FIGURE-BASED POLITICS IN LABORER VOTING BEHAVIORS : A Study of Semarang Industrial Laborers In the 2004-2014 Elections

This study is aimed to explore further how the laborers in Semarang determined their political choices in the 2004-2014 presidential and legislative elections. As we know, many studies have been conducted to discuss laborers and politics. But few of them discussed laborer voting behaviors in the election. This study uses a quantitative approach and survey method to interviewed 394 industrial laborers in Semarang city. By using psychological and rational choice models, this study resulted in two major findings. First, the emergence of political consciousness among industrial laborers in the electoral politics. Second, the dominant influence of psychological factors on laborer voting behaviors. From these findings, I argue that the quality of democracy among laborers is relatively good and that figure-based politics, even among laborers, tends to be found wider and seems to be the “new style” of democracy in Indonesia.


Introduction
This study is aimed to explore the laborers' voting behavior in the election.More specifically, this study was designed to investigate more deeply how the laborers determined their political choices in the 2004-2014 legislative and presidential elections.As Hadiz points out, studies of labor have been done quite a lot by experts, such as Ingleson (1986); Shiraishi (1990); Tanter (1990), andDeyo (1989) (Hadiz, 2005).However, most of the studies, are more dominated by the study of labor movements and their advocacy on labor issues.There are only few studies have discussed the laborer political behavior and the elections.
The study conducted by Ford, Caraway, and Nugroho (Caraway, Ford, & Nugroho, 2014) and Savirani (Savirani, 2015) discussed more about the laborers strategy of entering formal political institutions through elections, rather than studies that discussed their voting behavior in elections.Meanwhile, even the study conducted by Sholihin (Solihin, 2009) and the Reynald (Reynald, 2012) were about laborers' voting behavior, but they have not discussed voting behavior in detail and were not fitted with Taruno, Figur-Based Politics in Laborer Voting Behaviors: A Study of Semarang Industrial Laborers In the 2004-2014 Elections multivariate analysis to capture the explanatory effect on the local elections.
Whereas, the study on voting behavior which used multivariate-quantitative analysis as conducted by Gaffar (Gaffar, 1992), Kristiadi (Kristiadi, 1993), and Mujani et al (Mujani, Liddle, & Ambardi, 2012) proved to be more reliable and convincing in explaining the dynamics of voting behavior.
The study about laborers' voting behavior in the election is interesting to be done because it has at least two important reasons.First, labor is one of the marginal groups that is often ignored in the political processes.According to Rueschmeyer, laborers have enormous potential in process of democratization and are the main pro-democracy forces (Asgart, 2009).Second, the number of laborers in Indonesia is overwhelming (around 41 million people), but this group does not have a political channel that represents their interests.The absence of a formal political channel and the polarization of several labor unions leaves a riddle, where did the laborers vote in the election?
As it is known, Indonesian voting behavior is tending to be psychologically instead of sociologically.One of the interesting findings of the survey results conducted by Universitas Gadjah Mada (UGM) and University of Oslo (UiO) in 2013 about Power, Welfare, and Democracy (PWD) was that figure-based politics in Indonesia is getting stronger.Figure-based politics is a type of politics that focuses on figures and tends to ignore organizations (including parties) in mobilizing support (Santoso, et al., 2014).This is marked by the rise of local actors who became leaders and public officials even without strong bases of political party support.
The rise of figure-based politics cannot be separated from the tendency of Indonesian voting behavior which is strongly affected psychologically, especially figure and party identity (Mujani & Liddle, 2007) (Mujani & Liddle, 2010).This was at least seen from Mujani and Liddle's findings in the 1999 and 2004 elections in which figure and party's identity, the two main variables of voting behavior in psychological models, were two important factors in shaping the political preference of Indonesian voter.Hapsari when examined the voters in Jakarta, Bogor, Tangerang, dan Bekasi in 2009 presidential election, also found that the candidate's personality or role of personality has a strong influence in shaping voting behavior (Hapsari, 2010).
There are two other important psychological variables are party identity (party ID) and the role of mass media.Party identity is a psychological situation, a feeling of close, supportive or loyal, or an attempt to identify oneself to a political party.In some countries, such as Germany, the influence of party identity is even greater than the effect of political figures (Kaase, 1994;Rose andSuleiman, 1980 in Mujani andLiddle, 2006).While the role of the mass media, although the influence is not too large, it is still one aspect that also influence one's political choice.According to Barelson et al and Lazarfeld et al, in addition to encouraging individuals to vote (in elections), mass media also play a role in crystallizing and reinforcing someone's choice rather than altering it (Firmanzah, 2008).
In addition to the psychology aspects, the Indonesian voting behavior is also inseparable from the influence of voters' rational perspective.As the selfinterest axiom concept introduced by Down in the rational-choice approach, the voter's rationality assumes that someone's political choice is also based on the assumption of rationality as well as modern economic theory, that is, the calculations that will give him/her the greatest advantage (Dalton & Wattenberg, 1993).In Indonesia, although not a determinant factor, the voter rationality has a significant and positive influence on the party and president choices in the 2004 and 2009 elections (Mujani, Liddle, & Ambardi, 2012).This influence was seen in Megawati's defeat in the 2004 presidential election and SBY's victory in the 2004 elections as well as the 2009 elections.Does the explanation of voting behavior as described above also occur on the laborer?More specifically, what were the dominant factors which influenced a laborer to vote in the 2014 presidential election?Then after knowing the factor, what was that dominant factor and its effect in shaping laborer's political choices in the 2014 presidential election?
For this purpose, the author proposes general hypothesis that among industrial laborers, the character of the presidential candidate, the laborers' party identity, the role of the mass media, and the programs of presidential candidate, also become the important factors in explaining their voting behavior in the 2014 presidential election.In more details, the hypothesis is described as follows; 1.The presidential candidates are the most dominant factor in shaping the political choice of industrial laborers in the 2014 presidential election.2. The laborers' party identity is also a dominant factor in shaping the political choice of industrial laborers in the 2014 presidential election.3. The role of the mass media (the intensity of following the news or political information through the mass media) also influences the political choice of industrial laborers in the 2014 presidential election.4. The programs and political contracts of presidential candidates also influence the political choice of industrial laborers in the 2014 presidential election.
Voting behavior is a complex action, therefore it requires the right approach to investigate.Combining the two approaches as practiced by Gaffar (1992) and Kristiadi (1993) was a manifestation of an effort to understand the dynamics of voter behavior in a better way.Therefore, this study was conducted by using a psychological and a rational choice approach.Psychological approach is used as a framework to see the influence of the presidential candidates' figures, the role of the mass media, and the identity of the party on the voting for presidential candidate.The rational choice approach is used to see how the work program influences the voting for presidential candidates.
Taruno, Figur-Based Politics in Laborer Voting Behaviors: A Study of Semarang Industrial Laborers In the 2004-2014 Elections The voters' behaviors in voting for a presidential candidate are genuinely influenced by numerous factors.The use of both approaches is based on the reality that today's political preferences cannot be separated from the influence of presidential candidates' figures, the role of the media, and the individual party identity.Moreover, in the context of laborer voters, I assumed that the rationality of laborers is one of the factor in determining their political preferences.I assumed the programs of presidential candidates (presumed as the rational considerations) also affect the preference of presidential candidate.That factor did not only influence the vote, but also was the most dominant factor in influencing the political preference of the laborers.

Sampling and Data Collection
This study was conducted using a quantitative approach.The data was collected with a survey method in February-March 2015.The survey research method is a study that takes samples from one population and uses a questionnaire as a basic data collection tool (Singarimbun & Efendi, 1989).The field survey was conducted in several locations of industrial companies spread in Semarang City, Central Java Province.Based on data from Semarang City Manpower Office, the population of industrial companies was 315 companies with 97.723 laborers (35,096 male laborers and 62,627 female laborers) in 2014.Respondents were selected from several industrial companies as samples.
The number of samples weredetermined with the minimum sample size formula (Snedecor & Cochran, 1967); From the calculation of the formula above, obtained a minimum number of samples of 382 laborers.The author fulfilled the total sample to 400 laborers.
Samples was taken using multi stage sampling method, where the first step was to determine the company as the primary sampling unit and the second step selected the respondents randomly from the selected company.The determination of the number of companies was done systematic random sampling and each selected company was given a quota in accordance with the number of laborers in each type of company (micro, small, medium, and big companies).
Determination of respondents at selected companies was done randomly according to the proportion of male and female laborers.
However, due to obstacles in the field, the number of interviewed respondents was only 394 workers.The sample distribution of respondents by type of company can be seen in Table 1.Furthermore, against several workers who were selected as respondents, interviews were conducted using questionnaires.By conducting this interview, the sample is given a list of questions related to their political participation and preferences and the underlying reasons for a laborer to determine their political preference on a political party or presidential candidate.
The data was processed using SPSS (Statistical Package for Social Sciences) version 13.0.The analytical techniques were descriptive and inferential analysis.Descriptive analysis was used to explain the data by tables and graphs more easily, while inferential analysis (logistic regression) was performed to examine the significance of the effect of explanatory variables on the preference of candidates in the 2014 presidential election.Logistic regression analysis is used to explain the relationship between response variables in the form of dichotomic / binary data with explanatory variables of interval and or categorical scale (Hosmer & Lemeshow, 1989).In this study, the response variable used was the preferences of a presidential candidate in 2014 election.
The measurement of variables in this study largely adopted the research instrument ever used by Mujani et al (Mujani, Liddle, & Ambardi, 2012) to measure political participation, political preference, as well as the measurement of other variables with some changes.
The dependent variable used in this research was the participation of industrial laborers in the 2004-2014 elections, that is whether they follow the election or not.Furthermore, if he/she followed the election, then I also asked the preference of political parties or presidential candidates who became their choice in the 2004-2014 elections.This variable, either participation or presidential choice, was measured using a nominal scale.
Meanwhile, the independent variables suspected to influence the voter behavior of industry laborers consist of 4 variables.These four explanatory variables are; the character of the presidential candidate, the laborers' party identity, the role of the mass media (the intensity of following the political news through mass media), and the programs of presidential candidates.Measurements of all independent variables were done by using nominal scale.Examining the influence of figure is done by limiting the dimensions of presidential candidates into 5 important aspects, namely the ability to overcome the problems of the nation, the ability to fulfill the rights of the laborers, the aspect of honesty and the nature of trust, the attention aspect to the laborers, and the favorite aspects.Prospective candidates who get a positive assessment of the workers on the five aspects, then categorized as a figure.
Meanwhile, the examination of the influence of party identity, was based on its proximity to one of the parties supporting the presidential candidate of his preference/choice.Labor party identity for the Prabowo-Hatta voters was marked by its proximity to the Gerindra, Golkar, PAN, PPP, PKS or PBB.On the contrary, for Jokowi-JK voters, it was marked by its proximity to the PDI-P party, Nasdem, PKB, or Hanura.
Furthermore, the examination of the influence of mass media was based on the extent to which an intense laborer follows the news or political information through mass media, whether television, radio, internet, or social media.Categorization of intensity on the variables of this mass media was not only based on access to news or political information, but also the extent to which laborers were interested in news or political information from the mass media.
Finally, examining the effect of the program candidate, based on the extent to which laborers were aware of the work program and or political contract of the presidential candidate.In cases where a labor only knows briefly about a certain candidate's (through very few) programs or political contract of a certain presidential candidate, the labor was considered to know the program candidate or the political contract Table 2. Laborer respondents' profiles based on several variables Source: data processing result

Results and Discussions The Profile of Industrial Laborers
The description of industrial laborers who became the sample can be seen in Table 2. From the 394 respondents interviewed, 36.8 percent (145 laborers) were male and 63.2 percent (249 laborers) were female.The laborer's profile was dominated by those in the range of 20-30 years, married, graduated from high school/equivalent, owning a house, and earning more than the provincial minimum wage.Laborer Participations in the 2004-2014 Elections.
Political participation is one of form of democracy.Moreover, scholars who study political participation, as Verba, Schlozman, and Brady, believe that political participation is the essence of democracy (Mujani, Liddle, & Ambardi, 2012).According to those scholars, the participation of citizens is the heart of democracy.However, among the many forms of political participation, participation in elections is a form of the most elementary political participation in democracy (Mujani, Liddle, & Ambardi, 2012).
In the period of 2004-2014 elections, the level of industrial laborers' participations in the elections in Semarang was quite high.In line with the Mujani's opinion above, where the level of voter turnout is an indicator of the quality of democracy, I found that the quality of democracy among labors in Semarang is quite good.This is reflected in the high level of laborer participation in the last three elections.Gradually, most laborers already engaged actively participate in the elections, both legislative and presidential elections.In fact, when the general voter turnout level has decreased, the voter turnout of laborers in Semarang precisely showed an increasing trend.
Both in legislative and presidential elections, the level of industrial laborers' voter turnout showed an upward trend.It was the opposite of general voter turnout which showed a downward trend.In the 2004 election, where the general voter turnout was very high (up to 84.9 percent), the voter turnout of industrial laborers in Semarang was very low, below 50 percent.addition, they also wanted to elect a better government (26.49percent).This certainly had an enormous meaning for laborers that through the presidential elections, they hoped for a better government, a government who would bring a fundamental change and defend their rights.For industrial laborers, the momentum of the 2014 presidential election was seen as an opportunity to figure out labor issues.Aside from political awareness and their hope for a positive change, the upward trend of laborers' voter turnout was also due to their belief in the importance of the right presidential candidate who could defend their interests.Not a few laborers responded that their reasons for participating in the elections were because they idolized the figures of certain presidential candidate, they wished Indonesia to be led by a strong and firm presidential candidate, and they considered that their chosen presidential candidate's figure was a symbol of change and spirit.In short, the high participation of the industrial laborers in the three last elections, both in the legislative and the presidential elections (mainly in the 2014 election), was a sign of a new political consciousness where elections are believed to have great significances of laborer political struggle in the formal political institutions.

Laborer Political Preference in the 2004-2014 Elections
Reflecting on the voting result, PDIP apparently managed to be the labor party in Semarang.Table 3 shows that the PDIP was the party most widely chosen by the industrial laborers in the 2004 -2014 elections.Interestingly, although the Democratic Party was the winning party in the 2009 general election in Semarang, this party did not get a majority vote among the industrial laborers.PDIP obtained the most vote of the industrial laborers with 47.6 percent.The labor support for PDIP persisted until the 2014 elections, where the dominant vote for PDIP reached 66.8 percent.Was the high achievement of PDIP votes among industrial laborers due to the Jokowi Effect?Based on the data (Table 3), I found that the Jokowi Effect had an influence on the achievement of PDIP votes among industrial laborers.With 66.8 percent, close to the great number of labor support for Jokowi as the presidential candidate (79.29 percent), I believe that the phenomenon of the Jokowi Effect really occurred among industrial laborers, although the effect was not too significant.The influence can be seen from the increasing votes for PDIP among laborers after the decreasing number at previous elections.In the 2004 and 2009 elections, the preference of the political party was not in line with the preference of the president they wanted.From the total votes of laborers who chose PDIP in the 2004 election, about 27.16 percent gave support to SBY-JK of the Democratic Party.Meanwhile, the industrial laborers who consistently supported PDIP candidates (Mega-Hasyim) was only about 72.83 percent.The same thing could be seen in the 2009 election, in which more than a third (35.29 percent) of laborers who chose PDIP also supported SBY-JK compared to those who voted for Mega-Prabowo of PDIP.In short, the symptoms of split-ticket voting (-split ticket voting is a symptom of non-linear support between the elite decision of a political party and its electorate.For example, in the presidential election, party A constituents actually vote for presidential candidates promoted by party B. The emergence of the symptoms of split ticket voting one of them caused by the low of party identity) occurred in the 2004 and 2009 presidential elections.
Different things occurred in the 2014 presidential election.By observing the preference of the party and a presidential candidate, it seemed the symptoms of split ticket voting among laborers did not occur.Consistently, the laborers chose their presidential candidate from the same party they voted for in the previous legislative elections.5 shows the result of explanatory variables simultaneously.According to the table, we can see the probability value of presidential candidates, for both the voters of Prabowo and voters of Jokowi, were under 0.05.Since the probability value was under 0.05, we denied zero hypotheses which meant that four explanatory variables, such as figures of presidential candidate, laborers' party identity, the role of mass media and the programs of presidential candidates, simultaneously influenced the votes for presidential candidates of 2014.
Due to the result above, we could conclude that in Semarang laborers' circumstances, voting behavior was influenced by some variables, such as figures of presidential candidates, party identity, the role of mass media and the programs of presidential candidates.Simulta-neously and convincingly, those four variables had influences in shaping the laborers' political choices.Thus, the early hypothesis of this research was proven.This result is in line with the findings of Mahmud (Mahmud, 2009), Mujani (2010Mujani ( , 2012)), dan Esita (Esita, 2013), where those four variables significantly influence someone's political choices.The next question was: how were the real influences (effects) of each explanatory variable in shaping votes?

Figures Effect
The quality   7 shows the influence of laborers' assessment of the quality of candidates' figures in the 2014 Presidential Election.For sure, the quality of candidates' figures gave positive influence for the voters.The more positive laborer points of view caused the bigger probability of certain candidates to be voted for by laborers.Influence of figures' quality, for both Prabowo voters and Jokowi voters, was seen as the strongest thing that shaped presidential preference.If a laborer was pleased with certain candidate, it would be more likely for him/her to vote for that candidate.It happened on the contrary when a laborer had negative point of view for certain candidates' figures: he or she would have a bigger chance of not voting for this candidate.How much was the influence of candidates' figures on shaping a laborer's political preference?As explained above, the influence of candidates' figures quality in shaping laborers' political preferences was truly strong.For the Prabowo voters, his A different thing was seen in the way of laborers choosing Jokowi as a presidential candidate.Although his influence was strong, the figures effect of Jokowi was lower than the effect of laborers' party identity.It means, for the Jokowi voters, a closeness factor to the party that carried Jokowi became the most dominant factor for shaping the laborers' political choice.The figure effect of Jokowi followed.Apart from other factors, the laborers who positive points of view about Jokowi would lead them to vote for Jokowi up to 77 percent.
These two results confirm that the figure effect, as Mujani (2006;2010) and Hapsari (2010) have captured, is still being an important factor in shaping the political choices of Indonesian voters.These findings also reinforce the phenomenon of figure-based politics which was revealed by the PWD project (2014).This figure-based politics, actually does not only occur in developing countries such as Indonesia, but also in some developed countries such as United Kingdom, Canada, and Australia, where political parties have been relatively institutionalized (Winham and Cunningham, 1970;Graetz and McAllister, 1987;Bean and Mughan, 1989;Stewart and Clarke, 1992;Mughan, 2000in (Mujani & Liddle, 2007).Apparently, this figure-based politics will still be a new style of Indonesian voting behavior.

Party Identity Effect
Party identity is psychological factor, which is a feeling of being closer, supporting or faithful, or an effort to selfidentify with a certain political party.The party identity that is within the meaning on this study refers to party identity based on the party coalition that supports its presidential candidate.The findings of Mujani et al (2012) in the 1999 and 2004 elections confirmed that voters who identify with the PDIP, more than 82.3 percent finally chose Megawati in the first round of the 2004 elections.While voters who identify themselves with the Democratic party, more than 85.4 percent finally chose SBY in the 2004 elections.
For the laborers who voted Prabowo, this party identity was marked as their closeness to some parties, such as Gerindra, Golkar, PAN, PPP, PKS and PBB.On the other hand, for the Jokowi voters, this party identity was marked as their closeness to one of the parties that supported Jokowi-JK, such as PDIP, Nasdem, PKB and Hanura.
Table 8 shows the cross tabulation between laborers' party identities with their preference of presidential candidates.Generally, a laborer's party identity had a closed relationship to the choice of presidential candidates.Among those who felt close and became the partisans of parties that supported Prabowo, almost all (90 percent) of them chose Prabowo.There were just 10 percent of Prabowo partisans who voted for Jokowi.The same point happened for laborers who stated that they were partisans of parties which supported Jokowi on the elections, where almost all of them voted for Jokowi.The influence or effect of laborers' party identity on the 2014 Presidential Election also seems strong (Table 7).Once a laborer's party identity (of a certain political party that supported a presidential candidate) got stronger, the chance of voting for the related candidate also got the same strength.It also happened on the contrary: when a labor's party identity to the party that supported a certain candidate was weak, it caused the laborers to not vote for the candidate involved.For the voters of Prabowo, the effect of party identity was not as large as the variable of figures' quality.On other side, for the voters of Jokowi, the variable of party identity became the strongest part in shaping preference in the 2014 presidential election.
Conclusively, apart from all other factors and their influences, the laborers who had a party identity of supporting Prabowo, such as Gerindra, Golkar, PAN, PPP, PKS or PBB had 36 percent who truly voted for Prabowo in the Presidential Election.This also happened for Jokowi voters.Apart from other factors, the laborers who a had party identity of supporting Jokowi, such as PDIP, Nasdem, PKB or Hanura would push them up to vote for Jokowi and it got 96 percent (Table 7).

Mass Media Effect
One thing revealed in this study was about the role of mass media, such as television, newspaper, internet, and social media.Did they bring any effect in changing a laborer's political view in the election?Specifically, did the laborer's intensity for following political news and information through mass media also influence their preference in the 2014 Presidential Election?According to Price who was quoted by Mujani et al, ideally, democracy demands competent citizens, namely citizens who understand enough or have enough information about some problems in which they are involved to decide something (Mujani, Liddle, & Ambardi, 2012).
From total laborers who stated that they often followed much political news or information through mass media, there were 56,1 percent who stated that they were interested and strongly Taruno, Figur-Based Politics in Laborer Voting Behaviors: A Study of Semarang Industrial Laborers In the 2004-2014 Elections 129 interested in political or public news.When examined deeper, the laborers who were actually interested in politics were just about 6.1 percent.Whereas the rest, which was about 50 percent, were the laborers who were not interested in the politics, but they had curiosity about the latest political news and information or governmental issues.This finding became much clearer when it was related to the 2014 Presidential Election.
It was not the same as the findings before; the influence of mass media on the 2014 Presidential Election was actually not too strong.Although it was not too strong, both the voters of Prabowo and Jokowi had the same viewpoint, which was that the mass media variable was too significant in shaping political preference (Table 7).The interesting point was that for both models of voters, mass media variable had a smaller effect than the three other variables.Besides, the effect of the mass media variable also had negative effect in shaping choices in the 2014 Presidential Election.
This negative effect means that laborers who started having more political awareness (marked by spending more time to follow the political news and information through mass media) compared with the laborers who did not have any political awareness did not vote as much in the 2014 Presidential Election.It means, there were tendencies among the laborers, who were often following the political news by television, newspaper, radio and social media, to not vote at last.Most of the news that was dominated by negative information, especially in social media, emotionally influenced laborers.It caused many laborers who had positive views about both presidential candidates to doubt, turn away not to vote for either of the candidates.
Apart from other influential factors, the mass media effect had different influence on the laborers.The laborers who came in having political awareness would push them up and vote for Prabowo for 1 percent.On the contrary, at the same condition, the laborers who had political awareness were pushed up and voted for Jokowi for 23 percent (Table 7).

Work Programs Effect
The work program of presidential candidates (rationally reviewed) represented the policies and promises of presidential candidates that would be done when the candidate became the chosen one.By investigating this aspect, the voters were actually considering both pros and cons that would result if a certain candidate was chosen as president.The implication was each voter would give his or her vote to the presidential candidate that would bring a bigger benefit to him/her (Newman & Sheth, 1985).In the tradition of rational voters, a voter would choose a political party or a candidate that could maximize advantages (at least for self-interest) and minimize the disadvantages.
There was interesting nuance related to the 2014 Presidential Election in the laborers' circumstances.Besides the disunity of laborer votes, there were other interesting things to look at, such as the welfare offering from both presidential candidates.Both Prabowo-Hatta and Jokowi-JK promised and committed to fight for laborers' rights that seemed slighted lately.For Prabowo-Hatta, the promises and policies that related to the laborers' rights would be done by fulfilling 10 laborer and citizen pursuits; such as increasing the wages, revising KHL components and eliminating the outsourcing system.Jokowi-Jusuf Kalla promised and committed to the laborers' rights on 3 proper things: proper work, proper wages and proper living for the laborers.The question was, were the political promises from both sides of important to the laborers' consideration to vote in the 2014 Presidential Election?
From the data tabulation is seen that work programs from the presidential candidates did not have any strong relation to the presidential choices.That was why the relationship between a candidate's work program and laborer preference is not seen clearly.The work program effect became less significant in explaining laborers' votes.Thus, this finding complements the finding before: the psychological effect became the most dominant factor in shaping laborers' political preference in the 2014 Presidential Election.

Laborer Voters Have Not Been Rational Yet?
One of the conclusions of the study of Indonesian voting behavior in the 2004 and 2009 elections is that voters in Indonesia have become more rational and have made political choices rather than on primordial grounds (Mujani, Liddle, & Ambardi, 2012).Although not a dominant factor, voter rationality is one of the factors that reinforce someone's political choices.This rationality was seen in Megawati's defeat in the 2004 presidential election and SBY's victory in the 2004 and 2009 presidential elections.
However, according to the data, I assume that laborers' rationality in the 2014 presidential election was not shaped.Decisively, industrial laborers in Semarang chose the president based more on consideration of figure and party identity.Rationality, as I assumed, did not have great influence on their presidential choices.If there was a little bit of rationality, especially for Prabowo voters, it was not yet the dominant factor in shaping political choices.This rationality was still defeated by the figures factor and partisan behavior.
By this finding, I concluded that the characteristic form of laborer voters in Semarang was divided into two types.First, the characteristic form of Prabowo laborer voters tends to be in the psychological, and a little bit rational.Second, the characteristic form of Jokowi labors' voters was not rational but very much psychological.It could be concluded that in this 2014 Presidential Election, the laborer voters in Semarang have not acted rationally yet, but rather acted traditionally.

Conclusion
This study examined the laborers' voting behavior in the elections, particularly in the 2014 Presidential Election.By using psychology and rational choice approaches, this study found two main results.The first was the rise of the laborers' political awareness.It was marked by increasing laborer voter turnout in the Election of 2004 up to 2014, where the rate was growing larger Taruno, Figur-Based Politics in Laborer Voting Behaviors: A Study of Semarang Industrial Laborers In the 2004-2014 Elections 131 This increase of laborer's participation could be read as the increasing of laborers' political awareness in the election.
Due to the participation level being one of the indicators of quality of democracy as Mujani et al (2012) suggest, I argue that the quality of democracy that was shown by the laborers in Semarang was relatively good.Although in the reality, this quality needed special attention, whether they voted just for 5 yearly "ritual" events or based upon proper knowledge.
Second, the domination of the psychological factor of laborer voting behavior in Semarang was marked by the strength of candidates' figures and party identity in shaping their political choices.This finding did not confirm my assumption that laborer political choices is based on rational consideration.In fact, more laborers voted for candidates based on psychological consideration, which was dominated by the candidates' figures, their party identity, and mass media effects rather than program policy.The figures factor, however, still became the most dominant factor in shaping the voters' behavior in Indonesia.Thus, figure-based politics, even among laborers, tends to be found wider and seems to be the new "style" of democracy in Indonesia.
However, this condition started to change in the 2009 and 2014 elections.During this period, voter turnout among industrial labors showed a continually upward trend, both in legislative and presidential elections.Based on the interview results, most laborers admitted that their participation in the presidential elections was a form of citizen participation.In 48-Based Politics in Laborer Voting Behaviors: A Study of Semarang Industrial Laborers In the 2004-2014 Elections 123 Figures EffectThe quality of candidate's figures in this study was limited by four dimensions: competences, integrity, empathy, and favorite dimension.Based on the data, majority of laborers in Semarang stated that both Prabowo and figure's quality became the dominant point over the other effects.The laborers who had positive points of view about Prabowo, would push them to vote for Prabowo in the election up to 39 percent.A different thing was seen in the way of laborers choosing Jokowi as a presidential candidate.Although his influence was strong, the figures effect of Jokowi was lower than the effect of laborers' party identity.It means, for the Jokowi voters, a closeness factor to the party that carried Jokowi became the most dominant factor for shaping the laborers' political choice.The figure effect of Jokowi followed.Apart from other factors, the laborers who positive points of view about Jokowi would lead them to vote for Jokowi up to 77 percent.These two results confirm that the figure effect, asMujani (2006;2010) andHapsari (2010) have captured, is still being an important factor in shaping the political choices of Indonesian voters.These findings also reinforce the phenomenon of figure-based politics which was revealed by the PWD project (2014).This figure-based politics, actually does not only occur in developing countries such as Indonesia, but also in some developed countries such as United Kingdom, Canada, and Australia, where political parties have been relatively institutionalized(Winham  and  Cunningham, 1970; Graetz and McAllister,  1987; Bean and Mughan, 1989; Stewart  and Clarke, 1992; Mughan, 2000in (Mujani & Liddle, 2007).Apparently, this figure-based politics will still be a new style of Indonesian voting behavior.

Table 1 . Population distribution and sample by company type
Taruno, Figur-Based Politics in Laborer Voting Behaviors: A Study of Semarang Industrial Laborers In the 2004-2014 Elections

Source: data processing result
What was the influence of candidate's figures on the voting of the 2014 Presidential Election?Table