Political parties in troubled times: economic crisis and voter’s perceptual bias of parties’ ideology in Europe

To what extent did the 2010 economic crisis in Europe accentuate voters’ perceptual bias of parties’ ideological positions? This article investigates the perceptions of voters concerning the left–right positioning of parties by testing the displacement theory effects. Two moments in time are considered: before and after the economic crisis emerged (2009 and 2014), covering a sample of seven European countries (countries most and least affected by the crisis). It is argued displacement effects have increased after the emergence of the crisis, especially right-wing contrast effects in those countries most affected. Analysis using micro-level data from the European Election Studies (2009–2014) and party-level data from the Chapel Hill experts’ survey (2006–2014) supports the study’s primary argument. The findings presented below show general support for these propositions, making an important contribution regarding applying the displacement theory to different economic contexts, and globally to the study of democratic accountability.


Introduction
The recent economic recession in Europe is presumed to have had important functional consequences at different levels of European political systems. For instance, it seems to have increased the risk of a lack of governability and instability in cabinet formation; intensified and strengthened citizens' non-conventional modes of political participation; decreased turnout; augmented the fragmentation and polarization of party systems; and contributed to the growing political distrust and dissatisfaction (Ponticelli and Voth 2011;Bosco and Verney 2012;Blyth 2013;Monastiriotis et al. 2013, p. 356;Rüdig and Karyotis 2013;Teixeira et al. 2015). Of these consequences, those focusing on the relationship between voters and parties are particularly important since that link is at the core of democracy. One dimension of this link that has been neglected among recent research concerns the public's political perceptions and their accuracy.
Citizens' perceptions about the positions of parties and party leaders frequently define the political preferences and electoral choices of citizens (for a literature review, see Fortunato and Stevenson 2013, p. 459). In this regard, perceptions of the left-right shifts of parties have been shown to be related to the left-right shifts of the supporters of these parties (Adams et al. 2011(Adams et al. , 2014. Additionally, as Granberg and Holmberg state, in a "well-functioning democracy, there should be some observable linkages between where parties are located, where people perceive the parties to be, and where people place themselves. This should hold true regardless of whether we are dealing with abstract dimensions, such as left-right" (2002, p. 3). To sum up, how the public form and change their perceptions of parties' ideological positions are crucial to democratic accountability (Endersby and Hinich 1992;Ansolabehere and Jones 2010).
Most of the research on this topic has focused on the test of displacement theory effects. Borrowed from psychology, this theoretical approach basically states that individuals are motivated to agree with liked candidates and to disagree with disliked candidates, and, as a consequence, party supporters tend to pull their party closer to their own position (assimilation effect) and to push opposition parties away from their own position (contrast effect). The conclusions reached so far in testing this theory have been affected by some degree of ambiguity (see Granberg 1987 in particular, but also Listaug et al. 1994;Merrill et al. 2001; an exception to this trend is Drummond 2010). Hoping to add clarification regarding the ambiguity of the findings, the first main goal of this research is to discover whether the perceptions of the ideological positioning of parties by European voters are consistently explained by displacement theory effects, in a comparative, longitudinal and updated setting.
In addition, the occurrence of an economic crisis in Europe since 2010 provides a unique opportunity to measure the economic context influence on perceptual bias. As far as we are aware, the influence of such a context on voters' political perceptions has never been assessed. The second goal posed in this study is, then, to assess to what extent, following the economic crisis, there were changes in voters' patterns of ideological perception of the political parties. In specific, did displacement effects intensify to the left or to the right? And can we observe significant differences in the countries most affected compared to those least affected?
This study focuses on an analysis of seven European party systems involving 36 political parties, contributing to overcome the limitation of previous research that was mostly supported by case studies. As we aim at assessing the influence of the economic crisis, this sample encompasses countries that were most-Portugal, Greece and Ireland-and least-Austria, Germany, Denmark and Sweden-affected by the crisis in Europe. The reasons for choosing this sample of countries are twofold. First, considering the economic and financial indicators (GDP growth rate, unemployment rate, people at risk of poverty or social exclusion, and government gross debt 1 ) Portugal, Greece and Ireland are among the countries where the crisis impacted most severely. Other countries could potentially be added to this group, such as Spain, Italy, Cyprus and several Eastern European countries. However, because of limited data, Spain is only partially dealt with (data at the party level is only available on both dates for the Spanish Socialist Workers' Party and the Popular Party), while Italy and Cyprus could not be included (because, respectively, data are only available at both dates for the Democratic Party, and data on parties' positions are only available for 2014). Moreover, due to historical and cultural differences that make comparisons based on the left-right positioning with other European countries unwise, the countries of Eastern Europe were excluded. Also regarding those economic and financial indicators, the remaining countries in the sample are among the least affected by the crisis-Austria, Germany, Denmark and Sweden-and serve as a comparison with the first group. Although other countries in Europe could as well be included, as many countries in Europe were not significantly affected by the crisis, the requirement of parsimony in such an in-depth analysis recommends the limitation of the sample size. Besides that, the purpose of the comparative analysis is to contrast the countries most affected by the crisis with the least affected, in order to assess its effects, supporting on a similar number of countries in both groups. Since there are only data available for three cases among the countries most affected, adding more cases from the group of the least affected countries would not add much to the purposes of the study. Therefore, we consider that this sample of four countries provides a valid and balanced representation of the countries least affected by the crisis to compare with the most affected.
Second, this sample of countries concomitantly allows conducting the research at the micro and party-level, which is required to test displacement effects in citizens' perceptions of parties. For each country, individual data on parties' perceived left-right position and parties' actual position was required at two moments: before and after the crisis emerged. This is very demanding data that are not available for all European countries limiting the cases (countries and parties) qualified to be included in the study. Our sample is based on countries that fulfill the criterion of being among the most and least affected by the economic crisis, for which both citizens' and parties' data were available at both points, and which have a reasonable number of respondents.
After exploring the main theoretical and empirical contributions to the study of political perceptions and supporting the hypotheses, we will present the data and analytic choices. Following this, we will discuss the comparative analysis of voter perceptions of the left-right positions of parties both before and after the crisis emerged.

Mass-based political perceptions: literature and hypotheses
Partisanship has traditionally been considered a key variable in explaining perceptual bias, being defined as a "perceptual screen through which the individual tends to see what is favorable to his partisan orientation" (Campbell et al. 1960, p. 133; reiterated by many others, e.g.: Bartels 2002;Evans and Andersen 2006;Tilley et al. 2008;Gerber and Huber 2010, p. 153, pp. 155-156). When individuals are asked to position candidates or parties with respect to political matters without having the information necessary to perform that task (which is not unusual; see, for example, Zaller 1992), they will use other available information as a basis for reaching a conclusion (Sniderman et al. 1991, pp. 70-119;Popkin 1994). Political parties play an important role in this process, acting as privileged cue sources for deriving political perceptions (Granberg 1987, pp. 40-41;Woon 2007, pp. 1-2; Ansolabehere and Jones 2010), being therefore a reason for misperceiving and distorting parties' positions.
Supporting this idea, the displacement theory implies that individuals engage in an assimilation process in respect to their candidate's position (they distort their candidate's position in the direction of their own in order to assure consonance); and they engage in a contrast process regarding the position of other candidates (they distort the candidate's position away from their own in order to assure dissonance) (Granberg 1987). Providing evidence of the contrast effect, earlier research has shown that left-wing voters place right-wing parties further to the right than they truly are, while right-wing voters put left-wing parties further to the left than they really are (Granberg 1987;Merrill et al. 2001;Woon 2007). Although research has acknowledged displacement effects are generally to some degree observable and that assimilation tends to receive greater support, findings also tend to show a lack of consistency (Granberg 1987;Gerber and Green 1999;Granberg and Holmberg 2002;Merrill et al. 2001;Bartels 2002). In this vein, various authors found partisanship has produced mixed results with regard to bias perceptions (Granberg 1987;Listaug et al. 1994;Gerber and Green 1999, pp. 203-206;Granberg and Holmberg 2002, pp. 10, 16-17).
Therefore, the first step here is to generally test displacement effects among European parties, which is an original comparative setting regarding studies on the topic. The first hypothesis is thus the following: H1 Comparing to other parties, voters are expected to perceive their own party closer to their own left-right position than the party actually is and, concomitantly, right-wing voters are expected to perceive left-wing parties to the left of their real position; and left-wing voters are expected to perceive right-wing parties to the right of their real position.
Although several studies have leaned on the analysis of perceptions on issues related to the economic performance of parties in government (for example Jerit and Barabas 2012, p. 673;Soroka et al. 2015), little is known about the extent to which the economic context contributes to shaping the political perceptions of voters. The intervention of international lenders, the so-called Troika [the European Commission (EC), the European Central Bank (ECB) and the International Monetary Fund (IMF)], limited party and government responses, undermining their understanding with the voters, to the extent that it is expected that there will be significant changes to the patterns of perception of party ideology in those countries that have been bailed out, and that such changes have increased perceptual bias of parties' positions.
On the one hand, it is reasonable to assume that the process whereby voters' political perceptions are distorted is potentially more salient in adverse circumstances like those prevailing in the bailout countries during this period. Social psychologists, and more recently political scientists, have been arguing that the potential for perceptual bias is enhanced in a context of ambiguous information (see, e.g., Fiske and Taylor 2013, Chaps. 7-8;Maravall 2013, pp. 28-40), which is supposed to have increased in a context of crisis. In this regard, we already know that the severe worsening of economic conditions and austerity policies led parties to change their previous policy positions, increasing electoral polarization (Bosco and Verney 2012;Magalhães 2014;Tsatsanis et al. 2014) and pushed parties and governments away from their voters and their electoral promises (Teixeira et al. 2015;Freire et al. 2016). As measurement bias present in evaluations by individuals is not constant over time (Duch et al. 2000), and voters alter their perceptions as a result of changes in the behavior of parties and contextual information (Adams et al. 2014), it is to be expected that voters' perceptions of the parties will modify as a result of the new social and economic context. This is likely promoted by the changing and ambiguous information of such a troubled moment, generating propitious conditions to aggravate perceptual bias of voters over parties.
Additionally, such a context, dominated by a climate of increasing polarization of the political debate, is expected to have strengthened voters' feelings of sympathy or antipathy toward political parties. Such feelings are likely to be conditioned by party loyalties (see, e.g., Duch et al. 2000;Gerber and Huber 2010;Tilley and Hobolt 2011), or by ideological attachments (see, e.g., Nyhan and Reifler 2010), that tend to bias political perceptions. This is so because, although some argue that in a context of little political information and involvement citizens are able to take valid decisions, taking information cues, such as their own party's position (Kuklinski and Quirk 2000;Lau and Redlawsk 2001), there can be a misuse of heuristics leading to distorted political judgments (Kuklinski and Quirk 2000). The strength of ideological predispositions in the formation of perceptions is remarkable, likely overcoming the influence of real information. Evidence of this strength is that when presented with corrective information that runs counter their predispositions, individuals with ideological attachments fail to update their beliefs (Nyhan and Reifler 2010). Perceptual bias following the emergence of the crisis is thus expected both toward contrast and assimilation among the most affected countries.
On the other hand, the burst of the economic crisis abruptly increased citizen's distrust toward the national political institutions among the most affected countries. Political parties are considered among the less reliable political institutions after the crisis emerged, with countries such as Portugal, Spain, or Greece exhibiting in 2014 a percentage of citizens admitting to trust parties below 10% (the European average was 20%). 2 This growing distrust combined with the prevalence of low levels of political information among western European citizens (e.g., Zaller 1992) contributes to an environment favorable to contrast effects in perceiving political parties' positions. Mistrust entails disbelieving the message of parties and thus the distortion in perceiving their real political positions is enhanced. Low levels of political information potentially fuel this process of distortion as it is a favorable condition to promote the misuse of heuristics (Lau and Redlawsk 2001). In this vein, bias increase is especially expected among the most afflicted countries in the sense of contrasting perceptions when compared to assimilation, not only explained by the growing popular distrust affecting their political parties, but also by a general dissatisfaction regarding political leaders and institutions (Teixeira et al. 2015). Additionally, given that the policies of austerity are ideologically positioned to the right and that they have predominantly been supported by right-wing parties, the intensification of perceptual bias is in particular expected among the left-wing voters (promoting rightwing parties' bias).
To sum up, the two additional hypotheses are the following: H2 An increase in parties' perception bias is expected in the bailout countries (southern European countries and Ireland) in 2014, in comparison with non-bailout countries (central and northern European countries), and H3 This bias increase is expected to be more important regarding the right-wing (as compared to left-wing) contrast effects.

Data and methods
In exploring voter perceptions of the ideological position of European parties, ideology is assessed using an 11-point left-right scale relying, respectively, on voters' and experts' surveys. Although cross-validation on the use of left-right scales to estimate parties positions using expert surveys or manifestos' data have shown high correlation, attesting both approaches' reliability, using the former seems to present some advantages over the later regarding the validity in the ideological placement of parties (Volkens 2007, pp. 109-112). Some problems of validity may also arise from the comparison of voters and experts on the left-right scale (Lo et al. 2014).
As the left-right scale represents an abstract ideological continuum, the comparison between citizens and elites might be flawed since both actors' interpretations of such a continuum might have different meanings. For instance, voters' self-placement on the left-right scale is affected by a strong variation in the associations that they establish with the abstract concepts of left and right, related to characteristics such as education, potentially biasing the measurement of ideology among voters (Bauer et al. 2017; see Bauer et al. 2017, p. 575 for a literature review on the criticisms about the use of this scale). Therefore, high correlations between elites' and electorates' positions do not necessarily mean greater proximity between them (Powell 2000, p. 94). Additionally, the use of experts' positioning of parties as a proxy of their effective position also raises methodological issues. On the one hand, this option is dependent on the expert's interpretation of the party's position, and his/her perception is not bias-free. Indeed, expert surveys have been shown to be somewhat ideologically biased along the left-right dimension (Curini 2010). Experts' interpretations are not, therefore, synonymous of the parties' real positions. On the other hand, expert's surveys usually are based on low sample sizes, such as Chapel Hill which is used in this research. As a consequence, results should be read carefully, as some level of bias may underlie this measurement.
Notwithstanding these limitations, the literature broadly suggests that left-right dimension captures the comparative ideological position of citizens and elites reasonably well (e.g., Powell 2000, pp. 162-163;McDonald and Budge 2005, pp. 31-38, 228). It has also been shown that voters can place themselves quite accurately both within ideological families and on the left-right scale (Klingemann 1995, p. 192), and also tend to know where to place the major parties on the left-right scale in relation to the others (Thomassen and Schmitt 1999, p. 196). Subsequently, left-right positions have been used in many studies of mass public and political elites (e.g., Granberg 1987;Powell 2000;McDonald and Budge 2005; for a literature review on the validity of the use of left-right positioning, see Van der Brug and Van der Eijk 1999, p. 130). Furthermore, partisan perceptions of the parties' political shifts have been proven to be consistent with expert perceptions of these shifts (Adams et al. 2014). For that reason, the use of experts' interpretation of parties' ideological positions as a proxy of parties' real positions is considered as adequate to this study.
Supporting on previous research (e.g., Granberg 1987;Merrill et al. 2001;Drummond 2010), displacement effects are measured as follows. The assimilation effect corresponds to the difference between party voters' perception of their and the other parties left-right position and their left-right self-positioning. If the perceived distance is substantially smaller regarding their own party (tending to zero) than the other parties, we conclude by the existence of an assimilation effect. The contrast effect corresponds to the difference between voters' average perception of parties In order to broaden the number of cases to be included, parties with 24 or more respondents in at least one of the years studied were considered, for which there was data for both 2009 and 2014. Although the respondents' sample size within each political party is not always as high as desirable, low sample sizes are common in studies of public-elite comparisons.

Perceptions and perception bias
In order to assess how accurate respondents are at placing parties, Figs. 1 and 2 depict the agreement between where voters place their parties on the left-right dimension, and where the parties actually stand in 2009 and 2014. A total of 38 parties, corresponding to all the parties in the seven countries (plus Spain) for which we have data for the 2 years and regarding voters and party positions, are included in this analysis.
We can draw three principle conclusions from the figures. First, and most importantly, the positive and strong linear relationship between voter perceptions of their parties' positions and these parties actual position in both years, meaning that voters are quite accurate in ideologically placing their parties. The second relates to voter recognition of the center-right parties' concentration (especially after the emergence of the crisis), compared to a lesser concentration-even a relative polarization-on the left. This suggests that voters are able to reasonably recognize the ideological position of their parties, even when in 2014 parties became slightly less dispersed across the left-right spectrum (the R 2 increased from 83 percent in 2009 to 88 percent in 2014). The exception to this trend of concentration is the set of left-wing parties that radicalized their ideological positions after (and probably as a consequence of) the bursting of the crisis. That linear relationship is not, however, free from bias. The third and final conclusion is that there is a slight bias to the left and to the right in the perceptions voters have of the parties: in both years the parties on the left were perceived to be slightly less anchored to the left than they actually were, while . Assuming the tendency for party ideological stability over time (Budge 1994), which implies that when changes occur they are gradual, using that formula we expected to reach a close position to the parties' real positions in 2009. As the economic crisis emerged in Greece toward the end of 2009, followed by the burst of the crisis in Ireland, still in 2009, and then in Portugal in the beginning of 2010 (and in Spain afterward), the assessment of parties' positions by experts using data from 2010 is not free from contamination from the effects of the crisis in these countries. We expect, however, that such contamination is low. The full effects of the crisis were only felt after the agreements for external assistance were signed and the subsequent implementation of austerity programs (agreements were signed in May 2010 in Greece, in November 2010 in Ireland, and in May 2011 in Portugal). Considering that the reaction to party changes is not likely to be immediate (as discussed by Fortunato and Stevenson 2013, pp. 460-461, regarding the public in general), we may expect that the contamination of 2010 is not likely to be significant enough to undermine the use of these data. in 2014 there is also a similar bias to the right, suggesting the electorate thought the parties of the right were slightly less anchored to the right than they actually were. Consequently, the electorate's perception in 2014 has reinforced the trend of political parties toward the center.
In sum, the data points to the prevalence of relatively accurate voter perceptions, although there is a slight bias toward strengthening the position of the parties on the center-left (in 2009) and center-right (in 2014). With regard to the accuracy of these perceptions, these results are in line with those reached in other studies, either in Europe (Van der Brug and Van der Eijk 1999), or the USA (Sniderman et al. 1991, pp. 93-119).

Assimilation and contrast effects among European voters
With a view to testing the consistency of the effects of assimilation and contrast in comparative terms at the European level, Figs. 3, 4 and 5 show the percentage of party conformities with these effects. For both effects, the percentage of conformity results from the following calculation: the number of cases in which the effect is verified, divided by the total number of potential cases, multiplied by 100. The  (2009)(2010)(2011)(2012)(2013)(2014) assimilation effect is always observed when party voters perceive their party to be closer to their own ideological position (with a difference between voters' perception and position tending to zero) than its real position, and than that of any other parties. In turn, the contrast effect occurs on the left-wing when right-wing voters distort the position of the parties of the left, pushing them further to the left than their actual position; and it occurs on the right-wing when voters for left-wing parties distort the position of right-wing parties, pushing them further to the right than  (2009)(2010)(2011)(2012)(2013)(2014). Source: EES (2009, 2014) and Chapel Hill Experts Survey (2006, 2014 their actual position (see detailed data per party in each country in Online Appendix  Tables A1-A7).
Consolidating previous research, and according to one of the effects in the first hypothesis, the assimilation effect is generally verified in all of the countries and for almost all of the parties, tending to increase in 2014 (a sum of results is depicted in Fig. 3; see Online Appendix Tables A1-A7 for detailed information on countries and parties). Voters tend to perceive their parties ideologically in a way that is almost a juxtaposition from their own left-right position and, in general terms, that are closer to their own position than the party's actual position (with differences between both that are close to zero). Only occasionally is this tendency not verified, and that is often due to the predisposition of voters to place themselves in positions that are more central than that of their party 4 (see, for example, Converse and Roy 1986, pp. 593-629;Thomassen and Schmitt 1999, pp. 191, 198-199). Figure 3, which shows the percentage of conformities with the assimilation effect, reveals that Austria, Sweden and Denmark had the highest observed percentage of this effect in 2009 (and Ireland the lowest). Five years later, the presence of this effect tended to reinforce or maintain among the European political parties in the sample. Among the bailout countries, Portugal and Greece evidence substantially high increases of the effect, and Ireland exhibits the most important increment in our sample of countries, which suggests that some influence of the crisis on the tendency of voters to assimilate their party's ideological position to theirs might be at play.
In order to measure the increased significance of this effect, we first calculated a correlation between the left-right position of voters in the party and the verification of an assimilation effect (the unit of analysis is in all cases the party). We observed in 2009 that the correlation was positive and of a reasonable magnitude (Pearson's r = 0.37, p < 0.05, n = 36). In 2014 this relationship was substantially strengthened (Pearson's r = 0.53, p < 0.001, n = 35), signifying an intensification of assimilation by the voters of our sample of countries. If we look only at the countries that were hardest hit by the crisis, despite the limitations of such small samples, the effect is surprisingly significant and robust, strengthening in 2014 (in 2009: Pearson's = 0.51, p < 0.074, n = 13; and in 2014: Pearson's r = 0.80, p < 0.001, n = 12). Note that between 2009 and 2014 there was a general shift of voters to the left: only in nine of the 36 parties studied was this shift not observed (see Tables A1-A7 in the Online  Appendix). Such a change may have contributed to an increase in the number of occurrences and the intensity of the assimilation effect, given that we know that the party elites generally tend to be further to the left than their voters (e.g., Thomassen and Schmitt 1999, pp. 192-193).
The expression of the contrast effect is much smaller than the assimilation effect, which supports earlier conclusions (Merrill et al. 2001;Drummond 2010, pp. 717-726). We particularly note a prevalence of the contrast effect among right-wing parties compared to left-wing parties: left-wing voters more often distort the position of right-wing parties (Fig. 4), placing them further to the right than their true position, especially among the European countries most afflicted by the crisis and especially after its emergence (which might have been triggered by a genuine move of right-wing parties to the right, as was the case in Portugal, see Freire et al. 2016). Much less relevant, and opposing these results, contrast effects among left-wing parties (Fig. 5) are more salient for central and northern European countries, being reinforced in 2014.
In Portugal and Greece in 2014 (in the latter a continuation of what had happened in 2009), the contrast effect is omnipresent among left-wing voters toward right-wing parties (Fig. 4), meaning the systematic displacement of the position of right-wing parties to the right. The existence of right-wing governments in both countries that have implemented packages of unpopular austerity policies imposed by the Troika will have contributed to the expression of this effect. Ireland, where the government that took office in 2011 was also headed by a rightwing party, Fine Gael, in alliance with Labour, experienced the greatest rise in the prevalence of this effect, passing from a nonexistent effect in 2009, to a percentage of more than 70% of displacement of Irish right-wing parties in 2014 (which also happens regarding Irish left-wing parties, although to a lesser extent-see Fig. 5).
Irish party system is generally center oriented (the most to the center in our sample of countries), providing only modest ideological differentiation among parties. Voters, on their turn, in a probable reaction to this low differentiation, tend to perceive parties as being centrally placed in ideological terms (Marsh and Mikhaylov 2014). Findings for Irish voters contrast effect in 2009 seem to mirror this tendency (that is also observable in Table A6 in Online Appendix). This seems to have changed in 2014 with the move of the most centrist Irish parties (Labour and Fianna Fail) even further to the center, although voters perceive them as more polarized than they really are. The slight increase in the polarization of the parties in the extremes: Fine Gael moved to the right and Sinn Féin to the left, may also have contributed to the general perception of parties as more polarized in 2014 (see Table A6 in Online Appendix).
On the left-wing parties the contrast effect is less evident and reveals disparate tendencies (Fig. 5). The fact that right-wing parties have been preponderant in government during the years studied may help explain the prevalence of the right-wing contrast effects (of the 56 parties in government in 2009, 44 were on the right, and of the 77 parties in government in 2014, 53 were on the right).
In terms of the intensity of the effect, the correlation between the ideological positioning of left-wing voters and the existence of a contrast effect in the perception of the parties of the right is revealed to be both robust and negative in 2009 (Pearson's r = −0.47, p < 0.01, n = 45), maintaining this correlation, albeit slightly weaker, in 2014 (Pearson's r = −0.30, p < 0.05, n = 44). This supports the prevalence of a contrast effect to the right. In none of the years were the correlations statistically significant among right-wing voters, which confirms the relative unimportance of the contrast effect on the left-wing parties.
In sum, the assimilation effects predominate among European parties (and are clearly greater than the contrast effects), showing an increasing trend in 2014 in the majority of the European parties analyzed, including the countries most afflicted by the crisis. Following the growing trend in Europe, the contrast effect among rightwing parties is intensified in most of the countries in 2014, leading to this effect being extended to all Portuguese and Greek political parties. The contrast effect on the left-wing parties tends to be smaller (or nonexistent) among southern European countries than in the other countries in the sample in 2014.
Hypothesis 1, which aimed at testing the presence of displacement effects among European parties, is confirmed in a general manner in both years. The existence and slight accentuation of assimilation effects across most of the European parties are noticeable, and although contrast effects are not so high and regular, they were also widely noticed. The prevalence of contrast effects of left-wing voters toward rightwing parties is manifest among the most affected countries (having increased or consolidated in 2014), while the contrast effects of right-wing voters toward right-wing parties are as a whole less significant, having grown in the countries less affected by the crisis.
Besides confirming the presence of displacement effects among European parties, data also reveal that such effects are neither homogenous over time nor across countries. Findings suggest that the economic crisis has contributed to consolidate or intensify the contrast effect, especially that of left-wing voters vis-à-vis right-wing parties in those countries in which the Troika intervened (but not in the remaining countries, where a left-wing bias took place, although with much lower expression). A rightward move of right-wing parties as a consequence of the need to implement austerity measures might have contributed to the relevance of this effect. In Portugal, this move was observed in 2012 in the PSD, the leading party in the right-wing coalition, regarding economic policy issues (Freire et al. 2016, pp. 250-251). In the same vein, recent research on Portugal has shown that the government that took office in 2011 was able to use the crisis to benefit its own right-wing agenda, in the sense that the implementation of Troika recommendations corresponded to a great extent to structural reforms that the party leaders had long wanted for the country (Moury and Standring 2017).
These results partially corroborate hypothesis 2 that anticipated an increase in parties' perception bias in the bailout countries in 2014, in comparison with nonbailout countries (the hypothesis is not supported for left-wing parties' contrast effect, and is only partially supported regarding assimilation effect). Hypothesis 3 is fully supported as right-wing parties' contrast effects had a significant increase in 2014 in the bailout countries.

Contrast effects among government and opposition parties
As a validation of the scale of the right-wing perceptual contrast, we now look to the parties in government as compared to the main parties in opposition. Having in mind that the most important parties in the system generate stronger displacement effects (Drummond 2010, pp. 724-726), the fact a party is in government or opposition must enhance the effects of voter displacement. This is expected to be more evident in the context of an economic crisis that particularly impacted mainstream traditional parties (Bosco and Verney 2012). Therefore, looking at these parties should make it possible to highlight the importance of contrast effects among the bailout countries.
The analysis relies in a linear regression analysis in which the dependent variables correspond to the difference between the perception of the left-right position of the parties (in government and opposition) and their actual position (considering the voters of all parties other than the party electorate). Positive values mean individuals perceive government or main opposition parties to be to the right of their real position. The independent variables are: the left-right position of voters for the various parties, with the exception of the party included in the dependent variable, and a dummy for the countries most affected by the crisis. Table 1 shows the importance of the left-right position of party voters in explaining the perceptual contrast in relation to the parties in government and the main party in opposition both before and after the crisis. The negative signs of the correlations signify that the position of party voters on the left-wing is associated with the bias of government parties or in the main opposition party to the right, and vice versa. This trend is noticeable for governments in both years, but not for opposition parties. This finding generally supports that contrast effect prevails in the individuals' perception of the parties of government.
In Portugal, Greece and Ireland, the displacement of party positions is especially significant in 2014, for both the parties in government and the main party in opposition, which confirms our expectations regarding the potential effect of the crisis in the minds of individuals. After the crisis, the governments of these countries and the main party in opposition tend to be viewed as being to the right of their effective ideological position, confirming our previous conclusions.
But are contrast effects especially relevant regarding right-wing governments? Since splitting governments to the left and to the right implies a significant decrease in the samples' (already small) size, we perform an analysis simply focused on the Table 1 Explaining perceptual bias (contrast effect) regarding parties in government and the main party in opposition, OLS Regression Models (dv: voters' perception minus real position). Source: EES (2009EES ( , 2014 and Chapel Hill Experts Survey (2006,2010,2014) Values are beta-regression coefficients (significance in brackets) relationship between the dependent variable-the difference between the perception of the left-right position of the government and its real position-and an independent variable-the distance between the individual's ideological position and government's actual ideological position. We found negative significant correlations for right-wing government in both years, stronger in 2014 (2009: Pearson's r = −0.37, p < 0.014, n = 44; 2014: Pearson's r = −0.55, p < 0.001, n = 55). These correlations generally mean that the more individuals are to the left of governments, the more they perceive them to be positioned to the right, distorting their real position. This corroborates hypothesis 3 that states that right-wing contrast effects are especially expected toward right-wing parties. Negative significant correlations were only found for left-wing governments in 2014, suggesting a left-wing contrast effect, although the size of the sample is too small to allow drawing any solid conclusion.

Conclusions
Democratic theory generally suggests people will vote for candidates or parties with programs that most closely match their policy preferences. The accuracy of voter perceptions about the position of parties is therefore an important requirement for democratic accountability. The general purpose of this research was to assess voter perceptions of the left-right position of parties by testing the assimilation and contrast effects before and after the economic crisis emerged (2009)(2010)(2011)(2012)(2013)(2014). Findings show that the assimilation effect increased significantly after the crisis, even in those countries most affected by it. To a lesser extent, the contrast effect also increased in 2014. It is especially high among left-wing voters in Portugal, Greece and Ireland, whose perceptions pushed right-wing parties to the right of their real position. In 2014, Greece and Portugal were the only countries in which the effect was evident among the entire left-wing electorate, with the greatest increase being witnessed in Ireland. Right-wing voters in those countries do not seem to equally distort left-wing parties' positions: left-wing contrast effects are more relevant in the remaining central and northern European countries, especially in 2014. In line with these results, in Portugal, Greece and Ireland, the displacement of government party positions to the ideological right reaches significance after the crisis confirming the general expectation of the potential effect of the crisis in distorting individual's parties' perceptions. The economic crisis seems to have increased the importance of the right-wing parties' contrast effect, particularly in those countries in which the Troika intervened: a product of the implementation of right-wing policies that call for extensive cuts to social programs and privatizations, and which were mostly supported by right-wing parties. The economic context seems, therefore, to be important for reaching an understanding of bias in the political perceptions of individuals.
More and broader comparative analysis on the topic would enlighten these findings. To better estimate the importance of economic factors shaping the perceptions individuals have of political parties, in future research, it would be especially interesting to understand to what extent the conditions and economic perceptions of individuals affect their political perceptions in a cross-national context.