LITHUANIAN PARLIAMENTARY ELECTIONS : IMPLICIT AND EXPLICIT ATTITUDES OF YOUNG ADULTS TOWARD THE POLITICAL PARTIES

Our study is directed toward the examination of attitudinal grounds of political decision-making among young (≤ 24 years old) Lithuanian voters in the national parliamentary elections. The focus is on their attitudes regarding politically specific evaluative dimensions (competence, honesty and leadership). To estimate the prognostic values of explicit/implicit attitudes toward different Lithuanian political parties, we asked participants to fill out a questionnaire and to perform three sets of ST-IATs respectively. Due to the low voting diversity among participants, an elaborate data analysis could only be conducted regarding two Lithuanian parties: the Liberal Movement (LRLS) and Homeland Union – Lithuanian Christian Democrats (TS-LKD). The results of our study suggest that explicit attitudes are more important than implicit attitudes in predicting voting behavior. We found that positive explicit and positive implicit attitudes toward the LRLS competence and leadership, respectively, predicted the decision to vote for this party (reference group: non-voters). On the same note, our findings suggest that positive explicit and implicit attitudes toward the TS-LKD honesty predicted the decision to vote for this party (reference group: non-voters). Positive


Introduction
Lithuania has one of the lowest voting turnouts in national parliamentary elections over the last two decades (e.g., in 2004, the overall voting turnout was 46.08%, in 2008 -48.95%, in 2012 -52.93% and  in 2016 -50.64%) compared to other European Union countries. 1 The situation is even worse among young (≤ 24 years old) voters with the overall voting turnout of 18.9% in 2012 and 30.5% in 2016.Suchan unpromising record of turnout history in Lithuania is one of the key aspects stimulating research on factors that predict voting behavior.This research trajectory is aptly elaborated by Lithuanian researchers 2 as well as those from other countries. 3Amongst other important factors, the following appear to be studied the most: sociodemographic characteristics, political sophistication, partisanship, values and attitudes.The purpose of the current study is to examine the association between young adults' voting choice and their attitudes toward Lithuanian political parties as a marker of sociopolitical information processing.It stands as a natural continuation of research 1 Eurostat, "Voter Turnout in National and EU Parliamentary Elections," 2016, <http:// ec.europa.eu/eurostat/tgm/table.do?tab=table&plugin=1&language=en&pcode=tsd go310>, 2017 04 25.

2
Žiliukaitė R., "Kas Lietuvoje (ne)balsuoja?Dalyvavimo rinkimuose veiksniai", Ramonaitė A. (ed.), Kaip renkasi Lietuvos rinkėjai?Idėjos, interesai ir įvaizdžiai politikoje, Vilnius: Vilniaus universiteto leidykla, 2014, pp.65-88; Ramonaitė A., "Ar Lietuvos rinkėjas prognozuojamas?",Ramonaitė A. (ed.), Kaip renkasi Lietuvos rinkėjai?Idėjos, interesai ir įvaizdžiai politikoje, Vilnius: Vilniaus universiteto leidykla, 2014, pp.261-278.3   Bühlmann M., Freitag M., "Individual and Contextual Determinants of Electoral Participation," Swiss PoliticalScience Review 12 (4), 2006, pp.13-47.on attitudinal grounds of political decision-making of Lithuanian voters in the national parliamentary elections. 4From the political science point of view, attitudes are usually investigated as directed toward values and (or) value orientations regarding political stance, past and various global issues revolving around political campaigns.For example, Ainė Ramonaitė and Mažvydas Jastramskis5 found that attitudinal dimension explaining the Soviet past was the most powerful predictor of voting behavior in the 2012 Lithuanian parliamentary elections.Other attitudinal dimensions, such as the liberalconservative and economic left-right, had little to no explanatory power due to the low cohesion between attitudes within dimensions.These results pose a natural need to search and explore other attitudinal dimensions that would reflect the cognitive map of the typical Lithuanian voter more accurately.Recently, there have been attempts to undertake this task.For example, by applying the Q-sort method, Ieva Petronytė6 extracted three novel attitudinal dimensions, i.e.: free expression ("laisvos saviraiškos pasakojimas"), firm management ("tvirtos rankos perspektyva") and traditional morality ("tradicinės moralės perspektyva").These and other similar studies treat attitudes as cognitive structures, directed toward values and (or) value orientations, disregarding attitudes that are directed toward political parties' or their members' characteristics.To the best of our knowledge, the latter approach to attitudes has not been studied extensively.Another 4 Ramonaitė A., Jastramskis M., "Vertybės ir įsitikinimų struktūros", Ramonaitė A.  (ed.), Kaip renkasi Lietuvos rinkėjai?Idėjos, interesai ir įvaizdžiai politikoje, Vilnius: Vilniaus universiteto leidykla, 2014, pp.126-144; Ramonaitė A., Žiliukaitė R., "Explaining Partisan Loyalties in Lithuania," Lithuanian Political Science Yearbook 2008, Vilnius: Institute of International Relations and Political Science, Vilnius University, pp.11-31; Petronytė I., "Ideologinės visuomenės politinių nuostatų mozaikos Lietuvoje", Politologija 4 (72), 2013, pp.156-196; Ramonaitė A., "Kas ilgisi sovietmečio?Ekonominio statuso, socialinės aplinkos ir vertybinių nuostatų įtaka sovietinei -antisovietinei skirčiai Lietuvoje", Sociologija: mintis ir veiksmas 2 (33), 2013,  pp.265-285.limitation of all studies in the context of Lithuanian research on attitudes is methodological in nature.The main concern is exclusively an explicit measurement of attitudes applying self-report methods, such as interviews and questionnaires, which induce analytic information processing.Comprehensive empirical evidence in cognitive as well as social psychology suggests that a person can operate with two types of attitudes, i.e.: 1) Explicit attitudes that are formed and activated deliberately under the operation of analytic information processing; 2) Implicit attitudes that are formed and activated involuntarily under the operation of intuitive information processing. 7It is important to note that dual attitudes toward the same phenomenon are not necessarily the same in valence and (or) strength.For example, one can implicitly hold strong negative attitudes but explicitly express weak positive attitudes toward the left-wing parties.While this dichotomous conceptualization of attitudes has not been applied to the context of Lithuanian research on voters' behavior, the last decade marks its extensive application among foreign researchers.The current study is the first attempt to investigate explicit and implicit attitudes and their predictive power to vote for political parties in the Lithuanian parliamentary elections in the context of Lithuanian research on voters' behavior.The focus is on the youngest age group of the Lithuanian electorate, because it is known that these people are most susceptible to political attitude change during their early adult years and susceptibility drops off immediately thereafter. 8Thus, the results of our study would be of practical importance regarding efforts to shape attitudes of young non-voters.
The following sections of the introduction are devoted to present and discuss research on explicit and implicit attitudes outside the context of Lithuanian research, elaborating on three important aspects: 1) A general concept of dual attitudes and its relationship with political behavior; 2) The relationship between dual attitudes and political sophistication; 3) A brief introduction to the method designed to measure implicit attitudes.

Dual Attitudes and Political Behavior
Most recent models of political cognition 9 suggest its dual nature, namely, the way it operates with the help of explicit (conscious, slow, effortful) as well as implicit (often unconscious, fast, effortless) cognitive processes.In fact, authors of different dual-process theories agree on the fact that the operation of human cognition is based on two types of information processing, i.e., analytical (explicit) and intuitive (implicit). 10There is a great deal of evidence supporting the idea that intuitive information processing triggers implicit learning and subsequently forms implicit knowledge and implicit attitudes, 11 whereas analytic information processing triggers explicit learning and forms explicit knowledge and explicit attitudes. 12Thus, most 9 Kim S., Taber C. S., Lodge M., "A Computational Model of the Citizen as Motivated Reasoner: Modeling the Dynamics of the 2000 Presidential Election," Political Behavior 32 (1), 2010, pp.1-28; Erisen C., Lodge M., Taber C. S., "Affective Contagion in Effortful Political Thinking," Political Psychology 35 (2), 2014, pp.187-206.recent attempts to measure analytical and intuitive information processing in the field of political psychology have focused exclusively on explicit and implicit attitudes, respectively.This is only natural, because voting cannot be operationalized in terms of rightness and is usually based on subjective values and attitudes toward certain political parties or political candidates.In other fields, (e.g., medicine, economics or in court of justice), however, the measurement of dual information processing is more likely to be based on the rightness of made decisions, because subjective evaluations have little to no importance.
Political behavior in general and voting in particular are often seen as actions requiring deliberation and analytic reasoning.Therefore, the significance of implicit measures in this field was deemed rather low.However, recent findings suggest the additive, predictive importance of explicit as well as implicit measures, explaining political behavior. 13Explicit attitudes are typically measured with selfreport questionnaires.In turn, implicit attitudes can be measured with various subliminal and sequential priming methods (e.g., the Implicit Association Test, Go/No-Go Association Test, Affect Misattribution Procedure).14One of the most widely used methods in measuring implicit attitudes is the Implicit Association Test (IAT).In their meta- analysis, Greenwald and colleagues 15 found that the IATs perform particularly well for the prediction of political behavior.For example, Malte Friese, Matthias Bluemke and Michaela Wänke 16 found that implicit attitudes improved the predictive power to vote in the 2002 German parliamentary elections when added to explicit measures.Luciano Arcuri and colleagues 17 got similar results showing a significant predictive validity of implicit attitudes among both decided and undecided voters in the 2005 Italian local election in the Veneto region.Consideringonly the decided voters, the tendencies are clearcut: explicit measures predict voting behavior better than implicit attitudes. 18Taken together, most of the evidence suggest that explicit attitudes explain a greater part of a variance in voting behavior. 19owever, all these studies have one important limitation: explicit and implicit attitudes are measured as directed toward parties and their members in general, without specifying particular characteristics of interest.Therefore, results are very broad, in a sense that it is not possible to point out which characteristics of the parties or their members are important tovoters.The current study is focused on explicit and implicit attitudes regarding three evaluative dimensions, i.e., competence, honesty and leadership.These evaluative dimensions were chosen due to the following reasons.First of all, it is agreed that the quality of a politician or a political party can be defined on the basis of its competence and honesty 20 .Secondly, leadership remains one of the most established topics in the field of research regarding candidates' preference 21 and (or) liking. 22

Dual Attitudes and Political Sophistication
Political sophistication is a composite concept that draws on the general concept of expertise.Overtime, there were many attempts to conceptualize political sophistication, 23 all of which describe the same variable, i.e., the quantity and organization of a person's political cognitions, 24 resulting in in-depth political knowledge.
Expertise in a field of medicine or law is closely related to the occupational experience, as building knowledge in these domains requires systematic practice.In a case of a citizen's political expertise, it is more sensible to measure political interest, 25 because it provokes systematic exposure to political information.Scarce evidence on dual attitudes and political sophistication 26 suggests that the use of complex decision rules predict the voting choice of sophisticated subjects better compared to the choices of unsophisticated subjects.Thus, it can be hypothesized that explicit attitudes (the activation of analytic information processing) predict the voting choices of highly sophisticated subjects better compared to low sophisticates.Charles Taber and Marco Steenbergen 27 also found that simple decision rules predict voting choices equally well for sophisticated and unsophisticated subjects.From a standpoint of dual-process theories of higher cognition, it appears as a natural tendency, because the simplest decision-making rules (e.g., heuristics) are schematic structures, developed through the systematic repetition of certain information in an analytic fashion.Building on that, it can be hypothesized that implicit attitudes (the activation of intuitive information processing) predict the voting choices of high and low sophisticates equally well.These questions still remain open within literature; therefore, our study is aimed to fill this gap.

Measuring Implicit Attitudes: The Single-Target Implicit Association Test
The IAT is a reliable, valid and flexible tool to assess implicit attitudes. 28The IAT is designed to measure the strength of associations between a dichotomous target concept (e.g., men vs. women) and its evaluations (e.g., good, bad).It is achieved by asking participants to sort various stimuli related to the target concept into evaluative categories as quickly and accurately as possible by pressing different keys.Although the IAT has a fairly high reliability (Cronbach's α values range from 0.7 to 0.9), 29 the data must be interpreted with caution, as it is not absolute and only reflects a relative comparison between two target categories (e.g., men vs. women, black people vs. white people etc.). 30This shortcoming is important when dealing with implicit measures in a political domain.For example, the evaluation of multiple target concepts in using IAT would require numerous pairwise comparisons.As Bluemke and Friese 31 stated, the evaluation of Germany's complex political structure would need to compare liberals to conservatives, social-democrats, socialists and environmentalists at the same time.This can be applied to Lithuania's multi-party parliamentary system as well.Daniel Wigboldus, Rob Holland and Ad van Knippenberg 32 offered a solution to this problem by removing one of the target categories and keeping two attribute categories, thus creating a Single-Target Implicit Association Test (ST-IAT) variant, which diminishes the arbitrary influence of the contrast concept.Compared to the IAT, the psychometric characteristics of ST-IAT have been studied only fragmentary.Nevertheless, available evidence suggests it has fairly good reliability (Cronbach's α = 0.72), convergent validity (r = 0.34-0.43)and prognostic validity. 33The removal of contrast target category in ST-IAT opens up

32
Wigboldus D. H. J., Holland R. W., van Knippenberg A., "Single Target Implicit  Associations," unpublished manuscript, 2004.new prospects to measure implicit attitudes toward one desired target across several specific dimensions instead of an abstract dichotomous (positive/negative) dimension. 34All the more so, Bluemke  and Friese's (2008) 35 study has shown that serial position effects do not affect the ST-IAT when several target objects are evaluated in succession.To the best of our knowledge, no study to date has tried to undertake this task.
* * * Considering all of the above, the aim of this study is (1) to examine young adults' implicit and explicit attitudes toward the Lithuanian political parties across three evaluative dimensions (i.e., competence, honesty and leadership) by controlling for political sophistication as an expertise marker and (2) to estimate the prognostic values of these attitudes with regard to voting for the Lithuanian political parties on the 2016 Lithuanian parliamentary elections.

Methods
Participants.During the period from 16 September 2016 to 8 October 2016, 91 (23.6% males, 76.4% females) students of Vilnius University participated in this study (M age = 20.9years, SD = 2.22).It is important to note that the sample is biased toward an educated, urban electorate.This bias seems to be inevitable, because according to The Lithuanian Department of Statistics, over 90% of young (from 18 to 24 years old) adults are educated (i.e., have acquired secondary or higher education).This social group, amongst young adults, usually favors right-wing parties, e.g., the Liberal Movement or the Homeland Union -Lithuanian Christian Democrats.Therefore, we expect the sample to be biased toward these political parties as well.

34
Time spent for numerous pairwise comparisons can be traded-off for an in-depth evaluation of one target across several specific evaluative dimensions.
Procedure.At first, participants answered four questions that measured their political interest and six questions about the structure and procedures being carried out in the Lithuanian parliament, which measured political knowledge (for the questionnaire, see Appendix No. 1).Together, these measures indicated participants' level of political sophistication. 36Later on, participants were asked to complete another questionnaire and evaluate six major Lithuanian political parties 37 in regards to their competence, honesty and leadership on a 7-point scale ranging from "1 = poor fit" to "7 = excellent fit," which measured their explicit attitudes toward the political parties across these three evaluative dimensions (for the questionnaire, see Appendix No. 2).After this task, participants completed three sets of six ST-IATs (one set for one evaluative dimension).An example of one set of six consecutive ST-IATs, together with category assignment and stimulus proportions across the task, is presented in Table No. 1.
In each ST-IAT set participants started with a practice block of 12 trials, in which they had to discriminate between positive and negative words (e.g., in a set of ST-IATs measuring implicit attitudes toward the political parties' leadership, words were associated with positive and negative leader attributes).Training trials are intended to make sure that participants clearly understand and are able to discriminate words with different valence.An example of one training trial is presented in Figure No. 1.   36   Each of the four questions that measured political interest had an equivalent weight (i.e., a one-fourth) on the variable's estimate (it ranged from 0 to 0.5).In turn, each of the six questions that measured political knowledge also had an equivalent weight (i.e., a one-sixth) on the variable's estimate (it also ranged from 0 to 0.5).An estimate of political sophistication (it ranged from 0 to 1) was derived by adding the estimates of political interest and knowledge.

37
In total, there were 14 candidate political parties participating in the Lithuanian parliamentary elections of 2016.According to the results of the Lithuanian parliamentary elections from 2012 and the ratings of the second half of the year 2016, only six political parties exceeded the 5% limit of the total votes cast to be eligible for a seat.Afterward, participants completed one block of 30 trials in which positive attributes and stimuli representing target political party were assigned to the left-key (i.e."E") and negative attributes to the right key (i.e., "I").In the third block, this combination was reversed.An example of single trials in initial and reversed blocks are presented in Figure No. 2. It is known that stimuli categorization becomes faster and more accurate when one and the same key is pressed by grouping two cognitively closely related phenomena.Thus, a faster and more accurate categorization of stimuli in the second/third block respectively show positive/negative attitudes toward the target political party.Procedure was repeated with all six political parties summing up to 13 blocks in total.Both evaluative categories were represented by six words (for the list of all evaluative words used in ST-IATs, see Appendices, Table 1a), while political parties were represented by two pictures 38 (party emblem and the picture of the chairman) and one text stimuli (name of first vice chairman/chairwoman) (for pictorial and text stimuli representing political parties, see Appendix 2).Each stimulus was presented at least three times, adding up to 30 trials per one block.Target stimuli, coupled and uncoupled evaluative stimuli occurred in a ratio of 9:9:12 trials, leading to a proportion of left-hand and right-hand responses of 4:3 in one and 3:4 in the other combined block.
ST-IATs were administered using the Inquisit 5 Lab software.Each of the three ST-IAT sets as well as six consecutive ST-IATs in one set were presented in individually randomized order.
Finally, participants were asked to answer a question "Whom will you vote for in the Lithuanian parliamentary elections on 9 October 2016?", which measured voting intention.After the parliamentary elections we contacted all participants and got information about their actual vote choice.Study was carried out in a laboratory in groups of five-seven participants at a time.Overall, participation in a study took about 40 minutes.

Results
Preliminary analyses.In total, 50 measures were assessed, and these can be grouped into the following four categories: 1) Implicit and explicit attitudes toward three evaluative dimensions across six 38 Note: identical pictures of political parties' emblems and chairmen were used as stimuli in the questionnaire and all ST-IATs.political parties; 2) Overall implicit and explicit attitudes toward six political parties (averaged score across evaluative dimensions); 3) Vote choice; 4) Political sophistication.
Data acquired from the ST-IATs were prepared as follows.First of all, to calculate the ST-IAT d-scores, we used an improved scoring algorithm with a built-in error penalty, developed by Anthony Greenwald and colleagues. 39Moreover, trials with response latencies below 300 milliseconds and above 5000 milliseconds were discarded from further analysis.It should also be noted that two participants who had made 20% or more errors in at least one of 12 combined blocks on any set of ST-IAT were omitted.The resulting d-score ranges from -2 to 2, providing information about the direction and strength of the implicit attitude toward political parties across a certain evaluative dimension.
Data on explicit attitudes were standardized by calculating z-scores.The score of political sophistication (range from 0 to 1) was derived by adding up scores of political interest and political knowledge, which were weighed equally before the calculation.
All participants, who had the intention to vote for a certain party or not to vote at all (n = 60), declared that they made the decision in accordance with their intentions on the election day.Therefore, the decided voters' intentions coincided with their actual decisions.Undecided voters (n = 29) indicated whether they voted for a certain party (48.27%) or did not vote at all (51.73%).The vast majority of young adults participating in this study chose to vote for the Liberal Movement and Homeland Union -Lithuanian Christian Democrats (see Figure No. 3).These results are only natural, as these rightwing parties appeal most to the young, educated, urban electorate.Therefore, further data analysis is restricted to these two parties.39   Greenwald A. G., Nosek B. A., Banaji M. R., "Understanding and Using the Implicit Association Test: I.An Improved Scoring Algorithm," Journal of Personality and  Social Psychology 85 (2), 2003, pp.197-216.Explicit as well as implicit attitudes toward political parties overall across all three evaluative dimensions did not predict (p > .05) the intention of the young adults to vote in the 2016 Lithuanian parliamentary elections (see Appendices, Tables Nos.1b and 1c).

Implicit and Explicit Measures and Voting Choice for the Liberal Movement
We begin our analysis by examining attitudes toward the Liberal Movement (LRLS).First of all, we tested whether attitudes toward the LRLS predict the decision to vote for this party (reference group: voters who decided to vote for other parties).As can be seen from Table No. 2, only positive explicit attitudes predicted decision to vote for LRLS.This effect was evident in both conditions, i.e., taking explicit attitudes individually as well as controlling for implicit attitudes.An additional analysis showed that these effects remained the same even when controlling for political sophistication and interaction between explicit attitudes and political sophistication.Moreover, the effect of interaction between explicit attitudes and political sophistication was not statistically significant (see Appendices, Table No. 2a).Finally, we found that when all three evaluative dimensions (explicit attitudes related to competence, honesty and leadership) were taken as separate variables, none of them predicted (p > .05) the decision to vote for this party.We continue our analysis by examining whether attitudes toward LRLS predict the decision to vote for this party (reference group: voters who decided not to vote).As can be seen from Table No. 3, positive explicit attitudes predicted the decision to vote for LRLS.This effect was evident in both conditions, i.e., taking explicit attitudes individually as well as controlling for implicit attitudes.An additional analysis showed that these effects remained the same even when controlling for political sophistication and interaction between explicit attitudes and political sophistication.Moreover, the effect of interaction between explicit attitudes and political sophistication was not statistically significant (see Appendices, Table No. 3a).Implicit attitudes also predicted decision to vote for the LRLS but only in a case of not controlling for explicit attitudes (see Table No. 3).An additional analysis showed that this effect disappeared when controlling for political sophistication and interaction between implicit attitudes and political sophistication (see Appendices, Table No. 3a).
Finally, a more detailed analysis across three evaluative attitudinal dimensions yielded interesting results: positive explicit attitudes toward the LRLS competence and positive implicit attitudes toward the LRLS leadership predicted the decision to vote for this party.

Implicit and Explicit Measures and Voting Choice for the Homeland Union -Lithuanian Christian Democrats
We continue our analysis by examining attitudes toward the Homeland Union -Lithuanian Christian Democrats (TS-LKD).First of all, we tested whether attitudes toward TS-LKD predict the decision to vote for this party (reference group: voters who decided to vote for other parties).As can be seen from Table No. 4, positive explicit attitudes predicted the decision to vote for the TS-LKD.This effect was evident in both conditions, i.e., taking explicit attitudes individually as well as controlling for implicit attitudes.An additional analysis showed that only the former effect remained the same when controlling for political sophistication and interaction between explicit attitudes and political sophistication.In this case, the effect of interaction between explicit attitudes and political sophistication was not statistically significant (see Appendices, Table No. 4a).Implicit attitudes also predicted the decision to vote for the TS-LKD, but only in a case of not controlling for explicit attitudes (see Table No. 4).An additional analysis showed that this effect disappeared when controlling for political sophistication and interaction between implicit attitudes and political sophistication (see Appendices, Table No. 4a).Finally, a more detailed analysis across three evaluative attitudinal dimensions yielded interesting results: positive explicit attitudes toward the competence of TS-LKD predicted the decision to vote for this party.
We continue our analysis by examining whether attitudes toward TS-LKD predict the decision to vote for this party (reference group: voters who decided not to vote).As can be seen from Table No. 5, positive explicit attitudes predicted the decision to vote for the TS-LKD.This effect was evident in both conditions, i.e., taking explicit attitudes individually as well as controlling for implicit atti- tudes.An additional analysis showed that these effects remained the same even when controlling for political sophistication and interaction between explicit attitudes and political sophistication.Moreover, the effect of interaction between explicit attitudes and political sophistication was statistically significant (see Appendices, Table No.  5a).The beta weight of the interaction variable is negative, indicating the increase of explicit attitudes' prognostic power on the decision to vote for the TS-LKD with decreasing political sophistication.Implicit attitudes also predicted the decision to vote for the TS-LKD but only in a case of not controlling for explicit attitudes (see Table No. 5).An additional analysis showed that this effect disappeared when controlling for political sophistication and interaction between explicit attitudes and political sophistication (see Appendices, Table No Finally, a more detailed analysis across three evaluative attitudinal dimensions yields interesting results: positive explicit attitudes toward the honesty of TS-LKD and positive implicit attitudes toward the competence and honesty of TS-LKD predicted the decision to vote for this party.

Discussion
The results of our study indicate that, when taken separately, explicit and implicit attitudes predict the decision to vote for both parties of interest.However, implicit attitudes cease to have their significant predictive importance once included in the same model with the explicit attitudes.These findings are consistent with already existing evidence, which suggests that explicit attitudes play a more important role in predicting voting behavior. 40Additionally, a closer inspection of beta weights in Tables Nos.1-4 reveals that the prognostic power of attitudes to vote for the parties of interest was higher when target groups were pitted against those who decided not to vote compared to those who decided to vote for other parties.Thus, even though we measured the party-specific attitudes, it is plausible that these attitudes may, at some small level, reflect the overall attitudes across all political parties.This presumption would need further testing by increasing the size of the sample, as it may allow for data collection on the attitudes toward all Lithuanian political parties.
Next, we hypothesized that implicit attitudes predict the voting choices of high and low sophisticates equally well.One would expect this hypothesis to be true, because implicit knowledge, as well as implicit attitudes, are treated as a product of intuitive information processing, 41 while the scarce evidence in literature suggests that simple decision rules predict voting choice equally well for sophisticated and unsophisticated subjects. 42No bold conclusions can be drawn regarding this hypothesis, because implicit attitudes only predicted decision to vote for one party of interest, i.e., the TS-LKD.More importantly, when political sophistication and its interaction with implicit attitudes were included in the model, implicit attitudes no longer predicted the decision to vote for this party.Plausible explanations for these findings may be related to the measurement of political sophistication.For example, one of its aspects -political knowledge -was measured explicitly; thus, it can be speculated that low levels of explicit political knowledge signify intuition-dominating information acquisition, which fosters implicit knowledge and implicit attitudes toward political issues.
We also hypothesized that explicit attitudes predict voting choices of highly sophisticated subjects better compared to low sophisticates.This hypothesis is based on the fact that analytic information processing triggers explicit learning, thus forming explicit knowledge 41 Evans, 2008.and explicit attitudes. 43In addition, Taber and Steenbergen 44 findings suggests that the use of complex decision rules predict the voting choice of sophisticated subjects better compared to the choices of unsophisticated subjects.The results of the current study neither support nor go contrary to our expectations.Regarding the LRLS party, the inclusion of political sophistication and its interaction with explicit attitudes did not yield any additional significant effects.The same pattern is evident regarding the TS-LKD party (reference group: voters for other parties).Interestingly, we found that the inclusion of political sophistication and its interaction with explicit attitudes did have an additional negative effect comparing TS-LKD supporters and non-voters.The increase of explicit attitudes' prognostic power on the decision to vote for the TS-LKD with the decreasing political sophistication can be explained in terms of the override of intuition to the benefit of analytic information processing.Voting comes with a big responsibility; therefore, voting decisions ought to be made with extra care.This is even more true when considering young, inexperienced citizens, who presumably are more likely to become involved in rumination before making a final decision.Thus, rumination may interfere with intuitive processes, as it fosters analytical information processing. 45inally, our findings shed some light on attitude specificity regarding three different evaluative dimensions, i.e., competence, honesty and leadership.To the best of our knowledge, all studies in the field of political psychology 46 to date had measured implicit/explicit attitudes as very general, dichotomous, positive-negative variables. 47e succeeded in determining specific attitudes toward the parties of interest that had a predictive importance to voting for these parties.For example, positive explicit attitudes toward the competence of LRLS and positive implicit attitudes toward the leadership of LRLS predicted the decision to vote for this party (only when compared with voters who decided not to vote).These results may have strong implications for any future campaigns of the Liberal Movement, which should be directed toward the non-voting electorate by taking action in upholding the party's competence and leadership.Considering the Homeland Union -Lithuanian Christian Democrats, both positive explicit and implicit attitudes toward the party's honesty predicted the decision to vote for this party (also only when compared with voters who decided not to vote).Therefore, it can be presumed that those who decided to vote for the TS-LKD and those who decided to vote for any other parties share similar respective attitudes toward the TS-LKD.This premiseis somewhat unorthodox, because it is known that the TS-LKD is viewed quite differently by its partisans and non-supporters (the latter tend to evaluate the party very negatively).A plausible explanation for this rests upon the concept of partisan loyalty -an attitude encompassing a devoted attachment to the political ideals, which is expected to grow with political experience.Ainė Ramonaitė and Rūta Žiliukaitė provide evidence that young (22-25 years old) voters have the lowest level of partisan loyalty (only 5%) compared to other age groups. 48Even though the authors did not provide any information about level of partisan loyalty among young the TS-LKD electorate, it is very likely that the participants in our study were floating voters due to their young 47 Attitudes are treated as phenomena having very general, dichotomous, positivenegative evaluation categories.Moreover, attitudes are measured using stimuli (e.g., words: love, poison, joy, stink), which has little to no semantic connection to politics.age.In that case, our findings might not represent the full picture of young voters' attitudes toward the TS-LKD due to the psychological principle of the susceptibility of the attitudes to change during the years of early adulthood. 49Nevertheless, this explanation need further confirmation by carrying out additional studies controlling for party loyalty.Finally, positive implicit attitudes toward the TS-LKD competence predicted the decision to vote for this party (compared with both voters who decided to vote for other parties and those who decided not to vote).Thus, the Homeland Union -Lithuanian Christian Democrats should reflect upon the possibility of directing their future campaigns in the following two directions: (1) toward the nonvoting electorate by emphasizing the party's honesty and (2) toward all voters by upholding the party's competence.Taken together, these findings provide a meaningful framework for party image formation as well as its further research.
To conclude, the results of our study support the idea that voting behavior can be best predicted with the help of explicit rather than implicit attitudes toward political parties.
Limitations.We would like to comment on the two limitations of the present data.First, all data collection was carried out in the lab, which impeded the opportunity to recruit a larger and more politically diverse sample (there was only enough data for analyzing two political parties out of six that were included in the study).Further research should aim at recruiting participants on the internet, as it is also possible to administer the ST-IATs via the internet network.Second, in our ST-IATs, parties were represented by pictures and names of the most well-known party members.However, we only included the emblems of the political parties and the pictures of their members as representative information in a questionnaire to measure explicit attitudes.The inclusion of names of party members would have contributed to the better conceptual equivalence of the implicit and explicit measures.Finally, the results cannot be generalized neither to the Lithuanian population as a whole, nor to the young voters, as the sample is biased toward the young, educated, right-wing supporters.Nevertheless, it lay ground for the further experimental research in the field of electoral behavior in Lithuania.

Figure No. 1 .
Figure No. 1.An example of one training trial in a set of ST-IATs designed to measure attitudes toward the competence of political parties (Note: correct responses on the trial require one to press the left "E" key).

Figure No. 3 .
Figure No. 3. Number of participants who voted for a certain party or did not vote.0 5 10 15 20 25 30 35

Table No . 1. Category assignment and stimulus proportions across blocks on one set of six consecutive ST-IATs for an exemplary participant. Block Task description Left-key response Right-key response Number of trials Positive Negative Party
Abbreviations: LSDP -the Social Democratic Party of Lithuania, TS-LKDthe Homeland Union -Lithuanian Christian Democrats, TT -Order and Justice, LVZS -the Lithuanian Farmers and Greens Union, LRLS -the Liberal Movement, DP -the Labor Party.

Table No . 1c. Implicit measures. Results of logistic regression: prediction of decision to vote.
Table No. 2a.Results

of multinomial logistic regression. Prediction of voting choice for the LRLS: interaction with political sophistication (reference group: decision to vote for other parties).
Table No. 3a.Results