.2 is more theoretically grounded as reduction of unemployment fits better with the description of economic, rather social dimension which pertains to social freedoms. The three retained factors from Model 2.2 exhibit good internal reliability (between α = .71 and α = .84), and are relatively independent of traditional political orientation along the left-right divide (see Table 3). The model exhibited good fit in both subgroups (those who reported their position on the left-right spectrum and those who did not), with negligible changes after constricting specific parameters, suggesting that the dimensions are invariant of the ability of left-right placement (see Appendix for detailed results).

Values as predictors of political orientations
To collect evidence of model validity, we explored the relatedness of various dimensions of political orientation with personal values of participants. As values were not measured using a standard questionnaire within the FES Youth Study framework, we observed the correlations on two levels: individual items and four motivational value dimensions based on Schwartz theory (see Table 4). Similarly to previous findings with general Slovenian population (Piurko, Schwartz, and Davidov 2011), the traditional left-right dimension of political orientation was mainly unrelated to each value orientation. An exception was a low correlation between right placement and the importance of being wealthy, low negative correlation with the importance of achieving something when choosing a job and conservation values (those who reported of higher justification of using their connections placed themselves slightly more to the right). While the correlations of the three-   dimensional model were not high either, the dimensions related to a range of values and value dimensions.
To test whether value orientations predict political orientations over and beyond control (demographic) variables, we tested two multiple regression models. In each, demographics were entered in step one, and four motivational dimensions of values at step two.
The summary of both models for each dimension of political orientation is presented in Table 5, while the tables with unstandardised regression coefficients for each of the predictors of political orientation are presented in Appendix. In both models, control variables had low predictive power and accounted for between 2% and 6% of the variance in political orientation (the model was not significant for international dimension). Entering values significantly improved predictions for all dimensions of political orientation; however, the share of variance explained by the models remains modest.
Looking into the significance of particular unstandardised estimates for models containing control variables and values, we found that scores on left-right dimension are significantly associated with gender (males reporting their placement slightly more to the right) and conservation values that were slightly higher in those placed more to the right.
For the proposed three-dimensional model, being female and subjective importance of self-transcendence values were linked with higher scores on the social dimension (i.e. social left), while living in a smaller settlement and perceived importance of self-enhancement were predictive of lower scores on the international dimension. Conversely, scores on the economic dimension were significantly predicted by gender, age, and attained educational level (females and older participants were placed slightly more toward the economic left while those with higher education more towards the right). Furthermore, the importance of self-direction was predictive or more right-wing placement (less government intervention in the economy), and the importance of self-transcendence values was predictive of more left-wing placement on the economic dimension.

Discussion
With a relatively high percentage of youth who had problems placing themselves on the traditional spectrum between left and right in previous studies of Slovenian youth (Kirbiš and Flere 2011;Kirbiš and Zagorc 2014;Naterer et al. 2019), its usefulness in describing the political orientation of youth today is questionable at best. Previous findings in an adult sample of Slovenians (Piurko, Schwartz, and Davidov 2011) further point to a low correlation between personal values and political orientations, which goes against the notion of political orientation being a representation of one's values in political settings (Feldman 2003;Caprara et al. 2006;Schwartz, Caprara, and Vecchione 2010). To question whether both issues described could be tackled using a more nuanced approach to political orientation, we tested a three-factor model of political orientation (Marks et al. 2006;Bakker, Jolly, and Polk 2012).
The results show that the proposed model is stable with all the participants (including those who were unable to place themselves on a single dimension), and could be further tested by utilising a wider arrange of included issues and more clearly defining the international dimension. While the scores on economic and social dimensions were correlated positively, the scores on the international dimension were, surprisingly, negatively correlated with both social and economic left (while positively with right orientation along the single dimension). This gave us some thought, as one would expect that the support for social justice, for example, would be correlated negatively with opposing immigration or that strengthening of the military power would be negatively related to the need for economic security. While such results might point to the somewhat generalised nationalistic orientation of the Slovenian youth, other authors found that nationalistic ideas in Slovenian youth are comparably low, while at the same time, youth show high support for the European idea (Lavrič, Tomanović, and Jusić 2019;Naterer et al. 2019). This might suggest that negative correlations between dimensions are instead due to the specific wording of the items and thus a methodological artefact. The participants were not rating whether they support or oppose each of the policy issues, but rather to what extent the government should pursue this issue. Furthermore, the variability of ratings was relatively lowthe participants mainly provided high ratings for all the items (a similar issue was previously also observed in measuring of values; see Schwartz 1994).
Regarding the correlations with the left-right political orientations, the scores on each of the dimensions were relatively independent of scores on the traditional left-right dimension, which points to the fact that the meaning of left and right might indeed be highly individual in the Slovenian context (e.g. some might place themselves according to their economic positions, while others according to their position regarding the role of church, tradition or their support for same-sex marriage). Furthermore, the low correlations also point to the fact that by using each of the models, we gather qualitatively different data. The proposed model is thus not merely an extension of the widely used left-right dimension. Researchers in fields that relate to politics should thus in the future be mindful of how fine-grained information they seek by utilising different models of political orientation (e.g. are they interested in a more general and abstract placement along the left-right divide vs. more fine-grained placement on economic left-right dimension). This is especially important in youth whose understanding of politics and perceptions of political issues are distinct from other age groups (Walker and Iverson 2016;Büttner et al. 2019;Pew Research Center 2019).
Considering our other research interest, linking political orientation to the underlying personal values, even for those participants who did report their stance on the left-right spectrum, their placement was mostly unrelated to their values. This again raises the question of what exactly is the meaning of either left or right on an individual level and how it relates to collective perceptions (see also Jost, Federico, and Napier 2009). Still, the left-right placement was related to conservation values, which fits the traditional division between change and stability (Jost, Federico, and Napier 2009), but this could suggest that left-right continuum is perhaps best suited for describing only this part of political orientation, and not the entirety of worldviews present in young population.
On the other hand, while noting that variance accounted for by personal values is still modest (but higher than for the left-right dimension), scores on all three dimensions were accounted for by value orientation over and beyond the demographic characteristic of the participants. The theoretic linkage between self-transcendence (i.e. universalism and benevolence) and caring for nature, supporting the rights of females, minorities, and the importance of community was also reflected in our regression model. Same values were linked with individuals' placements on the economic scale. Those supporting a stronger role of the state in the economy, which includes providing social and economic security to citizens, reported higher self-transcendence values. In addition, self-direction values were more prominent in those supporting less government involvement in politics, again supporting the theoretical description of the dimension of individuals and companies taking the initiative in economic matters.
For the international dimension, living in a larger settlement was predictive of higher scores on the dimension, perhaps as the youth living in cities as compared to those, living in smaller settlements, come into contacts with a broader range of people and ideas, including from abroad and thus support the international agenda. Moreover, the importance of self-enhancement values was also indicative of lower scores on the international dimension (i.e. less support for internationalisation and more support for strengthening of national identity and military). There is no clear explanation for this relationship, which should be further explored in future research. Perhaps there is a correlation between valuing self-enhancement and strengthening of the national identity or the country in which one resides (e.g. nation-enhancement). Looking into specific items comprising the value scales, one could expect conservation values (namely the value of security) could be an important predictor of lower scores on the international dimension; however, this was not the case. Perhaps this could be attributed to an underrepresentation of this value in our measurement or the fact that striving for security could be achieved by both strengthening the international relations or by strengthening the national defensive systems. Regardless, the international dimension remains the most ambiguous of the three proposed dimensions and could benefit from clearer operationalisation and further inquiry.
In future research, the choice of values used in predicting scores on political dimension scales could be expanded to include also the more specific socio-political values (e.g. Thorisdottir et al. 2007;Schwartz, Caprara, and Vecchione 2010;Naterer et al. 2019) and the culture-level normative values (Schwartz 1999) the shared ideas about what is good, right and desirable in a society in general, not only one's own life, as is the case with personal values. If political orientation is considered an individual adaptation of the ideology and culture-level norms regarding the organisation of the society that people are motivated to construe and follow based on their psychological needs, interests or desires (i.e. the concepts of motivated social cognition and elective affinity; see Jost et al. 2003;Jost 2009), culture-level values could indeed be a much stronger predictor of political orientations when compared to basic (personal) values (Feldman 2003).

Limitations and future directions
Related to the above statement, we note that that the way values were assessed is an important limitation of the present study. The four motivational groups of values were not balanced in terms of items pertaining to each of the subordinate values, significantly lowering the variance of data, and possibly also the relatedness to political orientations. To address this issue, we also computed correlations for each of the used items with all the dimensions of political orientations. Those paint a more complex picture that should be further explored using one of the standard measures of basic personal values (e.g. Portrait Values Questionnaire; Schwartz et al. 2001). Moreover, as reported by, for example, Schwartz, Caprara, and Vecchione (2010), it would make sense to also explore how specific socio-political values are related to political orientations, as some of the basic values (e.g. hedonism or stimulation) have little to no clear representation in actual political interventions.
Still, we believe our results are significant in showing that there is some benefit in measuring political orientation using multiple dimensions or sampling the ideological stance of participants on more concrete issues. Besides the higher share of variance that is explained by participants' personal values, such assessment is also less abstract. This is, of course, beneficial for every participant, but especially crucial in youth, as in this developmental period abstract thinking is not yet fully developed (Yurgelun-Todd 2007). Moreover, youth is still in the process of political socialisation (Walker and Iverson 2016), and is yet to form concrete and lasting representations of where in the political space they standhowever complex it may ultimately be.
The model should, however, be tested across age-groups and cultures as the constellation of specific issues may be different in other demographic groups or social systems, especially those where the divisions in political space are more stable and less complex (e.g. two-party systems, traditional democracies with parties who have decades-long continuity).

Data availability statement
Data used in this research is part of the FES Youth Study SEE 2018/19 and is accessible through the website of Fredric-Ebert-Stiftung (https://www.fes-soe.org/features/youth-studies/).

Compliance with ethical standards
As no new data was collected within the study, the study was granted an exemption from acquiring ethics approval by the institutional research committee of the first author's institution (Ethics Commission of the Faculty of Arts, University of Ljubljana, No. 200-2020).