UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Responding to ideological change : partisan sorting in contemporary democracies, 1996-2015 Major, Tiana 2016

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


24-ubc_2016_november_major_tiana.pdf [ 386.41kB ]
JSON: 24-1.0314146.json
JSON-LD: 24-1.0314146-ld.json
RDF/XML (Pretty): 24-1.0314146-rdf.xml
RDF/JSON: 24-1.0314146-rdf.json
Turtle: 24-1.0314146-turtle.txt
N-Triples: 24-1.0314146-rdf-ntriples.txt
Original Record: 24-1.0314146-source.json
Full Text

Full Text

     RESPONDING TO IDEOLOGICAL CHANGE: PARTISAN SORTING IN CONTEMPORARY DEMOCRACIES, 1996-2015  by  Tiana Major   A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF ARTS  in  THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES  (Political Science)   THE UNIVERSITY OF BRITISH COLUMBIA  (Vancouver)   August 2016   © Tiana Major, 2016 ii  ABSTRACT This paper examines whether shifts in a party’s ideological positioning affect their own partisans’ left-right self-placements. When faced with a change in their party’s position, partisans are likely to adapt to this change, either through switching partisanship to another party, or updating their own ideology to match that of the party. Together, this process of partisan sorting leads to greater ideological alignment between parties and partisans. This paper examines whether there is evidence of partisan sorting in 23 contemporary democracies from 1996 to 2015. Using the Comparative Study of Electoral Systems and Manifesto Project data, the analysis finds evidence that partisan sorting may be occurring, but only in polarized party systems.   iii  PREFACE This thesis is original, unpublished, independent work by the author, Tiana Major.   iv  TABLE OF CONTENTS ABSTRACT…………………………………………………………………………………........ii PREFACE………………………………………………………………………………………...iii TABLE OF CONTENTS………………………………………………………………………....iv LIST OF TABLES………………………………………………………………………………...v LIST OF FIGURES………………………………………………………………………………vi INTRODUCTION…………………………………………………………………………...........1 LITERATURE REVIEW…………………………………………………………………………3      Partisanship…………………………………………………………………………………….3      Ideological positioning…………………………………………………………………………5      Perception of changes in left-right positioning………………………………………………...6      Partisan sorting: reacting to shifts in left-right positioning……………………………………8 DATA AND METHODS………………………………………………………………………..12 ANALYSIS AND RESULTS……………………………………………………………………15 CONCLUSION…………………………………………………………………………………..23 REFERENCES…………………………………………………………………………………..25   v  LIST OF TABLES Table 1     Regression results for changes in perceived party position…………………………..19 Table 2     Regression results for change in partisan self-placement…………………………….20 Table 3     Regression results for changes in perceived party position (by polarization)………..20 Table 4     Regression results for change in partisan self-placement (by polarization)………….21   vi  LIST OF FIGURES Figure 1     Correlation between real and perceived party change……………………………….15 Figure 2     Correlation between perceived party change and partisan change………………….17 Figure 3     Correlation between real party change and partisan change………………………...18   1  INTRODUCTION How do partisans react to changes in their party’s ideological position? There is an extensive literature on why parties place themselves in a certain position on the left-right ideological spectrum (e.g. Downs, 1957; Rabinowitz & Macdonald, 1989; Adams & Somer-Topcu, 2009b). Of course, the core elements of a party’s ideology plays a large part in where they place themselves, but there remains a range of possible positions; parties shift left or right between elections based on what they believe will bring about electoral success. This paper seeks to examine whether partisans perceive shifts in their party’s ideological positioning, and whether they react to such changes. Partisans are essential to every viable party, as they provide a level of electoral stability and core voters (Dalton & Wattenberg, 2000). If a party’s position is not congruent with that of their partisans, and their partisans are aware of this discrepancy, they risk alienating their core voter base. If partisans perceive a shift in the ideology of the party, there are two main ways they can react in order to keep their partisanship and ideology in alignment: the first is to update their own ideology to be more congruent with that of the party (party persuasion), and the second is to change their partisan attachment to a different party which better represents their ideology (party switching). Taken together, these processes are known as partisan sorting: the act of bringing partisanship and ideology into alignment (Fiorina & Levendusky, 2006; Levendusky, 2010b). Much of the research on partisan sorting has been done in the United States, and while these findings are valuable and may suggest a similar phenomenon is possible in other countries, expanding this literature is certainly necessary to better understand the relationship between party positions and partisan ideology. Many scholars have argued that American partisanship is 2  particularly durable; almost regarded as an inherited identity which changes rarely, and only under extreme cross-pressures. For this reason, and the relative uniqueness of the party system, the issue merits testing in other democracies. If the nature of partisanship varies between countries, then we might not necessarily expect this sorting phenomenon to occur elsewhere. Understanding how changes in party positioning affect partisans is also important in evaluating the health of a democracy. If parties can change their ideological position however they wish and citizens do not perceive such changes, this would provide a quite dismal view of the capacity of the electorate to be aware of their political environment and make informed decisions. Furthermore, whether partisans adapt to such changes in ideological positions also has normative consequences. Most people would agree that in an ideal democracy, citizens would all make informed voting decisions based on which party will best represent their interests. It would be quite troubling if parties and their partisans have ideologically divergent views, since governing parties would not even represent the views of those that elected them. This paper first reviews the existing literature on the relationship between party and partisan ideological positioning. This is followed by a description of the data and methods used in the subsequent analysis. In the analysis, I first examine whether partisans perceive changes in left-right party positioning, as conveyed by their campaign manifestos. Second, I analyze whether partisans adapt to such changes in positioning. Since the two main ways partisans can adapt – party persuasion and party switching – appear the same at the aggregate level (partisan ideology tracks that of the party) this paper will not attempt to discern which of these activities is primarily responsible for partisan sorting. The paper concludes with a discussion of the implications of these findings. 3  LITERATURE REVIEW Partisanship In order to properly examine whether and how partisans adapt to party position change, we must first understand the dynamics of partisanship. There are two main competing views of partisanship. The first, known as the Michigan model, views partisanship as a durable psychological attachment which is stable, generally acquired during the formative years and passed down in families, similar to religion and other values (Campbell et al., 1960; Miller & Shanks, 1996; Green, Palmquist & Shickler, 2002). In this view of partisanship, it is one of the most important factors in shaping how individuals interact with their political environment, comparable to a lens through which individuals perceive their environment – shaping preferences, perceptions and reasoning (Ibid.). There is plenty of evidence that citizens’ political attitudes and perceptions are strongly shaped by their existing partisanship (e.g. Evans & Pickup, 2010; Westen et al., 2006; Taber & Lodge, 2006). Campbell et al. describe it is as a “perceptual screen through which the individual tends to see what is favorable to his partisan orientation” (1960, p. 133). In terms of psychology, partisanship invokes motivated reasoning – an individual seeks out information that confirms their existing partisan beliefs, and discounts information that does not (Westen et al., 2006; Taber & Lodge, 2006). The second view of partisanship is as a ‘running tally’ of political attitudes and perceptions (Fiorina, 1981). Voters are constantly updating these tallies in their mind, and the party that is most favourable is the resulting partisan attachment. This is a very different conception of partisanship than the durable identity view. In this view, partisanship is no longer the cause of biases in perceptions and attitudes, but rather they are the same thing. A person’s partisanship is nothing more than an expression of this running tally, which is simply aggregated 4  attitudes and preferences, thus partisanship and political perceptions tend to be aligned (Fiorina, 1981). In either view, partisanship is an important predictor of political behaviour; however, the views differ when it comes to the stability of these attachments. Whereas the Michigan model views partisanship as difficult to change once established, the other suggests that it is constantly being re-evaluated in response to new political information. These two views of partisanship lead to diverging expectations on the primary cause of partisan sorting. If partisanship is stable and shapes attitudes, then people would be more likely to update their own ideology when faced with a discrepancy between their party’s positioning and their own. On the other hand, if partisanship is simply an expression of a person’s past attitudes, then they would be more likely to update their partisanship to match their own ideology. While American partisanship seems to align closely with the first view of partisanship as a durable psychological attachment, this may not be true of all countries. Some of the strongest objections to the identity view came from other countries where it did not seem to apply. Taking Canada as an example, several scholars have argued that Canadian partisanship is much weaker and more flexible than American partisanship (Jenson, 1975; LeDuc et al., 1984). Indeed, LeDuc et al. claimed that partisanship in Canada is affected by many of the same factors as vote choice, and thus it is just as volatile – it would “travel with the vote” (1984, p. 478). There is also evidence that party labels are not effective cues in shaping opinions in Canada, which could suggest that the messages parties are sending about their issue positions are not being absorbed by voters (Merolla et al., 2008). This would make it unlikely that partisan sorting could happen in Canada since voters do not seem to be very aware of the parties’ positions. In many European countries, scholars have voiced similar concerns about the applicability of a conception of 5  partisanship as a persistent identity (e.g. Thomassen, 1976; Butler & Stokes, 1969; Borre & Katz, 1973). These differences make partisan sorting interesting to examine in different contexts. Ideological positioning The main focus of this paper is to determine whether shifts in a party’s ideological position lead to shifts in their partisans’ own self-placement. This concept of an ideological placement is thus important to discuss before proceeding further. One popular conception of overall ideological positioning is the left-right dimension. This is a scale that places parties at a certain point based on their policy positions. The classic issue that represents this continuum is the degree of government intervention in the economy (more intervention is more left). However, there are many other issues that typically factor into this one-dimensional scale; for example, a traditional conception of morality is often considered a value of the right. The Manifesto Project, which I draw on later for the analyses, incorporates thirteen such issue dimensions into their left-right placement (Volkens et al., 2015). However, while the operationalization of the left and right is meant to represent consistent and meaningful variation, the public’s understanding of the left-right dimension has been shown to be limited and context-dependent (Inglehart & Klingemann, 1976; Zechmeister, 2006; Neundorf, 2009). Citizens’ understandings of the left and right are often more related to specific parties and figures than substantive policy dimensions, which is shaped by the national political discourse (Zechmeister, 2006). This may call into question the validity of using subjective left-right self-placements as a measure of an individual’s ideological position, as their placement may be influenced by their own relation to the non-policy elements of the left and right. However, predicting self-placements using an objective measure of party positioning (from manifestos) should produce a conservative estimate of how much people perceive and react to 6  ideological change, since the measure includes issues that may actually be irrelevant to the understanding of the left and right in a given country. Perception of changes in left-right positioning Parties change their ideological position through changing policy positions on issues that factor into the left-right dimension. A change in policy position can be observed in several ways. Speeches, interviews and advertisements can indicate policy positions to listeners, and this can be secondarily transmitted to others through peer interaction and media. The actions of a party also indicate positions, particularly for governing parties who enact policy, but also voting records of members of parties not in government. One fairly comprehensive source of a party’s documented policy positions is their campaign manifesto. During an election period, parties often publish a manifesto in which they outline their policy positions on a multitude of issues. In an ideal democracy, citizens would have complete and accurate information to draw on when making political judgements, and so would perceive any changes that parties make in their left-right positioning indicated in this manifesto. However, it is clear that reality does not allow for such perfect information. This leaves us with the question of whether we can reasonably expect partisans to perceive shifts in party positioning from their manifestos. There are a few reasons why we might think that changes in positioning of a party’s manifesto would not be noticed by their partisans. First of all, it is highly doubtful that many people actually read a party’s manifesto in an election, as it has been shown that citizens are rather uninformed and inattentive to politics (e.g. Converse, 1964; Gidengil, 2004). Even among partisans, who should reasonably be more attentive and interested in their party’s manifesto, there is likely only a very small minority who have read it carefully enough to place the policy position of the document. Even if we make the implausible assumption that partisans do read 7  their party’s manifestos, there is also much more going on in the political environment than just campaign manifestos, so it would seem that these documents may not weigh heavily in people's minds when evaluating a party’s position on the left-right spectrum. By contrast, there are several reasons to expect that partisans might actually perceive such shifts. Firstly, a manifesto is an expression of what a party is trying to ‘sell’ in the election – the policies that they believe will be most attractive to some portion of the electorate. A manifesto sets the overall themes of the campaign, and thus a party's manifesto position is likely to correlate with the position that is conveyed elsewhere - in the media, speeches, advertisements, etc. (Adams et al., 2011). Another reason to think that people may be attuned to the positions of party manifestos is the fact that campaigns heighten awareness and interest in political information (Adams et al., 2011). During campaigns, citizens are more attentive to politics, so if there was any time when people would perceive changes in a party’s ideological placement, it should be during campaigns. However, there is also evidence that the electorate takes time to update their judgements of the parties after they make shifts in ideology, leading to a lagged effect of party ideology on voter placements of the party (Adams & Somer-Topcu, 2009a). This makes sense because even when attentive, individuals take time to update their beliefs; attitude change often does not take place instantaneously. A shift in ideological position of a campaign manifesto also is not directly perceived by the public. The public perceives the position of a party through other activities, such as speeches, media attention, and discussion with others. If we think of a shift in ideological position as aggregated bits of information that each indicate a ‘left’ or ‘right’ position, then it will take time for the public to absorb all of the bits of information and correspondingly adjust their perceptions of the parties. 8  H1: Partisans update their views of the party’s ideological position in response to shifts in left-right tone of their manifestos. H2: The effect of lagged party shifts is stronger than the effect of the most recent party shifts on shaping partisans’ party placements. Partisan sorting: reacting to shifts in left-right party positioning If partisans do perceive changes in party positioning, whether from manifestos or the broader political environment, how might they respond to these changes? One possibility is that they will not respond at all: they simply retain their partisanship and personal ideology unchanged. However, there is evidence that opinions and beliefs are much more malleable than that (e.g. Zaller, 1992; Broockman & Butler, 2014). There is reason for this as well; people naturally try to avoid cognitive dissonance – the psychological discomfort associated with holding conflicting attitudes or beliefs (Westen et al., 2006). If a citizen perceives a disconnect between their personal ideology and their party’s position, they are likely to attempt to rectify it.  The two ways this can happen are to shift personal ideology to align it with their partisanship (party persuasion), or change their partisanship to be more consistent with their own ideological position (party switching). Together, this increased alignment of partisanship and ideology is known as partisan sorting (Fiorina & Levendusky, 2006; Levendusky, 2010b). Some scholars argue that the first method of sorting – persuasion – occurs more often (Layman & Carsey, 2002), and others believe that party switching occurs more often (Abramowitz & Saunders, 1998). It could also depend on the issues that are important to the partisan; Layman and Carsey (2006) find that if a person is aware of different party positions on an issue that is salient to them, this can lead to issue-based party switching. However, if a person is aware of 9  party differences, but does not find the issue salient, then he will change his position to align with his partisanship (Layman & Carsey, 2006). Another distinct possibility if a partisan is not willing to adapt their own ideology to match the party’s, instead of switching their identification to another party, they may also choose to become non-partisan. In fact, this is not an unlikely option; Dalton and Wattenberg (2000) find that individuals in many countries are becoming increasingly non-partisan. Furthermore, since partisanship is an identity, it may be difficult to immediately change your attachment from one party to another. It is a process that would likely occur over time – to first become non-partisan, then later to slowly start identifying with another party. However, because this process leads to the same outcome as party switching – citizens who are ideologically furthest from the party no longer identify as partisans, thus increasing the alignment of the party’s ideology and that of their partisans – I will consider this to be a type of party switching. Because the mechanism of partisan sorting is based on the cognitive dissonance that occurs if one were to hold beliefs conflicting with an identity, we should expect to see partisan sorting occur in all democracies. That is, if a party changes position and partisans perceive it, they will either update their ideology or their partisanship in order to remain aligned with their party. However, there are several theorized reasons why we might expect to see greater levels of partisan sorting in certain contexts. The main proposed cause of the high degree of sorting in the US is elite polarization; parties are becoming more ideologically divergent and consistent, sending clearer cues to partisans of what their issue positions and ideology should be (Levendusky, 2010b). Another possible cause is the changing media environment; media exposure today is more accessible and customizable because of the Internet. This leads to increased exposure to party positions, especially those of one’s identified party. With greater 10  exposure to media informing partisans of these positions, they are more aware of party positions on salient issues, thus inducing partisan sorting (Levendusky, 2010b). The major underlying mechanism that is thought to provoke greater alignment of personal ideology and party ideology is an awareness of divergent or polarized party positions on relevant issues (Fiorina & Levendusky, 2006; Levendusky, 2010b). Because we expect partisans to be somewhat aware of their political environment and perceive changes in their party’s position, I expect to find some degree of partisan sorting across most countries. Furthermore, I expect that in party systems with higher levels of polarization, the degree of sorting will be greater. This is because where the parties are more polarized, clearer cues are being transmitted to partisans about what their issue positions should be. H3: Changes in left-right partisan ideology follow changes in left-right party positions. H4: Changes in partisan ideology more closely correspond to changes in party positioning in more polarized party systems. In many studies of partisan sorting, scholars find that there is a lagged effect of partisan sorting; that is, partisans do not immediately align their ideology and partisanship (Adams et al., 2012). This is consistent with the finding that partisans do not immediately update their placements of the parties; if a lag is necessary for people to even perceive a shift in ideology, it should also be required for them to react to it. As discussed earlier, we can think of the transmission of information as partisans receiving pieces of information that signal either ‘left’ or ‘right’. It takes time for these bits of information to accumulate enough to change a person’s beliefs about the position of a party. It could then take even longer for a person to update their own ideology or switch parties. It is likely that people do not immediately update their ideology 11  or switch parties the instant they perceive a shift in their party. They may live with a certain amount of cognitive dissonance or skepticism about their party while the indications of movement continue to accumulate. At a certain point, the amount of evidence that the party has shifted is large enough that it would cause significant cognitive dissonance, and then drive the individual to either update their own views or change their partisanship.  It is therefore appropriate to introduce a lagged party shift variable in order to account for the gradual process of sorting. This lagged shift is equal to the difference in party position at election t-2 to election t-1 (in other words, the change that led to the party’s position in the previous election). However, since this change occurred several years ago, and the party could have started moving in a different direction since then, the lag may be too long to capture the desired effect of people taking time to update their beliefs and sort. H5: Lagged party shift should have a greater effect on partisan ideological shift than current party shift.   12  DATA AND METHODS To test these five hypotheses, I use data from the Comparative Study of Electoral Systems (CSES) and the Manifesto Project. I use the four modules of the CSES that are currently available, which includes data from 53 countries which held elections between 1996 and 2015. I matched up the parties that were asked about in each CSES study with the party-level data from the Manifesto Project. This left 520 observations of parties in 81 elections that were held in 23 countries. From the CSES, I obtain individual-level measurements of partisanship and ideological placements. Respondents were asked about partisanship with the following pair of questions: “Do you usually think of yourself as close to any particular political party? What party is that?” To measure ideological placements, respondents were given a scale from 0 (most left) to 10 (most right) and asked where they would place themselves on that scale, followed by each of the parties that were deemed relevant in that election (the 6 most popular parties plus an optional three additional parties, determined by the national election study team). From these variables, I construct partisan self-placement by taking the average of partisan self-placement scores for each party in an election. Similarly, to get a measure of perceived party placement, I take the average score of where partisans placed their own party on the left-right scale.  I use the Manifesto Project to get data on party ideological change. In the project dataset, a measure of left-right party positioning called “RILE” is provided. The project hand-codes party manifestos by classifying each sentence in the manifesto into one of several categories. Each category corresponds to a certain position on an issue; for example, “Traditional Morality: Positive” would be a count of the number of sentences in the manifesto that are supportive of traditional morality. The RILE score is computed by adding together all ‘right’ category 13  mentions in a party’s manifesto, and subtracting all ‘left’ category mentions, divided by the total number of sentences in the manifesto: 𝑅𝐼𝐿𝐸 =𝑅 − 𝐿𝑁 However, this score has been criticized for several problems, and I am therefore using a modification of the RILE score as proposed by Lowe et al. (2011): 𝑅𝐼𝐿𝐸 = log𝑅 + .5𝐿 + .5 This modification corrects the dependence on the number of neutral sentences in the manifesto, and the logged measure aligns better with human perception. It is particularly applicable for this analysis since their measure is claimed to be more stable over time, and correlates better with expert surveys (Lowe et al., 2011).  I also use manifesto data to construct the measure of party system polarization.  Dalton (2008) proposed a measurement that is based on the positions of each party in the system, weighted by the vote share they receive. The equation is: 𝑃𝑂𝐿𝐴𝑅𝐼𝑍𝐴𝑇𝐼𝑂𝑁 = √∑ 〖((𝑟𝑖𝑙𝑒𝑖 − 𝑤𝑚𝑒𝑎𝑛100)2∗ 𝑣𝑜𝑡𝑒𝑠ℎ𝑎𝑟𝑒𝑖〗)𝑛𝑖=1 where wmean is the average left-right position of parties, weighted by vote share. The analysis uses Ordinary Least Squares regression to test the hypotheses. Unfortunately, adequate panel data is not available to assess how partisans respond to party ideological change at the individual level.  This limits the analysis to examining only whether the general partisan sorting phenomenon is occurring, and not determining whether the effect is 14  caused more by partisan switching or party persuasion, since both activities related to partisan sorting appear the same at the aggregate level – partisan ideology tracks that of their party (Adams et al., 2012).  Another problem this analysis encounters is with establishing causality. There are two possibilities that could be causing a correlation between partisan and party positioning at the aggregate level: either partisan sorting, or elite responsiveness to partisan ideology. This makes the causal direction of the relationship difficult to determine; for this reason, and for the theoretical reasons discussed earlier, I include a lagged party shift variable which may help to determine the direction of the relationship. I also include country fixed effects in order to control for general patterns occurring in each country. For instance, if all parties and partisans in a country are gradually moving to the left over time, this effect on partisan positioning will be attributed to the country, and not each separate party’s movement. The first question of whether partisans perceive changes in left-right position of their parties is thus tested using the following model: 𝑃𝑒𝑟𝑐𝑒𝑖𝑣𝑒𝑑 𝑠ℎ𝑖𝑓𝑡𝑡 =  𝛽0 + 𝛽1(𝑝𝑎𝑟𝑡𝑦 𝑠ℎ𝑖𝑓𝑡𝑡) +  𝛽2(𝑝𝑎𝑟𝑡𝑦 𝑠ℎ𝑖𝑓𝑡𝑡−1) + 𝛽3(𝐶𝑜𝑢𝑛𝑡𝑟𝑦 1) + ⋯ + 𝛽𝑛+3(𝐶𝑜𝑢𝑛𝑡𝑟𝑦 𝑛) where perceived shift is the average party placement at time t, minus the average placement at t-1 (and similarly for party shift). The second question of whether partisan ideology and party positioning vary together is tested using essentially the same model but with partisan change in position becoming the dependent variable: 𝑃𝑎𝑟𝑡𝑖𝑠𝑎𝑛 𝑠ℎ𝑖𝑓𝑡𝑡 = 𝛽0 + 𝛽1(𝑝𝑎𝑟𝑡𝑦 𝑠ℎ𝑖𝑓𝑡𝑡) +  𝛽2(𝑝𝑎𝑟𝑡𝑦 𝑠ℎ𝑖𝑓𝑡𝑡−1) + 𝛽3(𝐶𝑜𝑢𝑛𝑡𝑟𝑦 1) + ⋯ + 𝛽𝑛+3(𝐶𝑜𝑢𝑛𝑡𝑟𝑦 𝑛) 15  ANALYSIS AND RESULTS To begin the analyses, it is useful to examine each of the main variables and their bivariate relationships. First I examine the correlation between real left-right changes in party positioning and partisans’ perception of these changes. From the figure below, we can see that there is a relatively weak (r=.14) but positive correlation between real changes and perceived changes, which is statistically significant. It is important for the partisan sorting hypothesis that there is at least a loose relationship between real left-right change of parties and their partisans’ perception of such change. If there was no relationship, it would not be reasonable to expect partisans to react to ideological changes, since they would not even consciously perceive them. Figure 1: Correlation between real and perceived party change 16  Secondly, I examine whether partisans believe they are sorting; that is, do partisans’ self-placements move with their changing perceptions of the party? The correlation in Figure 2 clearly demonstrates that they do. With a Pearson’s correlation coefficient of .77, we can say with confidence that partisans certainly perceive themselves to stay aligned with their party. If they believe the party has moved left or right, the partisans also place themselves more left or right in congruence with their party. This shows that at least part of the sorting mechanism is at work; people are attempting to avoid any cognitive dissonance that would occur by placing themselves and their party at ideologically divergent points. 17  Figure 2: Correlation between perceived party change and partisan change  Third, I examine the relationship between real party change, and partisan self-placement – the essence of the partisan sorting phenomenon. From the graph below we observe a relatively weak (r=.16) but positive correlation between real party change and partisans’ change in self-placement. While not very strong, the correlation is nonetheless statistically significant at the 95% confidence level.     18  Figure 3: Correlation between real party change and partisan change   Now I move on to testing the five hypotheses presented earlier. First, I examine whether partisans perceive changes in their party’s left-right position, as presented in their campaign manifesto. Table 1 shows the results of the model using partisan placements of their respective parties as the dependent variable. The results do not provide any evidence that partisans perceive changes in their party’s positioning from their manifesto. This suggests that partisan sorting should not be occurring – if partisans cannot perceive changes in their party’s position, it is impossible for them to respond to such movement. However, there may nonetheless remain a link between partisan and party positioning; both could respond to external pressures in a similar manner, or it could be partisan ideology that leads the party to change their position.  19   I should note however, that these results do not prove that partisans do not perceive shifts in their parties. If I run the model without lagged party shift, the effect of the most recent party shift is somewhat significant (at the 90% confidence level). This difference is likely due to the loss of observations when including the lagged variable. Unfortunately because I only started with an average time series of 3.5 elections for each country, including a variable that calculates a change drops this to 2.5 usable observations, then including a lagged change reduces this further to an average of 1.5 complete observations of a party. This consequently reduces the model from 240 to 125 degrees of freedom.  Table 1: Regression results for changes in perceived party position Variable  Change in perceived party position (0-10 LR scale) Intercept -.188 (.175) Left-right party shift (in RILE score) .015 (.072) Lagged left-right party shift (in RILE score) .043 (.061)    Observations 125 R-squared .116   To test the third and fourth hypotheses (partisan ideology tracks party ideology and lags behind actual shifts), I run the model with partisan ideology as the dependent variable. The results shown in Table 2 do not provide any support for the partisan sorting hypothesis. This is not surprising since there was little evidence that partisans perceived changes in their party’s position. However, the model run without the lagged variable again becomes significant (this time lagged party shift is significant at the 95% level). It seems that if partisans were to more readily perceive changes in party placement from their manifestos, they would likely adapt to the change, since we observed a strong correlation between perceived party change and partisan self-placement. However, the results shown here do not necessarily provide support for the partisan 20  sorting hypothesis. While we observed some significance in the models without including the lagged variable, we cannot place much weight on the results without this, since the lag is the only way the analysis can attempt to establish causality. Without the lag, the causal relationship could just as easily be reversed: parties responding to their partisans’ preferences. Table 2:  Regression results for change in partisan self-placement Variable Change in partisan self-placement (0-10 LR scale) Intercept -.074 (.154) Left-right party shift (in RILE score) .075 (.063) Lagged left-right party shift (in RILE score) .025 (.053)   Observations R-squared 126 .168   To test the fourth hypothesis that partisan sorting occurs to a higher degree in more polarized systems, I categorize each election as one which had either high or low polarization, based on whether it is above or below the median value on the polarization index. I re-run the analyses with these two samples, shown in Tables 3 and 4. Table 3: Regression results for changes in perceived party position (by polarization) Variable Change in perceived party position (0-10 LR scale): Low polarization systems Change in perceived party position (0-10 LR scale): High polarization systems Intercept -.246 (.183) -.125 (.159) Left-right party shift (in RILE score) -.235 (.154) .130 (.070)* Lagged left-right party shift (in RILE score) -.038 (.152) .075 (.060)    Observations R-squared 50 .118 70 .241 Note: *p<.10 21  Table 4: Regression results for change in partisan self-placement (by polarization) Variable Change in partisan self-placement (0-10 LR scale): Low polarization systems Change in partisan self-placement (0-10 LR scale): High polarization systems Intercept -.217 (.173) -.042 (.124) Left-right party shift (in RILE score) -.158 (.148) .139 (.055)** Lagged left-right party shift (in RILE score) -.003 (.146) .027 (.047)    Observations R-squared 50 .170 70 .326 Note: **p<.05   From these results it appears that partisan sorting may be occurring, but only in polarized party systems. In Table 3 we see that in a party system with lower than average polarization, partisans have difficulty perceiving changes in their party’s position – the variables are not significant, and even the direction of the coefficients is negative.  Unsurprisingly, the results are similar for the test of whether partisans respond to party changes. However, the story becomes very different when we look at countries with a higher level of polarization; partisans seem to both perceive changes in their party’s position and react to them. This supports the partisan sorting hypothesis, and suggests that it is caused by an awareness of polarized party positions. This makes sense because when a party system is more centripetal, the policy positions of the different parties may be more difficult to distinguish; all parties occupy only a small section of the left-right spectrum so perceiving changes within this space is more difficult.  The average magnitude of party shift was 0.5 in either direction on the RILE score. Based on the coefficients in the model of polarized systems, this would result in .065 change in the average placement of the party according to its partisans, and a .070 change in the average 22  partisan self-placement (on a 0-10 scale). This is a very small movement on average, but where we see the largest changes in RILE score (+/- 4), this amounts to a half-point change in partisans’ evaluations of the party and themselves. To check that the results presented in this section are robust, I run all of the above regression models without notable outliers (parties in Mexico, Greece, France and Ireland) and find very similar results.   23  Conclusion When parties change their left-right ideological placement their partisans are left to respond to these shifts. If partisans are not aware of these shifts, then it is unlikely that they would change their own ideology in response. However, if they are aware of party shifts and choose to respond by aligning their own partisanship and ideology, there are two options: switch parties (or become non-partisan) to match their ideology, or adapt their ideology to match existing partisanship. These activities, collectively known as partisan sorting, are theorized to occur due to increasingly clear and consistent messages sent by parties; people become more aware that their partisanship and personal ideology are inconsistent and thus bring the two into alignment (Levendusky, 2010b). I find that whether partisans perceive and adapt to changes in their party’s position does depend heavily on how polarized the system is. In party systems with higher than average polarization, I find evidence that partisans perceive these changes and react to them, thus keeping partisan and party ideology in alignment. In lower-polarization systems however, there is little evidence that partisans perceive changes in their party’s position, and thus they do not respond to the changes. However, I do find that partisans believe that they are continuously aligned with their party; there is a very strong correlation between changes in partisan self-placement and their placement of the party. This suggests that sorting might occur even in lower-polarization systems if partisans were to perceive changes in their party’s positioning. Perhaps repeating this study with a different measure of real party placement, such as text analysis of speeches to the public, would find a more perceptive citizenry. These findings have normative implications for an evaluation of contemporary democracies. We find here limited evidence of partisan attentiveness to their party’s position. 24  This provides a rather dismal view of the capabilities of the electorate; partisans should presumably be the ones most aware of the party’s position, and even they cannot perceive changes to it, except when there is adequate polarization in the system. The remaining non-partisan portion of the electorate are likely even more unaware of changes in parties’ positions. While the congruence between party ideology and voter ideology is important for a healthy democracy, there may also be consequences to partisan sorting. Because of elite polarization and the resulting clearer partisan cues, partisan sorting has been shown to increase partisan bias, anger and protest activity, while actual mass positions remain relatively moderate (Mason, 2014). However, partisan sorting may also have the benefit of mass attitudinal consistency; people have more consistent views because they are aligned with party positions (Levendusky, 2010a).  The major limitation of this paper is the lack of panel data to study partisan activity at the individual level. This would allow for a much more nuanced analysis of partisan perception and behaviour in the face of change. If evidence of sorting is found, it would also be worth testing the extent to which it is caused by party switching or party persuasion, which would be possible only with panel data.    25  REFERENCES Abramowitz, A. I., & Saunders, K. L. (1998). Ideological realignment in the U.S. electorate.Journal of Politics, 60(3), 634-652. doi:10.2307/2647642 Adams, J., & Somer-Topcu, Z. (2009). Moderate now, win votes later: The electoral consequences of parties’ policy shifts in 25 postwar democracies. Journal of Politics,71(2), 678-692. doi:10.1017/S0022381609090537 Adams, J., & Somer-Topcu, Z. (2009). Policy adjustment by parties in response to rival parties’ policy shifts: Spatial theory and the dynamics of party competition in twenty-five post-war democracies. British Journal of Political Science, 39(4), 825-846. doi:10.1017/S0007123409000635 Adams, J., Ezrow, L., & Leiter, D. (2012). Partisan sorting and niche parties in Europe. West European Politics, 35(6), 1272-1294. Adams, J., Ezrow, L., & Somer-Topcu, Z. (2011). Is anybody listening? Evidence that voters do not respond to European parties' policy statements during elections. American Journal of Political Science, 55(2), 370-382. Borre, O., & Katz, D. (1973). Party identification and its motivational base in a multiparty system: A study of the Danish general election of 1971. Scandinavian Political Studies,8(A8), 69-111. doi:10.1111/j.1467-9477.1973.tb00485.x Broockman, D. E., & Butler, D. M. (2014). How do citizens react when politicians support policies they oppose? Field experiments with elite communication. Butler, D., & Stokes, D. E. (1969). Political change in Britain: Forces shaping electoral choice. New York: St. Martin's Press. 26  Campbell, A., Converse, P. E., Miller, W. E., & Stokes, D. E. (1960). The American voter. New York, NY: Wiley. Carsey, T. M., & Layman, G. C. (2006). Changing sides or changing minds? Party identification and policy preferences in the American electorate. American Journal of Political Science, 50(2), 464-477. doi:10.1111/j.1540-5907.2006.00196.x Converse, P. E. (1964). The nature of belief systems in mass publics. In D. E. Apter (Ed.),Ideology and discontent. London, UK: Free Press of Glencoe. Dalton, R. J., & Wattenberg, M. P. (2000). Parties without partisans: Political change in advanced industrial democracies. Oxford, UK: Oxford University Press. Dalton, R. J. (2008). The quantity and the quality of party systems: Party system polarization, its measurement, and its consequences. Comparative Political Studies, 41(7), 899-920. doi:10.1177/0010414008315860 Downs, A. (1957). An economic theory of democracy. New York, NY: Harper and Brothers. Evans, G., & Pickup, M. (2010). Reversing the causal arrow: The political conditioning of economic perceptions in the 2000–2004 U.S. presidential election cycle. The Journal of Politics, 72(4), 1236-1251. Fiorina, M. P., & Levendusky, M. S. (2006). Disconnected: The political class versus the people. In P. S. Nivola & D. W. Brady (Eds.), Red and blue nation?: Characteristics and causes of America's polarized politics, Volume 1 (pp. 95-110). Stanford, CA: Washington: Brookings Institution Press and Stanford: Hoover Institution Press. Fiorina, M. P., Abrams, S. J., & Pope, J. C. (2006). Culture war?: The myth of a polarized America (2nd ed.). New York: Pearson Longman. 27  Fiorina, M. P. (1981). Retrospective voting in American national elections. New Haven, CT: Yale University Press. Gidengil, E. (2004). Citizens. Vancouver, Canada: UBC Press. Green, D. P., Palmquist, B., & Schickler, E. (2002). Partisan hearts and minds: Political parties and the social identities of voters. New Haven, CT: Yale University Press. Inglehart, R., & Klingemann, H. (1976). Party identification, ideological preference and the left-right dimension among Western mass publics. In I. Budge, I. Crewe, & D. Farlie (Eds.),Party identification and beyond: Representations of voting and party competition(pp. 243-273). London, UK: Wiley. Jenson, J. (1975). Party loyalty in Canada: The question of party identification. Canadian Journal of Political Science, 8(4), 543-553. doi:10.1017/S000842390004628X Layman, G. C., & Carsey, T. M. (2002). Party polarization and "conflict extension" in the American electorate. American Journal of Political Science, 46(4), 786-802. Layman, G. C., Carsey, T. M., & Horowitz, J. M. (2006). Party polarization in American politics: Characteristics, causes, and consequences. Annual Review of Political Science,9, 83-110. doi:10.1146/annurev.polisci.9.070204.105138 LeDuc, L., Clarke, H. D., Jenson, J., & Pammett, J. H. (1984). Partisan instability in Canada: Evidence from a new panel study. The American Political Science Review, 78(2), 470-484. Levendusky, M. S. (2010). The partisan sort: How liberals became Democrats and conservatives became Republicans. Chicago, IL: University of Chicago Press. Levendusky, M. S. (2010). Clearer cues, more consistent voters: A benefit of elite polarization.Political Behavior, 32(1), 111-131. doi:10.1007/s11109-009-9094-0 28  Lowe, W., Benoit, K., Mikhaylov, S., & Laver, M. (2011). Scaling policy preferences from coded political texts. Legislative Studies Quarterly, 36(1), 123-155. Mason, L. (2014). "I disrespectfully agree”: The differential effects of partisan sorting on social and issue polarization. American Journal of Political Science. doi: 10.1111/ajps.12089 Merolla, J. L., Stephenson, L. B., & Zechmeister, E. J. (2008). Can Canadians take a hint? The (in)effectiveness of party labels as information shortcuts in Canada. Canadian Journal of Political Science, 41(3), 673-696. doi:10.1017/S0008423908080797 Miller, W. E., & Shanks, J. M. (1996). The new American voter. Cambridge, MA: Harvard University Press. Neundorf, A. (2009). Growing up on different sides of the wall – A quasi-experimental test: Applying the left–right dimension to the German mass public. German Politics, 18(2), 201-225. doi:10.1080/09644000902870834 Rabinowitz, G., & Macdonald, S. E. (1989). A directional theory of issue voting. The American Political Science Review, 83(1), 93-121. Taber, C. S., & Lodge, M. (2006). Motivated skepticism in the evaluation of political beliefs.American Journal of Political Science, 50(3), 755-769. doi:10.1111/j.1540-5907.2006.00214.x Thomassen, J. (1976). Party identification as a cross-national concept: Its meaning in the Netherlands. In I. Budge, I. Crewe, & D. Farlie (Eds.), Party identification and beyond: Representations of voting and party competition. London: Wiley. Volkens, A., Lehmann, P., Matthieß, T., Merz, N., Regel, S., & Werner, A. (2015). The Manifesto Project dataset - codebook. Manifesto Project (MRG / CMP / MARPOR). Version 2015a. Berlin: Wissenschaftszentrum Berlin für Sozialforschung(WZB). 29  Westen, D., Blagov, P. S., Harenski, K., Kilts, C., & Hamann, S. (2006). Neural bases of motivated reasoning: An fMRI study of emotional constraints on partisan political judgment in the 2004 U.S. presidential election. Journal of Cognitive Neuroscience,18(11), 1947-1958. doi:10.1162/jocn.2006.18.11.1947 Zaller, J. (1992). The nature and origins of mass opinion. Cambridge, UK: Cambridge University Press. Zechmeister, E. (2006). What’s left and who’s right? A Q-method study of individual and contextual influences on the meaning of ideological labels. Political Behavior, 28(2), 151-173. doi:10.1007/s11109-006-9006-5  


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items