Impact of Campaign Appearances on Key Constituencies
Disclaimer: This dissertation has been submitted by a student. This is not an example of the work written by our professional dissertation writers. You can view samples of our professional work here.
Any opinions, findings, conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of UK Essays.
Why do people vote the way they do and what can candidates do to sway the votes of those that go to the polls? Do people cast their votes based on the policies advanced by a candidate, rational self-interest, or political campaigns and the get-the-vote-out efforts (e.g. campaign appearances, advertising, door-to-door canvassing, leaflets, phone banks, electronic mail)? Which one of these influences most the number of votes a candidate gets in elections? Recent studies in voting behaviour point out to an increased importance of get-the-vote-out strategies. The purpose of the present paper is to explore the effects of one type of get-the-vote-out efforts - more precisely, local visits by the candidate or incumbent in a certain constituency - and the strategies politicians employ in order to increase the number of votes they get.
I take as a case study the Romanian 2009 presidential election. I look at whether the abovementioned type of get-the-vote-out efforts made a difference in the number of votes the candidates got. Social scientists have recently started to pay increasing attention to matching in an attempt to infer causation based on experiments that rely on observational studies. In order to test my hypotheses I use matching as a main method. In addition to it, I conducted content analysis on printed and audio-visual media and run several OLS regressions. The results show that only one candidate's campaign appearances were marginally significant in statistical terms and did have an effect on the number of votes that candidate got.
Voting means freedom of expression in terms of political views; in any democracy, it is the ultimate method through which a mature, informed, law-abiding individual expresses his/her decision about how he/she wants to be ruled. It is about representation and participation; it is the assumption that those casting the vote are politically knowledgeable and want to express that knowledge (Popkin vs. Converse, Popkin, 2006). Thus, one of the key elements that candidates to public offices need to bear in mind is the fact that they must persuade the average individual not only to participate in the electoral process (and exercise their Constitutional right to freedom of expression) but also to cast a vote in their favour.
The standard literature on voting behaviour lists as factors that influence the voters' preferences for one candidate or another, elements such as: the position of the candidate on certain issues and their respective policy priorities, ideological attachments, rational self-interest (which candidate's policies best fit the voter's private interests) Kim 2009 and Sears et al. 1979, discussants (the political preferences of people one esteems, trust etc.), and charisma (Popkin 2004, Lau and Redlawsk, 2006). However, these elements fail to explain much of the realities in newer democracies, for instance post-communist countries in Central and Eastern Europe such as Romania. Even in old democracies, researchers have started to focus more on political campaigning and advertising as factors that make individuals cast a vote for a particular candidate. The broad category of political campaigning includes: campaign appearances, advertising (TV spots and billboard ads), door-to-door canvassing, leaflets, phone banks, direct and electronic mail. This approach has proven to be a promising stream in voting behaviour research, as many studies conducted in the U.S. show. A somewhat similar study was conducted after the 2000 election in the U.S., by King and Morehouse (2005), their aim being to demonstrate that the Gore Mississippi River trip of August 2000 was paramount in 'moving' voter preferences for this candidate in the states included in the itinerary of the trip - this being a more productive campaign scheme than television ads and media consulting services.
The natural question that emerges thus is to what extent political campaigning has an effect on voters' preference for one candidate or another. The present paper aims at exploring this question by focusing on the case of the Romanian presidential election of 2009. Of all the abovementioned types of political campaigning I decided to take into account the candidates' electoral visits in different towns, the main reason being the fact that this the most commonly used in Romania. Therefore, the main research question that this paper raises is: Do campaign appearances have an effect on the number of votes a particular candidate gets in elections?
To answer this question I test one main hypothesis:
H1: In towns where a particular candidate already enjoyed a high number of supporters, that candidate's electoral visit led to an increase in the number of votes he got.
Several additional hypotheses are tested, although they do not refer strictly to the core research problem - the effect of making appearances on the number of votes a candidate gets in election.
H2: Candidates organizing electoral visits in constituencies leads to an increase in the turnout of a particular election.
H3: Candidates organizing electoral visits in constituencies lead to an increase in the added number of votes of those respective candidates.
In order to test these hypotheses I use matching, a statistical method that compares groups (in this particular case, the number of votes each candidate got in the towns where they made an electoral appearance against the number of votes they got in the towns where they did not make an electoral appearance) on measurable parameters. The mentioned parameter should be as closely resembling (if not equal) as possible. Thus, the paired towns have similar (sometimes even identical) values for the confounding variables, the difference in the number of votes each candidate got being therefore attributed to the electoral visits of that said candidate. While this method has been extensively used in medical or economics researches, recently political scientists started using it on a more extensive basis, especially when it comes to observational or experimental studies.
In addition to this, I conducted content analysis in order to identify the towns where the presidential candidates made appearances. I analyzed two national newspapers (chosen based on the total circulation number) and one national TV station. I crosschecked the information derived from the media against the information provided on the personal web pages of the candidates. Furthermore, I ran several OLS regressions in order to test whether the hypotheses hold.
In what regards the structure, the paper is divided into four main sections as follows: the first offers an overview of the existing literature in the field of voting behaviour; the second deals with the theoretical background and the methodology employed; the third introduces the data used and presents the analysis, whereas the fourth section discusses the main findings and proposes directions for further research.
The novelty that this paper brings consists in applying a relatively new approach in voting behaviour research - the effects of political campaigning on voters' choice for a particular candidate - to a country that previous studies have tackled very little. In addition to this, matching represents an innovative method able to provide a more in-depth and meaningful insight in this particular field, still underused by social sciences researchers.
LITERATURE REVIEW, METHOD AND THEORY
This chapter will focus on a brief overview of the current literature in the field of voting behaviour with an emphasis on campaign events used by candidates to persuade voters to cast a vote for them, while highlighting at the same time the aspects of the theory that are of crucial importance for the present study, and propose a way to analyze them further.
Together with voting, electoral campaigns represent the main tool through which citizens in representative democracies assess the suitability of a candidate running for office. This allows that candidates, in their turn, can use the same tool in order to attract a larger and broader audience whose political views they can thus hope to shape into favouring them above all other candidates (Arceneaux, 2010).
For the purpose of the present study, it is important to point out from the very beginning, that the scholarly literature on electoral campaigning and voter behaviour focuses on two main aspects of campaigning. The first deals with campaign appearances and local visits in key constituencies (Holbrook 2002, 1996, Campbell 2000, Shaw 1999, Jones 1998 cited in King and Morehouse 2005), while the second with television advertisements (Shaw 1999, Freedman and Goldstein 1999, Ansolabehere and Iyengar 1995, Finkel 1993 cited in King and Morehouse 2005). The two aspects mentioned above have been deemed as the two most important in impacting voters' behaviour in terms of choosing a candidate and/or voting on Election Day. Yet it has been argued by scholars (Fowler et al. 2002 and Joslyn and Ceccoli 1996) that in order for a candidate's visit or ad to have a lasting impact on the voter, the specific voter needed to have a previous lingering inclination towards that specific candidate (cited in King and Morehouse, 2005).
As I have previously mentioned, the literature on elections and voting behaviour, focuses primarily on types of voting or factors that influence voting, i.e. economic voting, instrumental or expressive voting, self-interest voting, issue voting and so on. The focus is thus on who votes (Wolfinger and Rosenstone 1980, Sigleman, Roeder, Jewell and Baer 1985, Verba, Nie and Kim 1978) or on what makes individuals vote. More precisely, the focus is on what are the influences of how individuals vote and how a candidate can persuade voters to cast a vote in his/her favour. Relatively recent studies on campaigning and campaign management (Feddersen and Pesendorfer 1997, Dutta, Jackson, Le Breton, 2001, Thurber & Nelson, 2004, King and Morehouse, 2005, Gerber and Green, 2008) illustrate that campaigns usually aim at influencing the reasons individuals have for voting in a particular way, with a strong emphasis on manipulation and appeal to emotions. Still a clear-cut connection between different types of campaigning, the symbolic factor and the rationality of the voter is yet to be found - especially in what concerns new democracies, such as Romania.
In order to explore the issue of how Romanian leaders carried out campaigns in the last 20 years since the revolution it needs to be pointed out that their main focus was, as mentioned above, to appeal to the emotions of the individual. They achieve this through the symbolic over-flooding of messages (Sears, Hensler and Speer, 1979, Pippidi, 2004) in the printed press and the audiovisual, in the speeches they hold during TV debates and/or rallies, or in their campaign appearances in different constituencies. In the U.S. case (the 2000 U.S. elections); King and Morehouse point out that the Democrats presidential campaign made use of campaign appearances as a tactic to generate positive media coverage. Through these campaign appearances the Democrat candidate managed to 'energize the base', while it also generated free advertising and appealed more to the local community since is their media reporting the news. King and Morehouse go on to underline that: "local newspapers and television stations are eager to cover campaign events and they tend to approach politics with less cynicism than one finds among the national press corps" (King and Morehouse, 2005). Thus they emphasize that candidates should in fact choose key swing states for campaign appearances and attract the local media to generate free coverage both locally and at national level.
Although both in 2004 and 2009 Romanian presidential candidates made use of electoral visits in order to gain media coverage, the Romanian media landscape is quite different from the American one. In contrast to the U.S. where local media is powerful and autonomous, the Romanian local media is mainly an extension of the national media. Most national TV stations and newspapers or radios have local versions for most of the towns or counties. Consequently, the news coverage these local media do is still in accordance to the articles printed at the national level - at least in the case of high level importance elections.
Referring to the pieces of news that voters use to inform themselves about candidates and campaigns, recent research showed that (Baum, 2006) soft news impact to an important degree inattentive individuals by making them change their preferences depending on the cues they get from that type of news. Following the idea presented by King and Morehouse (2005) that local media is and should be used by politicians to get their message across for free during elections to as many individuals as possible, it also follows that candidates should and must tailor their message and their campaign strategies depending on the different societal groups they aim at reaching (Baum, 2005). Since most supporters of parties build their political knowledge and political preferences based on the cues they get from different tabloids or entertainment shows - yet still do this on a rational manner (Jerit, Barabas, and Bolsen, 2006) - it is all the more clear why Romanian politicians use the sensational factor in their speeches and why they centre their campaign discourse more on rhetoric than ideology and concrete policies. The average voter tends to watch more soft news than hard news, as the former promote a humanized version of politics and politicians, emphasizing the personality and character of a candidate (as well as his/her shortcomings and mistakes). Thus, instead of the policies, the voter ends up identifying with the candidate and vote more based on feelings and emotions rather than ideological attachment or policy preferences. Jerit mentions that "even if learning from this medium is largely passive and unintentional, individuals may obtain enough information to function as monitorial citizens (Schudson, 1998)" (Jerit et al., 2006).
Another relevant aspect involving the literature on campaigning and campaign appearances promoted through national and local media is presented by Zaller. His main argument, that "mass communication is a powerful instrument for shaping attitudes ... and [that] it exercises this power on an essentially continuous basis" (Zaller, 1996, p. 18) strengthens the argument made by King and Morehouse (2005) and mentioned above. By meeting with the local press and presenting their platform they insure that even those not present at the rallies or not watching the debates might still find out about what their campaign platform is.
Lastly, still on the issue of media and campaign appearances, Natalie Stroud's article points out that there is in fact a relationship between selective exposure and political attitudes (the example she gives is that those who watched "Fahrenheit 9/11" were significantly more negative towards the Bush administration; not only that, but they also were more prone to start political discussions on the topic, as opposed to those who did not see the film). Thus, those exposed to such means of communication, tend to be more politically active and to engage in political discussions due to the attitude polarization of the extreme media messages. For the present paper, this argument can be interpreted in the light of the local visits made by candidates. Given the trail of articles following the candidates' visits, the previous argument supports the statement that visits generated local and national coverage in the press for the candidates. This in turn generated debates between the supporters and even the detractors of those said candidates, ultimately leading to their being 'moved' towards the polls on Election Day.
Directly related to turnout though, Powell (1980, 1986) was the first to look at vote turnout in an analysis that span over twenty-nine democratic countries, between 1958 to 1976, in an attempt to answer why is turnout high in some countries and low in others. He found that there are certain 'mobilizing voting laws': compulsory voting and the fact that the governments assume responsibility for registering individuals on lists for elections, as well as strong-line party alignments (Crewe, 1981, also lists strong-line alignments) that increase turnout. Consequently, for the present study, one can take from Powell's study the fact that having a 'party with strong-line alignments' (Romania already has a system where the government assumes responsibility for registering the eligible voters on lists), that has kept a constant line in politics over time, and that enlists as a candidate a charismatic individual has more chances to move voters and persuade them to cast a favourable vote for that candidate, given that the party has a campaign that ensures its candidate makes appearances in key constituencies.
Other similar studies include those made by Jackman (1987), Blais and Carty (1990), Black (1991) or Franklin (1996), but all dealing with turnout and what influences higher or lower turnout - either the electoral system, compulsory voting, degree of disproportionality of electoral outcomes, unciameralism vs. bicameralism, postal voting, Sunday voting, number of polling days. While focusing on all the abovementioned factors that influence turnout the referred studies either eliminate or loose sight of the campaign effect on the number of votes a candidate gets. This is why the present paper, following these studies, in an attempt to complete the picture presented by them, focuses on the effect of campaign visits on the number of votes candidates got in the visited constituencies. Another aspect to bear in mind is that the previously mentioned studies were focusing on old democracies, while this paper analyses the election in an East-European new democracy.
Method and theory
As this paper is using matching as a main methodological tool - constituencies where the candidates made appearances are matched with similar constituencies where those candidates did organize electoral visits - a closer look it is needed in order for the analysis to be better understood.
"Matching has been proposed as a non-parametric solution to problems of bias that emerge in observational studies" (Rosenbaum and Rubin, 1983, 1985 cited in Arceneaux et al. 2006). However, scholars are split between the issue of matching having a bias or not, especially given the fact that it is nonetheless a method that allows for unobserved differences between groups to remain unnoticed. The literature on the topic, to date, focuses on the performance of matching estimators but uses 'experimental benchmarks', and not observational data (Dehejia and Wahba 1999; Heckman, Ichimura, and Todd 1997, 1998; Heckman, Ichimura, Smith, and Todd 1998; Smith and Todd 2003 cited in Arceneaux et al. 2006). "The matching process identifies treated individuals [in this case, towns] who share the same background characteristics as untreated individuals [towns]. It is hoped that after matching on covariates, any remaining difference between groups can be attributed to the effect of the treatment" (Arceneaux et al. 2006).
This is one of the reasons why this paper will look at the performance of matching estimators in explaining if campaign appearances generate increases in a candidate's obtained number of votes in those constituencies visited but will use observational data instead of experiments.
Another aspect to bear in mind when conducting an electoral campaign focused research is the strategic nature of campaigns, for as Arceneaux highlights, selection biases may appear in voter exposure to campaign stimuli therefore creating biased estimates of campaign effects (Arceneaux, 2010). This is why it would prove interesting to look at the campaign trail for each candidate and highlight the strategy used for 'getting-out-the-vote' in the constituencies visited. Yet the present study will not deal with this aspect extensively since the main issue that it addresses remains the ability of campaign appearances to persuade voters to cast a favourable vote for the candidate making the appearance. In terms of time and geographical parameters, the focus here is on the Romanian presidential elections of 2009. In order to counteract the possible matching bias that the literature on the topic mentions (Arceneaux et al. 2006), a comparison between the number of votes the candidates considered for this study got in 2009 and the number of votes their parties got in the 2008 parliamentary election is introduced as an independent variable (the comparison is made for each of the towns included in the study and their matches).
For the purpose of this study it is also of great importance to shortly note the manner in which the candidates lead their campaign. Pippa Norris mentions that most candidates are vote-maximizers, following set patterns when setting their electoral agendas, and that they usually involve putting issues before voters (Pippa Norris, 2004). Therefore, a candidate's electoral agenda should focus on issues that are crucial points for the constituency he is running for. For the presidency this means centring their agenda on issues that interest most of the country. Consequently, the agendas should follow the lines of thought of the majority of voters without transforming the speech into pure rhetoric, demagogy or empty promises. As the following sub-chapter will highlight this is usually not the case for Romania and for the Romanian electoral campaigns.
Social networks researchers (Valdis Krebs, 2004) suggested that there are certain key facts to take into account when addressing the issue of building the electoral campaign in such a way that will mobilize voters and persuade them to vote for a particular candidate. One is getting a charismatic figure as candidate, someone that could be perceived as a role model. A second one would be using candidate appearances (or party supporters) to get the votes of the undecided - this has been a technique also used by the Gore campaign of 2000 (King and Morehouse, 2005), whereas the third would be using campaign appearances to consolidate the faith and votes of a constituency that already supports that particular candidate. For the Romanian 2009 electoral campaign the latter two mentioned strategies were used - using campaign appearances to either convince the undecided or consolidate the faith and votes of a constituency - though with a higher emphasis on the latter.
As far as the present paper is concerned, it is also important to look at the campaign agendas and stump speeches the candidates tailored for each visited constituency. It is noteworthy to point out that, contrary to what Rose and Haerpfer (1994) highlighted for Eastern European voters, voting in Romania is based on socio-tropic evaluations. In Romania, as in most Eastern European new democracies, the main problems governments are faced with are political and economical. Due to the negative views regarding the state of economy, individuals tend to vote economically. In other words, this means that they hold the president responsible for the state of the economy - even if to a lesser extent than they view it as the parliament's responsibility. Having in mind the aspects mentioned above, one of the independent variables chosen for this study is the unemployment rate, at county level, this helping pinpoint the level of development of the region. The assumption for that choice was that the better developed the region was, the more people it has employed in different industry branches, the higher the income of the inhabitants is (this may also mean better education for the voters), and the better the ability to assign responsibility correctly and therefore, make more informed decisions when casting their vote.
Further on, I looked at previous voting patterns for those constituencies where the candidates made campaign appearances, in order to establish whether there is a recurring pattern in the voting habits of the inhabitants of the area, and if the appearances influenced it one way or another. This is needed to show whether political candidates chose some constituencies with the intention to get those voters to go to the polls and vote for them, the choice being done strategically or not. As mentioned before, this paper will not attempt to deal with the large issue of political campaigning as a whole. It will examine a particular aspect of it - candidate appearances, whether it is rallies (events where candidates meet with their constituents), or meetings with the media in a certain constituency. The latter aspect of campaigning is a practice commonly used in Romanian elections, where candidates meet with the local media and thus get free advertising in the local media as well as coverage at the national level. Yet it is unclear if this aspect is at the back of the minds of the candidates when doing it.
Theoretical Background for Romania
In post-Revolution Romania, too often party members change their affiliation; floating from one party to another, only to adapt to the fact that their former party lost elections or simply because the party no longer meets their needs and political aspirations. Each election, the electorate is faced with incumbents or new candidates who have both tailored their political rhetoric to the current political persuasion of those who have nominated them so that they can secure elections without having a clear and strong ideology. Consequently, politicians are transformed into clients of the parties or party officials and vertical accountability in office is void of its initial meaning (Pippidi, 2004).
This system has been maintained in place by a dual executive that doesn't share power but functions on a servitude basis and on the basis of repetition - a bicameral legislative framework, where the two chambers are equal, elected in the same way, with almost identical functions and where "both deputies and senators ... cannot be bound by instructions of their constituencies" (Ludwikowski, 1996, pp. 129).
All these have been a staple of the post-communist elections in Romania and have eroded the quality of the democratic process. As a consequence, both presidential and parliamentary elections are heavily dominated by populist discourses. In 2009, the main parties having candidates running for the presidential office were the Social Democratic Party (PSD), the Democratic-Liberal Party (PD-L), and the National Liberal Party (PNL). The first one, the Social Democratic Party (PSD), is considered to be the heir of the former Communist Party following the revolution of 1989, and as such maintained itself as the largest and best institutionalized one (Curt, 2007). However, the alleged failures of the 1992-6 and 2000-4 governments placed it on the second position in the preferences of the people.
For the purpose of this study, I have chosen to exclude from the analysis candidates of other parties or independent candidates that also ran for the presidential office. That is based on the assumption that their importance in the preferences of the majority of the electorate was secondary compared to the candidates of the three parties mentioned above.
DATA DESCRIPTION AND ANALYSIS
This section of the paper focuses on the collected data, more precisely the sources of the data, the method of collection, as well as how it was used further in the analysis.
Given that before I have mentioned the tendency of the Romanian voter to base political decisions on the state of the economy and on feelings (the persuasiveness of the candidate playing a key role here), it will facilitate the understanding of the analysis to point out that voters also tend to be influenced by other factors (since they don't appear to vote by making evaluations, appraisals or weighting their decision). This is why the speeches of the candidates during their campaign appearances in the thirty towns used in the analysis were centred on aspects such as the economy and the recent economic crisis, family values, wages and the incompetence of the other candidate versus their own competence superiority. By choosing these facts to centre the speeches on, they tackled some of the key aspects that influence the Romanian electorate: the socio-economic background, their social environment, self-interest, the charisma of the candidate, and the fear of change that the elections will bring changes in government that will prove to the detriment of the country.
The socio-economic background is important for both the Romanian voter and the candidates since it is a well known fact that most of the Romanian middle class families tend to vote for coalitions or for democrat-liberals, as these give them the confidence that their income will stay at the same level; while the working class families, for example tend to vote for socialists or social-democrats.
The social environment also is a crucial factor, since family values are still of paramount importance in Romania. Individuals therefore, tend to vote in high percentages, the same way their parents vote (Hatemi et al., have already shown that parents have a major role in determining the initial political direction of their children especially if the children still live at home which is the case for most Romanians, thus the previous inference gets higher support.)
Individuals also tend to focus on only one or two key policy areas (issue publics - as they tend to acquire information about candidates and elections based on their interest in particular issues) and inquire only about issues they are primarily affected by or interested in (Young Mie Kim, 2009). Therefore, candidates tailor their speeches based on the issue that is more pressing for the electorate they visit. Seeing things in this light, Feddersen and Pesendorfer (1997) argue that though individuals focus on one or two policy areas makes the electorate as a whole better informed collectively about what the alternatives and best outcomes are, for this particular case, at an individual level, the choice might still not be the best one.
On the other hand, yet still highly related to the previous statement, it is not clear if self-interest is necessarily what guides the Romanian voter. On these lines, Sears et al. (1979) highlight that self-interest guides the vote choice together with rational choice and that it is defined as minimizing losses for private well-being. But for Romania in 2009, the former correlation, that self-interest in terms of voting goes hand in hand with rational choice, is not necessarily always the case. What is more important is the security of the job, maintaining lower taxes, the ability to pay-off debts or receiving credits to pay-off debts, free medical care, a welfare state on the lines of the socialist credence. Subsequently, as long as a candidate promises the electorate the fulfilment of these, and appears credible (and here the charisma of the key party leader plays a very important role) in his promises, rational choice reasoning for casting a vote is rarely used. Lastly, I would also argue that, at least for Romania, candidates, but especially incumbents, use fear of change to influence voters and to move them to polls. Here fear of change is understood in the sense of apprehension towards everything and anything that is new, that might change the previous order, the previous system and its institutions, that might demand for different behaviour.
Having all of the above in mind, and going back to the purpose of the paper, to conclude, the main issue this paper looked at was if by making campaign appearances in different constituencies, and using the abovementioned influencing factors, the candidates for presidency persuaded voters and managed to get them to cast a favourable vote; that is to say, if the campaign appearance of a candidate in a particular constituency increased the numbers of vote he got based on that visit alone. By comparing the presidential candidates' obtained percentage results for this election, for each visited town, with the percentages gotten in the previous elections by the candidates' parties, I attempt to eliminate the bias and control for other factors that might have influenced the results.
Since there was no previous database to select the variables from, I have compiled my own database (divided into two smaller datasets) compiling data from different official sources: the National Institute for Statistics (INS), the Romanian National Office for Employment (ANOFM), the Ministry of Regional Development and Tourism.
The variables used are described below:
City is the variable that lists the big towns where the candidates made electoral appearances as well as the towns where they didn't. The first dataset was used for the un-matched OLS regressions, while the second dataset was used for the matched OLS regressions and lists only those towns where the candidate made appearances together with their matches.
The Population size variable, lists the total population for each of the chosen towns, as listed in the last performed census (2002). This variable was included firstly because it provides for one of the economic variables needed and, secondly, because it helps to match towns on two important characteristics: number of inhabitants as well as size of town - one can thus state if a town is relatively large, medium or a small one and thus track the campaign pattern in selecting towns for each candidate. For the 2009 elections it was evident that the candidates chose to tour mostly big towns, out of which the majority were in fact county capitals. From a strategic point of view, choosing only (or mostly) county capitals and thus big towns to visit offers the advantage of reaching a larger number of voters with one visit - instead of touring numerous small towns or villages to get the same numerical result - yet it presents the disadvantage that it doesn't necessarily reach the target of mobilizing voters and moving them to the polls. Individuals in big towns have a larger, more diverse pool of informational means to choose from when collecting campaign information as opposed to individuals in small towns or villages. Also, the educational level of the individuals from big towns is significantly higher thus they are more 'educated' in the electoral process and more likely to vote than the individuals from, say a village.
All of the abovementioned arguments make all the more sense in light of the theoretical background presented in the previous section, where I have detailed the other possible influencing effects of why people vote with a certain party in Romania, i.e., the socio-economic background, the environment, self-interest, knowledge of one issue, charisma of the candidate, fear of change. Consequently, candidates could have made use of their charisma and the self-interest of individuals in smaller towns on one issue, during possible electoral visits in the effort to increase the number of votes they would get and move the undecided voters to the polls.
The Percent of Hungarians variable, lists the percentage of Hungarian inhabitants in the visited towns and the ones that were not visited but were matched on the socio-economic variables (the data being also taken from the official website of the Romanian census). This continuous variable was taken into consideration for this analysis for two reasons: first because knowing what percentage of the population is of what ethnicity might shed light on the reasons candidates had for choosing certain towns to tour in (thus, matching those towns with similar ones where the candidates did not go, could prove easier and better suited for the final analysis); second, because scholars have long pointed out that minorities tend to respond negatively to elections where the candidates do not represent them directly or have not included in their campaign issues that affect the said minority directly and therefore they abstain from voting.
A side reason for looking at the ethnicity of the inhabitants for each town visited, for the Percent of Hungarians variable, was to determine also whether candidates focused solely on towns where the majority of the population is Romanian, or if they also visited towns with a preponderantly Hungarian population. Here the underlying line of thought is that they would not consider making electoral appearances in towns where the population is preponderantly Hungarian, as they might feel it is not a vote maximizing strategy - such towns would have higher turnout for the Hungarian party, the Democratic Union of Hungarians in Romania or UDMR, in Romania anyway.
The Unemployment variable looks at the percentage of unemployed individuals from the visited towns and the ones that were not visited but were matched on the socio-economic variables. The data for this variable have been taken from the official website of the Romanian National Office for Employment (ANOFM) and were provided only at county level. Still, given that the candidates made appearances mostly in big towns, county capitals, where most of the workforce is concentrated, it is of little consequence to the overall results of the analysis that the unemployment rates are taken at county level. The discrepancies rely more between regions of development and counties than towns in the same county.
The usefulness of the variable for the matching process is given by the fact that, for Romania, towns from the same region but from different counties might have different unemployment rates. This is the case for the North-West region where the town of Cluj-Napoca is highly developed and has a very small unemployment rate, whereas a town from a near-by county, but from the same region, Bistrita, is notably less developed and with a considerably higher unemployment rate. This is also the case for the South-Oltenia region where the discrepancies lay between the towns of Pitesti and Calarasi or the South-East region and its towns, Bacau and Vaslui. Therefore, in order for the matching process to be as accurate as possible in connecting towns on very similar characteristics, this variable is needed, for it refers at both the economical and the development aspect.
Lastly, I have looked at the history of voting for the towns the candidates visited. Thus I have complied from the Romanian Central Electoral Bureau's website (BEC) the data percentages for the last parliamentary election of 2008 as well as for the 2009 presidential election. The variables that resulted (turnout08 and turnout09) represent the turnout for the towns included in the analysis (in percentages); turn08all and turn09all, represent the added number of votes got by the three candidates (in percentages), while the variables number of obtained votes 08 - PSD, number of obtained votes 08 - PDL, number of obtained votes 08 - PNL, number of obtained 09 - PSD, number of obtained votes 09 - PDL and number of obtained votes 09 - PNL, respectively, list the total number of votes for the 2008 and 2009 elections got by each respective party/candidate, the social democrats (PSD), the democrat liberals (PDL) and the liberals (PNL). The percentage of the number of votes each candidate's party got in the 2008 elections was used as an independent variable in the OLS regressions and was essential in determining if the results presented by the regressions and the matching were not biased and in fact due to completely different factors.
The research design that this paper follows, was developed in three major steps: during the first stage I conducted a content analysis study of two major media outlets and one television channel in the country (focusing on town level appearances). The analysis was limited in time, focusing only on the thirty day period of the campaign, as the Romanian electoral law stipulates that only this time period before Election Day should be employed by the candidates to conduct their electoral campaign. The purpose of the content analysis was to identify, first of all, the main areas of interest for each of the three candidates, and, second of all, to help match the places where candidates made appearances to similar places where they did not. Thus I looked at those constituencies where candidates have been reported to have conducted their electoral touring and made appearances either in front of their electorate or holding press releases with the local media.
The content analysis was conducted for the period starting on October 23rd and ending on November 21st, 2009, for Realitatea TV (Reality TV) news channel and two national newspapers, Jurnalul National (The National Journal) and Adevarul (The Truth). The selection of the two newspapers was done based on which two newspapers had the highest reported circulation nationwide for the previous year as stipulated by BRAT (the Romanian Bureau for Circulation Audit) as well as which of them actually had available archives. Table 1 shows the circulation for the newspapers and the distribution rate for the year 2009.
While I have used the online version of the newspapers, which might imply some accuracy issues, I must underline that the content of the printed version of the newspapers is almost identical with the online one, at least for the two selected cases. Given that I have tracked high level importance pieces of news, I have all the more reasons to expect that the articles were also posted on the webpage of the newspapers. Still, to insure that by using only two newspapers and a TV station as sources for the content analysis I do not get biased results, either because I was using the online versions of the outlets and especially given that the Romanian media is known to have a bias towards one party or another, to a greater or lesser extent, and also to control for the relatively small newspaper sample, I compared the results against the party websites where the candidates posted their campaign itinerary. Table 2 shows the distribution of campaign appearances for each candidate.
I have excluded from the content analysis the capital, Bucharest, and it's surrounding area, Ilfov, for two reasons: the difficulty to find any relevant matches and the relevance for the entire analysis (candidates always organize at least one appearance, usually more, in the capital and its surrounding region). The development regions refer to the seven Regions of Development as presented on the website of the Ministry of Regional Development and Tourism.
The three main candidates made electoral appearances in a total of thirty big towns (most of them county capitals), out of which, only three have been chosen by all to hold an event, Timisoara, Buzau and Pitesti. These three towns, interesting enough, are situated in two of the most economically developed regions of Romania, the South-Muntenia and the West regions.
The thirty towns visited were compared based on demographics such as the age of the population, number of inhabitants, ethnicity percentages (here mostly focusing on the Hungarian population of the towns since they constitute the largest minority in Romania), the unemployment rate or division into regions of development.
As mentioned in the previous chapter dealing with the theoretical background, having in view the situation of the country's political system (the change in the electoral system), the developments since 1989, and the voting behaviour predispositions of the Romanian voter, as presented above, I expect that campaign appearances did in fact increase the number of votes each candidate got in the visited towns yet not by high percentages.
Subsequently to deciding which would the variables be, I have proceeded to match the thirty towns where the candidates made electoral appearances with towns similar on the socio-economic variables - the population size of the towns where candidates made an appearance, the seven regions of development (as established by the NUTS system mentioned in footnote5), the unemployment rate in percentages, the percentage of ethnicity, the overall turnout and the number of votes the candidates got in the towns visited. While finding matches for the towns where the candidates made electoral appearances I took into consideration the known fact that most of the social democrats' electorate is made up of individuals leaving in the rural parts of the country thus I have looked for matches in similar regions and with similar unemployment rates for those towns where only the social democrat candidate went. There is another rationale for this that appeared more clearly after the content analysis was performed and after I have compiled the results of the 2008 parliamentary elections. Most of towns visited by the social democrat and democrat liberal candidates were towns in which their party won the majority of votes in the past two elections.
Afterwards, it proved more rewarding for the scope of the present study to conduct several OLS regressions, both for the matched dataset and for the un-matched one. The reason behind choosing this method was to make as accurate as possible predictions, i.e. forecasting results based on the proposed model for further analysis, through fitting the predicted model to the observed data. Therefore I have performed five OLS regressions for the matched dataset and five for the unmatched one, first for the effect on the total number of votes cast in the 2009 elections, second for the effect on the cumulated percentage of the number of votes for the three candidates, then, individually, for each candidate's electoral score.
RESULTS AND FURTHER DISCUSSION
As it was shown in the previous section, after conducting the matching for the thirty cases - the town where candidates made electoral appearances - the results can be interpreted through a comparison between the two datasets, the matched and un-matched one. This chapter will look at the findings of the analysis through the prism of the expected results expressed in the first part of the paper, then will propose further avenues for the usage and expansion of the study.
Expected Results and Actual Findings
As stated in the introduction, the main hypothesis advanced by the present paper is: In towns where a particular candidate already enjoyed a high number of supporters, that candidate's electoral visit led to an increase in the number of votes he got.
After running the analysis, the hypothesis failed to be rejected for only one of the three candidates considered for the study, the social democrat leader Mircea Geoana. The results show that in most of the towns visited by this candidate the number of votes he managed to get was significantly higher than the number of votes his party got in the 2008 parliamentary elections. For example, in Buzau, the capital of Buzau county, a town where for the previous elections the social democrats (PSD) won the majority of the votes (both at the town and county level) and where the mayor is also a social democrat, the fact that Mircea Geoana made an electoral appearance increased the number of votes he obtained. The same observation can be made for a number of other towns where the social democrats have organized appearances for their candidate - Braila, Botosani, Pitesti, Targu Jiu, Targu Bujor. This is visible from the interaction between the variable recording the number of votes obtained by the party in the 2008 parliamentary elections and the one recording the number of votes obtained by the candidate in the presidential elections, which was included as independent variable in the analysis.
The analysis consisted in several OLS regressions ran on two datasets, one containing all the medium-size and big towns in Romania, the other containing only the towns where the candidates made appearances and their respective matches. I chose to present the results for both datasets, the reason being that for the unmatched dataset they turned out to be more significant in statistical terms, whereas the matched dataset has the advantage of controlling for the other factors that might have influenced the outcome of the elections, thus making the results of the analysis more accurate.
In addition to testing the main hypothesis, I chose to look at the effect of electoral visits on the turnout of the 2009 presidential elections as well as on the added number of votes of the three candidates considered for the study. The results of the analysis as well as their interpretation are presented in the following section.
Interpretation, Comparison and Further Application
This final section of the paper will detail the findings of the analysis while also proposing further avenues for expanding the research. Since I have mentioned before the apprehension scholars still have about the possible bias of matching, I will put forward, in the last sub-section, other possible influencing factors that might have tilted the results in the same direction.
Interpretation of the matched and un-matched OSL regressions
The results of the OLS matched regression from table 3.1 below indicate a high value of the R² thus showing that the model has a good fit in terms of explanatory power. The variables recording the size of the town (expressed in number of inhabitants), the percentage of Hungarians in those towns, and the level of unemployment have a significant effect. These results highly support the initial assumption that the selected variables are in fact relevant for the analysis. Although all the control variables turned out to be significant, the main independent variable is not significant, this meaning that the candidates having organized electoral visits or not does not have an effect on the turnout of the elections.
The second column shows the results for the un-matched dataset. While the sample size is three times bigger than in the previous sample the population size (size of town) variable looses its significance and so does the variable unemployment. The percentage of Hungarians in each town remains significant, though the significance level decreases. This was to be expected since the previous statements about minorities hold in this case too.
When the dependent variable changes form the 2009 total turnout to the added number of votes the three candidates got together, the results are different. As table 3.2 shows, the variable recording the size of town is not any longer significant and the same applies to the variable recording the percentage of Hungarians. Here the explanation might be that given the existence three alternatives, that do not represent their interests directly, the members of the minority tend to be less interested in the electoral battle and its outcome. Thus, the effect of campaigning in a town where the majority of the population is represented by a minority, in this case Hungarians, does not have a significant effect on turnout.
Table 3.2 below indicates that the OLS regression having as dependent variable the added number of votes the three candidates considered for this study obtained also yielded few significant results. For the matched dataset the only variable that has an effect is the one recording the level of unemployment, which, given the context of the country and the towns included in the analysis, is not surprising. As the matched dataset has included in the analysis only those towns the candidates visited and their matches, and candidates chose big towns, most of which are highly industrialized (hence a lot of the inhabitants work for state owned enterprises thus have their jobs more or less guaranteed), again it is understandable why the unemployment percentages would have a significant effect. For the unmatched dataset, the fact that the analysis included a larger variation of towns in the regression - the number of included cases is almost triple in size in comparison with the one for the matched dataset - did not add much to the significance levels except that, in this case, the percentage of minorities in the towns has an effect on the added number of votes the three candidates got. For this regression the most important observation is the fact that, while some of the independent variables remain significant, the main independent variable, the number of candidates visiting, is not significant, this meaning that the candidates having visited a town or not does not have an effect on the total number of votes the three candidates got together in that respective town.
Leaving aside the effects of campaign appearances on turnout and the added number of votes of all the three candidates, the following results refer strictly to the main research question I pose in the beginning of the paper and at the advanced hypothesis. More precisely, I now turn to the effects of campaign appearances on the number of votes each candidate got separately in the 2009 election.
Table 3.3 below shows that the social democrats' candidate electoral visits had a marginally significant effect on the number of votes he obtained in the elections. For both the matched and the unmatched dataset the high R² indicates that the model has high explanatory power. These findings indicate support for the hypothesis advanced in the introduction, for the independent variable of interest, the number of candidates visiting, is significant. A possible explanation is that the social democrat candidate's campaign was concentrated in those towns where he already had substantive support and that by doing so he managed to persuade voters to cast a vote in his favour on Election Day.
Table 3.3.1 below indicate the results of the regression analysis if an interaction - between the number of votes got by the social democrats in 2008 in a particular town and whether or not the social democrat candidate organized an electoral visit in 2009 - is included as an independent variable. The two variables having a joint effect besides their independent one suggests the fact that organizing electoral visits in those towns where there was already considerable support for the candidate had a positive effect on the number of votes the candidate obtained in the election. Again, this finding gives support for the advanced hypothesis.
The results are even more significant if the interaction is included for the un-matched dataset which points to one of either two possibilities, either Mircea Geoana's campaign was strong enough to mobilize supporters and voters from towns that he did not visit (which is a statement both difficult to prove at this present moment and also highly unlikely) or the high number of votes the social democrats got in the 2008 parliamentary elections in the towns where the candidate did not made an appearance in the 2009 tour still has a high influence on the overall effect. This last statement is supported by the high level of significance as presented for the turnout for the 2008 elections for the social democrats in table 3.3.1.
As distinct from the results just presented, as tables 3.4, 3.4.1, 3.5, and 3.5.1 show in the following pages, the campaign appearances for the incumbent Traian Basescu as well as those for the liberal candidate, Crin Antonescu, had no significant effect on the number of votes the two candidates obtained in the 2009 elections. It has to be pointed out that the direction of the effect, namely having organized visits in a particular town led to an increase in the number of votes obtained by the candidate, is the expected.
The results below can be understood through the prism of the fact that, since Traian Basescu was the president at the moment of the campaign, voters were familiar with his platform and his political trajectory in terms of where he stood on various issues thus his electoral visits did not influence the turnout as he was bringing nothing new to the audience. Also the result can be linked with the relatively limited number of towns the democrat liberal toured in comparison to his opponents, thus the spread of towns visited by the democrat liberal was significantly smaller than that for the social democrat. It is to be noticed that the variable with the most significant effect in both the matched and the unmatched dataset is the number of votes obtained by the liberal democrats in the 2008 elections. Also, as compared to the explanatory power of the previous model, the R² has decreased.
The previous six tables show that there is in fact support for the main hypothesis of the paper, that candidates organizing visits in towns where they have a higher number of supporters did in fact register an increase in the number of votes they got in elections. While this finding is only significant for one of the three candidates, the social democrat, this supports the statement that he organized his campaign appearances in those towns where he got the highest support in the previous election. The other two candidates therefore, it can be said, either organized events choosing towns on a random basis or specifically targeted those towns where their support was low in the effort to change that.
The variable recoding the size of town (expressed in number of inhabitants) is significant for the social democrat candidate and could be understood as having an effect especially if it is seen in ideological terms. Social democrat parties are usually worker parties and, as mentioned previously, in the case Romania, the social democrats are known to have high supporters in the ranks of both the rural and working class population. Since, based on the content analysis, Mircea Geoana made campaign appearances in highly industrialized towns, were most of the population works in the public sector in state owned companies, or in towns were more than 50% of the population earns its leaving from agricultural related jobs, it is understandable why this variable would have an effect.
A downward slope in the appeal for the president Traian Basescu has been cited by opinion polls on a recurrent basis (see appendix 2, 3 and 4), this being one of the possible explanations for the low significance in the scores he got. Another possible reason might be the fact that the liberal democrat candidate made electoral appearances in a limited number of towns and where he didn't particularly command high numbers of supporters.
The second hypothesis put forward was the campaign appearances have influenced the number of votes a candidate gets overall in a particular election. This might be dependent on the number of supporters the party or candidate has in a particular town but also on how spread throughout the country are the visited towns. The hypothesis is supported also by the above tables that presented the interactions. Thus the case of Mircea Geoana, for whom the effect of the electoral appearances is more significant from a statistical point of view, chose towns where he enjoyed higher support. Table 4 below confirms this expectation and shows that from among the three candidates only one did in fact visit in his electoral campaign tour towns where his support was high during the past elections and thus managed to mobilize more its traditional supporters (Mircea Geoana). Hence, for him it becomes clear that visiting towns where his party managed to win the majority of votes during the previous election boosted his support by probably also mobilizing the undecided. As for the other candidates, especially for Crin Antonescu, choosing towns where towns where they did not traditionally had strong support did not prove to be an efficient strategy.
A possible explanation for the significant results obtained by only one candidate, as shown in the table, could be that, Mircea Geoana made campaign appearances in a larger array of towns - as shown by the standard deviation - where he had larger support. Traian Basescu on the other hand did not, based on the same result of the standard deviation (which show a larger spread for the social democrat, 15, as opposed to 8 for the democrat liberal). Thus the assumption could hold since, as mentioned before, Geoana did in fact visit a larger number of towns than Basescu. But again, further research needs to be conducted in order to explore more this possible explanation.
Discussion and Further Assumptions
This paper attempted to show that organizing campaign appearances in key constituencies has an effect on the average voter. That is to say that it influences individuals' decision to go vote for a particular candidate. The analysis conducted focused on one country, Romania and on its latest presidential elections, the 2009 ones. Accordingly, the main hypothesis for this study was that in towns where a particular candidate already enjoyed a high number of supporters, that candidate's electoral visit led to an increase in the number of votes he got. Several additional hypotheses were tested, although they do not refer strictly to the core research problem - the effect of making appearances on the number of votes a candidate gets in election. Thus the secondary hypotheses were that candidates organizing electoral visits in constituencies leads to an increase in the turnout of a particular election and, that candidates organizing electoral visits in constituencies lead to an increase in the added number of votes of those respective candidates.
The analysis performed, matched the towns where the three main candidates organized campaign events with similar towns where the candidates did not organize any such appearances, controlling for the number of votes the parties of each candidate obtained in the previous election, the size of town expressed in number of inhabitants, the rate of unemployment expressed in percentages and the percentage of Hungarians in each town (as Hungarians represent the larges minority in Romania). The matching was done to ensure that, first of all the matched towns and the relevant studied cases were as similar in the variables which may be related to the variable studied but also in order to determine if the variable of interest, the number of votes each candidate got, was in any way influenced by the campaign appearances. The study thus tried to show that candidate appearances in various constituencies during the 2009 presidential elections in Romania have tilted the balance in those respective constituencies for the particular candidate and increased the overall number of votes he got.
The analysis showed that organizing campaign appearances in towns where the candidate has more supporters increased the number of votes got by the candidate. This was most evident and significant from a statistical point of view only for the social democrat candidate. Even if the results of the OLS regression were marginally significant for Mircea Geoana, they do show that there is an effect on the total number of votes he got while results remained not significant for the other two. These findings must be understood through the prism of both the diversity of the towns visited by the social democrat candidate, i.e. diversity here to be understood as diversity in terms of the towns being from different regions of the country (covering all seven regions of development) as well as the number of supporters he had in each town visited. As the main hypothesis suggested, candidates' appearances had an increasing effect on the number of votes that candidate got if the town chosen had a high number of supporters.
The fact that none of the other two candidates managed to produce the same effect on the number of votes they got, only goes to prove that their campaign was not centred on those areas where their supporters highest in number. For the democrat liberal candidate, Traian Basescu, as it was mentioned throughout the paper, the decrease in appeal, as shown by polls conducted prior to elections, can be considered as one of the reasons for the lack of significance in the OLS regression results. Yet, based on the analysis the evident reason is the smaller number in towns visited as compared to his social democrat opponent and also the lack of a high number of supporters in those towns where he organized campaign appearances.
For the liberal candidate, Crin Antonescu, the lack of significance of the results, i.e. the lack of any significant effect of the visits on the number of votes he got is als
Cite This Dissertation
To export a reference to this article please select a referencing stye below: