CREST. Follow my Leader? A Cross-National Analysis of Leadership Effects in Parliamentary Democracies. By John Curtice and André Blais - PDF

CREST RE FOR RESEARCH INTO CENT ELECTIONS AND SOCIAL TRENDS Working Paper Number 91 September 2001 Follow my Leader? A Cross-National Analysis of Leadership Effects in Parliamentary Democracies By The

Please download to get full document.

View again

of 20
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.

Essays & Theses

Publish on:

Views: 58 | Pages: 20

Extension: PDF | Download: 0

CREST RE FOR RESEARCH INTO CENT ELECTIONS AND SOCIAL TRENDS Working Paper Number 91 September 2001 Follow my Leader? A Cross-National Analysis of Leadership Effects in Parliamentary Democracies By The Centre for Resear ch into Elections and Social Trends is an ESRC Research Centre based jointly at the National Centre for Social Research (formerly SCPR) and the Department of Sociology, University of Oxford A b s t r a c t It has been argued that an increasing focus on leaders in election campaigns together with a decline in partisanship amongst voters means that even in parliamentary elections voters are now more likely to vote on the basis of their evaluations of leaders rather than parties. This presidentialisation of parliamentary elections might be particularly thought likely to have happened where a majoritarian rather than a proportional electoral system is used. These claims are tested using data from the first wave of the Comparative Study of Electoral Systems Project. No evidence is found to support the claim that voters are more likely to vote on the basis of leader evaluations in countries where levels of party identification are low or where a majoritarian electoral system is in place. It is suggested these negative findings may have been obtained because the presidentalisation thesis may have taken insufficient account of the ability of leaders in parliamentary democracies to shape the strategy and appeal of the parties that they lead. 1 Introduction It has become quite common in discussions of how the electoral process is developing at the turn of the millennium to suggest that parliamentary elections have become presidentialised. It is argued that although parliamentary elections were once devices that either secured the election of a legislature whose party political composition is broadly representative of public opinion or else the election of a government formed a single political party (or a pre-election coalition) they have now become an occasion for voters to choose between alternative candidates for prime minister. This change has come about, it is argued, both because of changes in the way that elections are fought and in the motivations that voters bring to the ballot box. In making this claim it is also sometimes argued that some forms of parliamentary election are more susceptible to this trend than others. We have already referred to the traditional distinction between elections being about the election of a representative legislature or about the election of an alternative government. This distinction is of course embodied in the difference between proportional electoral systems that allocate seats in the legislature such that the proportion of seats each party wins is at least broadly proportional to the share of the overall vote that they win, and majoritarian systems that either formally or in practice tend to ensure that bigger parties, and mostly importantly of all the biggest party, secure a significantly larger share of the seats than they do of the votes such that one party may well indeed be able to win an overall majority (Lijphart, 1999). Elections conducted under the latter type of electoral system have traditionally been regarded as a choice between alternative governments and it is these that it is now suggested are the ones that are most susceptible to the trend towards presidentialisation. 2 Although such claims may be commonly made (Bean and Mughan, 1989; Glaser and Salmon, 1991; Mughan, 1993; Mughan, 1995; Swanson and Mancini, 1996), they have been little tested. They of course require two rather different types of analysis. The argument that parliamentary elections have become more presidential requires us to examine the relationship between leader evaluations and voting behaviour over time. Although some attempts have been made at this within individual countries, the analysis of the impact of leader evaluations on voting behaviour in parliamentary democracies (as opposed to the presidential system of the United States) has been notable for its paucity rather than its extent (McAllister, 1996). It certainly has not been systematically examined across a range of polities. Meanwhile the claim that elections conducted under majoritarian electoral systems are particularly likely to have become presidentialised requires the systematic collection of data on leadership evaluations and voting behaviour across countries with different voting systems. To our knowledge at least such data have simply not existed until now. This paper is a first step at filling some of these gaps. It uses data on leadership evaluations, party evaluations and voting behaviour collected on a systematic basis across a range of countries the Comparative Study of Electoral Systems project ( to provide a direct test of the claim that elections conducted under majoritarian electoral systems are more presidential than those conducted under more proportional systems, together with an indirect test of the claim that parliamentary elections in general have become more presidential. In the event we find little support for either argument. We suggest that the importance of leaders in the electoral process lies not, as the presidentialisation thesis implies, in judgements that voters make about leaders independently of their parties but rather in the judgements that voters make about parties as a result of the messages that are conveyed about those parties their leaders. 3 The Presidentialisation Thesis As we noted in the introduction there are two strands to the argument that parliamentary elections have been turned into presidential contests. One refers to developments in the way that elections are fought while the other refers to changes in the motivations that voters bring to the ballot box. We will describe each of these briefly in turn. The first argument is essentially a claim that parliamentary elections are increasingly being fought as if they were presidential contests. The demands and opportunities created television in particular have ensured that the reporting of election campaigns has come to be concentrated increasingly on the activities of leaders. Television, it is argued, needs relatively accessible visual images to project messages, and the personality of a politician provides an image that no party manifesto can ever hope to match. At the same time, television cannot necessarily afford to have camera crews following a wide range of leading party politicians on the campaign trail and thus they tend to focus on the activities of the leader. The parties themselves respond to these pressures focusing their campaigns on their leader, whose image and personality can after all be conveyed across the nation television in a manner that cannot be achieved any other means of communication. The parties may even agree to their leader participating in a televised debate with all the other party leaders, debates that are similar in style to those which have now become a fixture of US presidential elections, while the print media are not immune to the increased focus of election campaigns on leaders (Dalton, McAllister and Wattenberg, 2000: 51). In short, election campaigns have in effect become gladiatorial contests between the party leaders fought out on the small screen. 4 The second argument in contrast refers to relatively well worn themes about how voters motivations and behaviour have changed. At its heart is the claim that modern electorates have experienced a process of partisan dealignment. According to this thesis, voters now largely lack the strong emotional attachment to a political party that many of them once enjoyed, an attachment that helped bring about the development of party identification theory (Campbell et al, 1960). For voters with a strong party attachment or identity, political parties were a vital cue, shaping their views about both policies and leaders. Thus, for example, they would be inclined to favour a particular policy position if it were adopted the political party with which they identified, while they would be likely to oppose it if were proposed some other party. And equally, they would tend to like a party leader, irrespective of their personal qualities, if that leader were the leader of their own party, and to dislike them if they were leading a different party. For so long as voters had motivations like these, leadership evaluations could have little or no independent impact on voting behaviour. They were after all simply derivative of party identification. But if, as has been widely argued is the case (Dalton, 2000; Franklin, Mackie and Valen, 1992; Schmitt and Holmberg, 1995), fewer voters now have a strong party identification, not least perhaps because in the television age voters are exposed to a wider range of messages than was previously the case, then other possible influences can be expected to play a role in their decision about how to vote. Amongst those other influences might be what they think about the issues at stake or of the merits of the alternative leaders with whom they have been presented. And of course they have every incentive to focus on the latter if indeed the first argument that leaders themselves have become more prominent in election campaigns is correct. 5 Even so, there may be institutional constraints on the degree to which evaluations of leaders rather than, say, party positions on issues, can fill the void left the decline of party identification. Indeed even some of the strongest advocates of theories of partisan dealignment have expressed doubts about the degree to which parliamentary elections as opposed to presidential elections either have or are likely to become beauty contests between the leading candidates for prime minister (Dalton and Wattenberg, 2000). But if that view is correct, the institutional constraints would appear to be greater in parliamentary elections fought under a majoritarian electoral system than in the case of those fought under more proportional ones. Elections fought under a majoritarian electoral system can in many respects appear to be functional equivalents of presidential contests. If such a system facilitates the dominance of electoral representation two parties (or at least two blocks of parties) then the election is likely to determine who will hold executive office. There is no need for any process of post-electoral coalition bargaining to determine who will grasp the reins of power. In this situation, elections are less about who will provide representation in the legislature and more about who will form the next government. And of course the most powerful office in that government is the post of prime minister. So we might therefore anticipate that under majoritarian electoral systems at least evaluations of leaders have acquired an importance in determining how voters behave that they may well still not have where a more proportional electoral system is in use. However, we should also be aware of another perspective on this question. Perhaps what matters most in determining the importance of leader evaluations is not the method of election, but rather the size of the party. After all, even when a proportional method of election is in use, when a number of parties are able to secure significant representation in 6 the legislature, and when post-election coalition bargaining is likely to have to take place, it is still likely to be the case that only the leaders of the larger parties have any realistic prospect of becoming prime minister. So if voters do have a particular interest in the qualities of who is to hold the highest political office in the land, that could well mean that while they take into account what they think of the leaders of large parties they take far less account of the qualities of the leaders of smaller ones. So we might find that evaluations of the leaders of big parties matter more than those of small ones, and indeed that this is true irrespective of the electoral system that is in place. Data and Method To test these propositions we use data from the Comparative Study of Electoral Systems (CSES) project ( This project is a voluntary collaboration between national election studies across the world. Each country devotes about ten minutes of questionnaire time to asking a module of questions in a common format while at the same time also collecting a range of commonly coded socioeconomic background information. In each case the data are collected as soon as possible immediately after an election has been held. This means of course that rather than being undertaken at one point in time the data collection is spread across a number of years. In our case we are analysing data collected as part of the first wave of collaboration that took place between 1996 and This first wave of collaboration was designed to focus on three subjects. These were the impact of different electoral systems on the way that people vote, the role of social cleavages in voting behaviour, and what influences people s evaluations of their system of government. This paper is an example of the first of these. Like the project as a whole, it gains its analytic purchase on its subject the opportunity that is created to examine how relationships between variables collected the surveys vary according to the 7 institutional context of the participant countries. In other words we can analyse how far electoral systems affect the impact of leadership evaluations on the way that people vote examining how far the relationship between leadership evaluations and vote varies between countries according to the kind of electoral system that they employ in their parliamentary elections. The CSES project collected three simple but crucial pieces of information of relevance to our purposes here. First, respondents were asked to state on a scale ranging from 0 to 10 how much they liked or disliked each of the main party leaders. Normally, evaluations of up to six party leaders were collected in each country, with information being gathered on each of the party leaders whose party enjoyed significant representation in that country s legislature. The relationship between these evaluations and vote then provides us with our crucial indicator of the impact of leadership evaluations on vote. Second, using exactly the same scale the CSES project also asked respondents how much they liked or disliked each of the main political parties. We have seen that the presidentialisation thesis argues that evaluations of leaders comprise one of the factors at least that has come to replace evaluations of parties in voters decisions. This implies that rather than just looking at the variation between countries in the relationship between leadership evaluations and vote, we should be examining the variation in the relative importance of leadership and party evaluations. By deploying the party like and dislike data as well, we can adopt this approach. Finally, the CSES project also collected data on party identification in a systematically comparable manner across countries. It is this information that provides us with an opportunity to implement an indirect test of the claim that evaluations of leaders have become more important because of a decline in the proportion of voters who feel a sense of party identification. Of particular interest to us is the proportion of people in each 8 country who claim a party identification. If the presidentialisation thesis is correct and applies to parliamentary elections then we would expect that leadership evaluations would be more important in those countries with relatively low levels of party identification. The presence or otherwise of a party identification was collected the CSES asking respondents, Do you usually think of yourself as close to any particular political party? Our indicator of the level of party identification in each country is the proportion that responded positively to this question. 1 By now it should be apparent to the reader that these data allow us to do more than examine how the relationship between leader evaluations and vote varies between countries. After all we have information on leader and party evaluations for each of the main parties within each country. And as we have already acknowledged such evaluations may well be more important to voters in deciding whether to vote for a big party whose leader has a realistic chance of becoming Prime Minister than it is in deciding whether to vote for a small party whose leader does not have any such prospect. It might be the case too that a small party leader finds it more difficult to convert personal popularity into votes because of strategic disincentives to voting for a small party. We have then a data structure that allows us to examine how the role of leadership evaluations in the voting decision varies according to the characteristics of both countries and parties. To exploit this opportunity we adopt a very simple two stage approach. In the first stage we acquire an estimate of the impact of leadership evaluations on vote for each party in each country. This is done undertaking a simple linear regression of vote, dichotomised as 1 if the respondent voted for the party in question and 0 otherwise, against leader evaluations using the individual level data for the relevant country. We 1 Those who said no were asked a follow-up question, Do you feel yourself a little closer to one of the political parties than the others, but we have taken no account of the incidence of these leaners in our analysis. 9 also acquire a similar estimate for the role of party evaluations undertaking exactly the same procedure for these. Together the resulting two equations equip us with two simple linear regression coefficients for each party in each country, one of which measures the strength of the association between leadership evaluations and vote, the other the strength of the association between party evaluations and vote. 2 It is these coefficients that then become the focus of our analysis in the second stage. We analyse how the coefficients vary according to the characteristics of the party and of the country in question. In addition to the level of party identification for each country (derived from the relevant CSES data set) we have also coded each country s legislative electoral system according to whether it is primarily majoritarian, proportional or mixed. The last of these categories typically refers to those countries in which some seats are allocated according to a majoritarian principle, some a proportional method and where the latter are not allocated such as to overcome the disproportionality generated the outcome in the former. 3 Details of how we have classified each country s electoral system can be found in Table 1 below. 4 Meanwhile we also collected for each party the share of the vote that it won at the election to which the CSES data refers, there enabling us to distinguish between big and small parties. 2 We are course aware that normally a logit or probit procedure would be preferable in the analysis of a dichotomous dependent variable. Linear regression gives us however a more easily interpreted metric for the second stage of our analysis described in the next paragraph. 3 Our principle source of information on the characteristics of each country s electoral system was Reynolds and Reilly (1997), supplemented where necessary consultation with a range of appropriate Internet sites. 4 We also classified countries according to the maturity of their democracy in case this might have influenced or confounded our results. For example, we might anticipate that leader evaluations are more important in countries where democracy has not been in place for long and where as a result voters do not have a clear impression of the standpoint o
Related Search
Similar documents
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks