Issue Convergence and Polarization in the US:
Highlighting Extremism through Dialogue
Karmen Kert (11105585) Master’s Thesis Supervisor: Joost van Spanje
University of Amsterdam Graduate School of Communication Master’s Programme Communication Science
Word count: 6404 03-02-2017
Abstract
What happens if Democrats and Republicans converge on policy positions or on the attention they give to policy issues? Do they, as a result, come across more moderate or can we expect more extreme rhetoric? The existing research has shown that two phenomena – political polarization and issue convergence have simultaneously been present in US politics. While the public has been witnessing a political divergence of Democratic and Republican parties, both parties’ attention to policy issues has greatly overlapped. With this research I looked into how the co-occurrence of the two separate concepts could be interrelated. By aligning measures of issue convergence and polarization I detected no expected recurring relational patterns nor did I find any converse associations. The findings lead me to draw three inferences: the connection between the two concepts is missing altogether, there is a more complex association or the measures utilized were not sufficient enough. I propose that a link might be projected when mediation analysis with negativity and salience is employed.
Issue Convergence and Polarization in the US: Highlighting Extremism through Convergence
Since the 1990s, the United States has dealt with a growing polarization of political parties as well as the electorate (Abramowitz & Saunders, 2008). Scholars measuring polarization have found that with each consecutive election year, the public has faced increasingly extreme presidential candidates in terms of their political dispositions (Bartels, 2016). Especially after the 2016 presidential election, the cleavage between Democrats and Republicans seems to be wider than ever before.
However, a recent article in Foreign Affairs (Zakaria, 2016) points out that Democrats and Republicans are actually not as at odds with each other as one might assume. For
instance, in the 1990s the Democratic president Bill Clinton employed relatively centrist economic policy positions, while the latest election showed how the Republican candidate Donald Trump, although rhetorically preaching for opposite solutions, proposed economic formulas rather similar to Hillary Clinton’s, the Democratic candidate. Therefore, amid polarization, the two sides have actually been converging on some of their policy positions.
In fact, the observed polarization could be conveyed by the use of extreme language rather than diverging policy positions (Gentzkow et al., 2016). For example, although Republicans have chosen the term “illegal aliens” and Democrats have opted for a milder version “undocumented workers”, Barack Obama has deported more people than any other president in the history of the US. This suggests that parties are communicating extreme views by opting for rhetoric that makes them look further apart than they actually are in terms of policy positions. This contradictory co-existence of rhetorical extremism and converging policies begs the question of why both sides appear more extreme than they actually are.
A parallel concept of issue convergence, which has characterized US presidential elections at least since the 1960s, can offer an explanation. The study by Sigelman and Buell
(2004) demonstrates that throughout presidential elections (in the period 1960-2000) candidates of both parties have shown a similar degree of attention towards the same policy issues. For example, in the 1992 campaign Bill Clinton and George H. W. Bush overlapped by 76 per cent in their policy issue attention. Moreover, the overlap of candidates’
attentiveness was around 70 per cent across multiple presidential elections. Although the degree of issue convergence does not indicate an overlap of actual policy positions, the logic of utilizing extreme rhetoric during election can be significant nonetheless.
Since elections tend to include topics that the public deems important, avoiding these issues could cost votes (Ansolabehere & Iyengar, 1994). Candidates are ought to show a similar degree of interest in prominent policy issues such as defense, abortion and taxes. Therefore, during elections, even if sharing policy positions or simply being compelled to address the same issues, both sides need to be discernable in order to maintain electoral appeal. One possibility to differentiate themselves can be the employment of extreme rhetoric. Indeed, what we have seen from election campaigns is that when both sides address the same issues their statements become more extreme. An example from the latest presidential
election in 2016 illustrates that conjecture. During the presidential announcement speech, Republican candidate Donald Trump spoke the following infamous words.
“When Mexico sends its people, they’re not sending their best/.../They’re bringing drugs. They’re bringing crime. They’re rapists. And some, I assume, are good people.”
In contrast, Democratic candidate Hillary Clinton, while addressing the same topic at a campaign stop in Nevada, chose a different route.
“I will fight for comprehensive immigration reform, and a path to citizenship for you and for your families across our country.”
By combining the above-mentioned parallel trends of polarization and issue
convergence, the following question arises. In general, do rival presidential candidates come across more extreme in their campaign statements when engaging in a dialogue? The aim of this thesis is to establish whether during elections, regardless of actual convergence of policies, the polarized aspects of policies become more pronounced when attentions collide. Therefore, I propose the following research question:
To what extent, and how are issue convergence and polarization related in US presidential elections?
This research question is important if converging on issues indeed compels parties to adopt extreme rhetoric. For instance, if the Democratic or Republican elite appears to be increasingly divided, it polarizes the partisan part of the public even further (Layman & Carsey, 2002; Zaller, 1992). Also, as I will argue below, polarization per se can be an incentive for convergence. Therefore, with this thesis I am not searching for a causal relationship but rather exploring if the two separate concepts are interrelated.
In answering this question, I will expand the research on issue convergence as well as contribute to the polarization scholarship by adding a new perspective on how extremity of political parties is conveyed. By investigating polarization and issue convergence jointly, I will offer a fresh point of view according to which the two concepts are seen as linked in the context of elections.
In order to answer the research question, my argument unfolds in four sections. First, I will expand on the theoretical underpinnings of issue convergence and polarization. Second, I will explain the methodology needed to corroborate the hypotheses. In the third section, by observing the ebb and flow of the two concepts, I will show that the separate phenomena do not occur in a pattern that would indicate a relationship. In the fourth section of the thesis, I will analyze the results and propose avenues for future research.
Theory
In this section, I will discuss the concepts of issue convergence and polarization. In the first half, I will look at how convergence compares to theories of optimal campaign strategies – the issue ownership theory (Petrocik, 1996) and the dominance principle (Riker, 1993) – and also how it comes about in the first place. The second half of this section concentrates on polarization and how it can be theoretically linked to issue convergence.
Issue convergence
According to the issue ownership theory (e.g. Budge, 2015; Petrocik, 1996), to win elections political parties should avoid addressing the same issues and instead emphasize those that serve them best. For example, the issue ownership theory states that each party owns issues in which the public sees them as the most competent (Petrocik, 1996): Democrats are seen most capable in dealing with social issues, while Republicans are perceived by voters as having the best record in foreign policy. In order to fare best in elections, both parties should stick to their own issues and circumvent the ones that are more advantageous to the other party (Budge & Fairlie, 1983). Similarly, Riker’s (1993) dominance principle proposes that parties should avoid issues that accentuate the opposing party’s strengths and call
advocate the eschewal of dialogue. The result of successfully adopting these strategies is that voters do not get a clear understanding of where parties stand on similar policy issues.
An alternative perspective argues that even if the avoidance of dialogue is strategically the preferred choice, the actuality of elections forces political parties to trespass issues
(Damore, 2004; Sides, 2006, 2007). This trespassing of issues is also known as issue
convergence (Sigelman & Buell, 2004). Research on issue convergence has demonstrated that the initial starting point, which assumes that political parties tend to talk past each other (i.e. by not addressing the same issues), does not reflect the real election setting. Research has shown that parties do address the same issues (Banda, 2015; Green-Pedersen & Mortensen, 2010; Kaplan et al., 2006). Thus, the theory of issue ownership and the dominance principle might explain the deviations from the more prevalent tendency to converge (Sigelman & Buell, 2004). Moreover, I would argue that not only does convergence represent the rule rather than the exception it also mirrors the course of election campaigns. While the ownership theory explains the initial strategy deemed most beneficial for the parties, convergence illustrates the extent to which this strategy holds up in reality.
Cause of Convergence
Although issue convergence research has only demonstrated the overlap of issues addressed by both candidates during elections, these findings have not uncovered reasons for this phenomenon. However, there are scholars who have provided some evidence and
suggestions.
Sides (2006) has elaborated on why parties tend to trespass each other’s issues by utilizing Ansolabehere and Iyengar’s (1994) “riding the wave” strategy. Even if parties have initially chosen to address the issues that could benefit them the most, focusing on the selected issues is not always an option. During elections, candidates often need to address
issues that the public considers important, even if it means talking about issues that are owned by the competing party. If a candidate does not address issues deemed crucial by voters, he or she could be perceived as dismissive of public concerns, and could subsequently be penalized with a lower vote count (Sides, 2006). Therefore, the convergence of issue attention may be the result of candidates being forced to “ride the wave” rather than abiding by the dominance principle or the ownership theory. In fact, Kaplan and colleagues (2006) found that
convergence could be predicted by the salience of issues.
Another rationale for convergence is the existence of certain universally commendable goals. These issues are also called valence issues (Stokes, 1966). Valence issues are usually assessed to be either exclusively positive or negative. For example, corruption represents an issue that everybody agrees on – less corruption is always desirable. However, the majority of policy issues elicits disagreement and thus can be addressed from different positions – these issues are called position issues (Stokes, 1966) or non-consensual issues (Kaplan et al., 2006). Abortion, government spending and gun laws are some of the issues regularly subject to impassioned disputes. During elections, parties tend to clash on position issues rather than valence issues. While the latter should foster more convergence, that assumption has not been confirmed (Kaplan et al., 2006). Rather, convergence could be determined by the campaign context, which encourages competitors to differentiate themselves to appeal to the electorate.
According to Downs (1957), competing parties in a two-party system should move towards the center where swing voters are. The underlying logic is that the largest amount of votes lies in the middle among median voters. Based on this theory, both parties should adopt moderate policy positions. However, by appealing to the median voter, parties suppress the differences they have and force voters to make decisions with no clear alternatives (Sigelman & Buell, 2004). Similarly, Banda (2015) has argued that by doing so, parties may risk losing the support and votes of their active supporters. That seems to be the case in US presidential
elections. Bartels (2016) has shown that with each year, Republican and Democratic presidential candidates are catering more for the preferences of partisans rather than swing voters. Therefore, while elections are characterized by issue convergence, alongside both parties are increasingly appealing to the partisans.
An additional reason for convergence may be that candidates want to contrast their positions to the opponent’s or respond to the attacks coming from the other side of the aisle (Banda, 2015). Likewise, Sigelman and Buell (2004) explained the possible cause for convergence with a tit-for-tat sequence (p. 659). A “move” from one side elicits a
“countermove” from the other side, which in turn leads to “a continuous process of mutual adjustment exemplified by the instant response” (p. 660). In fact, issue convergence has been shown to occur more often in election settings with higher degree of competitiveness (Banda, 2015; Kaplan et al., 2006). In other words, convergence is a result of, as well as impetus for competition: Campaigns force candidates to highlight contrasts between themselves and their rivals, which in turn draws out conflicts that contribute to convergence.
According to Sides (2006), when trespassing issues owned by the competing party, candidates frame and provide solutions to them based on their ideological traditions. Hence, convergence forces two sides to accentuate their ideological leanings. As a consequence of issue trespassing, “both sides may “diverge” in their rhetoric on issues” (p. 431). This divergence of rhetoric was recently further explored by Gentzkow and colleagues (2016). They used US Congress as an example of how both parties have employed distinct language to differentiate their policy positions.
In conclusion, issue convergence is evoked not only by voters’ interest in certain issues, but also by conflicts that elections induce. This means that during the course of elections both sides are required to talk about the same issues, advantageous or not. By “riding the wave” or being part of a tit-for-that sequence, parties try to emphasize the
differences between them by stressing their respective ideological inclinations in regards to the same issues – a process that leads us to the concept of polarization.
Polarization & Convergence
The concept of political polarization has been dissected in multiple ways. There are scholars who claim that polarization reflects divergence from ideological center towards extremes (Fiorina & Abrams, 2008). Others insist that polarization mirrors the consistency of partisanship and political views (Abramowitz & Saunders, 2008). Affective polarization is said to contrasts the feelings towards the out-party to feelings one has for a party he or she belongs to (Iyengar et al., 2012). Then, there is a perceived polarization that indicates how far the public considers political parties to be polarized (Hetherington, 2008; Westfall et al., 2015).
Whichever the measure or perspective, there is a general consensus among scholars that political parties in the US are polarized. Republicans have become more conservative in their political views and Democrats have drifted towards more liberal stances (Fiorina & Abrams, 2008). However, some findings attest that the electorate is subject to a similar trend too (Abramowitz & Saunders, 2008). In fact, it is the partisan section of the electorate that has become “polarized on all fronts” – regarding divergence, consistency, perception and
affection (Lelkes, 2016:402). Hence, addressing the polarized public instead of moderate voters (Downs, 1957) is not a risk per se. This view is further corroborated by recent research, which showed that adopting extreme positions does not result in a smaller vote count (Cohen et al., 2016). In other words, the US electorate has been “buying” the extremity that
politicians are “selling”.
With this growing extremity that parties have been projecting, the electorate has been placing presidential candidates of Republican and Democratic parties further apart on policy
scales (Bartels, 2016). Also, both Republican and Democratic identifiers have grown to increasingly dislike each other (Iyengar et al., 2012). Moreover, polarization might not reflect the actual policy positions and instead could be the result of using polarized language when presenting policy views (Gentzkow et al., 2016). This is crucial since the way political parties come across affects the way the public adjusts their own views. If both parties persuade their partisan supporters of increasing differences, the latter attune to this trend (Layman & Carsey, 2002), even if the differences are more moderate than the language has led to believe.
However, the party polarization does not have to be negative in itself. Dalton (2008) has pointed out that highly polarized systems help highlight the differences between parties and bring about clear alternatives. Evidently, political parties agree with that too. Spoon and Klüver (2015) pointed out that political parties in Europe are willing to concentrate on polarized issues in order to clearly position themselves for voters. They also demonstrated that parties tend to tackle issues on which the public is already polarized. This gravitation towards polarization is also beneficial for voters, since it supports them in their decision-making (Dalton, 2011).
Hetherington (2001) has showed that polarization among Congressmen and -women has helped citizens identify parties’ positions. Also, as mentioned above, Democratic and Republican presidential candidates have become more concerned about partisans than moderates (Bartels, 2016). Thus, in order to choose a candidate, citizens need to be provided with clear alternatives from which they can choose the one that is closest to them
ideologically and in policy matters. However, voters cannot make necessary comparisons without both parties addressing the same issues (Banda, 2015).
Since research has demonstrated an existence of dialogue between the opposing parties, it is important to examine how much of the existing polarization presents itself through this dialogue. I will base my reasoning on Sides’ (2006) argumentation, which stated
that if both parties happen to or are forced to address the same issues their ideological leanings become more pronounced. Therefore, I would argue that the greater the overlap of issue attention, the more candidates are forced or able to accentuate their positions and the more they come across as extreme in their policy positions, as well as ideological
identification.
Hypothesis 1: The higher the degree of issue convergence during US presidential elections, the more Republicans and Democrats are perceived as extreme in their policy positions.
Hypothesis 2: The higher the degree of issue convergence during US presidential elections, the more Republicans and Democrats are perceived as extreme in their ideological
identification.
Moreover, since the public is responsive to the polarization of elites (Layman & Carsey, 2002), I would expect the ideological consistency and affective polarization to increase among the electorate when convergence is high.
Hypothesis 3: The higher the degree of issue convergence during US presidential elections, the higher is the affective polarization among the electorate.
Hypothesis 4: The higher the degree of issue convergence during US presidential elections, the stronger is the ideological consistency among the electorate.
In addressing these hypotheses, I will examine possible constellations of issue convergence and polarization. I will show how issue convergence can be inter-related with polarization of political parties as well as the public.
Method
In order to answer the research question and to examine the hypotheses I worked with two separate data sets. Secondary data was taken from the American National Election Studies (ANES) and served the purpose of indicating polarization. Primary data comprised of statements that I collected from the New York Times (NYT) and was utilized to measure issue convergence. The aim was to explore the extent to which the issue convergence and polarization co-occur under the circumstances of presidential elections. According to
Abramowitz and Saunders (2008), since 1992 polarization among the political elite has grown dramatically, and for that reason I concentrated on a time frame starting in 1992 and ending in 2012. Because the ANES had only conducted a Pilot Study for 2016, that year was not
included in the analysis.
To detect the inter-relation on a more detailed level I concentrated on three policy issues Defense, Government spending, and Guaranteed jobs and living. These topics have constantly been part of the ANES surveys and therefore correspond to the selected time frame. I am focusing on how many times both candidates addressed those selected issues, and how much their respective parties were perceived by the ANES respondents as polarized in regard to those issues. To provide a more comprehensive overview I paralleled more than one measure of polarization with overall issue convergence.
Secondary Data and Measures
The ANES data provided items to compute different polarization measures. I concentrated on 3 different measures: perceived polarization, affective polarization and consistency of ideology.
For the perceived polarization of political parties I used overlap coefficients computed by Lelkes (2016). These coefficients were calculated using ANES Time Series dataset. The coefficients represent the degree of overlap or common ground between the Republican and the Democratic parties. Lelkes assessed the perceived polarization between the two parties on several policy issues (including those covered in this thesis) and ideological identification. All policy issues as well as ideological identification were measured on a 7-point scale in the ANES surveys (For the exact wording, see Appendix A). The formula for the overlap coefficient (OC) is
where f(x) represents the distribution of ideology or placement on policy issue scales among Republicans, and g(x) represents the distribution of ideology or placement on policy issue scales among Democrats. An OC of 1 indicates a total overlap and hence no polarization and an OC of 0 points to a complete separation and therefore high polarization. Overlap
coefficients computed by Lelkes (2016) can be found in Table 1.
Affective polarization was computed by utilizing feeling thermometer items from the ANES surveys from 1992 until 2012. In these surveys participants were asked to place various societal groups on a scale from 0 to 100. Groups towards whom a respondent felt warm or positive feelings were placed between 50 or higher, while groups that elicited negative feelings were placed between a range of 0 to 50.
Table 1
Perceived Polarization of Republican and Democratic parties, 1992–2012 (overlap coefficients) 1992 1996 2000 2004 2008 2012 Defense 0.60 0.68 0.69 0.44 - 0.61 Gov. Spending 0.55 0.45 0.46 0.53 - 0.37 Jobs/Living 0.51 - 0.57 0.46 - 0.38 Ideological identification 0.68 0.66 0.54 0.71 0.82 0.72
To compute the measure I discerned two groups: respondents who identified
themselves as or were leaning towards Democrats and the ones who identified with or leaned towards Republicans (For the exact wording, see Appendix A). I focused on how both of these groups felt towards the Republican and the Democratic parties and their presidential candidates. To measure affective polarization in-groups and out-groups were identified: Democratic identifiers’ (including leaners) in-group was the Democratic party (or candidate) and out-group was the Republican party (or candidate), and vice versa. In case of high polarization, the feeling thermometer score for in-group is supposed to be higher than for the out-group.
First, I subtracted Democratic and Republican identifiers’ out-party mean scores from their in-party mean scores. Next, I took an overall average of both groups’ mean scores. In the end I had two affective polarization scores for each election year. The first represented an average score Democrats (and leaners) and Republicans (and leaners) combined gave to an out-party compared to an in-party. The second reflected the same, but towards the Republican and Democratic presidential candidates (Figure 1). The higher the score the more affectively polarized the public is.
The ideological consistency measure, which was used to validate the forth hypothesis, was computed by correlating ANES respondents’ self-placement on 7-point ideological
identification scale with their position on 7-point liberal-conservative identification scale (For the exact wording, see Appendix A). It was the same measure Abramowitz & Saunders used in their 2008 study. The higher the correlation (Pearson r) the stronger the ideological consistency is (Figure 2).
Figure 1: Affective polarization: Mean difference between in- and out-group, differentiated by presidential candidates and parties, 1992–2012.
Figure 2: Ideological consistency: correlation of party identification and liberal-conservative identification, 1992–2012. 0.00 0.20 0.40 0.60 0.80 1992 1996 2000 2004 2008 2012 Correlation (Pearson’s r) 0 10 20 30 40 50 60 1992 1996 2000 2004 2008 2012
Primary Data and Measures
Data I collected comprised of statements made either by presidential candidates or their spokespersons. I extracted all the statements from news articles that were published in the New York Times – the newspaper of record – from 1992 until 2012, and between the Labor Day and the Election Day (as in Sigelman and Buell, 2004). All the quotes and paraphrases from candidates and spokespersons were considered as statements. In other words, a statement could have been just a one sentence or an entire paragraph that used paraphrases and/or direct quotes from a candidate about a certain issue.
All the articles from which the statements were taken met the following criteria: Referenced an election, included at least one of the presidential candidates, and were
published on the National Desk section. The selection of articles was limited even further by the emphasis on certain topics. Selected articles were either about Defense, Public Finance, Welfare or Job Creation. The reason I focused on these subjects was that they correspond to the three selected issues provided by the secondary data (see Table 2). The search in
LexisNexis produced 354 articles, minus three articles that were transcripts of debates.
Table 2
ANES Issues in the Questionnaire Compared to Search Subjects in LexisNexis
Not all the articles included relevant statements. In the end, I extracted 762 statements by presidential candidates or their spokespersons from 244 news articles published in the NYT during presidential elections from 1992 until 2012. Each statement was coded (For the ANES issue topics in the questionnaire: NYT news subjects in LexisNexis:
Defense spending “Defense”
Government Services-Spending “Public Finance”
Codebook, see Appendix B) either to have been said by a Republican presidential candidate (or his spokespersons) or by a Democratic candidate (or his spokespersons). Also, each statement was categorized under one of the following topics: Defense, Government Spending or Welfare/Job Creation (Krippendorff’s alpha (N =78) = 0.94). The distribution of statements in regard to the three topics can be found in Table 3 and the number of statements by each presidential candidate in Table 4.
Table 3
Frequencies of Issues in Presidential Campaign Statements, 1992–2012 (number of statements & percentages)
Table 4
Candidates’ Issue Attention in Presidential Campaigns, 1992–2012 (number of statements)
Issues: 1992 1996 2000 2004 2008 2012 Defense 34 (19.2) 7 (12.3) 26 (22.4) 224 (72.0) 3 (4.1) 3 (10.7) Spending 83 (46.9) 30 (52.6) 85 (73.3) 66 (21.2) 46 (63.0) 16 (57.1) Jobs/Living 60 (33.9) 20 (35.1) 5 (4.3) 21 (6.8) 24 (32.9) 9 (32.1) Total: 177 57 116 311 73 28 1992 1996 2000 2004 2008 2012 Issues: R D R D R D R D R D R D Defense 10 24 4 3 16 10 108 116 3 0 2 1 Spending 41 42 14 16 52 33 33 33 27 19 8 8 Jobs/Living 29 31 6 14 0 5 8 13 13 11 6 3 Total: 80 97 24 33 68 48 149 162 43 30 16 12
In order to calculate the extent both sides converged on a specific issue, I adapted the measure proposed in the study by Kaplan and others (2006, p. 735):
(1−|(MDi − MRi )/(MDi + MRi )|)∗100
where MDi and MRi represent the number of times Democratic and Republican presidential candidates or their spokespeople addressed a certain issue. For example, if a specific issue was mentioned 65 times by a Democratic candidate and 45 times by a Republican candidate, the convergence score for that issue would be (1–|(65 – 45)/(65 + 45)|)*100= 82%. This measure, as pointed out by Sigelman and Buell (2004) does not help to uncover the issue-positioning strategies of both candidates. However, because the aim was to show whether both sides addressed the same issue to a similar degree, the opted measure served its purpose.
In addition, to measure the overall issue attention overlap in three selected topics (Defense, Government spending, and Guaranteed jobs and living), I used a measure proposed by Sigelman and Buell (2004):
where PD and PR were the percentages of Democratic and Republican presidential candidates’ total attention allocated to a particular issue. The absolute differences between both
candidates’ issue attention were summed over all n of the potential issues in a campaign. For the measure to range between 0 and 100 the sum was divided by 2. Subtracting from 100 converted the measure to one of similarity rather than dissimilarity (Sigelman & Buell, 2004). Scores of specific issue and overall convergences can be found in Table 5.
Table 5
Issue Convergence in Presidential Campaigns, 1992–2012 (percentages)
Issues 1992 1996 2000 2004 2008 2012
Defense 58.8 85.7 76.9 96.4 0.0 66.7
Spending 98.8 93.3 77.6 100.0 82.6 100.0
Jobs/Living 96.7 60.0 0.0 76.2 91.7 66.7
Overall 87.8 82.6 89.6 97.3 93.0 83.3
To answer the research question and verify the hypotheses two separate steps were taken. First, I compared the overlap coefficients of perceived polarization on each of the selected issues to the respective issue convergences. I paralleled the two measures onto a graph to analyze the co-occurrence. That action generated three graphs for each issue
(Defense, Government spending, and Guaranteed jobs and living). For example, on the topic of Defense I measured how much both parties’ candidates addressed defense issues during presidential campaign (issue convergence), and compared that to the degree of overlap of both parties’ positions on a defense scale (overlap coefficient).
Second, I focused on the overall issue convergence and how it is related to various polarization measures. I set side by side the overall convergence and the following
polarization measures: the overlap coefficients of perceived polarization on
liberal-conservative scale, the two sets of affective polarization scores (parties and candidates), and the ideological consistency correlation scores.
To conduct the analysis and compare convergence to different polarization measures, all the scores were converted to range from 0 to 1.
Results
The aim of the thesis was to discern the ebb and flow of polarization and issue convergence, and to detect relational patterns between the two concepts. Because of the
limited number of data points, it was impossible to subject any of the hypotheses to formal testing. For that reason, the research question and the hypotheses are answered by describing the recurrent pattern of polarization and issue convergence.
The first hypothesis expected a high level of perceived polarization to co-occur with high issue convergence. I explored this association by comparing the convergence on the three selected policy issues with the overlap coefficients adopted from Lelkes’ study (2016).
On the subject of Defense, across most of the observed election years (excl. 2004), overlap coefficients remained around 0.60, while convergence fluctuated from 0.60 to 0.96 (Figure 3). Overall, convergence and overlap coefficients took on similar values, which indicates that the degree of overlap of issue attention is equal to the overlap of perceived policy positions. However, in 2004 the expected relationship occurs: the highest score of issue convergence (0.96) is accompanied by the lowest overlap coefficient (0.44), indicating high polarization.
Figure 3: The co-occurrence of issue convergence and polarization (overlap coefficient) on the topic of Defense, 1992–2012.
0.00 0.20 0.40 0.60 0.80 1.00 1992 1996 2000 2004 2012
Government spending graph (Figure 4) demonstrates that across the election years presidential candidates have converged on the topic of government spending to a high degree (lowest convergence being 0.78). Overall, compared to the subject of Defense a high level of issue convergence on government spending is accompanied by smaller overlap coefficients, which indicate higher polarization. However, the convergence and polarization measures increase and decrease separately without adhering to the expected pattern. Therefore almost none of the selected election years offer support for the first hypothesis. The closest result is provided by the 2012 election: With 100 per cent of convergence, there is the lowest degree of perceived overlap (0.37) between the two parties.
Figure 4: The co-occurrence of issue convergence and polarization (overlap coefficient) on the topic of Government Spending, 1992–2012.
On the subject of Guaranteed jobs and living (Figure 5) Republicans and Democrats converge to a high degree (0.97) in 1992, while their perceived positions on policy overlap by half (0.51). This result in itself supports the first hypothesis. However, other election years do not yield confirming results. In 2000, when there is zero issue convergence, both parties’
0.00 0.20 0.40 0.60 0.80 1.00 1992 1996 2000 2004 2012
distribution overlaps by more than half (0.58) – a result that counters the initial assumption. Also, another incident of low convergence in 2012 (0.67) coincides with high polarization (overlap coefficient 0.38). Thus, on a subject of Guaranteed jobs and living, the observed years offer contradictory findings.
In conclusion, observing issue convergence and perceived policy polarization across elections does not provide clear patterns to neither support nor contradict the first hypothesis.
Figure 5: The co-occurrence of issue convergence and polarization (overlap coefficient) on the topic of Guaranteed Jobs and Living, 1992–2012.
Other hypotheses (H2, H3, H4) predicted a high degree of issue convergence to go hand in hand with high degree of perceived ideological polarization, increased affective polarization and stronger ideological consistency. The results show that none of the polarization measures decline or increase in consistent patterns that would indicate the expected co-occurrence with issue convergence (Figure 6).
0.00 0.20 0.40 0.60 0.80 1.00 1992 2000 2004 2012
Regarding the perceived ideological polarization (H2), most of the election years do not reveal an expected ebb and flow. When the overall convergence goes up polarization goes down (from 2000 to 2004). During some of the election years if convergence declines (1992 to 1996 and 2008 to 2012) polarization goes up, which is the opposite of what the second hypotheses presumed. However, some of the election years partly demonstrated the expected relation. For example, in 2000 when convergence goes up (0.90) compared to the previous election year, the perceived ideological polarization increases as well (0.54). Similarly, in 2008 as convergence declines (0.93) polarization does too (0.82). Therefore, while a few of the election years provide some support for the second hypothesis, there is no consistent pattern between the overall convergence and the perceived ideological polarization to validate it.
Figure 6: The co-occurrence of issue convergence and various measures of polarization, 1992–2012. 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 1992 1996 2000 2004 2008 2012 Affective polarization (candidates) Affective polarization (parties) Ideological Consistency Issue Convergence Perceived polarization (overlap coefficient)
The third and forth hypotheses expected affective polarization and ideological consistency measures to rise and descend in concert with issue convergence. In most of the instances the drop in convergence is paralleled with an increase in the polarization measures. For example, negative feelings towards out-party steadily climbs from 1992 to 2012, while convergence fluctuates across elections. Also, contrary to expectations, in 2000 when issue convergence moves upward, ideological consistency and affective polarization (candidates) decline. However, from 2000 to 2004 when the convergence measure ascends (from 0.90 to 0.97) so does affective polarization of presidential candidates (from 0.30 to 0.41).
Furthermore, the co-occurrence of the overall issue convergence and affective polarization (candidates) continues to 2008 when both measures go down (0.93 and 0.33, respectively). The few corroborative instances were not sufficient enough to substantiate the third and forth hypotheses.
In conclusion, while polarization has steadily increased by most measures since 1992, issue convergence has alternated across elections. Therefore, paralleling the two phenomena did not produce consistent patterns and offered only a few examples of the expected
associations. Hence, none of the four hypotheses were corroborated by the results.
Discussion
With this thesis I investigated the connection between issue convergence and
polarization. I hypothesized that with high issue convergence different types of polarization become more pronounced. By addressing the inter-relation on a single issue and on an overall issue attention level, I offered a detailed and a comprehensive view of the association. The results however did not validate any of the hypotheses. The presumed relational pattern was missing on a single issue and on the overall convergence level. None of the polarization measures demonstrated the expected ebb and flow in relation to convergence. Instead,
fluctuating convergence was accompanied by a steady growth of almost all the polarization measures (except the perceived polarization). Moreover, I did not find a consistent converse pattern either: i.e. high convergence coinciding with low polarization. These results suggest that either there is no link between the two phenomena or the connection is more complex than initially expected.
I conjectured that the nexus between issue convergence and polarization is the use of polarized rhetoric. However, the rhetoric was an implied component of the association and was not measured nor incorporated as a factor into the analysis. This assumption led to a few predicaments.
First, even if both sides do apply a more polarized rhetoric, it might not be as
detectable to the public as expected. For example, when Republicans talk about “Obamacare” and Democrats about “Affordable Care Act”, it might not be evident to the public that the two sides hold opposing positions on the same issue. Studying the opinions of a more
well-informed part of the electorate could produce results closer to what was hypothesized. Second, the extreme rhetoric might not actually be present in each time rival presidential candidates enter a dialogue. Since the convergence measures did not incorporate the rhetoric used, it is unclear whether the association between the two concepts comes about via extreme rhetoric. For example, rival candidates can employ either moderate or extreme language when converging on a topic and therefore polarization can be disguised. Thus, the quantity of overlap in itself does not best reflect the way polarization is conveyed. Instead a more detailed study of how both sides converge might be more revealing.
The presence of negativity in dialogues might be a better indicator. Previous research has shown that campaigns become more negative with increasing frequency of attacks (Geer, 2010), which in turn increases affective polarization among the electorate (Iyengar et al., 2012). For example, in a campaign where high issue convergence is coupled with a low
degree of negativity, polarization might not be as pronounced as it would be in a campaign where both sides hardly converge but show a lot of negativity.
In their research, Sigelman and Buell (2004) not only measured the convergence on policy, but also on campaign (campaign finance, dirty tricks etc.) and candidate issues
(ideology, behavior, record etc.). If convergence on the latter types of issues were additionally paralleled with polarization, maybe the co-occurrence would be more in line with the
proposed hypotheses. While negative advertisements attacking policy positions have been more common (Geer, 2010), the share of attacks on character has been notable too. For example, during 2012 presidential elections, 42 per cent of advertisements addressed the character of a rival (Benoit & Glantz, 2015). To illustrate further, during 1992 campaign Republicans focused heavily on Bill Clinton’s draft record, and in 2008 they tried to underline Obama’s lack of experience. Therefore, if a large portion of attacks is on character, the
concentration on policy issues would not fully reflect campaigns and therefore would not be sufficient to indicate polarization.
The concentration on three policy positions brought forth a few other problems. The selected campaign topics might not have been salient enough across the set of elections and due to practical considerations other prominent campaign topics were excluded from analysis. As was argued by Sides (2006), if both sides have to address the same issue (e.g. because of its salience), they also highlight their differences. In 2004 presidential candidates John Kerry and George W. Bush addressed the issues of guaranteed jobs and living to a similar degree; however, the issue might not have been in the forefront. That in turn could mean that both sides did not deem it necessary to address the topic in a polarized manner. As a result, the exact extent of complete issue convergence could not be captured.
Furthermore, since I measured issue convergence by extracting statements from the New York Times, I offered a filtered perception of the issue attentions. Extracting statements
directly from campaigns (as Kaplan et al., 2006) or from different media would have produced a more comprehensive picture. Also, because of time constraints, I was able to cover only six elections. With more data points I might have detected the expected or contradictory patterns and could have paired it with formal testing.
In addition, the overlap coefficients borrowed from Lelkes (2016) reflected the electorate’s perception towards the parties not their presidential candidates. However, since I measured issue convergence of presidential campaigns, the perception about the candidates would have been more appropriate. As Bartels (2016) demonstrated, presidential candidates have become increasingly more extreme and, in fact, Democratic candidates have become even more extreme than their own partisan base. Also, my own research showed that the assessment of parties and their candidates do not overlap entirely: i.e. with affective polarization, the feelings towards candidates were overall more negative than towards the party. Therefore, equating parties and their candidates can skew research outcome.
The results of this thesis revealed an interesting pattern: The perceived polarization measure – the overlap coefficient – did not align with other polarization measures. Instead of steady increase, as expected from previous research (e.g. Fiorina & Abrams, 2008, Iyengar et al., 2012), the overlap coefficients (of a single issue as well as of ideological identification) showed a more inconsistent pattern. The probable explanation is that the coefficients represented the views of both partisans and moderates, while the other measures were computed by differentiating ideological leanings. In fact, Lelkes (2016) has pointed out that perceived polarization is entirely prompted by partisans.
The assumptions that undergird this thesis are based on previous research about the two parallel phenomena that are present in the US. Therefore, since the premise and the measures were US-specific, the generalizability to other systems is questionable. I would
argue that when there are more than two parties involved the nuances of this relation multiply and issue competition as well as polarization become more complex.
Since the results were contradictory and did not validate nor offer any alternative patterns, I cannot conclude that the two phenomena are interrelated. Having said that, there is a possibility of a link between the two concepts. Regrettably, the measures I chose were not sufficient enough to prove that connection. For future research I suggest an extension of Sigelman & Buell (2004) research in its entirety, including policy issues as well as campaign and candidate issues. Another option would be to conduct content analysis, which would not only record the overlap, but also the salience of issues, and the presence of negativity or conflict. A mediation analysis via negativity and salience could uncover the missing link between convergence and polarization.
References
Abramowitz, A. I., & Saunders, K. L. (2008). Is polarization a myth?. The Journal of Politics, 70(02), 542-555.
Ansolabehere, S., & Iyengar, S. (1994). Riding the wave and claiming ownership over issues: The joint effects of advertising and news coverage in campaigns. Public Opinion Quarterly, 58(3), 335-357.
Banda, K. K. (2015). Competition and the dynamics of issue convergence. American Politics Research, 43(5), 821-845.
Bartels, L. M. (2016). Failure to Converge Presidential Candidates, Core Partisans, and the Missing Middle in American Electoral Politics. The ANNALS of the American Academy of Political and Social Science, 667(1), 143-165.
Benoit, W. L. (2001). The functional approach to presidential television spots: Acclaiming, attacking, defending 1952–2000. Communication Studies, 52(2), 109-126.
Benoit, W. L., & Glantz, M. (2015). A functional analysis of 2008 general election presidential TV spots. Speaker & Gavel, 49(1).
Budge, I. (2015). Issue emphases, saliency theory and issue ownership: a historical and conceptual analysis. West European Politics, 38(4), 761-777.
Budge, I., & Farlie, D. J. (1983). Explaining and predicting elections: Issue effects and party strategies in twenty-three democracies. London, England: Unwin Hyman.
Cohen, M., McGrath, M. C., Aronow, P., & Zaller, J. (2016). Ideologically Extreme Candidates in US Presidential Elections, 1948–2012. The ANNALS of the American Academy of Political and Social Science, 667(1), 126-142.
Dalton, R. J. (2008). Citizenship norms and the expansion of political participation. Political studies, 56(1), 76-98.
Dalton, R.J. (2011). Left-right orientations, context and voting choices. In R.J. Dalton & C.J. Anderson (eds), Citizens, context and choice: How context shapes citizens’ electoral choices. Oxford: Oxford University Press.
Damore, D. F. (2004). The dynamics of issue ownership in presidential campaigns. Political Research Quarterly, 57(3), 391-397.
Downs, A. (1957). An economic theory of political action in a democracy. Journal of Political Economy, 65(2), 135-150.
Fiorina, M. P., & Abrams, S. J. (2008). Political polarization in the American public. Annual Review of Political Science, 11, 563-588.
Geer, J. G. (2010). Fanning the flames: The news media’s role in the rise of negativity in presidential campaigns. Joan Shorenstein Center on the Press, Politics and Public Policy Discussion Paper Series.
Gentzkow, M., Shapiro, J. M., & Taddy, M. (2016). Measuring polarization in high-dimensional data: Method and application to congressional speech (No. w22423). National Bureau of Economic Research.
Green- Pedersen, C., & Mortensen, P. B. (2010, June). Issue competition and election campaigns: Avoidance and engagement. In ESF Workshop on Policy Agendas, Political Parties and Their Positions, Manchester.
Hetherington, M. J. (2001). Resurgent mass partisanship: The role of elite polarization. In American Political Science Association, 95(3), 619-631.
Hetherington, M. J. (2008). Turned off or turned on? How polarization affects political engagement. Red and blue nation, 2, 1-33.
Iyengar, S., Sood, G., & Lelkes, Y. (2012). Affect, not ideology a social identity perspective on polarization. Public opinion quarterly, 76(3), 405-431.
Kaplan, N., Park, D. K., & Ridout, T. N. (2006). Dialogue in American political campaigns? An examination of issue convergence in candidate television advertising. American Journal of Political Science, 50(3), 724-736.
Layman, G. C., & Carsey, T. M. (2002). Party polarization and party structuring of policy attitudes: A comparison of three NES panel studies. Political Behavior, 24(3), 199-236.
Lelkes, Y. (2016). Mass Polarization: Manifestations and Measurements. Public Opinion Quarterly, 80(S1), 392-410.
Petrocik, J. R. (1996). Issue ownership in presidential elections, with a 1980 case study. American journal of political science, 825-850.
Riker, W. H. 1993. “Rhetorical Interaction in the Ratification Campaigns.” In Agenda Formation, ed. William H. Riker. Ann Arbor: University of Michigan Press. Zakaria, F. (2016, December/November). Populism on the March: Why the West Is in
Trouble. Foreign Affairs, 95(6), 9-15.
Sides, John (2006). “The Origin of Campaign Agendas”, British Journal of Political Science, 36(3), 407-436.
Sides, John (2007). “The Consequences of Campaign Agendas”, American Politics Research, 35(4), 465-488.
Sigelman, L., & Buell, E. H. (2004). Avoidance or engagement? Issue convergence in US presidential campaigns, 1960–2000. American Journal of Political Science, 48(4), 650-661.
Spoon, J. J., & Klüver, H. (2015). Voter polarisation and party responsiveness: Why parties emphasise divided issues, but remain silent on unified issues. European Journal of Political Research, 54(2), 343-362.
Stokes, D. E. (1966). Some dynamic elements of contests for the presidency. American Political Science Review, 60(01), 19-28.
Zaller, J. (1992). The nature and origins of mass opinion. Cambridge university press. Westfall, J., Van Boven, L., Chambers, J. R., & Judd, C. M. (2015). Perceiving political
polarization in the United States: Party identity strength and attitude extremity exacerbate the perceived partisan divide. Perspectives on Psychological Science, 10(2), 145-158.
Appendix A Questions in ANES Guaranteed jobs
Some people feel that the government in Washington should see to it that every person has a job and a good standard of living. (1972–1978: Suppose these people are at one end of a scale, at point 1). Others think the government should just let each person get ahead on his/her own. (1972–1978: Suppose these people are at the other end, at point 7. And, of course, some other people have opinions somewhere in between, at points 2,3,4, 5, or 6.) Where would you place the (Democratic Party/Republican Party) on this scale? (1. Government see to job and good standard of living 2. 3. 4. 5. 6. 7. Government let each person get ahead on his own).
Defense spending
Some people believe that we should spend much less money for defense. (1996, 2004: Suppose these people are at one end of a scale, at point 1.) Others feel that defense spending should be greatly increased. (1996, 2004: Suppose these people are at the other end, at point 7. And, of course, some other people have opinions somewhere in between at points 2, 3, 4, 5, or 6.) Where would you place the (Democratic Party/Republican Party) on this scale? (1. Greatly decrease defense spending 2. 3. 4. 5. 6. 7. Greatly increase defense spending.)
Government spending
Some people think the government should provide fewer services, even in areas such as health and education, in order to reduce spending. (1996 AND LATER: Suppose these people are at one end of a scale, at point 1). Other people feel that it is important for the government to provide many more services even if it means an increase in spending. (1996: Suppose these people are at the other end, at point 7. And, of course, some other people have opinions
somewhere in between, at points 2, 3, 4, 5, or 6.) Where would you place the (Yourself/ Democratic Party/Republican Society) on this scale? (1. Government should provide many fewer services: reduce spending a lot 2. 3. 4. 5. 6. 7. Government should provide many more services: increase spending a lot).
Party identification
Do you think of yourself as closer to the Republican or Democratic party? (1. Strong Democrat 2. Weak Democrat 3. Independent– Democrat 4. Independent–Independent 5. Independent–Republican 6. Weak Republican 7. Strong Republican)
Partisanship
Party Identification of Respondent- Summary 3-Category. Collapsed from item Party identification. 1. Democrats (including leaners) 2. Independents 3. Republicans (including leaners)
Ideological identification
We hear a lot of talk these days about liberals and conservatives. I’m going to show you (1996 AND LATER: Here is) a seven-point scale on which the political views that people might hold are arranged from extremely liberal to extremely conservative. Where would you place the (Yourself/Democratic Party/Republican Party) on this scale? (1. Extremely liberal 2. Liberal 3. Slightly liberal 4. Moderate, middle of the road 5. Slightly conservative 6.
Appendix B
Codebook for Content Analysis Only one option can be selected for each item,
1. Coder ID: Initials of the coder, in case coders have similar initials an identifying number is added.
2. Statement: Copy and paste the statement found from the news article 3. Year the statement appeared in NYT: Write a number
4. Who does the statement belong to – Democratic or Republican candidate/spokesperson?
Spokespersons can be vice-presidential candidates, campaign managers and surrogates who promote the message and speak on behalf of presidential candidate.
1. Democrat 2. Republican
5. What is the main issue of the article: 1. Defense
2. Government spending 3. Welfare/Job creation
