Young People's Perceptions of Youth Unemployment: Insights From 11 European Countries

Youth unemployment has been an issue in European countries for many years. However, the attention paid to it by policymakers has varied over time, and there are high cross‐country variations in both the size of the phenomenon, representations of it, and policy interventions. This study adds an intra‐country component to the country‐comparative dimension and assesses the factors affecting how young adults perceive youth unemployment. From a theoretical perspective, we postulate that the perception of youth unemployment as an issue depends on both sociotropic and egocentric evaluations. To address these research questions, we analyse data from the Cultural Pathways to Economic Self‐Sufficiency (CUPESSE) dataset, which comprises responses from more than 20,000 young adults (aged 18–35) from 11 European countries (nine European Union member states together with Switzerland and Turkey). The empirical analysis is based on multilevel modelling and reveals that the problem perception varies both across countries and within them following the hypothesised pattern. The findings show that two factors are particularly important for explaining young people’s perception of youth unemployment as a problem: first, whether they experienced their parents being unemployed when growing up, and second, whether their friends are unemployed.


Introduction
Policy studies have offered several insights into the factors that initiate a policy process and shape the interactions between the participating policymakers and the resulting policy design (Howlett & Mukherjee, 2014;Howlett & Rayner, 2014).An analytical lens that has received noticeable attention in recent years is the multiple streams framework, which, inter alia, stresses the importance of problems for policymaking and alludes to the existence of constellations in which solutions look for problems (Kingdon, 2014).The direction of the relationship between problems and policymaking in the multiple streams framework deviates significantly from the classic understanding of policymaking as a process that aims to solve problems.While the classic perspective may initially appear less exciting, it does raise intriguing questions, such as whether it is only factual problems that matter for policymaking or also issues that are merely perceived as problematic.
The latter is especially relevant for determining which stages of the policy process are affected not only by problems but also by the perception thereof.
In this article, we engage with the factors explaining how issues are perceived in order to prepare the ground for studies that incorporate problem perception more systematically into their analysis of policymaking and governance than is currently the case, that is, with the notable exception of the literature on political agendas (Green-Pedersen & Walgrave, 2014).We contend that we need to improve our understanding of how exactly the relationship between the perception of problems and policymaking plays out in varied contexts so that we can better understand how issues are governed.
To this end, we concentrate on an issue that has been a problem of varied extent across the EU member states and its neighbouring states and which affects a particularly vulnerable group: youth unemployment.
Typically, young people living in Southern Europe-i.e., Greece, Italy, Portugal, and Spain-have experienced difficulties in transitioning from education to work (Berlingieri et al., 2014).However, the financial crisis hitting Europe from 2008 onward aggravated the labour market situation of the youth not only in Southern Europe but also in European states with traditionally low unemployment rates (O'Reilly et al., 2015;Tosun et al., 2017).Youth unemployment therefore struck us as an issue that may yield valuable insights into the factors that determine whether it is perceived as problematic.Furthermore, the fact that unemployment refers to an individual state but is meaningful at the societal level qualifies it for gaining insights into whether the problem perception is driven by sociotropic evaluations, egocentric evaluations, or both.Researchers have evaluated this for perceptions of the economic performance of governments (e.g., Lewis-Beck & Stegmaier, 2013) and of the EU (Hooghe & Marks, 2005), but less so for other types of issues.
Given our research interest, this analysis is guided by the following question: Which factors determine the problem perception of youth unemployment among young people in various European countries?
To examine this, we use a dataset produced by the collaborative research project Cultural Pathways to Economic Self-Sufficiency (CUPESSE; Tosun, Arco-Tirado et al., 2019).The dataset is based on a survey administered in 11 European countries and includes measurements of how young people (aged 18-35) perceived youth unemployment in 2016.The multilevel models show that the problem perception varies both across countries and within them, with the place of residence being one of the factors explaining this variation.The analysis shows that variables capturing the egotropic dimension and the sociotropic dimension matter in explaining how young people perceive youth unemployment.
The remainder of this study unfolds as follows.We first present some background information on the issue of youth unemployment in Europe, then turn to the theoretical considerations and hypotheses that guide our analysis.Subsequently, we provide information on the database and the operationalisation of the key variables.
Finally, we present and discuss our empirical findings and offer some concluding remarks.

Setting the Stage for the Analysis
This study strives to contribute to various literature strands but primarily to the one focusing on the governance of the school-to-work transition (STWT).In essence, the STWT is the process that occurs toward the end of adolescence and the beginning of adulthood when individuals make a transition from schooling to employment.These processes are shaped by numerous factors, from individual ones to macro-level factors such as the policy and institutional arrangements in place for governing the STWT; the latter tend to be path-dependent and therefore difficult to modify (Walther, 2006;Simões, Tosun, & Rocca, 2022).Smooth STWTs are typically marked by an awareness of the demands of the labour market, investment in human capital, and arrangements that facilitate the economic self-sufficiency of young people (Brzinsky-Fay, 2014).Chevalier (2016) adds to these the type of youth social citizenship (familialised vs. individualised) and youth economic citizenship (encompassing vs. selective).
The STWT has received increased attention in both academic and public debates since mounting evidence has shown that a smooth STWT is becoming the exception; the majority are characterised by individualisation, fragmentation, and delays (Arnett, 2007).The prolongment of the transition process, the uncertainty associated with it, especially for individuals belonging to vulnerable groups (Masdonati et al., 2022), and the delayed entry into the labour market can have detrimental impacts on individuals but also adversely affect society as a whole (Ralston et al., 2022).It follows that policy interventions are needed to improve the STWT and must comprise measures targeting multiple policy sectors and ones that form multi-actor networks, as these will ensure that the policy measures are properly implemented (Trein & Tosun, 2021).
Evidently, the most important experience hampering a successful STWT concerns unemployment.
As documented in the literature, youth unemployment reached its peak in 2013 when the unemployment ratio for individuals aged 15 to 24 years reached about 10% as an EU average (Tosun, Hörisch, & Marques, 2019).The unemployment ratio is lower than the unemployment rate, which is more frequently reported by Eurostat and other statistical offices.However, unemployment ratios, which are the number of unemployed divided by the number of employed, unemployed, and economically inactive, are considered to provide a more accurate measure in the European context because those not searching for full-time work are included in the denominator, which reflects the situation of students (O'Reilly et al., 2015).
Figure 1 consists of bar graphs for the EU member states as of 1995 (Figure 1a) as well as those that joined the EU after 2004, alongside Switzerland and Turkey (Figure 1b).The latter two countries are included not only because they are covered by the CUPESSE dataset but also because they have close political and economic relationships with the EU.The bar graphs gauge the unemployment ratio for these 30 countries in 2006 (the first year for which data were available for all countries) and 2016 (the year in which the CUPESSE survey was fielded).(2023).
In 2008, the economic and financial crisis erupted, which had the most drastic impact on Greece, pushing the country to the verge of bankruptcy.As Figure 1 shows, the crisis also resulted in a sharp increase in youth unemployment not only in Greece (Bitzenis et al., 2013), but also in other South-European countries which had high youth unemployment ratios even before the crisis: Croatia, Cyprus, Italy, Spain, and Turkey.
In fact, Spain and Cyprus are the two countries that experienced the sharpest increase in youth unemployment.
Thus, one take-away from Figure 1 is that youth unemployment increased especially in countries with high unemployment levels already.
However, youth unemployment also increased in countries where young people were not used to it, such

Hypotheses
The public perception of issues is an element of several theories and empirical analyses in political science.
One piece of literature that has been particularly instructive elaborates on the question of how support for or opposition to European integration is formed.Within this literature, in their influential study, Hooghe and Marks (2005) summarise three lines of reasoning on how attitudes towards the EU are formed.The first one is grounded on economic theory and postulates that individuals evaluate the economic consequences of the EU for themselves and for the groups of which they are part, which then shape their attitudes.The second line of reasoning is based on psychology and contends that individuals' social identity affects their attitudes towards the EU.The third argument put forth by the authors is that political cues mediate the effect of economic evaluations and social identity.
The main takeaway from the study by Hooghe and Marks (2005) for our research question is that the perception of an issue is determined by the perception of its consequences for both individuals (egotropic dimension) and the group of which they are part (sociotropic dimension).The distinction between egotropic and sociotropic concerns is firmly rooted in the literature on political attitudes (Watson et al., 2022) and electoral choices and how they are affected by perceptions of the economy's performance (Lewis-Beck & Stegmaier, 2013), but it is equally applicable to other phenomena where individuals can be expected to form their perception on the basis of personal or collective experience (Shehata, 2021).
Unemployment is a situation which individuals can either experience themselves or indirectly via their family, friends, or community or via reporting in the media.This makes unemployment an ideal candidate for examining whether the perception of youth unemployment as an issue depends on sociotropic, egocentric, or both types of evaluations.
The most obvious, egocentric variable shaping the perception of unemployment as a problem is that the individual concerned is affected by it, which is the relationship that our first hypothesis postulates.Research using survey data has shown that the personal situation affects how individuals perceive welfare policies, for example (Schuck & Shore, 2019), which aligns with our focus on unemployment: H1a: Unemployed individuals are more likely to perceive unemployment as a problem.
In our understanding, the egocentric scope entails the young people's own situation.However, the literature has also stressed the importance of the formative years between childhood and adolescence for the development of (political) attitudes.With regard to the issue of unemployment, research has found, for example, that paternal unemployment has both short-and long-term implications for their children (Rambla & Scandurra, 2021).The evidence for enduring effects suggests that experiences gathered during the formative years may affect how individuals perceive issues (Neundorf et al., 2013).Consequently, we hypothesise that individuals who grew up with unemployed parents are (still) more likely to consider it as a problem: H1b: Individuals who grew up with unemployed family members are more likely to perceive unemployment as a problem.
The distinction between the collective and the personal level is not easy to make and depends on the specific research interests.In the present case, we consider having friends who are unemployed as a factor that is still very close to individuals and could affect them personally.Therefore, we treat the employment situation of the individuals' friends as belonging to the egotropic scope: H1c: Individuals with unemployed friends are more likely to perceive unemployment as a problem.
The sociotropic scope, as we define it, reaches beyond the personal situation or environment and refers to a higher level of aggregation, one which entails that individuals cannot observe issues directly and media information plays an important role in obtaining information (on this, see Shehata, 2021).In light of the variation in the characteristics of labour markets both at the national and sub-national levels, we contend that both levels matter for how individuals perceive problems.
Our conceptualisation of the sociotropic scope also takes into account the argument by Hooghe and Marks (2005) that evaluations of issues which result in attitudes can be subjective as well as objective.Departing from this argument, the following two hypotheses postulate the existence of a connection between subjective and objective assessments of a problem.These hypotheses differentiate between the country-level situation for forming attitudes and the subnational one.Depending on many things, such as whether the individual concerned consumes mostly news media that have country-wide or regional coverage, there could be a difference in which of the two levels is more important for perceiving unemployment as a problem: H2a: The higher the level of unemployment in the country in which individuals live, the higher the likelihood of them perceiving it as a problem.
H2b: The higher the level of unemployment in the subnational entity in which individuals live, the higher the likelihood of them perceiving it as a problem.
The first two sets of hypotheses concentrate on the distinction between egotropic and sociotropic evaluations of issues as put forth by the literature on political attitudes and behaviour.The third hypothesis is motivated by the literature in political science on populism and by studies in economic and political geography.The latter is important since it makes the case that there exist "places that don't matter" (Rodríguez-Pose, 2018) and that the populations of these feel disadvantaged, even if objectively they are not.Research on populism has shown that this sense of receiving little social recognition is one of the reasons why right-wing populist parties have been successful among rural populations in EU member states such as France and the Netherlands (Mamonova & Franquesa, 2020).
One of the reasons why people vote for extremist parties is that they believe to perceive a problem that mainstream parties have not addressed (Havlík & Voda, 2018).Also, employment opportunities are scarcer in rural areas, which is one of the reasons why living in them results in higher migration rates (Weiss et al., 2021).
It is possible, therefore, that the individuals' living circumstances shape their perception of unemployment.
Put differently, individuals living in rural areas may experience more unemployment and therefore indicate it as a problem.However, as argued by the abovementioned literature, it is equally possible that individuals living in rural areas have a biased perception of unemployment due to a general sense of being left behind.
In other words, it is possible that members of rural communities tend to overestimate a problem relative to the objective levels of unemployment.Both theoretical perspectives suggest that the place of residence may matter for how individuals perceive issues related to the country's socio-economic situation, which includes the question of how rampant unemployment is: H3: The higher the degree of ruralisation in areas in which individuals live, the higher the likelihood of them perceiving unemployment as a problem.

Operationalisation of the Key Variables
To test our hypotheses, we use data from the EU-funded research project CUPESSE, which was originally carried out by Tosun, Arco-Tirado et al. ( 2019) and now provides a publicly available dataset (see Tosun et al., 2018).The dataset offers a wide range of indicators in relation to the labour market participation of young Europeans and to related concepts such as their work values (Cemalcilar et al., 2019;Kittel et al., 2019;Kraaykamp et al., 2019;Weiss et al., 2021).
The sample on which our analysis draws includes 20,008 observations.These were collected for respondents aged between 18 and 35 years at the time of the fielding of the survey, between February and April 2016, and cover 11 European countries: Austria, Czech Republic, Denmark, Germany, Greece, Hungary, Italy, Spain, Switzerland, Turkey, and the UK.The country selection reflects important dimensions of economic variation within Europe as well as variation in their political systems and (youth) welfare state arrangements.The latter in particular matters for how young people are affected by unemployment and includes STWT regimes (Chevalier, 2016;Dorsett & Lucchino, 2014;Simões, Tosun, & Rocca, 2022).The observations also comprise responses from the young people's parents, which is the most interesting feature of the CUPESSE dataset.In terms of sample size, the minimum requirement per country was 1,000 for young adult respondents and 500 for parents, with a reasonable proportion of fathers and mothers.
In nine of the 11 countries, the data were collected using an online survey (78.5% of the overall respondents), whereas in Hungary (6.5%) they were collected via computer-assisted personal interviewing and in Turkey via face-to-face interviews using paper and pencil (15.1%).Despite the different survey modes, the sampling frames were consistent.The companies carrying out the surveys were asked to provide a probability sample of individuals aged between 18 and 35 years that was representative of their employment status (employed, self-employed, unemployed, in education/training), NUTS 2 region, age group, education, and migration background/minority group membership.Even though the sampling frames are consistent, the regression models include controls for the three different survey modes used.
The outcome variable in this analysis indicates the extent to which the respondents consider youth unemployment as a major problem in their respective country of residence.Answers range on a five-point scale, from strongly disagree (1) to strongly agree (5).For a description of the outcome variable see Table A1 in the Supplementary Material.The data used for the analysis are weighted.
The first focal explanatory variable concerns the employment status of the respondents in the last month.
The CUPESSE survey operationalised this variable by using a multiple-choice question, resulting in a series of dummy variables, including unemployed, self-employed, in education, doing housework, etc.The next focal explanatory variable is structured similarly to the first one but with the difference that it captures the employment status of the respondents' mother or father when they were 14 years old.The third variable captures the number of friends that the respondents indicated to be unemployed and is measured on a five-point scale ranging from none of them (1) to all of them (5).
Both pieces of information are available for the level of the NUTS 2 regions from the Eurostat (2023) database.
The data for these two variables refer to the year 2016.The date of extraction is important for them since unemployment and the share of young people being classified as NEETs are susceptible to temporal variation, as shown in Figure 1.
To test H3, we employ a variable that differentiates between urban, peri-urban, and rural areas.The classification follows the approach presented by de Beer et al. ( 2014), who developed it in the framework of the EU-funded NEUJOBS project.For the descriptive statistics on the focal explanatory variables see Table A2 in the Supplementary Material.
Controls comprise variables gauging the respondents' socio-economic status, including their gender, age, education, marital status, having children, having caring duties, having had a paid job for more than one year in the past, being able to afford extras for oneself, self-defining as belonging to an ethnic minority, having been born abroad, employment status, personal values and subjective assessments (risk-taking/aversion, social trust, participation in voluntary associations, subjective assessment of personal health), information about parents (education, quality of relationship both with mother and father), the framework of reference (whether friends are in employment, in education, running businesses, or unemployed), survey mode (which differentiates between online, face-to-face interviews with computer-assisted personal interviews, and face-to-face interviews with paper and pencil), and regional and national contextual variables (gross domestic product per capita, in thousands and expressed in purchasing power standards).All controls are described in Table A3 in the Supplementary Material.
The empirical strategy is to estimate trilevel hierarchical models, with respondents nested in NUTS 2 regions, which are in turn nested in countries.Through this estimation approach, we ensure that we obtain unbiased standard errors.The regression models include a large number of covariates, but this poses no problem given the large number of observations, which range between 15,220 and 18,252 despite missing values.
The missing values mostly result from the fact that some of the covariates include information on the respondents' parents or which had to be supplied by the parents directly.Given that information about parents is missing for almost a quarter of the sample, we run models first without and then with such controls to check the robustness of the findings.We denote the latter set of models in Table 1 (and in Table A4 in the Supplementary Material) using "PAR" for "parents" in the name of the models.In addition, we first run the models without interaction effects, then add them step by step.Table 1 shows only the effects of interest for the three sets of hypotheses and without any categorical variables.

Empirical Findings
In the 11 countries under study, opinions on youth unemployment vary (see Table A1 in the Supplementary Material).The countries with the highest score for strongly agreeing that youth unemployment is a major problem are the Mediterranean countries (Greece 84%, Spain 78%, Italy 68%), followed by Turkey (51%) and Hungary (50%).The corresponding figures for agreement on youth unemployment being a major problem are 78% in the UK (56% agree and 22% strongly agree), 77% in Austria (53% agree and 24% strongly agree), 65% in Germany (43% agree and 22% strongly agree), and 59% in the Czech Republic (18% agree and 41% strongly agree).A much milder agreement is to be found in Denmark (13% agree and 37% strongly agree) and Switzerland (39% agree and 12% strongly agree).The countries with a higher proportion of young people who do not regard youth unemployment as a major problem are Switzerland (29%), Denmark (24%), Czech Republic (19%), and Germany (16%).Notably, the countries where Eurostat data indicate a high rate of youth unemployment (Greece, Italy, Spain, and Turkey) are those in which the CUPESSE dataset confirms that young people perceive unemployment as a major national problem.
An empty multilevel model indicates that 21% of the total variation is located across countries, 24% is due to variations at the regional (i.e., NUTS 2) level, and the remainder is located at the individual level.This leaves a lot of explanation to be given at each of these levels and indicates that hierarchical models are adequate for testing our hypotheses (see Table A4 in the Supplementary Material).
With respect to our first set of hypotheses related to the egotropic dimension, we can only partly confirm H1a.The egotropic stance of considering unemployment as more problematic when young people are unemployed has little empirical support.The effect is not significant when not controlling for parental traits and becomes significant at  < 0.10 afterwards.The impact of being unemployed is not significant at  < 0.10, while the effect of being employed increases the odds of considering unemployment as a problem by 1.16, being significant at  < 0.001 in all the models.Unemployed people are likely to show a .07increase on the five-point scale of considering unemployment as a problem.This means that unemployed individuals have a higher likelihood of considering youth unemployment as problematic by 1.75% more on the scale than employed people.This is basically null, so H1a does not hold.However, family (H1b) and friends (H1c) are relevant to young people's perceptions of unemployment.
If the respondent's father was unemployed when the respondent was aged 14, the score of perceiving youth unemployment as problematic is on average 0.24 lower than if the father was employed in the respondent's childhood.This reveals the importance of the father as a model.However, if the mother was unemployed at this time, it increases the perception of youth unemployment as being problematic by 0.12 points.In other words, the father's unemployment effect measures about a 16th of the total range on this problem perception scale, while the mother's effect is roughly two times lower.At the same time, it is noteworthy that in many of these countries (Turkey and the Southern European countries), the family model is one in which the father works and provides the main income, while the mother takes care of the family and children (Chevalier, 2016).This explains the greater influence of the father's unemployment on young people's perceptions that unemployment is a problem.These findings confirm H1b by showing that young people from families where parents have experienced unemployment are more likely to view unemployment as a problem.
The multilevel model also indicates the influence of friends' employment status on young people's perceptions of unemployment.Thus, the more unemployed friends the respondents have, the stronger their perception that youth unemployment is a major problem.This confirms H1c, indicating that young people with unemployed friends are more likely to perceive unemployment as a problem.
A one-point increase in country-level unemployment leads to a significant 0.03 increase on the four-point scale of the dependent variable, that is, 10% of the maximum variation.These results confirm H2a, indicating that the national level of unemployment strongly influences young people's perceptions of unemployment as a problem.
The subnational level is not so important here.The results show that regional unemployment has no effect on perceptions of youth unemployment as being problematic.The result is robust to adding interaction effects.This finding is plausible considering that regional-level variation is not equally high in the countries included in the analysis.Of these, Spain and especially Italy are the ones where youth unemployment has been subject to considerable cross-regional variation (Cefalo et al., 2020).Also, mass media tends to present the youth unemployment situation at the national level, not at the subnational one.Thus, the sociotropic dimension remains important for youth perceptions of unemployment, but largely at just the national level, which supports H2a but rejects H2b.
The rurality of the NUTS 2 region does not significantly influence the outcome variable.The interactions between the rurality and unemployment indicators also have no significant impact on the perceptions of youth unemployment as a problem.Although living in a rural area does influence youth unemployment (Simões, Erdoğan, et al., 2022), it does not influence a young person's perception of this phenomenon.This can also be explained by informal employment of young people in rural areas and the existence of various labour mobility programmes that can stimulate employment (Weiss et al., 2021).Furthermore, we conceptualised rurality at the regional level, not at the individual one.In other words, we know how rural a region is, but not whether the respondent lives in a more rural area.The skewed distribution locates only 10% of the respondents as living in predominantly rural regions (see Section A2 in the Supplementary Material).We conclude that the impact of a region's rurality is not significant, but we are left with the question of whether the actual location of the respondent in a rural area contributes to their assessment of youth unemployment as problematic.−24,029 −24,028 −24,028 −24,025 −20,131 −20,130 −20,131 −20,128 Notes: †  < 0.10, *  < 0.05, **  < 0.01, ***  < 0.001.

Conclusions
Most young Europeans live in consolidated democracies with well-functioning and affluent economies, which give them abundant opportunities for learning and working.However, even if by international standards young people living in Europe benefit from a favourable socio-economic and political context, it does not mean that all of them experience a smooth transition from school to work.Consequently, there is still room for improving the policy instruments implemented to improve the participation of young people in the labour market.However, policymakers act in a setting in which they are faced with more problems than they can possibly address, forcing them to prioritise certain issues over others (Green-Pedersen & Walgrave, 2014).
In this study, we put forth the assumption that policymaking requires problems to be treated as such, which involves the policymakers themselves but also the groups that will be adversely affected if no (further) policy action is taken.Against this background, we assessed which sociotropic and egocentric evaluations induce young people to perceive youth unemployment as a problem.
Our analysis showed it is predominantly the egotropic dimension which influences young people's perceptions of youth unemployment as being a problem.More precisely, we could support H1b, that young people who grew up with unemployed family members are more likely to perceive unemployment as a problem, as well as H1c, which postulated that individuals with unemployed friends are more likely to perceive unemployment as a problem.In contrast, we could not find support for H1a, which postulated that unemployed individuals are more likely to perceive unemployment as a problem.Of course, other individual factors affect this perception as well-gender, age, educational level, marital status, etc.-but the personal experience of having unemployed family members or friends is the most important factor.
With regard to the sociotropic dimension, our analysis revealed that the national level of unemployment is relevant to the perception of youth unemployment as a problem, as postulated by H2a.The influence of the national level in politics is reflected in the presentation of this issue by the media and policymakers as one of national importance.At the same time, the regional dimension of youth unemployment is relatively little observed and/or understood by the population of the countries covered by this study.Consequently, we had to reject H2b, which hypothesised that the higher the level of unemployment in the subnational entity in which individuals live, the higher the likelihood of them perceiving it as a problem.We consider this finding interesting because unemployment at the subnational level should be a more accurate indicator of the likelihood of individuals living in the region concerned experiencing unemployment given that labour markets vary across regions, which is particularly marked in Italy, for example.Nonetheless, the empirical data support the idea that the situation at the national level is more influential in shaping individuals' views.
The third hypothesis was motivated by the literature in political science on populism and by studies in economic and political geography.It postulated that the higher the degree of rurality in areas in which individuals live, the higher the likelihood of perceiving unemployment as a problem.Based on the empirical findings obtained, we had to reject H3.Nonetheless, our findings suggest that people observe problems accurately; for rural areas that tend to have higher unemployment levels, this entails increased efforts by policymakers to demonstrate that they are willing to address this issue.We consider this observation to be politically relevant because if people living in rural areas get discouraged by the labour market situation, this can aggravate their sense of marginalisation, subsequently inducing them, inter alia, to vote for populist or extremist parties, as we have been able to observe in several European countries (Mamonova & Franquesa, 2020).From this perspective, it is plausible and sensible that the EU Commission launched the Rural Pact in 2021 as part of its Long-Term Vision for EU's Rural Areas.In 2023, the corresponding Rural Pact community platform became operational, and now five community groups are working on different types of challenges rural communities face.
While the CUPESSE dataset is a valuable resource, it also comes with limitations, which have determined the scope and ambition of this study.First and foremost, the dataset does not contain any geoinformation, which would have facilitated a more compelling test of whether an individual's place of residence affects how they perceive youth unemployment.Here we could only rely on the rough distinction between urban, peri-urban, and rural areas.Another limitation concerns the current employment status of the young people's parents, which was gathered only for those parents who participated in the survey.Going forward, it would be useful to replicate the analysis using a dataset that contains these two types of information.
Another promising avenue for future research would be to compare the findings derived on the basis of data from 2016 with more recent data that include the period before and after the Covid-19 pandemic, which also had implications for the transition from education to the labour market (e.g., Brazienė et al., 2021).
In this context, it is conceivable that attitude formation works differently when individuals and society are dealing with health-related or other crises.
Therefore, we are aware that this study can only mark the beginning of a literature that investigates in a systematic fashion how individuals form their attitudes towards issues.Nonetheless, we are positive that the empirical findings presented here as well as the discussion of the limitations that characterise this study are promising points of departure for future empirical research.

Table 1 .
Multilevel analysis: Testing the hypotheses.