Where Have All The Voters Gone?

This Paper was given at the Political Marketing Conference, Bath Spa University College, 9 September 2004

Introduction

The decline of turnout in British elections in the last few years has created something approaching a panic in the political establishment and has left the Electoral Commission and other interested bodies with an acute problem in political marketing. What has resulted is both the commissioning of research to explore the reasons for non-voting and a rash of suggested solutions. This paper discusses newly-published evidence about the way the public views voting and how this is related to other attitudinal and behavioural characteristics commonly treated collectively as components of "activism" or "good citizenship". It also considers the implications of these and other findings for some of the suggested solutions to the turnout problem.

The analysis is primarily based on a survey conducted by MORI for the Electoral Commission and the Hansard Society in December 2003. Initial findings from that survey were published as An Audit of Political Engagement published by The Electoral Commission and the Hansard Society in March 2004 (Note 1); MORI then conducted further analysis of the data using multi-variate statistical techniques, which was reported to the Electoral Commission as Rules of Engagement? in August 2004 (Note 2).

The immediate reaction of many commentators after the record-low 59% turnout at the 2001 general election was to blame "apathy" among the electorate. But most of the available evidence was against this explanation -- for example, the proportion of the electorate saying they were "interested in politics" during the 2001 campaign was almost identical to the numbers who had said the same at earlier elections with much higher turnouts. On the other hand, the fact that the 59% who were interested in politics was the same proportion as the number who had voted suggested that the problem might have been in an unusual failure to engage the interest of those who were not generally "interested in politics". Research also indicated that far fewer voters than in other general elections over the previous 40 years felt that there was a significant distinction between the two major parties, and that there had been a fall in the number who said it was important to them who won. Connected with this, and presumably of importance in allowing it to have influence, was the finding that a significant section of the public did not agree with the statement that "It is my duty to vote"; this denial of voting as a civic duty was very much higher among the young than the old, suggesting the possibility that it might be a generational effect.

These were some of the considerations that were taken into account in the design of the MORI Political Engagement Poll, conducted in December 2003 for the Electoral Commission and the Hansard Society. It had several purposes. One was simply to establish benchmark measures of a number of aspects of political engagement which will enable the "democratic health" of the British system to be tracked in future years. But it was also intended that the findings should inform work by the Commission, the Society and others to facilitate engagement with, and participation in, the political process, as well as being (by publication of the findings) a means of prompting discussion by the media, political parties, the public and others of the issues involved.

Research findings

Levels of political engagement

The survey, which interviewed a representative sample of almost 2,000 adults across the United Kingdom (Note 3), made around fifty separate measurements of respondents' political engagement, which can be broadly grouped into three categories -- knowledge, attitudes and (reported) behaviour. It found a perhaps surprising degree of congruence between the "topline" percentages of those showing or claiming different characteristics of political engagement. This might seem initially to suggest that it is possible to give Britain a "democratic health check" using a single thermometer -- that political engagement can be reasonably be measured on a single scale, and that tracking the position of the median voter and of various percentiles on this scale will tell us all we need to know about how the position is improving or worsening.

All four indicators of knowledge -- real and perceived -- were consistent in finding between 42% and 45% of the public to be well informed. Two in five (42%) could correctly name their local Member of Parliament. Almost the same number (45%) could correctly answer four or more out of seven political knowledge "quiz" questions. Similarly, 42% of the public feel they know at least "a fair amount" about politics, though just 3% would say they know "a great deal", and 45% claim to know at least "a fair amount" about the role of MPs, with 4% saying they know "a great deal". Slightly more (50%) say they are fairly or very interested in politics.

These figures seem to suggest a falling-off of engagement in recent years. The number who can name their MP is lower than the number who were able to do so in the early 1990s, and the level of interest in politics is the lowest MORI has found since the question was first asked in 1973 (though since it had been measured as 58% as recently as May 2003, this might represent only a transitory change). In some ways, the "quiz" question answers were most alarming -- while the public performed respectably on those concerned with local government, barely a quarter of the public realised that the statement "There has to be a general election every four years" is untrue.

Most of the attitudinal measures on efficacy and satisfaction are at a similar or slightly lower level, four in five of the attitudinal measures finding the number with positive attitudes in a five-point band (36% to 41%): 36% agree getting involved in politics can really change the way the UK is run; 36% are satisfied with the way the Parliament works; 41% are satisfied with the way their own MP is doing his/her job; and 36% of the public feel the current system of government "works well", though only 2% see no room for improvement. The majority feel it could be improved "quite a lot" or indeed "needs a great deal of improvement". This level of satisfaction is higher than in the mid-1990s, but is considerably lower than the most recent previous measures.

Less satisfactory, though, is trust in politicians in general. Just over a quarter of the public (27%) say they would trust politicians a great deal or a fair amount; hardly anyone (1%) says they would trust them a great deal. In contrast, one in five (19%) say they would not trust politicians at all.

The participation measures found that the various benchmarks were clustered round three different levels, related to the degree of commitment or interest required. Voting and attitudes related to it produce the highest scores: two-thirds of adults (65%) put their likelihood of voting in an immediate general election at eight or more on a ten-point scale, similar to the proportion who say they feel a sense of satisfaction when they vote (63%), but substantially lower than the 74% who agree "It is my duty to vote".

Lower down the scale, two in five (38%) say they have discussed politics or political news with someone else in the last two to three years, 39% have signed a petition, and a slightly higher number, 44%, have either paid to join or donated money to some charity, campaigning organisation or political party. (However, donations to or membership of political parties make up only a fraction of this activity.)

But three more demanding indicators find participation around one in seven or one in eight: 13% have contacted an elected representative (at a local or national level) and some point over the last 2-3 years, 14% say that they have at least urged somebody outside their family to vote (but only 1% have taken an active part in a party's general or local election campaign, or been a candidate for office themselves). A higher proportion, one in six (18%), have boycotted products for political, ethical or environmental reasons.

Distinguishing the separate dimensions of political engagement

Illuminating as these initial findings were in themselves, further analysis is possible by using the techniques of multi-variate statistical analysis to link the various measures to each other and to the extensive demographic information also collected on the survey. This secondary analysis is aimed at deeper understanding of the relationship between the various attitudes and activities broadly described as political engagement or activism. In particular, the analysis considers how these relate to propensity to vote (Note 4), with a view to gaining an insight into how far efforts to increase engagement might hope to lead to increased turnout. It will also inform planning of future audit surveys, by making it clear which indicators are the most valuable to track over time.

The analysis consisted of three techniques, each of which build upon the previous findings -- factor analysis, regression ("key drivers" analysis) and cluster analysis ("segmentation").

The 49 measurements of aspects of engagement and involvement taken in the Political Engagement Poll can be reduced by factor analysis to 14 "dimensions" or underlying attitudes, each independent of each other. The analysis suggests that knowledge, behaviour and attitudes are distinct from each other; in particular, likelihood of voting is distinct from knowledge (real or perceived), and satisfaction with the performance of the system and the politicians. Nor do engagement and activism in the non-political sphere either follow from these factors or lead to voting.

The one dimension which combines attitudinal and behavioural factors, however, is that which encompasses voting -- voting in practice (or at least the report of it) and attitudes to voting in theory are related as a single dimension of opinion, rather than being distinct and independent of each other; but they are distinct from other apparently related factors such as interest in politics or activism in the non-"political" civic sphere.

Interest in politics proves to be most closely related to "claimed knowledge": the pattern of respondents' answers classifies it firmly with the five "How much do you know...?" questions. The fact that it is most closely related to perceived knowledge, not to satisfaction or voting, perhaps suggests that it is not a particularly useful measure in isolation as a benchmark of political engagement. Consequently, there might be less cause for alarm in the sharp fall in interest in politics noted above than might otherwise be assumed. (It is also worth noting that it is related to perceived rather than actual knowledge -- claimed interest in politics is not a particularly good predictor of high scores on the political "quiz".)

Relating voting to other aspects of political engagement

The next stage is to discover how the different dimensions identified are related to propensity to vote . Regression ("key drivers") analysis looks for patterns we would expect to find in the data if the variables are causally linked. In this case, we discover which other dimensions of political engagement are the best predictors of propensity to vote; while the analysis cannot prove that one causes the other or that they share a common cause, it can be used to predict how propensity to vote is likely to change if levels of political engagement change in other ways.

The factors most closely related to propensity to vote seem to be a belief that involvement in politics works ("efficacy"), which is much the most powerful influence, and a feeling of familiarity with political institutions and with politics itself ("claimed knowledge"). But only just over a third of the variation in propensity to vote can be explained in terms of other aspects of political engagement. Tackling political disengagement as distinct from disinclination to vote is not in itself, therefore, likely to be a panacea for low turnout, though it may help.

One finding with obvious implications is that "good causes activism" -- being prepared to help organise charity events, for example, or playing an active part in non-political organisations -- is not related to propensity to vote. These activities, which would normally be considered aspects of "good citizenship" and helpful in building up "social capital" do not, apparently, lead their participants to turn out at the ballot box.

Demographic factors (even newspaper readership) also contribute little to understanding likelihood of voting independent of other aspects of political engagement. Commitment to the various political parties, by contrast, is a highly significant factor -- propensity to vote is closely related to knowing which party one intends to vote for. While this is not remotely surprising, it emphasises how important to turnout is the success of the parties in appealing to the public, and voters' ability to discriminate between them.

The final step in the analysis is segmentation by cluster analysis, dividing the public up into natural groups with similar characteristics. It is particularly useful in planning marketing or political campaigns for identifying groups that would best be treated differently and can be separately targeted. The analysis identifies similarities between individuals based on all the different facets of political engagement; the clusters derived in this way turn out to have some distinctly different propensities to vote.

The most convincing clustering derived from the audit survey divided the public into nine groups, with widely differing propensities to vote.

A small group of highly-engaged Enthusiasts (6% of adults) stand out from the rest, and are probably the initiators and main driving force behind almost all political activity in Britain.

Of the remainder, those who generally vote can be divided into four groups of roughly similar size -- Foot soldiers, Contented voters, Protest voters and Bustlers. Each group has different motivations and a different view of the existing system, and each probably requires a different approach to reinforce and expand their voting turnout.

At the other end of the scale, while there is a substantial "don't know, don't see the point" group of the Utterly Disengaged who never vote, there is an equally large group of Passive onlookers who rarely vote even though they do see the point and admit they ought to turn out.

Implications and discussion

Political engagement has turned out to be a multi-dimensional phenomenon, with various facets apparently independent of each other. This suggests that the "thermometer" model is too simplistic: there is no single all-encompassing measure which can be read as a proxy for all the others and as being an indicator of Britain's civic health; nor, consequently, is there likely to be any "silver bullet" solution to tackle disengagement.

This has particular implications for the interpretation (and encouragement) of electoral turnout. Most theorists, and most politicians, take it as read that voting is a facet of good citizenship. Morally justified as they may be in holding this, it is clear that the public -- especially the young -- do not necessarily make the same connection. They may be good citizens in other ways, yet not feel that voting is an important part of that. (Many more of the young than of older citizens, for instance, do not accept that it is their duty to vote -- see table.) It is not simply a case of non-voters being on a lower rung on the citizenship ladder than we would like, they are on a different ladder altogether.

Q To what extent do you agree or disagree with the following statements? "It is my duty to vote"

Source: MORI/Electoral Commission/Hansard Society

The analysis suggests that turnout will not simply be increased by persuading the public to a greater recognition of their civic responsibilities. Politics, especially party or electoral politics, has become a world apart. The "disconnection" which researchers have adopted as a description of the public's opinions is just as evident in their reported behaviour. One key to voting behaviour is "efficacy" -- a belief in voting, in the importance of politics and of the worth of having a say in the way the country is run. These attitudes seem not to be primarily driven by demographic factors; they may be matters purely of temperament, but there must be a suspicion that they can be taught. If education in citizenship (whether formally to children or in a broader sense to the whole adult community) is to revive electoral participation, it looks as if it should specifically concentrate on these factors, and not assume that they are implied by a more general exhortation to shoulder the burden of citizenship and civic involvement.

The political parties have a clear role to play here, since it seems that the ability to choose between them is closely related to turnout; but further research may be indicated as this was not a factor which the Political Audit survey measured except by implication.

All politics is local?

One other conclusion that emerges, at least tentatively from the research, also has implications for tackling the turnout problem. Willingness to vote at local elections seems to be not a separate dimension from general election voting but merely another point on the same continuum. Those who vote in local elections are more intense holders of the same values as those who vote in general elections. If this is true -- and it must be admitted that the evidence that the Audit survey can offer on this point is comparatively weak, though it supports the hypothesis as far as it goes -- it implies that any trivialisation or setting apart of local elections is counter-productive for turnout.

Yet some of the solutions adopted or suggested involve precisely this. Local elections are used as testing grounds for experimental voting methods which may or may not work, with the intention of extending them to general elections if it proves safe to do so; the implication that the risk of damaging local elections is unimportant seems inescapable.

Other campaigners apparently ignore the existence of local elections altogether: the Votes at 16 campaign, for example, argued "if the voting age is not reduced, young people [emerging from citizenship education at age 16] will have to wait for up to seven years before they can put their knowledge into practical use", oblivious of the possibility of their voting at local elections. It may be that more consideration needs to be given to viewing the whole voting process -- national, local, devolved and European -- as a single entity to be promoted rather than splintered into its components. But, again, further research is needed here.

Why do we worry about turnout?

In any case, if there is a distinction between "good citizenship" and voting then it is no longer possible to avoid addressing the question of what we hope to achieve by encouraging turnout in the first place.

Do we believe that the act of voting is good in itself for the voter? That is potentially a strong argument, but one that is rarely made, and which would certainly be harder to establish if there is a complete disconnect between civic and electoral virtue.

Or do we argue that high turnout is necessary to the credibility of the political system and to the legitimacy of governmental power that claims to rest on a democratic mandate? This argument seems a specious one if the mandate is no more than a Potemkin village, maintaining a faзade of public acceptance regardless of the real level of informed consent. Expedients which inflate the turnout figures without increasing the number who cast a valid vote (Note 5), or which promote voting at the expense weakening its connection with voters' engagement, fail to tackle the underlying problems.

Encouraging voting by making it easier but simultaneously trivialising it might be seen as being within the same class of solutions. Compulsory voting would achieve the same, and arguably no more. (Its opponents would argue that it transfers the choice of government from those with an informed preference to those without one, while not necessarily doing anything to promote engagement, although its supporters might counter that its longer term effect is liable to be one of increasing interest among those who would not otherwise vote.) For that matter, weakening the enforcement of electoral law so that turnout is inflated by personation and vote stealing would also improve the figures, but few would argue that it would on those grounds be a welcome development.

If we believe that it really strengthens the functioning of a democracy to maximise real turnout by motivated and discriminating voters who will feel they have a responsibility for the outcome and a stake in the wider governmental structure, solutions that address symptoms rather than causes are inadequate. It is not enough to make people vote, what is necessary is to make them want to vote. If the aim of encouraging higher turnouts is understood on this more specific basis, then it is increasing engagement on a wider front which must be aimed at, and turnout viewed only as an important indicative symptom rather than the primary indicator of health.

Voting clearly cannot be divorced from "politics". "Politics" as the public prefers to understand that term is a minority activity, and is verging on being a minority interest. Worse, "politics" is that activity pursued by that widely-reviled creature "the politician". Hostility to the concept of politics presumably acts as a deterrent to involvement. Many of the public seem to see politics as an obstruction to, rather than the means of, proper government of the country.

Yet confront the public not with the generalised "politician" but a specific individual, especially one they are familiar with, and the hostility is much less pervasive. The data seems to be an illustration of two wider principles that MORI has frequently found to apply in all walks of life: that specific opinions of individuals tend to be more favourable than generalised views of institutions, and that "familiarity breeds favourability, not contempt". Although "claimed knowledge" (familiarity) and "satisfaction" (favourability) both appear to have only a minor effect on propensity to vote in the current political climate, it seems probable that a general improvement in the standing of politics and politicians, if it could be brought about, might change those relationships. It may also in this way be possible to re-establish a connection between voting, political engagement and other aspects of "good citizenship". The "rules of engagement" are not necessarily rules that are set in stone.

1. An Audit of Political Engagement (March 2004) can be downloaded from www.electoralcommission.gov.uk
2. Rules of Engagement? (August 2004) can be downloaded from www.electoralcommission.gov.uk
3. In Great Britain, interviews were conducted face-to-face, in home, as part of MORI's regular Omnibus survey, on 11-17 December 2003 [Political Monitor -- December 2003]. In Northern Ireland, interviews were conducted face-to-face, in home on 6-15 December 2003.
4. For this to be possible the factor analysis is re-run, excluding the voting behaviour variables as inputs; this eliminates one of the 14 dimensions (the "voting" one), but the remaining 13 are almost unchanged and can be used as independent variables in the regression analysis.
5. Recalculating turnout figures to include blank and spoiled ballot papers, as was done in the official turnout figures released after the 2004 local and European elections, would seem to fall into this category.