In Chapter 2 we established a rationale for the voting act that would make it plausible to assume that people are influenced by electoral context when deciding whether to vote, thereby laying the groundwork for resolving several of the puzzles outlined in Chapter 1. This context changes from election to election, depending on something we have called the character of elections – primarily their competitiveness. But we asserted in Chapter 1 that the context of the election has more impact on some cohorts (newer cohorts) than others (older cohorts). Before we can proceed to an analysis of the effects of the changing character of elections on turnout, therefore, we need to address the question of why and to what extent cohorts differ in their responsiveness to the changing character of elections. Until we have done so we will not be able to properly specify any model of turnout change.

The idea that generational replacement plays a role in turnout change is one that has not been extensively explored in the literature. Newly enfranchised individuals are known to be particularly open to recruitment by new parties and to be largely responsible for such changes as occur in the support for existing parties (Campbell et al. 1960; Butler and Stokes 1974; Nie, Verba, and Petrocik 1978; Inglehart 1977, 1990, 1997; Rose and McAllister 1990; Franklin et al. 1992; Miller and Shanks 1996).

In Chapter 1 we saw that the attempt to explain voter turnout without reference to the instrumental motivations of voters gives rise to a number of puzzles, and we asserted that those puzzles would be easier to resolve if we could derive a calculus of voting that made it rational for individuals to take account of the political benefits they might enjoy as a consequence of the electoral victory of one party or candidate rather than another.

It is perhaps somewhat brave of us to think that we can, in this book, resolve a paradox that has baffled some of the best minds in the profession over several decades. However, it is clear that contemporary political scientists know a great deal about the voting act that was not known to Riker and Ordeshook when they published their seminal article in 1968. With the benefit of this knowledge (and building on several advances in rational choice theorizing about individual decision-making in the context of collective action) we are able to put forward a rationale for the voting act that makes it quite understandable that voters would turn out in greater numbers in elections that aggregate-level findings tell us should indeed be high-turnout elections.

We acknowledge, as have many following Downs (1957) and Olson (1965), that voters face two collective action problems when it comes to voting: gathering enough information to form a preference among the candidates or issues, and expending the effort required to cast a ballot.

The puzzles alluded to in the Preface and Introduction to this volume come in two major varieties. The first are puzzles about voter turnout itself, among them, why do people bother to vote at all? Why is turnout so relatively stable over time (compared to the enormous differences we see between countries)? Why does it decline when it does? And why (in some countries) does it not decline at all – or even rise? The second variety of puzzles have to do with how to study voter turnout. For example, is turnout an aggregate-level or an individual-level phenomenon? Can we understand it best by studying turnout change over time? By studying differences in turnout among countries or by studying why some individuals vote while others do not? What can we learn from the fact that turnout varies more among younger cohorts and less among established cohorts?

In this chapter we take these and other puzzles one at a time and use the process of going through them as an opportunity to set out not only the puzzles themselves but also the major approaches that have been employed in past research for solving them.

Why So Much Turnout Change – or Why So Little?

It has already been pointed out that turnout decline is not ubiquitous. Turnout did rise in the 2000 U.S. presidential election, for example.

The findings of past chapters – and especially those of Chapter 5 – have provided us with the raw material we need to understand turnout change of all kinds, but in this chapter we will focus particularly on turnout decline. This is a topic that has much exercised commentators, as reported in Chapter 1. In Chapter 5 it was established that the causes of turnout change primarily have to do with the character of elections, not the character of society, so commentators who see in falling turnout a reflection on the civic-mindedness of citizens, or on their commitment to democracy, appear to be mistaken. Variables such as party attachment (Wattenberg 2000, 2002), trust in government (Dalton 1999), or union membership (Gray and Caul 2000) contribute nothing to the explanation of turnout developed in this book and tested in Chapter 5 (see Chapter 5, note 36).

So why has turnout declined? Evidently, the preferred model established in Chapter 5 plays out differently in different countries. We would certainly expect that changes in the responsiveness of the executive play a primary role in explaining Swiss turnout decline, while the abolition of compulsory voting must play a primary role in Dutch turnout decline. Nevertheless, the overall picture can still be painted in terms of all countries taken together provided that we bear in mind that certain developments take place (certain variables change their values) in some countries but not in others.

Voter turnout regularly makes news. Seemingly, whenever an election is held, the question comes up: How many people voted? Sometimes the turnout is unexpectedly high. Commentators were amazed at how many people stood for hours in the hot African sun waiting to vote in South Africa's first truly free and universal election. But it is rare to see stories about higher than expected turnout. More often we see stories that express concern at the fact that turnout is lower than expected – so much more often, indeed, that one might be forgiven for supposing that low or declining turnout was ubiquitous in contemporary democratic elections. One prosaic reason for this is the newsworthiness of turnout decline. Stable turnout is not news. Moderately increased turnout is not news. Low or declining turnout is newsworthy. So commentators draw attention to the level of turnout mainly when it is down. How many people are aware that turnout was higher at the American presidential election of November 2000 than at the previous presidential election, in 1996? Of course, the level of turnout in the more recent of those elections was overshadowed by its other, more newsworthy, features – butterfly ballots and such. But the lack of press attention given to increased turnout, when it occurs, is one reason why we have this general perception that turnout everywhere is in decline. What is true is that, whenever turnout is down, the decline makes news.

The time has come to put what we have learned into a model that attempts to explain turnout change over the longest possible period for the largest possible number of countries. Twenty-two countries have a record of elections held continuously since within one electoral cycle (generally four years) of the end of World War II, amounting to 356 elections in all. We need to study elections held continuously because, if any elections are missing, we can investigate neither cumulative effects nor influences like the habit of voting that are coded with reference to past elections. Many countries that started holding elections shortly after World War II do not meet the criterion of continuous elections. If we attempt to increase the number of countries beyond twenty-two, we dramatically shorten the total period over which we can make the comparisons. At the same time, the period cannot be extended to include years before World War II because so many countries had at least two electoral cycles without elections before and during that singular cataclysm (Austria, Britain, France, Germany, Italy, Japan, Luxembourg, The Netherlands, and Norway) or did not exist (Israel). The countries we can study from 1945 onward include those ten and also Australia, Belgium, Canada, Denmark, Finland, Iceland, Ireland, Malta, New Zealand, Sweden, Switzerland, and the United States. The period of our study ends in 1999 because data for some variables are hard to find for elections more recent than that.

Human beings, it has been suggested, have a “puzzle instinct” (Danesi 2002) – a fascination with puzzles and an aptitude for solving them. Academic research in most disciplines is all about puzzle-solving, but political science is perhaps unusual in being home to a great many puzzles that are of interest beyond the walls of academe. Bernard Grofman recently edited a book with the title Political Science as Puzzle-Solving (2001) whose premise was that interesting puzzles lead us to topics where the tools and skills of political science can be brought to bear, teaching us useful things about the world.

If the vexing questions of political science can be regarded as puzzles, the particular topic of voter turnout could be called the “grand enchilada” of puzzles in political science. As we will see in Chapter 1, almost everything about voter turnout is puzzling, from the question of why anyone bothers to vote at all to the question of why certain variables appear to explain voter turnout in some circumstances but not in others.

I became interested in these puzzles in the early 1990s as a by-product of my interest in elections to the European Parliament. Turnout in these elections is very low despite the fact that they occur in countries (of the European Union) where turnout levels in national elections are generally high.