Abstract and Keywords
The chapter starts by presenting the goals of the book in relation to the advancement of the study of party change, of political personalization, and of the relationship between them. It then examines the scope of the analysis (a comprehensive and integrative look at the two main phenomena, i.e., party change and political personalization), the level of analysis (country), and the research population (twenty-six countries). It notes that the study is about processes; it is not a static snapshot of the state of the parties or of personalized politics. It then presents the methodology that was selected: a broad-brush mapping of both phenomena rather than the use of proxies or of the best available data on a few indicators. Finally, the chapter presents a brief outline of the book.
The “century that has just started will be the age of personalization, just as the previous one was the century of mass collective actors—a trend that political science has a duty to consider with greater attention.” With these words Musella and Webb (2015: 226) end their introductory chapter to a special issue of Italian Political Science Review on “The Personal Leader in Contemporary Party Politics.” From a broader historical perspective, what we have witnessed in recent decades may be “the comeback of personal power” (Calise 2011: 3). Alternatively, perhaps, what we are currently witnessing is not a complete change or comeback but rather a synthesis of partisan and personalized politics.
This book will examine two of the most prominent developments in contemporary democratic politics: the change in party–society linkage and political personalization. The former is manifested in many ways: from changes in the party background of politicians to changes in the density of party membership; and from changes in party–interest group relationships to assorted aspects of voter behavior (such as electoral turnout and electoral volatility). We will refer to all these elements as pointing to an increase, decline, or no change in partyness.1 The latter development, political personalization, is the process by which the weight of the group (in this book, the political party) declines in politics, while the centrality of individual politicians rises. This phenomenon is multifaceted and is reflected in changes in political institutions, in the ways in which politics is presented and covered by the media, and in the behavior of both politicians and voters.
These two phenomena appear to be related: when parties decline, it would be reasonable to expect that the weight of individual politicians in politics will increase. Up until now, the analysis of these relationships was limited either to (p.2) specific aspects of these phenomena (Renwick and Pilet 2016) or to single-country studies (Wattenberg 1994). No study has empirically examined, in a comprehensive manner, the relationship between these two developments from a cross-national comparative perspective. Moreover, while decline in the party–society link may be a necessary condition for personalization, personalization is not a necessary outcome of party change, because actors other than individual politicians (e.g., the media, interest groups, courts) may take over some of the functions previously assumed by political parties. In addition, it is plausible to suggest that in some cases personalization will be channeled in ways that may not hurt the party or may even strengthen it.
The Goals of the Book
In these pages we endeavor to take the cross-national comparative study of party change, of political personalization, and of the relationship between them a step forward. With regard to party change, no study, to the best of our knowledge, offers a comparative cross-national estimation of variance in the levels and patterns of party change among countries.2 The literature on party change has thus far acknowledged the differences between countries but has tended to overlook them, sacrificing them to the “party decline” versus “party adaptation” debate, as it is usually framed (Dalton, Farrell, and McAllister 2011 and Mair 2013 are among the most prominent recent works on the subject).
Our aim is not merely to examine and integrate the existing data but to inject fresh cross-national comparative data into the study. This addition is particularly pertinent to the almost neglected aspect of the presence and success of national parties at local and regional levels. Beyond some single-case studies (Åberg and Ahlberger 2015; Brichta 1998), there are several comparative studies that throw light on this aspect (Dandoy and Schakel 2013; Detterbeck 2012; Reiser and Holtmann 2008a), but, to date, it has not featured in the general study of party change.
As for political personalization, the book presents a more comprehensive cross-national analysis than other studies, in terms of both number of countries and indicators (Downey and Stanyer 2010; Karvonen 2010). The literature offers some good theoretical reasons for expecting political personalization. The decline in the party–society linkage, the mediatization of politics (especially through television, with its visual emphasis) and the overall cultural process of individualization all suggest that we might find clear evidence of a general trend (p.3) toward political personalization. Yet most studies until now did not detect such evidence (Adam and Maier 2010; Karvonen 2010). By emphasizing variance among countries, this book will take the first step toward bridging the gap between the very good theoretical reasons we have to expect political personalization and the generally weak empirical findings. In addition, it will outline the most expansive cross-national comparison of online personalization to date. This area has been researched, but not in a comprehensive manner, which directly compares the activity of parties and politicians and how online personalization is consumed in twenty-five countries, as will be done here.
Finally, regarding the relationship between the decline of the party–society linkage and political personalization, many studies claim, logically, that party change breeds political personalization (Karvonen 2010; McAllister 2007; Webb, Poguntke, and Kolodny 2012). However, except for Wattenberg’s (1994) study of American politics, no work has empirically examined these relationships. Having completed the two preliminary tasks of measuring the levels of party change and measuring the levels of political personalization (each one important in its own right), the book is then able to systematically analyze the relationship between the two phenomena.
The three main research questions that guided us are as follows. First, are there significant differences in the levels and patterns of party change among democracies? If there are, how can we explain them? Second, are there significant differences in the levels and patterns of political personalization among democracies? Again, if there are, how can we explain these differences? Third, what is the relationship between party change and political personalization?
The emphasis on variance across states and on the development of the numerous aspects and dimensions of party change and political personalization exposes differences and leads to a richer and better informed discussion. In the case of party change, it offers an analytical framework that extends beyond the dichotomous debate of party decline versus party adaptation. In the matter of political personalization, as noted, the emphasis on variance helps in bridging between the high theoretical expectations and the disappointing empirical findings. As for the theoretically sound linkage between the two phenomena, this volume not only is the first to submit it to a comprehensive cross-national examination, but it also proposes a more nuanced understanding of this relationship.
The Scope and Level of the Analysis
Current comparative politics literature points to an array of recent developments in the field, for example on the decline in partisanship (Dalton and (p.4) Wattenberg 2000a; Mair 2013), the expansion of political opportunities beyond party politics (Cain, Dalton, and Scarrow 2003), the presidentialization of politics (Poguntke and Webb 2005a; Webb, Poguntke, and Kolodny 2012), the growth of power of independent lists in local politics (Reiser and Holtmann 2008a), changes in party membership (Van Haute and Gauja 2015; Scarrow 2015), institutional reforms (Bedock 2017), electoral reforms (Renwick and Pilet 2016), party reforms (Gauja 2017b), the opening up of leadership and candidate selection methods (Cross and Blais 2012; Cross and Pilet 2015; Hazan and Rahat 2010), and the changing relations of center-left parties with trade unions (Allern and Bale 2017). This book can be seen as an ambitious attempt to present an integrated analysis of all these developments as facets of two superdevelopments: party change and political personalization.
Party change and political personalization can be analyzed at various levels: the country level, the party level, and, in the case of personalization, the individual level. This study will focus on variance at the country level. It is based on the premise that in modern representative democracies political parties operate mainly in relation to other parties within their specific party system. That is, parties compete and cooperate with other parties at the national level and are deeply influenced by these interactions. Even in a globalized world, each country has its specific political, social, institutional, and economic order, which justifies its being treated as an autonomous political unit. It is not only the case that countries differ when comparison is at the country level (e.g., their national electoral systems); studies also recognize significant similarities between subsystems (regions, municipalities, parties) within a given country and significant differences between these subsystems and their counterparts in other countries. These within-country similarities stem from the fact that the subunits are nested within the national order. Indeed, in his concluding overview to the collection Organizing Political Parties, Katz (2017: 326) observes:
One important result common to most of the chapters in this volume is to highlight the importance of national differences, which virtually always overwhelm everything else. This is certainly evident in the various comparisons of nation versus party family as a distinguishing characteristic, but it can also be evident when either system-level variables or country dummies are included in explanatory models.
Having said that, it is of course also possible to study the phenomena both at the party level and at the individual level, and even to account for the influence of the country by including it as a control variable in the analysis. Kölln (2015a) has made a convincing study of party change at the party level and identified differences among party families. When studying political personalization, one may argue, for example, that radical and extreme right-wing parties typically (p.5) are highly personalized by comparison to parties from other families. Yet, in view of the task we set ourselves here—of looking at both phenomena from the broadest possible perspective—and in view of the limitations of the available data, we will focus on the country level.
The Research Population: Countries During a Specific Period
Our starting point is the early 1960s. This decade is relevant to both phenomena under investigation. It can be seen as representing a “golden age,” the heyday of the mass political party, when parties were the central political actors. The early 1960s can also be seen as the beginning of the new age of mass-political communication (marked by the famous Nixon–Kennedy televised debate in the United States), which is characterized by the rise of television and its personalizing influence (Blumler and Kavanagh 1999). Our end point is the most recent one we could manage: 2015. Unfortunately, many pieces of data that were collected fail to cover the entire span of about fifty years, but the 1960s is the initial reference point and the aspiration was to cover as many years as possible.
Our group of democracies, twenty-six in total, includes a core of veteran and established democracies. After all, we are dealing with processes whose starting point is in the 1960s; thus first- and second-wave democracies should be our main focus. This group of countries contains longtime European democracies (Austria, Belgium, Denmark, Finland, France, Germany, Iceland, Ireland, Italy, Luxembourg, Netherlands, Norway, Sweden, Switzerland, the United Kingdom) and veteran democracies elsewhere in the world (Australia, Canada, Israel, Japan, New Zealand). We also examine here early third-wave (Greece, Portugal, Spain) and late third-wave democracies (Czech Republic, Hungary, Poland).
We did not include presidential democracies in our study, because parties play a substantially different role in such systems of government (Samuels and Shugart 2010) and because the inherently personalized nature of such a regime would have overburdened the study. Most of the countries are purely parliamentarian, while a few belong, according to the various definitions of scholars, in an interim category.3 This implies that, by comparison to presidentialized (p.6) systems, and especially to the presidential regime types, parties should be relatively central actors in their politics, while levels of personalization can be expected to be lower (Samuels and Shugart 2010). Yet it emerges that these countries differ in many aspects, from their general institutional order (consensus versus majoritarian, federal versus unitary) and electoral system to their political cultures (Lijphart 2012). Overall, our group covers almost all cases of nonpresidential veteran democracies, together with cases of younger democracies. This variance in the potential independent variables allows us a fair chance to explain the varying levels of party change and personalization.
Important Note: Let’s Not Forget that we are Studying Processes
The aim of this book is to study processes, developments over time, and not a static condition such as the polities’ level of partyness or of personalized politics. It is highly likely that there is variance among countries regarding the starting points. In some countries, parties were the dominant actors in politics in the 1960s, while in others they were “just” central actors; similarly, some countries had a more personalized politics than others. This means, for example, that some countries that started from “higher” partyness points may have witnessed high levels of party decline and yet came up (at the end point) with a higher level of partyness than countries that experienced smaller changes. For example, 14.3 percent of the Danish electorate around 1960 were party members, by comparison to 4.1 percent around 2008, while only 2.7 percent of the German electorate in 1960 were party members, by comparison to 2.3 percent around 2008. Thus Denmark experienced a much steeper decline (in absolute and relative terms), yet retained a much higher level of party membership than Germany, which saw a much smaller decline. We should not only keep this in mind, but relate to it in our analysis of change.4
Due to lack of comprehensive data, many studies in political science (including some of our own) focus their efforts on the limited area where the lamp post shines its light. Scholars understandably prefer to probe those cases they are more familiar with and often limit themselves to the use of existing or more readily accessible data. To a certain degree, this is an inevitable result of real limitations of knowledge and resources. Yet a review of the literature reveals that there are studies that may be relevant to a wider approach to the phenomenon of party change and political personalization. For example, studies of political recruitment and regional and local politics can also contribute to the study of party change, while studies of institutional reforms at the state and party levels may benefit the study of political personalization.
Indeed, the strategy adopted here is one that applies the broadest possible perspective in terms of indicators of both phenomena. That is, instead of the few indicators of relatively high quality and specific properties that allow for some sophisticated analysis, this study looks at a wider set of indicators. These cover much more ground but are necessarily of a lower quality and compel us to process the data in a manner that will take their weaknesses into consideration.
Our data come from various sources: first, extensive data mining from works of other scholars; second, analyses based on available cross-national (e.g., Eurobarometer, European Social Survey, World Value Survey) and national political surveys (e.g., national elections studies); third, analyses based on newly available databases concerning aspects of party change (e.g., Members and Activists of Political Parties, MAPP); and, fourth, analyses of original material that include, where needed, updating the data, making additions to them, and also collecting and coding new data on our own, from scratch.
There were several challenges to our ambitious aim to cover as much ground as possible in terms of data gathering, coding, and analysis. First, the availability of data. For some countries, almost all data are available, at a rather satisfactory quality and with coverage of almost all the period in question, but for others this is not the case. We are studying here processes that developed over more than half a century. Some data from the past can no longer be gathered, whether or not they ever existed. Questions that were not asked in surveys decades ago are a clear case in point. The other side of the coin is that today’s trend of establishing online data sets and our newfound ability to conduct online searches for stats that are hiding in books and articles mean that we are much richer in data by comparison to previous generations of scholars, and even to ourselves just a decade ago.
(p.8) A second challenge regards the quality of the data. Some indicators are based on data drawn from electoral records (e.g., electoral volatility, turnout) that are of high quality. Others are based on surveys, and thus suffer to a greater or lesser degree from the problems that arise when surveying citizens, from ensuring the representativeness of the sample to phrasing questions in an optimal manner. And these are relatively easy to handle by comparison to cases in which one has to base the coding of the various aspects of the phenomena on scholars’ analyses of case studies.
The reliability of the data was a third challenge. A relevant example is the Pedersen (1979) index that is used to measure net electoral volatility and serves as a possible sign of party change. That is, when volatility increases, it is a sign of decline in party loyalty and one of the indicators of dealignment. This measure seems to be straightforward. All one has to do in order to compute it is to detract the percentage of votes each party won in election t from the percentage of vote it won in election t-1, add up all the products, and divide the sum by two. Yet the problem is that this computation requires us to face the complex reality of splits and mergers and in general the complicated issue of defining what constitutes a new party (Barnea and Rahat 2011). And then different scholars offer different answers to this question, if they address it at all.
A fourth challenge that is especially relevant to this study is comparability. Many of the codings of indicators are based on single-case studies or comparisons of several cases. These have to be “translated” and integrated into one database. Even when there are seemingly comparable data, for example of party membership, their nature may differ across countries and even within countries, in ways that affect the numbers (amount paid, obligations, etc.; see Scarrow 2015). In some cases, data are simply not comparable. For example, party membership in Iceland could not be included because parties seem to accumulate members—once a person joins, she is a member for life, without any obligations (Indriðason and Kristinsson 2015).
We must cope with these challenges, because our aim is to look at the phenomena from a wide angle rather than using the best available data as a “proxy” for the study. We thus looked for a way to include the less than perfect and much less than perfect data we gathered as well. This led us to search for a remedy in the field of qualitative comparative analysis, which proposes dealing with such difficulties by calibrating the data (Downey and Stanyer 2010). That is, we translated each indicator of party change or political personalization into a five-point scale (−2, −1, 0, 1, 2). In this way we limited our analysis to two levels of change for each direction (high or moderate increase or decrease) and to one level of no trend. In addition, we invested much effort in being transparent with our data, as is evident from the number and richness of our twenty-one appendices. We also came up with various other solutions to these problems, as will be elaborated in the sections (p.9) “Coping with the Challenges of Measuring Variance in Party Change” in Chapter 2 and “On the Processing of the Data” in Chapter 3, and in the discussion of each indicator in Chapters 3, 4, 7, and 8.
The logic of the structure of the book is very plain. Part I focuses on party change. It starts with a chapter that lays out the conceptual and theoretical foundations for the later chapters, which offer an empirical analysis of the phenomena. The end product is a broad cross-national analysis of party change that allows us to identify both the general trend and the variance among countries. Part II follows the same path, this time for political personalization. Part III examines the links between these two phenomena. It starts with a critical review of the literature and ends with an empirical analysis that uses the two data sets presented in the previous chapters in order to examine the relationship between party change and personalization.
Part I, then, examines party change. Its first chapter, Chapter 2, sets the theoretical and conceptual basis for the attempt made in the subsequent three chapters to analyze party change from a cross-national comparative perspective. Here we argue that, while the literature frames the study of party change around (1) a broad agreement concerning the very existence of a change and (2) a debate that revolves around the depth and meaning of this change (decline or adaptation), it tends to overlook cross-country variance. In order to map variance, a dozen viable indicators of party change are identified. The chapter also describes the methodological barriers for conducting research on longitudinal cross-country variance and the ways in which they were addressed. In addition, it discusses other indicators that were not included in the analysis, explaining why they were left out, and assesses the cost of their exclusion.
Chapter 3 starts with a brief explanation of data processing and then turns to examine nine of the twelve indicators of party change. These include widely used ones (e.g., party membership, electoral volatility, electoral turnout) as well as several that have been proposed and examined by only a few scholars (partyness of ministers and members of parliament, party–unions relationships, continuity of parties). All these indicators examine the direct and indirect links of parties with society. The indirect, mediated links include the relationship between the extra-parliamentary organization and the “party in government,” the party and its members, and the party and interest groups. The direct links with voters include voter attitudes toward parties, as well as patterns of voter behavior and the resulting party system. The significance of (p.10) each indicator is explained, its advantages and limitations are examined, and the trends over time for each indicator are discussed.
Chapter 4 detects patterns of stability and change in the strength of national parties at the local and regional levels. It starts by explaining why the presence of national parties at subnational levels should be regarded as an important indicator of partyness. It continues by justifying the expectations for the decline of national parties at these levels. It then turns to analyze the performance of the national parties. This is done using three different measurements: the proportion of local-level parties’ votes or seats in local elections; the proportion of their seats in the councils of the five largest cities in each country; and their proportion in holding the mayor’s position. In the regional arena, the indicator used is the change in the level of dissimilarity between voting in regional and in national elections.
Chapter 5, which concludes Part I, gives an integrative view of the dozen indicators of party change. This includes, first, a separate examination of the relationship between indicators within the three main types of links: through socialization (party background), through mediators (party membership, interest groups, local and regional government) and directly (voter stands and behaviors). Second comes an analysis of the findings per indicator and dimension. An analysis by country comes next. Finally, we look at the available data on the development of party switching over the years, in order to determine whether parties in some countries already ceased to exist and became just formal and nominal entities.
Part II examines political personalization. The opening chapter lays down the conceptual and theoretical basis for the analysis of political personalization from a cross-national perspective. It proposes a definition of political personalization and closely examines its core meaning, its broadness, and the implications of perceiving personalization as a process. A typology of personalizations is presented, differentiating between institutional, media, and behavioral types and between subtypes within each category. In addition, the chapter examines what the research literature has to say about the general causes of political personalization and about the relationships between its different types.
Chapter 7 identifies ten viable indicators that cover all types and subtypes of offline political personalization: institutional personalization of both governmental institutions (the electoral system and the executives) and nongovernmental institutions (political parties); media personalization, both uncontrolled (news coverage of politics) and controlled (the ways parties and politicians present themselves); and personalization in the behavior of politicians (legislative behavior) and of voters. The logic and significance of each indicator is explained, its advantages and limitations are examined, and trends over time are discussed. For each type of personalization, other potential indicators are mentioned and examined and their exclusion is explained.
(p.11) Next, Chapter 8 looks at online political personalization. After reviewing the study of political personalization online and its claims about the influence of online platforms on political personalization, it presents the results of an original research that compares parties, party leaders, and prominent politicians from twenty-five democracies. The study looks at both the production side (presence and publication of Facebook posts) and the consumption side (Facebook likes).
Chapter 9, which closes this part, presents an integrative analysis of the dozen indicators of political personalization (ten offline and two online). It looks at personalization per indicator and at the relationship between the three types—institutional, media, and behavior. It also compares the levels of personalization by country and attempts to explain them by looking at institutional and political cultural explanations. Finally, the chapter examines the claims that are raised in the literature about the consequences of political personalization.
Part III links party change and political personalization. It opens with a theoretical discussion in Chapter 10 that looks at their relationship from the perspective of the study of political parties and from that of the study of political personalization. It also examines the integrated perspective expressed in Wattenberg’s works (1991, 1994)—the only case in which both phenomena were given equal weight, without one overshadowing the other. It then turns to examine the challenges that were posed to the common wisdom of a zero-sum relationship between party change and political personalization and to the issue of the direction of the causal relationship between the two phenomena.
Chapter 11 presents an empirical analysis of the relationship between the two processes of party change and political personalization: is this indeed a negative relationship? Is it always a zero-sum relationship? It does so using the analysis of the two phenomena that was applied in the previous two parts and integrated in Chapters 5 and 9. It also examines the question of the direction of this relationship: does party change cause personalization, or is it the other way around?
The final, concluding chapter comprises an overview of the main findings presented in the book. It also proposes directions for future research in the three subject areas covered: party change, political personalization, and the relationship between them. Its final section is dedicated to the claim that personalization should be seen as a threat to the quality of democracy, indeed to its very existence, and also suggests ways to redirect personalized energies to the benefit of political parties and democracy. (p.12)
(1) We borrowed the term “partyness” from Dalton and Wattenberg (2000b), though they were not the first to use it (see, for example, Katz 1987). The term “partisanship” was rejected because it is used in the research literature either to denote the attachment of voters’ sentiments to parties or specifically in relation to party identity (see, for example, Dalton 2000; Dalton and Weldon 2007). Partyness is a wider concept that can be applied to voters and also to other elements, such as ministers’ and MPs’ party backgrounds, links of parties with interest groups, and the status of national parties at the local and regional levels.
(3) Following Samuels and Shugart (2010: 32–3), seventeen of the countries in our group were parliamentary (including Israel, except for the period 1996–2003), five were premier-presidential (prominent among them is France), and one, Austria, was president-parliamentary. The most presidential country, according to these authors’ categorization, is Austria, which, in its parliamentary de facto dynamic, is an outlier in their group. Switzerland, Iceland, and Luxembourg are not included in their categorization. Luxembourg can clearly be claimed to belong in the parliamentary group, while Iceland is more parliamentary than presidential. The exception, for all the types, is Switzerland, and its inclusion is beneficial, as demonstrated in the analysis, owing to its different features. It can thus be claimed that our group is about parliamentary democracies and additional regime types that tend to be parliamentary more than presidential.