CREST PAPER NO 64 : NEW LABOUR, NEW TACTICAL VOTING ? THE CAUSES AND CONSEQUENCES OF TACTICAL VOTING IN THE 1997 GENERAL ELECTION
BY
GEOFFREY EVANS, JOHN CURTICE AND PIPPA NORRIS
Introduction
The outcome of the 1997 election broke a number of records. More Labour MPs (419) were elected than ever before, surpassing even the party's victory in 1945. More Liberal Democrat MPs (46) were elected than at any time since 1929, more than doubling their numbers compared with 1992 and giving the party the kind of breakthrough of which the former SDP/Liberal Alliance had dreamed. Meanwhile the Conservatives were left with fewer seats (165) than at any time since the Liberal landslide of 1906.
Yet if we look at the outcome in terms of votes the performance of neither Labour nor the Liberal Democrats was at all historic. At 44.4 per cent, Labour's share of the vote was lower than at any election between 1945 and 1966. Its lead over the Conservatives was less than that secured by the Conservatives themselves over Labour in 1983. Meanwhile, the Liberal Democrats suffered their third successive drop in support. At 17.2 per cent, their performance was the second worst since the party resumed fighting elections on a nation-wide basis in 1974.
Evidently the Labour landslide and Liberal Democrat revival of 1997 were as much a product of the operation of the single member plurality electoral system as an indicator of the strength of their electoral support. Had the movement of votes since 1992 been uniform across the country as a whole, Labour would have secured a majority of 131 rather than 179. The Liberal Democrats, meanwhile, would have won just 28 seats rather than 46, leaving the Conservatives with no fewer than 208 seats rather than 165. One reason for this divergence appears to have been the degree of tactical (otherwise known as strategic) voting. A significant number of voters appear to have been willing to support whichever of the main opposition parties was best able to defeat the local Conservative.
Tactical voting between Labour and the Liberal Democrats is therefore central to understanding and explaining the outcome of the 1997 general election. Moreover, the analysis of tactical voting has significant implications for our understanding of theories of voter choice and party strategy, in Britain and elsewhere. This article therefore sets about two tasks. First, it establishes whether there was any change in tactical voting between 1992 and 1997 by examining both the pattern of constituency election results and individual-level survey data from the 1997 British Election Study. In particular, we assess whether these two very different sources confirm that there was an increase in the amount of tactical switching between Labour and the Liberal Democrats. Second, we consider what might account for any observed changes by evaluating the plausibility of several theories of tactical voting, theories which emphasise the importance of changes in the motivation of voters, party mobilization, the information environment, and party competition respectively. We conclude by considering briefly some of the implications of our findings.
Data
This article draws primarily on two distinct sources of information. The first is analysis of the aggregate election results at constituency level. This allows us to assess the apparent incidence of tactical voting, taking into account the variable electoral geography of Britain. The second is analysis of individual-level survey data from the 1997 British Election Cross-section Study. This allows us to examine voters' own reports of having voted tactically rather than simply inferring its existence from constituency-level patterns. This survey is the largest and most representative of a number of studies conducted as part of the 1997 British Election Study (BES) and is therefore the one best-suited for examining a relatively rare phenomenon (see Galbraith and Rae, 1989; Heath et al. , 1991; Johnston and Pattie, 1991; Evans, 1994) such as tactical voting.
The British Election Cross-section Study interviewed a random sample of 3,615 respondents throughout Great Britain, including 882 respondents in Scotland, an over-sampling designed to make it possible to analyse the distinct political situation there. This represented a response rate of 62 per cent. The interview, conducted face-to-face using computer-assisted interviewing, lasted on average one hour; in addition 85 per cent of respondents completed a self-completion supplement of 20 minutes length. Approximately 90 per cent of the fieldwork was completed within six weeks or so of the election. The results in this paper are based on a provisional weighting of the data which, inter alia, downweights Scottish respondents so that they form the same proportion of the sample as they do of the electorate.
The incidence of tactical voting in 1997
The first indication that we examine of the incidence of tactical voting in 1997 is the pattern of the constituency results.1 Table 1 analyses the average change in party support between 1992 and 1997 according to the tactical situation in a constituency.2 It strongly suggests that anti-Conservative tactical voting occurred on a significant scale. The table distinguishes six situations:
1. Lab-Con safe seats. Seats where Labour were first and the Conservatives second in 1992, but where the Conservatives won at least a third of the vote in 1992. This is, in effect, our control group against which we compare what happened in other situations. Given both the result in these seats in 1992 and the evidence of the opinion polls that Labour was well ahead nationally, there was little reason to doubt that Labour would win locally again in 1997, and thus little reason why people should vote tactically. We exclude from this group those seats where the Conservatives won less than a third of the vote in 1992 because Conservative support systematically fell by less than the national average where they had previously polled below that threshold (Curtice and Steed, 1997).
2. Con-Lab seats. Constituencies where the Conservatives were first and Labour second in 1992. These are seats where voters who were concerned to ensure the defeat of the local Conservative would have reason to support Labour rather than the Liberal Democrats.
3. LibDem-Con seats. Constituencies where the Liberal Democrats were first and the Conservatives were second in 1992, and where the Conservatives won more than a third of the vote. Although these seats were already in Liberal Democrat hands, given that the polls put Liberal Democrat support nationally lower than it was in 1992, Labour supporters in these constituencies might feel they needed to vote Liberal Democrat in order to keep the Conservatives out.
4. Con-LibDem marginals. Seats where the Conservatives were first and the Liberal Democrats second in 1992, and where the Liberal Democrats were both more than six per cent ahead of Labour and less than 30 per cent behind the Conservatives.
5. Con-LibDem safe seats. Seats where the Conservatives were first and the Liberal Democrats second in 1992, but where the Liberal Democrats were more than 30 per cent behind the Conservatives.
6. Three-way marginals. Seats where the Conservatives were first and the Liberal Democrats second in 1992, but where Labour were less than six per cent behind the Liberal Democrats and less than 36 per cent behind the Conservatives. These were in effect 'three-way marginals' where both opposition parties might claim to be best able to defeat the Conservatives.
We therefore make three distinctions amongst those seats where the Conservatives were first and the Liberal Democrats second in 1992. Given the position of Labour and the Liberal Democrats in the polls, people in these constituencies may not have been convinced that a vote for the Liberal Democrats was an effective means of defeating the Conservatives if the Liberal Democrats were starting off a long way behind or if Labour were also in close contention.
These different tactical situations exhibited very different patterns of party performance. Consider, first, those seats where the Conservatives started off first with Labour second. Labour's vote rose on average by over three points more in these seats than in those seats where they were already first. Meanwhile the Liberal Democrat vote fell by nearly three points more. It would appear that in seats where Labour started second to the Conservatives, around three per cent of those who turned out to vote opted for Labour rather than the Liberal Democrats because they wanted to try to ensure the defeat of the local Conservative.
Table 1
Change in Support (%) By Tactical Situation
Tactical Situation Con Lab LibDem N Lab seats; Con > 33.3% -12.6 +9.6 -0.3 (107) Con-Lab seats -12.6 +13.0 -3.0 (181) LibDem seats; Con > 33.3% -10.6 +9.6 +1.6 (8) Con-LibDem; Con lead < 30% -11.8 +6.5 +1.9 (80) Con-LibDem; Con lead > 30% -13.5 +10.0 -0.8 (60) Three-way marginals -11.6 +10.9 -2.3 (18)
Source: Curtice and Steed (1997)
Meanwhile we see the very opposite pattern in those seats where the Liberal Democrats were second to the Conservatives and less than 30 per cent behind. Here the Liberal Democrat vote rose by two points against the national trend, while Labour's vote rose by three points less than where it started first. It looks as though in these seats around two to three per cent of voters opted for the Liberal Democrats rather than Labour in order to try and defeat the Conservatives. This was most evident where the Liberal Democrats started off less than 15 per cent behind the Conservatives; in these seats the Liberal Democrat vote rose almost everywhere. In contrast, where the lead was between 15 per cent and 30 per cent their performance was more patchy. And where the Liberal Democrats were even further behind or where Labour were also in close contention, voters evidently did not see much reason to make a tactical switch to the Liberal Democrats. Indeed in a number of three-way marginals Labour seem to have been the beneficiaries of tactical switching. Overall, it looks as though voters needed rather more persuasion about the value of making a tactical switch to the Liberal Democrats than simply the claim that they were in second place last time.
Because it looks at how votes changed between 1992 and 1997, this form of analysis can only hope to identify apparent evidence of new tactical voting, that is the decisions of voters to vote tactically who did not do so at the previous election. Of course there may also be voters who voted tactically in 1992, or even earlier, and who continued to do so in 1997. Even so, at first glance this analysis appears to suggest that such new tactical voting was widespread in 1997. Yet we should be careful. Even within those constituencies where conditions evidently facilitated tactical voting we have found that only two to three per cent of voters acted anew strategically. And these constituencies themselves constitute only around a third of the total number of 641 seats (see the entries in brackets in Table 1). Thus in practice these results only imply that little more than one per cent of all voters voted tactically in 1997, having not done so in 1992. In addition, we should remember that some of the switching from Liberal Democrat to Labour in Con-Lab seats in Table 1 may arise from the decisions of Labour supporters to stop casting a tactical vote for the Liberal Democrats because their preferred party had regained second place in 1992. The only significant net increase in tactical voting that is necessarily implied by Table 1 is in switching from Liberal Democrat to Labour.
Is this confirmed by our survey results? Here our estimates of tactical voting will include those who may already have voted tactically in 1992 and decided to continue doing so in 1997. In order to establish whether the data confirm our expectations from Table 1 we need to compare the results obtained by the 1997 cross-section survey with similar figures obtained in 1992. The main indicator of tactical voting in the BES series is a question which has been asked as part of the cross-section election study at each of the last four elections. It asks:
Which one of the reasons on this card comes closest to the main reason you voted for the Party you chose?
I always vote that way.
I thought it was the best party.
I really preferred another party but it had no chance of winning in this constituency.
Other (Please specify).
Voters are classified as 'tactical' if they choose the response - 'I really preferred another party but it had no chance of winning in this constituency'. They are then asked which party they 'really preferred', allowing us along with information on how they actually voted to identify the direction of their tactical switch. Using this measure of tactical voting we find that in 1997 approximately ten per cent of voters can be coded as tactical, compared with nine per cent in 1992.3 This is the highest level of reported tactical voting since the question was first asked in 1983 (Evans, 1994; Heath et al., 1991).
Even so, the increase in reported tactical voting is no more (and no less) than the small one percentage point increase we anticipated. Has the pattern of tactical voting also changed in line with our expectations? Do we find that more people switched between Labour and the Liberal Democrats? Table 2 shows the proportion of voters in both the 1997 BES cross-section study sample and the equivalent 1992 sample who said that they had voted tactically, broken down by the kind of switch that they made.4 Comparing the two parts of the table, we can see that reported switching between Labour and the Liberal Democrats was higher in 1997 than in 1992. In 1992 only 1.5 per cent of voters reported making a tactical switch from the Liberal Democrats to Labour; in 1997 no less than 2.2 per cent did so. Similarly, the proportion saying they switched from Labour to the Liberal Democrats rose from two per cent in 1992 to 2.5 per cent in 1997. So, altogether, in 1997 1.2 per cent more voters reported switching between Labour and the Liberal Democrats than did so in 1992, again close to the estimate of one per cent more we derived from the election results themselves.
The rise in Labour/Liberal Democrat switching is more than enough to account for the total reported rise in tactical voting between 1992 and 1997. Other kinds of tactical voting did not increase. Indeed anti-Labour tactical switching clearly declined. In 1997, only 1.8 per cent reported switching in one direction or the other between the Conservatives and the Liberal Democrats compared with 2.5 per cent in 1992.5 Thus our two very different sources of evidence agree that there was a small rise in anti-Conservative tactical voting in 1997 compared with 1992 while other forms of tactical voting became, if anything, less common. On their own, both kinds of evidence may have their limitations; but the correspondence between their findings suggests that these limitations are not obscuring the truth. In 1997 tactical voting both became more common and changed its character.
Table 2
Reported Tactical Voting in 1992 and 1997
(a) Preferred party and actual votes of tactical voters 1992
(percentage of voters)
Actual Vote
Con Lab LibDem Other All
Preferred
party
Con - * 0.7 * 1.3
Lab * - 2.0 * 2.3
LibDem 1.8 1.5 - * 3.4
Other * * * - 0.8
All 2.1 2.3 2.9 0.4 7.7
(b) Preferred party and actual votes of tactical voters 1997
(percentage of voters)
Actual vote
Con Lab LibDem Other All
Preferred
party
Con - * 0.5 * 1.0
Lab * - 2.5 * 2.8
LibDem 1.3 2.2 - * 3.5
Other * 0.9 * - 1.4
All 1.8 3.4 3.2 0.3 8.7
Note: * indicates less than 0.5 per cent.
The number of voters engaged in tactical voting is still small. They matter not because of their numbers but because of their impact. Curtice and Steed (1997) estimate that at least 25 and perhaps as many as 35 seats were lost by the Conservatives as a result of new tactical voting. Using a different method of estimation Norris (1997a) suggests a very similar figure of 24. The two estimates are not only close to each other, but are far higher than those that have been made for the impact of tactical voting in previous general elections (see Curtice and Steed, 1992; Crewe et al. , 1992). Without tactical voting Labour would clearly still have won a decisive majority, but the strength of the Liberal Democrat parliamentary party would have been much reduced, and the Conservatives would have been significantly stronger. (Ironically, as Table 2 shows, the Conservatives continued to be net beneficiaries of tactical voting in terms of votes; but thanks to its geographical distribution they were clearly losers in terms of seats.) The 1997 British election clearly demonstrates how under the single member plurality system, tactical switching by a very small number of strategically placed voters can have a very big impact on the outcome in seats.
Explaining the rise of tactical voting
There are four possible explanations for this important development in British electoral behaviour. They emphasise the importance of changes in the motivation of voters, party mobilization, the information environment, and party competition respectively.
Changes in Voters?
Could it be that voters have simply become more willing to vote tactically? If people vote tactically, it suggests they are concerned instrumentally about the outcome in seats, and are willing and able to engage in a rational calculus to ascertain what might be the best way of achieving the outcome they desire (Cain, 1978; Cox, 1997). Thus an increase in the incidence of tactical voting has been interpreted as evidence that people have become more rational and less influenced by partisan loyalties than they were in the days of The American Voter (Campbell et al., 1960) or Political Change in Britain (Butler and Stokes, 1974). Why should voters have become more rational? The most commonly proffered explanations have been rising levels of education and the expansion of mass communications, which together have supposedly lead to the growth of an informed, participant citizenry (Dalton, 1996; Inglehart, 1997). Such citizens are thought to be more willing and better able to make the calculations needed to act strategically.
If this process accounted for changes since 1992, however, then we should expect to find a general increase in tactical voting of all kinds - between Liberal Democrat and Conservative as well as between Liberal Democrat and Labour. But as we have seen this was not what happened. Only a very specific form of tactical voting was more prevalent in 1997, that is switching between Labour and the Liberal Democrats. It looks, therefore, as though we need to look at the context in which voters made their decision in 1997 rather than at the motivations that they brought to the ballot box.
Changes in the Information Environment?
Alternatively, what may have altered in 1997 is the information available to voters about the strategic situation in their constituency. One potentially important piece of information provided by the media consists of the results of opinion polls. The 1980s saw national opinion polling reach saturation levels, while polls conducted in individual constituencies also became common. However, thanks to doubts about the accuracy of opinion polls, especially after the debacle of the 1992 election (Market Research Society, 1994), fewer polls were commissioned in 1997 than previously. Despite the campaign being a record six weeks in length, only 43 national polls were published compared with 57 in 1992 (Crewe, 1997). Moreover these polls were given less prominence when they were reported. Meanwhile, the publication of local opinion polls had already peaked in 1987 when at least 78 single constituency polls in 52 seats were commissioned. In 1997 in contrast only about 29 single constituency polls were conducted in 26 seats. There is, then, little reason to believe that voters were better informed by the opinion polls in 1997 than previously.6
Increased Mobilization?
But we should remember that political parties can also try to inform voters of the local tactical situation in their campaigning. And recent academic research has suggested that constituency campaigning may have a greater influence on electoral outcomes than had hitherto been appreciated (Denver and Hands, 1997; Pattie et al., 1994, 1995). There is evidence that in the 1997 campaign both Labour and the Liberal Democrats prioritized, professionalized and centralized their strategic attempts to gain swing voters in marginal seats to a greater extent than ever before. For two years before polling day a Labour task force was designed to switch 5,000 voters in each of 90 target marginal seats. Those identified as potential Labour converts were contacted by teams of volunteers on the doorstep and by canvassing run from twenty telephone banks around the country, co-ordinated by Millbank Tower. The Liberal Democrats also concentrated their campaigning resources on their target seats to a greater extent than previously (Norris, 1997b). One of the features of this targeted campaigning was the highlighting in campaign literature and elsewhere of claims about the likely outcome for different parties. Thus we might anticipate that certain forms of tactical voting might have become more common in 1997 because parties themselves had made greater attempts to stimulate it. If the efforts associated with Labour and Liberal Democrat targeting proved effective, we would anticipate that more people should have vote tactically in the targeted constituencies than elsewhere. However, the evidence is decidedly mixed.
Targeting does appear to have encouraged tactical voting for the Liberal Democrats. Amongst those seats where the Liberal Democrats started within 30 per cent of the Conservatives, the Liberal Democrat vote rose by four per cent in those seats which it targeted, but fell by 2.3 per cent where it did not. Labour's vote meanwhile rose by only 4.4 per cent in seats that the Liberal Democrats targeted compared with 11 per cent elsewhere.7
But in Labour's case targeting does not appear to have stimulated a higher level of tactical voting - or indeed had any discernible impact at all. Table 3 compares Labour's performance in those of its target seats where the party started off second to the Conservatives with its performance in those places where it started off second but which it did not target. The analysis is undertaken separately for London, the rest of the South East and the remainder of the country in order to take account of the generally higher swing against the Conservative government in London and the South East. As can be seen, Labour's performance in its target seats was little different from what happened in non-targeted seats. Voters apparently did not need mobilising locally to be persuaded to vote tactically for Labour.
Table 3
The Non-Impact of Labour Targeting
Change in % voting Labour
in:
Target Non-target
seats seats
London +15.9 (7) +14.6 (26)
Rest of South East +13.7 (29) +14.2 (29)
Elsewhere +11.8 54) +12.3 (54)
Note: Table confined to those seats where Conservative were first and Labour second.
Source: Curtice and Steed (1997)
Moreover, if access to information was important in explaining the incidence of tactical voting in 1997, one might anticipate that more voters would have been inclined to vote tactically at the end of the election campaign than at the beginning. After all, election campaigns are the time above all when parties attempt to impart information to voters. Yet the evidence from another part of the 1997 BES, that of a panel of voters interviewed during the course of the campaign as well as afterwards and known as the British Election Campaign Panel Study, fails to support this expectation. The proportion of the panel saying that they were likely to vote tactically when they were first interviewed during the election campaign was, at 9.7 per cent, almost exactly the same proportion who eventually did so. Equally, there is no sign either of any increase during the course of the campaign in the willingness of Liberal Democrat voters in particular to make a tactical switch to Labour.
Changes in Party Competition?
The final possible explanation is that the rise in anti-Conservative tactical voting reflects changes in the appeal of the parties, reflecting changes in their ideological positioning or their perceived competence. Previous research has suggested that people are most inclined to vote tactically if they strongly dislike one party while being relatively indifferent between the remainder (Heath et al., 1991; Niemi et al., 1992). This suggests that tactical voting might become more common either because electors become heavily disillusioned with one party and/or because they come to believe that there is little to choose between the remainder.
There is good reason to believe that this may have been the case in 1997. The outgoing Conservative government had, after all, been the most unpopular government in the history of opinion polling in Britain and it entered the election still in the electoral doldrums (Norris, 1997a). Meanwhile the two main opposition parties had moved closer together ideologically. Labour had come to endorse a wide range of constitutional changes, many of which had been long-standing Liberal Democrat policy. At the same time, under the leadership of John Smith and Tony Blair, 'New Labour' shifted towards the centre on many social and economic issues (Smith and Spear, 1992; Norris 1997c), abandoning Clause 4 and ruling out any increase in income tax rates to finance increased social spending. Indeed, if anything the latter policy put them to the right of the Liberal Democrats who, as in 1992, advocated increasing income tax in order to spend more on education. Meanwhile Labour's stance was echoed by the Liberal Democrats. Ashdown moved his party from an official position of 'equidistance' between Conservative and Labour to one which ruled out the possibility of supporting a minority Conservative government while leaving open the possibility of a deal with Labour. Moreover, just before the election Labour and the Liberal Democrats concluded a formal agreement on how best to implement the wide range of constitutional changes they now both favoured. In short, there is every reason to anticipate that many voters will have come to the conclusion that there was little to choose between the two opposition parties, while there was an awful lot to choose between either of them and the Conservatives.8
Our survey evidence certainly suggests that is precisely what happened. In both 1992 and 1997 we asked our respondents how much they were in favour of or against each of the main parties. As we can see from Table 4, the results obtained in 1997 were very different from those obtained five years earlier.
Table 4
Feelings Scales by Party Preference in 1992 and 1997
Rating of :
Con Lab LibDem
Con supporters 1992 4.30 1.86 2.94
1997 4.00 2.56 2.92
Lab supporters 1992 1.94 4.30 3.06
1997 1.90 4.40 3.38
LibDem supporters 1992 2.74 2.61 4.18
1997 2.37 3.28 4.19
All 1992 3.07 2.90 3.15
1997 2.60 3.62 3.38
Note: The table shows the average score given to the party named at the top of each column on a scale ranging from 1 (strongly against) to 5 (strongly in favour). A Conservative supporter is someone who either did not vote tactically and voted Conservative or someone who voted tactically and said the Conservatives were the party they most preferred. A similar definition applies to Labour and Liberal Democrat supporters.
The most striking change is in the views of Liberal Democrat supporters. In 1992 Liberal Democrat supporters were if anything slightly more opposed to Labour than they were the Conservatives. By 1997 they were much more favourably disposed towards Labour. Moreover, with their opinion of their own party little changed, the gap between their feelings towards the Liberal Democrats and their feelings towards Labour had narrowed significantly. The views of Labour supporters changed too. The average rating they gave to the Liberal Democrats rose by 0.32, greater than the 0.10 increase in the score they gave to their own party. In short, Labour supporters came to see less of a gap between Labour and the Liberal Democrats as well.9
Here then appears to be the most fruitful line of explanation for the rise in anti-Conservative tactical voting in 1997. As a result of the closer ideological proximity between Labour and the Liberal Democrats, together with the isolation and perceived incompetence of the Conservatives, more Liberal Democrat supporters were relatively indifferent between their own party and Labour, while disliking the Conservatives. Labour supporters were also somewhat less likely to feel there was a big difference between their party and the Liberal Democrats. As a result both sets of supporters were more willing to make a tactical switch in order to help defeat a local Conservative MP.
If this is indeed a sufficient explanation then two other things ought also to be true. First, those who actually made a tactical switch should on the whole have been those who were indifferent between Labour and the Liberal Democrats while disliking the Conservatives. This proves to be so. Among those Liberal Democrat supporters who voted tactically for Labour, there was on average an 'approval gap' of only 0.40 between the Liberal Democrats and Labour, while the Conservatives were extremely disliked (a gap of 2.76). Equally, among Labour supporters who made a tactical switch to the Liberal Democrats, the 'approval gap' between Labour and the Liberal Democrats was just 0.49, while the gap from the Conservatives was again extremely large (2.75).10
The second thing that should be true is that for any given pattern of likes and dislikes, the probability that a voter cast a tactical ballot should have been the same in 1997 as in 1992.11 In other words, amongst those who were relatively indifferent between Labour and the Liberal Democrats while disliking the Conservatives, the level of tactical voting should have been the same on the two occasions. This is largely what we find. In 1997 15 per cent of those who said they were in favour of both Labour and the Liberal Democrats (that is, gave the parties a score of four or five on our scale) and were against the Conservatives (giving them a score of one or two) voted tactically. This in fact is slightly below the equivalent figure of 19 per cent in 1992. What was different, as we would expect from Table 4, was that far more people belonged to this group. In 1992, only nine per cent of voters were in favour of Labour and the Liberal Democrats while being against the Conservatives; in 1997 no less than 22 per cent fell into that category.
So it looks as though the secret to understanding why anti-Conservative tactical voting increased in 1997 lies not in the social psychology of voters but rather in the actions of the parties. Voters did not enter the polling station with a new set of motivations that meant they were more willing or able to vote tactically. Rather the messages they had received from the parties had changed. Their reaction to that change was in fact highly consistent with the way they have behaved in the past.
Conclusion
We have provided considerable support for two propositions. First, in 1997 more people voted tactically in order to try to defeat their local Conservative candidate than did so in 1992. These people were still small in number but in terms of seats they had a significant impact. Second, this increase happened not because voters brought different motivations to the ballot box than five years previously, or because they were more informed or more effectively mobilized, but rather because they believed that the parties had changed.
There are at least two important implications that flow from these conclusions, one political, the other theoretical. First, in the forthcoming debate about the future of the single member plurality electoral system that has been promised by the new Labour government, one question we might ask of the existing system is whether it is desirable that the outcome of an election can be influenced to such an extent by the strategic manipulation of the few? Of course, we cannot assume that such patterns of tactical voting will necessarily be repeated in future; if the parties change by 2002 then so also presumably will the behaviour of voters. But the potential for such manipulation under the existing system has clearly been demonstrated.
Second, there is a tendency in the study of electoral behaviour to make inferences about the motivations of voters on the basis of evidence of changes in their behaviour. Thus, for example, voters may be deemed less willing to be loyal to a political party because they change their voting behaviour more often. Here we have demonstrated why such reasoning is potentially misleading. The act of voting is not simply the result of what the voter brings to the ballot box; rather it is the product of an interaction between parties and voters. With changing party images comes a changing set of tactical choices. We should not be surprised that 'New Labour' brought with it 'New Tactical Voting'.
ACKNOWLEDGEMENTS
We would like to thank our colleagues, Anthony Heath, Roger Jowell, Lindsay Brook, Alison Park, Bridget Taylor and Katarina Thomson for their contributions to the research reported here. The 1997 British Election Study is generously funded by the Economic and Social Research Council (ESRC) in collaboration with the Gatsby Charitable Foundation. The Centre for Research into Elections and Social Trends is core funded by the ESRC as part of its research centres programme. Much of the aggregate data analysis reported here was originally undertaken with Michael Steed for the 1997 Nuffield Election Study for which funding was provided by the Leverhulme Trust. None of the funders necessarily concur with any of the views expressed here, which are solely the responsibility of the authors.
NOTES
1. We focus in this paper only on tactical voting in Great Britain. The BES does not extend to Northern Ireland which has a completely different party system. Note also that the pattern of tactical voting in Scotland can be expected to be somewhat different because of the strength of Scottish National Party there.
2. The 1997 election was fought on new constituency boundaries. Our analysis relies on the estimates made by Rallings and Thrasher (1995) of what would have happened if the 1992 election had been fought on the 1997 boundaries.
3. These figures include respondents who did not choose this answer but volunteered other, apparently tactical, motivations for their vote. Such responses occurred in 0.3 per cent of cases in 1992 and 0.4 per cent in 1997. These respondents were not asked for their preferred party and are not included in the analyses which follow. People who reported voting against a particular candidate (0.7 per cent in 1992 and 1.2 per cent in 1997) are not here counted as tactical.
4. Excluded from the table are respondents who either refused to say who they had voted for or were unable to give the name of their preferred party, together with those who reported having voted for their preferred party despite giving a tactical answer to the 'reasons for voting' question. This explains why the sum of all the entries in the table for 1997 is 8.7 per cent rather than the ten per cent who said they voted tactically, and why the equivalent figure for 1992 is 7.7 per cent rather than nine per cent.
5. These observations - and others that follow - are also confirmed by an equivalent analysis of the 1997 British Election Campaign Panel Study which administered after polling day the same questions about tactical voting to a sample of 2,047 respondents who previously been interviewed on up to three occasions in the twelve months before polling day.
6. True, the well publicised Observer/Scotland on Sundaypoll in 16 constituencies undertaken the weekend before polling day may have had a significant impact in a handful of constituencies, most notably in Hastings and Rye and in St. Albans where, in contrast to the 1992 result, it suggested Labour was better placed than the Liberal Democrats to defeat the Conservatives. But there were simply too few polls in too few constituencies for this to provide any general explanation of the rise in tactical voting.
7. The Conservative performance in contrast was almost identical in seats the Liberal Democrats did target and in those they did not.
8. Compared with 1992, fewer Labour supporters were living in constituencies where their party was starting off third, while in contrast more Liberal Democrats were in that situation. But while this might help account for some of the increase in tactical switching from Liberal Democrat to Labour, it should have been accompanied by a reduction in tactical switching from Labour to the Liberal Democrats. As we have seen that did not happen. Moreover, the election results themselves suggest that Liberal Democrats living in constituencies where their preferred party started off third in 1992 as well as in 1997 were still more willing to vote tactically this time around.
9. Note also that Conservative supporters' dislike of Labour fell markedly between 1992 and 1997, giving them less reason to switch tactically to the Liberal Democrats in order to keep Labour out, a pattern which as we saw earlier was reflected in their behaviour.
10. Similarly, among Liberal Democrat supporters who voted tactically for the Conservatives, there was an 'approval gap' of only 0.33 between the Liberal Democrats and the Conservatives, while Labour were distinctly less popular (a gap of 1.34).
11. This assumes that the proportion of voters with any given pattern of likes and dislikes living in constituencies where it might make sense to vote tactically was not significantly different at the two elections.
References
Butler, D. and D. Stokes (1974), Political Change in Britain , London: Macmillan.
Cain, B. (1978) 'Strategic voting in Britain',American Journal of Political Science 22: 39-55.
Campbell, A., P. E. Converse, S. E. Miller, and D. E. Stokes (1960) The American Voter, New York: Wiley.
Cox, G. W. (1997), Making Votes Count, Cambridge: Cambridge University Press.
Crewe, I. (1997) 'The Opinion Polls: Confidence Restored?', Parliamentary Affairs 50: 569-585.
Crewe, I., Norris, P. and Waller, R. (1992) 'The 1992 general election' in P. Norris et al. (eds) British Elections and Parties Yearbook 1992, pp.xv-xxxvi. London: Harvester Wheatsheaf.
Curtice, J. and M. Steed (1992) 'Appendix 2: The Results Analysed', in D. Butler and D. Kavanagh, The British General Election of 1992, pp.322-62. London: Macmillan.
Curtice, J. and M. Steed (1997) 'Appendix 2: An Analysis of the Results', in D. Butler and D. Kavanagh, The British General Election of 1997, pp.295-325. London: Macmillan.
Dalton, R. (1996), Citizen Politics in Advanced Industrialized Democracies, Chatham, NJ: Chatham House.
Denver, D. and G. Hands (1997), Modern Constituency Electioneering, London: Frank Cass.
Evans G. (1994) 'Tactical Voting and Labour's Prospects' in A. F. Heath, R. Jowell and J. Curtice with B. Taylor (eds) Labour's Last Chance?, pp.65-84. Aldershot: Dartmouth.
Evans, G. and A. Heath (1993) 'A tactical error in the analysis of tactical voting', British Journal of Political Science 23: 131-7.
Galbraith, J. W. and N. C. Rae (1989) 'A test of the importance of tactical voting: Great Britain, 1987', British Journal of Political Science 19: 126-36.
Heath, A., R. Jowell, J. Curtice, G. Evans, J. Field and S. Witherspoon (1991), Understanding Political Change: The British Voter 1964-1987 , Oxford: Pergamon.
Inglehart, R. (1997), Modernization and Post-Modernization, Princeton, NJ: Princeton University Press.
Johnston, R. J. and C. Pattie (1991) 'Tactical voting in Great Britain in 1983 and 1987: An alternative approach', British Journal of Political Science 21: 95-128.
Market Research Society (1994), The Opinion Polls and the 1992 Election, London: Market Research Society.
Niemi, R., G. Whitten and M. Franklin (1992) 'Constituency characteristics, individual characteristics and tactical voting in the 1987 British General Election', British Journal of Political Science 22: 229-40.
Norris, P. (1997a) 'Anatomy of a Labour landslide', Parliamentary Affairs 50: 509-32.
Norris, P. (1997b) 'The Battle for the Campaign Agenda' , in A. King, New Labour Triumphs: Britain at the Polls, pp.113-44. Chatham, NJ: Chatham House.
Norris, P. (1997c), Electoral Change Since 1945, Oxford: Blackwell.
Pattie, C., P. Whiteley, R. Johnston and P. Seyd (1994) 'Measuring Local Campaign Effects: Labour Party Constituency Campaigning at the 1987 General Election', Political Studies52: 469-79.
Pattie, C., R. Johnston and E. Fieldhouse (1995) 'Winning the Local Vote: The Effectiveness of Constituency Campaign Spending in Great Britain, 1983-92', American Political Science Review 89: 969-83.
Rallings, C. and M. Thrasher (1995), Media Guide to the New Parliamentary Constituencies, Plymouth: Local Government Chronicle Elections Centre.
Smith, M. J. and J. Spear (eds) (1992), The Changing Labour Party, London: Routledge.
Return to CREST Page