This is a book about elections and voters. It is intended as a textbook for those who want a general introduction to the topic, but it is not first and foremost concerned with imparting exhaustive factual knowledge about the nuts and bolts of electoral systems, voting arrangements, party systems, and policy differences. Such a book would necessarily focus on only part of what we want to cover, and such books (each one dealing with only a part of our agenda) already exist.
This is first and foremost a book about the logic of representative democracy and about the role of the electoral process within this logic. It sees elections as opportunities for strategic action on the part of voters and politicians, and tries to explain how election outcomes should be understood as resulting from the interplay of preferences and strategies, which in turn are constrained and channeled by institutional arrangements and communication structures.
In it we hope to supply a picture of how electoral democracy works, together with an assessment of how well it works, using a complete (though not exhaustive) set of tools and theories employed at the cutting edge of political science research. In short, this is a book about what has been called “the wider agenda of electoral research” (Thomassen 2000) in which we have tried to integrate theories about specific aspects of the electoral process, including theories about electoral systems, coalition formation, voter motivation and mobilization, political communication, and so on.
We try to relate all these theories to oneanother in an encompassing view of electoral democracy. We also try to clarify some important reasons why politics in different countries has a different flavor: not because voters are different in different countries, and seldom because of differences in political leadership, but more usually because institutions and party systems are different.
Though there are of course differences in individual values, and historical differences with 191 contemporary resonance in terms of social structure, we see systemic characteristics as being fundamental in explaining why party systems differ from country to country, why political leaders approach elections differently, and why voters make their choices in different terms. Because the characters of electoral and other institutions are so fundamental, and because these institutions differ primarily between countries (and only secondarily over time), in our view elections and voters can only be understood comparatively.
That is what we try to do. The countries that we study are mainly established democracies, where the electorates of today have only democratic life experiences. However, from time to time we do refer to democracies that are not yet established, and in Chapter 7 we devote an entire section to the analysis of voting behavior in the countries of Central Europe that recently became members of the European Union.
The knowledge base on which we build this story is incomplete, but still we try to give a complete and coherent picture, filling in gaps by extrapolating from cognate knowledge and sometimes even by speculation (though we try to be absolutely clear about the basis in political science research for the claims that we make). As such, this book attempts to be an invitation to readers to think along with us and bring to our framework specific information that they may have about political systems and periods that we cannot provide in a volume of this compass.
So we provide a skeleton that can be fleshed out by the reader – or by an instructor using this book as a classroom text. 192 VII. Voter Orientations Over the course of the previous chapters it has become clear that the interaction between voters, parties and electoral institutions is a complex one. An election outcome can be understood from different perspectives. From one perspective, an outcome can be seen in terms of what voters do how they evaluate parties and candidates. Alternatively an election outcome can be seen in terms of what parties do – how they choose policies, exercise leadership, and strategically position themselves.
Yet again, an election outcome can be seen in terms of the institutional setting in which voters and parties interact. So an election outcome is the result of a dynamic interaction between voters, parties and institutions. We have already discussed institutions in Chapter 3. In this chapter we will consider changing voters and in the next chapter we will consider changing parties. Voters change in a number of ways and for a number of reasons. We already dealt with one of those ways (thermostatically) and one of those reasons (responsiveness to policy) in Chapter 6.
Given the important role we earlier gave to habits and routines, there might be thought to be a puzzle involved in understanding how public opinion can be as malleable as we saw it to be in Chapter 6. Part of the answer arises from changes that are always occurring in the composition of the electorate that we discussed in an earlier chapter, but the electorate also contains at any point of time a large number of individuals whose preferences are not (yet) locked down by forces of habit and inertia and whose outlooks, knowledge and skills are still in the process of being formed.
These provide the main source of malleability in voters’ behavior, the subject of this chapter. It is true that older voters may also be swung to support a different party but they are likely to later return to their previous party allegiance. In other words, voters with an established partisanship who are successfully induced to change their votes make poor converts and are unlikely to become committed to their new party. Thus it is predominantly in the malleability of younger voters that we find the potential for long-term change.
Attracting uncommitted voters is an important task for political parties and is responsible for constant jockeying among them, as they seek to position themselves to attract such voters; but the extent to which parties can adjust their policy positions is constrained by the party’s need to ‘keep the faith’ with their established supporters – generally older ones. These older voters have predispositions that may be oriented towards social groups associated with particular political parties, or focused on ideologies embodied in one or more political parties, or linked to specific parties per se.
In any case, because 193 these predispositions are so stable once they have been acquired, most parties at any given point in time have an established clientele whose votes they normally can count on (but whose concerns they must bear in mind). The decline of partisanship Because the potential for long-term change arises mainly from individuals who do not have a firmly established partisanship, the size of the group of uncommitted voters relative to the electorate as a whole is of fundamental importance.
In the days when party loyalties were largely established on the basis of social group memberships, newly adult voters essentially inherited their partisanship and entered the electorate with their loyalties almost fully formed. In such a world there was little room for change other than as made possible by enlargements of the electorate to incorporate new types of voters (as mentioned in Chapter 2). Once universal suffrage had been established it appeared as though party systems became frozen.
However, as described in Chapter 4, the declining power of social cleavages to structure partisanship that occurred in many countries in the last part of the 20th Century, also led in many countries to a decline in the strength of partisanship, with weaker effects of group loyalties giving rise to weaker identification with political parties.
In Britain, and presumably in most European countries, a decline in the strength of party loyalties followed immediately upon the decline in cleavage politics, giving rise in Britain to a phenomenon highlighted in the title of a widely read book, The Decade of Dealignment (Crewe, Alt and Sarlvik 1883), which focused on the declining strength of partisan identification with Britain’s two major political parties. We already asserted in Chapter 4 that the same thing would have happened in the United States, where the decline happened much earlier than in Europe. Figure 7. 1 shows this indeed occurring.
The chart plots the percentage of strong partisans in successive electoral cohorts over time. At the left of the chart we see the pattern for cohorts that were already members of the electorate (or who entered the electorate) in 1952, the date of the earliest US election study containing the appropriate measure. As we move rightwards across the chart we see additional cohorts of voters entering the electorate with varying levels of initial partisanship, and we also see the evolution of partisanship for each cohort.
The fluctuations in partisanship are partly random, due to the small number of people interviewed in each cohort at each point in time, so these fluctuations have been smoothed by taking overlapping averages of each trio of adjacent cohorts. The figure shows a progressive decline in the percentage of strong 194 partisans among cohorts that entered the electorate from 1948 to 1968, along with a temporary downward swing, in 1968 and 1972, among older cohorts – especially those that had not been members of the electorate for long enough for their initial level of partisanship to become locked down.
After 1968 there ceases to be a clear decline in initial partisanship (and might even have been a slight rise). It seems as though, whenever it began (presumably before the advent of academic election studies), the overall decline in the percentage of strong identifiers in the American electorate had come to an end by the time the first members of the post World War II baby boom generation reached adulthood in 1968 – though some studies have seen a particularly strong decline of party identification in the 1964 and 1968 cohorts (Nie, Verba and Petrocik 1979). % 48% S t r o 40% n 1944-52 g p 31% a r t i 21% s a Up to 1956 1952-64 1960-68.
1992-00 1964-76 1976-88 1972-80 1984-92 1948-56 1960-68 1952-68 1956-64 1988-96 1984-92 1980-88 1964-72 1968-76 1972-80 n s 13% 1952 1968 1992-00 1996-04 1988-96 1976-84 year 1988 2004 Figure 7. 1 Percentage US strong partisans by electoral cohort, 1952-2004 From about 1968 onwards, the percentage of strong partisans among young adults entering the electorate (shown by the starting height of each successive line in the graph) appears to be less than half what it was in 1948 and earlier (top line in the graph), and this lower partisanship among young adults has continued to characterize the American electorate in more recent years.
The corresponding drop in partisanship among older cohorts, which also occurred in 1968, was not so 195 long-lived, however. In 1988 party identification among cohorts that entered the electorate in 1960 or earlier returned to more or less its pre-1968 level – which became once again the level of party identification that successive cohorts appear to have reached (or to be in the process of reaching) as they age.
This brings us to an even more striking feature of the graph than the lower level of partisanship among young adults after 1968: the fact that the difference between younger and older adults that we see from 1968 onwards is rapidly made good as each cohort ages. Older voters today are no less partisan than they were fifty years ago, but young voters are much less so. So there are more weak partisans in the American electorate today than there used to be because younger members of the electorate have lower partisanship than used to be the case.
The ‘locking down’ of partisanship that occurs with increasing age, which we mentioned repeatedly in past chapters, turns out to be a much more marked feature of the contemporary American electorate than it was half a century ago; and this is probably the case in other countries also. The result is that the group of uncommitted young voters is much larger than it used to be, providing the raw material for greater volatility of election outcomes in the modern era (and greater potential for long-term change). The same pattern that we see among strong partisans in figure 7.
1 can also be seen among weak partisans and those who lean towards a particular party. All of them show the same evolutionary pattern over time. Generational replacement and electoral change Because an electorate is a constantly evolving entity, with voters leaving in a constant stream through incapacity and death and entering in a constant stream as children grow up to become young adults old enough to vote, it is not actually necessary for any individuals to change their mind about what they want politically in order for the electorate collectively to change its mind.
All that it takes is for the voters who are leaving the electorate to want different things than the voters who are entering the electorate. To the extent that there is a systematic difference (that is not an effect of aging) between the political complexion of those leaving and those entering the electorate, over time there will be a change in the political complexion of the electorate. This is obvious if we consider the extreme case of an electorate in which older people all support one party and younger people all support a different party.
Over time, support for the older voters’ party will dwindle to nothing as those voters progressively leave the electorate while support for the younger voters’ party will increase to encompass the whole electorate. This would take about 196 sixty years to accomplish – the time that the average voter remains in the electorate. In practice, generational change is seldom as dramatic as this, but even a quite small difference in political orientations between younger and older cohorts of voters can shift the electorate significantly to the benefit of the party or parties that are supported by younger voters.
In western countries in recent decades, at each election roughly 10 percent of the electorate has been new. If new voters favor a given party 30 percent more than the oldest voters do, then at each election the shift in votes will be about 3 percent in favor of the party supported by new voters. Elections are often won or lost by margins less than this (as discussed in Chapter 5 – see especially Figure 5. 1). If the difference between younger and older voters is only 10 percent then it will take three elections for a 3 percent shift in votes to occur.
Smaller differences between younger and older voters will slow down the speed with which political change takes place, but the changes induced by generational replacement are inexorable. The fact that they cumulate over time gives them immense power over the long haul. Political change and political realignments But generational replacement is surely not the only way in which the electorate changes. There is controversy about this in the political science literature, but in some countries it is clear that generational replacement is far from being the dominant force in electoral change.
Van der Eijk and Niemoller (1983) found that in the Netherlands party switching by established voters in consecutive elections caused three times as much political change as generational replacement. However, to the extent that this switching was due to older voters changing their party support, many of those changes are likely to have been short-term in nature, with converts being reconverted at subsequent elections. And much of the switching that occurred was limited to shifts within particular ‘party families’, as discussed in Chapter 4, rather than between parties with radically different policy postures.
Consequently, over longer periods of time the replacement effect is likely to have been greater, as was indeed established in a longer-term study by the same authors (van der Eijk and Niemoller 1992:268). But there is an important question as to whether the customary situation that we have described up to now in this chapter also applies at times of massive change in party support – changes that are known as ‘political realignments’ – that we mentioned in Chapter 2.
Realignments of this type are rare, but many suppose that they occur precisely because the normal forces for continuity in party support go into abeyance at such times. 197 A classic treatment of political realignment in the United States (Key 1955) suggested a typology of patterns of change that distinguished between elections that (a) maintain the alignment of political forces (such that the parties continue to be supported by their established clienteles), or (b) loosen the relationship of party attachments to a previous clientele (dealignment), or (c) alter those relationships (realignment).
Actual elections are not generally very easy to classify according to this typology, as they tend to exhibit mixtures of maintaining, dealigning, and realigning tendencies among different groups of people. Key suggested that realignment had to be preceded by dealignment — that the bonds of loyalty had somehow to be loosened so that they could be reconstituted in a different configuration yielding a new long-term equilibrium of party forces. However, no research has been able to establish that this is in fact what happened at the time of historical realignments in the United States and elsewhere.
As a case in point, one contemporary account of the 1930s realignment in the United States (Lubell 1952) reports that the author, after talking to hundreds of voters over the course of several years, had failed to find even one whose party loyalties had been weakened and transformed. The problem for political scientists is that no realignment of the classic type has occurred in any country since the advent of academic election studies based on random samples of the mass electorate.
By reconstructing electoral loyalties of the 1930s from academic surveys conducted in the 1950s, Kristi Andersen was able to validate Lubell’s failure to find voters who had changed their party allegiance during the US ‘New Deal’ realignment of the 1930s. Her reconstruction seemed to show that the realignment of the 1930s in the United States, took place as a result of the mobilization of large numbers of previously non-voting individuals rather than the conversion of existing voters.
The new voters, according to Andersen, were predominantly drawn from the pool of first- and second-generation immigrants who had acquired citizenship prior to the depression years but had not yet been mobilized to vote. This argument is also consistent with Lipset and Rokkan’s view of expansion of electorates as the motor of European electoral change in the late 19th and early 20th Centuries. Newly enfranchised voters provided time and again a pool of electoral newcomers available to be mobilized especially by new political parties.
According to this account, what President Roosevelt managed to do during the 1930s in the United States was to motivate people to vote for the Democratic Party who otherwise would not have voted at all. This interpretation of the 1930s realignment in the US has been contested, and it is hard to be definitive about what happened because the realignment occurred over the course of three 198 elections, which has allowed political scientists to argue about what should be considered the proper base from which to calculate the changes that took place (Erikson and Tedin 1981).
Arcane definitional problems remain unresolved, and different scholars today describe the events of the 1930s in different terms. Less ambiguity surrounds the British realigning election of 1945, when the Labour Party definitively replaced the declining Liberal party as major competitor to the British Conservatives. Because there had been no election in Britain since 1935 (an election that should have been held in 1939 was cancelled at the outbreak of the Second World War), the realignment of 1945 took place at a single election.
There is thus no argument about the base from which the realignment took place, and there has been no controversy regarding the only research so far conducted on the topic, which attempted to reconstruct the 1945 British electorate on the basis of 1960s election study data. The reconstructed data (Franklin and Ladner 1995) showed exactly the same pattern of change in the British realignment as had been found by Kristi Anderson for the earlier realignment in the United States: the major source of change was generational.
Citizens who had not been old enough to vote in 1935 were entirely responsible for the huge increase in votes for the British Labour Party in 1945. There is no need to suppose that normal bonds of party loyalty were somehow weakened through dealignment in order to explain the British realignment of 1945. It happened not because voters switched their party allegiances but because new voters were sufficiently numerous and sufficiently different from established voters to bring about the change in election outcomes.
Generalizing from the British and American cases, it would seem that realignments, even of the classic type, did not in fact require that normal forces of party loyalty go into abeyance. Instead, such political realignments appear to have been long-term consequences of electoral enlargements that increased the sizes and political distinctiveness of new cohorts, boosting enormously the normal effects of generational replacement.
When the franchise was extended to new groups of citizens, or (in the United States) when large numbers of immigrants started to take advantage of their recently acquired citizen status, the result could be large swings in the electoral fortunes of political parties. With the enfranchisement of virtually all adult citizens in established democracies in the contemporary world, there would seem to be little room for further realignments of this kind.
But the modern world is also different in another way from the world in which realignments of the classic type occurred. Because of the decline of cleavage politics, a window has opened during young adulthood in which partisan forces for most young adults are 199 much less than they used to be, as we saw earlier in this chapter. Because of this development, large swings in support are possible in contemporary electorates without requiring a corresponding enlargement of the electorate.
A tentative assessment of the sources of the realignment that is known to have occurred in the southern states of the USA after 1948, where a pre-1948 hegemony of the Democratic Party gave way to almost universal Republican electoral victories in later years, illustrates how this can happen. In Figure 7. 2 we show a subset of the electoral cohorts involved in the realignment that took place between 1948 and 1984 among white voters in the Southern United States (starting with the earliest election for which we have adequate data and continuing to include the Republican electoral victories of Ronald Reagan in 1980 and 1984).
We omit the election of 1972 and cohorts after 1968 (the missing year and cohorts would have muddied the picture because of the Watergate scandal in 1972-3 and the resulting Democratic victory in 1976). The cohort movements have also been smoothed by taking three-cohort overlapping averages in order to dampen fluctuations due to the small numbers of Southern whites in U. S. election studies. These cohorts show three clear patterns of movement over time.
The first is a swing to Republicans and back again (and then back to the Republicans a second time and the apparent start of a second reversal) among the oldest cohorts (those entering the electorate in 1956 and earlier). This pattern is typical of what we expect from cohorts who have acquired a partisanship – their members may be induced to defect from that partisanship, but only temporarily. Such movements contribute towards the “swing of the pendulum” that we will consider later in this chapter, but not to long-term electoral change.
The second pattern is of a progressive movement towards the Republicans among cohorts who entered the electorate in 1960 and later. Because the cohorts concerned began their move to the Republicans before they had had time to become locked down in their support for their initial party choice, it was still possible for them to vote for a different party and even to acquire a different partisanship – a pattern we expect from voters who are not yet locked down in their partisanship.
The third pattern is a clear pro-Republican movement among successive cohorts of voters, each of which either enters the electorate with a stronger initial Republican complexion than the previous cohort, or rapidly acquires such a complexion (in the case of the 1952 cohort) and maintains this cohort differential in succeeding elections.
Thus 200 1960-68 83. 5 Smoothed 3-cohort averages % 1956-64 R e p u b l i c a n x x 1952-60 1948-56 1944-52 1948 and earlier x x v o t e x x 38. 5 1952 1960 1968 year of study 1976 1984 Figure 7. 2 Republican vote among Southern white 1948-68 cohorts, 1952-1984 elections Source: American National Election Studies, 1952-1984, cumulative file. each line in Figure 7. 2, representing the extent of Republican voting in a particular cohort, acquires a position higher on the chart than the line for the previous cohort. This progression is what we expect to see during a period when some group (Southern US whites, in this case) is changing its partisan identity.
Each successive cohort reflects the trend more clearly than the preceding cohort, and the final line-up of these cohorts in 1984 has them clearly arranged in sequential order, with the 1960-1968 cohort as the most Republican at the top of the chart in 1984 and the 1948 and earlier cohorts as the least Republican at the bottom of the chart in that year. This example shows how a new long-term equilibrium can come about. It happens incre– mentally, requiring a number of consecutive elections to become manifest.
The search, especially among American political scientists, for “the next realignment,” while only considering pairs of consecutive elections, is therefore bound to be fruitless. In the absence of a massive influx of new voters (such as following a franchise extension), only over a sequence of elections can we expect to find sufficient electoral replacement to provide the basis for a realignment (if there is one). On the other hand, it also needs to be born in mind that in the modern word large swings in party support no longer mean what they meant before the decline of cleavage politics.
When party 201 preferences were largely locked down and large swings in support could only come from massive increases in the voting population, realignment really meant something. Today, large swings in party support no longer require a change in the complexion of the electorate – but the very flexibility of contemporary electorates that facilitates such large swings also makes it likely that they will soon be reversed. So the identification of large swings in party support with realignments of the party system no longer holds (Franklin and Hughes 1999). The hand of the past
We have seen that a voter’s early experiences tend to become ‘locked down’ with the passage of time. A cohort that gives strong support to some party when its members are young usually continues to do so in the future, and the reverse happens for a cohort that avoids voting for a particular party when its’ members are young. This results in a baseline vote in any election that is largely set by the electorate’s past history. The same is true for turnout. A high turnout era will tend to perpetuate itself among the cohorts that entered at that time, as will an era in which a turnout was low.
Only a portion of the electorate is truly free to vote or not, or to choose any one of the political parties competing for their votes. The rest is set in its ways and those settled ways reproduce in the present the political orientations of years gone by (cf Miller and Shanks 1996:2235). Because voters are most responsive to the politics of their formative years, those politics continue to affect the future for as much as sixty years ahead (the length of time the average voter remains in the electorate after first casting a ballot).
Of course, even established voters are not all immunized, and immunization can be overcome. Certainly, voters set in their ways of voting for a particular party may fail to go to the polls in a given election if the party really disappoints them, as we have seen. In these ways we can expect short-term deviations from the baseline, but those deviations are liable to reverse themselves in the near future. More permanent changes amongst formerly immunized voters occasionally occur too, given strong enough motivation.
Voters whose immunization is neutralized later in life again become free to adopt new partisan identities (as appears to have happened in France for many of the generation that supported left parties in the late 1960s – see Box 7. 1 on President Mitterand’s U-turn). Another example occurred when the Dutch Labour Party, in government in the early 1990s, was forced to take the lead in a neo-liberal restructuring of welfare arrangements in the country.
This caused widespread disenchantment particularly 202 amongst its oldest and most established supporters who withdrew their electoral support in sizably numbers, not to return again (Hillebrand and Irwin 1999). Similar problems are being felt at this time in Germany where the governing Social Democrats needed to bring the country into line with an already restructured pattern of welfare arrangements in neighboring countries, and instituted policy changes which helped in the rise of a new left-of-Social Democrats party (Die Linke). These are instances of changes in the baseline itself, which therefore have permanent effects on turnout or on party choice (or both).
Circumstances that may bring about an increase in the number of unimmunized voters can also include changes in the institutional context. The abolition of compulsory voting in the Netherlands in 1970 is a likely suspect in our attempt to explain the increasing volatility of Dutch voters since the 1970s (Irwin 1974; Wolinetz 1988); or large-scale changes in the set of choices on offer (the melt-down of the Italian party system in the 1990s), for example. Deep disappointment – such as was suffered by left voters in France at the time of President Mitterand’s fam