Psephology, a word both charming and antiquated, is the study of elections. Ivor Crewe, also charming but not so antiquated, is a studier of elections. The current president of Britain’s Academy of Social sciences and the master of Oxford’s University College, Crewe has long been a respected voice on politics in the UK, US and elsewhere, as evidenced by the acclaim his recent book with Anthony King, The Blunders of Our Governments, has received.
Here, in conversation with our Nigel Warburton, Crewe marshals that scholarship to divine some salient facts about predicting elections — an exposition that comes post-Scotland’s IndyRef and pre-US midterms. He argues that while current polling attempts to pick a winner, current polling studies is looking for the reason for the result. “The main reason,” he explains, “for studying voting patterns – voting behaviour – is to provide a much more accurate account of why elections turned out in the way that they did: why did one party win rather than another?”
Crewe formerly served as vice chancellor of the University of Essex from 1995 to 2007; the Crewe Lecture Hall at Essex is named for him and he was the founding director of its Institute of Social and Economic Research. He also edited or co-edited the British Journal of Political Science for more than 15 years.
To directly download this podcast, right click HERE and “Save Link As.”
Click HERE to download a PDF transcript of this conversation. The full text also appears below.
Social Science Bites is made in association with SAGE. For a complete listing of past Social Science Bites podcasts, click here.
David Edmonds: Before an election, some people make it their business to try and predict the results: who will vote for whom and why? After an election the questions are similar, but in the past tense: who voted for whom and why? Ivor Crewe has been involved in electoral studies for over four decades.
Nigel Warburton: Ivor Crewe, welcome to Social Science Bites.
Ivor Crewe: Thank you. I’m very pleased to be talking to you.
NW: The topic we’re going to focus on is psephology. Could you just begin by sketching what psephology is?
IC: Well, ‘psephology’ is a rather old-fashioned term now. It’s a bit of a fun term really for the study of elections. The origin of the word is ‘psephos’, which is the Greek for pebble. It was pebbles that were thrown into an urn in Ancient Greece – in Athens – when elections took place. I think, now, the word is used to describe a particular approach to study of elections based very much on constituency election results and the attempt to infer patterns of voting behaviour from the results that occur in particular constituencies particularly at general elections. Nowadays, serious social scientists don’t use the term and they would refer to Electoral Studies.
NW: There’s something about the word ‘psephology’, as well, that sounds a bit like soothsaying or clairvoyance – the sense in which you’re looking into the future. I know it’s based on empirical evidence, but there does seem a bit of a gamble involved.
IC: There are many in the media and amongst students, and amongst the many people who are just interested in elections, who think that the main purpose of studying voting patterns is to predict an election result. Actually, that’s not the main purpose of voting studies, and it’s not something that the analysis of voting is necessarily that good at doing. The main reason for studying voting patterns – voting behaviour – is to provide a much more accurate account of why elections turned out in the way that they did: why did one party win rather than another? Why did they win by a large majority rather than a small majority? And why did people with different backgrounds and different attitudes on the main issues of the day vote in the particular way that they did? Why did some people turn out to vote and others stay at home? That’s what electoral studies is good at explaining, but it does so retrospectively better than it does so prospectively.
NW: And how do you actually get the information that’s going to allow you to assess why somebody voted a particular way on a day?
IC: High quality advanced electoral studies, nowadays, are based on large national surveys. The best of them are based on what are called panel surveys: people interviewed more than once at intervals, as if they were members of a panel to which the interviewer came back periodically: perhaps a year before the election and then on the eve of the election, and then after the election – three or four times perhaps. And that is the most reliable way of tracing change in voters’ attitudes, in linking those changes in voters’ attitudes to political events – to the political environment – and then drawing conclusions about what made them vote the way they did.
NW: There’s one missing element, which is private, that’s particular to elections, in that we’re not supposed to be open about how we vote, or we don’t have to be open about how we voted. It’s a private affair whether we did what we said we did.
IC: That’s entirely true. Nobody’s obliged to say how they voted. We’ve had a Secret Ballot Act in operation since 1872. The large majority of people in Britain certainly are quite open about their political affiliations because they feel they’ve got nothing to be frightened of. Surveys always have to cope, however, with the fact that there will be a small proportion of prospective respondents to a survey who will not want to participate for a number of reasons. Or, who even if they do participate, will – again, for a number of reasons – not want to say how they intend to vote or how they actually did vote after the election has taken place.
NW: If I’ve understood you correctly, most of what you’ve described is quantitative data: you’re looking at large numbers of people and they’re answering a closed set of questions.
IC: Election surveys normally ask some so-called ‘open-ended’ questions. In open-ended questions the respondents can answer questions in their own words: they’re not forced to just tick one box from a small choice. That material is, in one sense, very valuable, because it gives one a much better insight into the way in which voters think about politics – how they conceptualise politics, a much better of appreciation of their depth of feeling about a political issue, for example.
But there are problems, and one is that it’s quite difficult to analyse that particular material precisely because you can’t put it into boxes, and you can’t very easily assign a weight or a number to the answers. But if I can just give you an example of how valuable such material can be: way back in the 1960s, the American National Election Study, which was a harbinger for the national election studies that are conducted here in Britain, asked a very simple question: “What do you like about – in the American case it is – the Democratic Party/the Republican Party?” “What do you dislike about the Democratic and Republican party?” We asked similar questions about the Conservatives, Labour, Liberal Democrat parties in the UK. And what those answers revealed was that the way in which most people think about parties is very different from the way in which the small minority of people who are really interested in politics – what you might call the political anoraks/the political class – think about parties. Very few ordinary respondents – people who don’t take a huge interest in politics – think of parties in terms of those parties’ policy positions. They think of parties in terms of whether the country’s going through good times or bad times; they, to some extent, think of parties in terms of the personalities of the leaders; or they may think of parties in terms of the social group that they think is protected or promoted by those parties. It’s much rarer for them, though, to make reference to anything that’s in the party manifesto.
NW: Everything we’ve talked about so far relies on the assumption that you can take a small part of the electorate – a representative sample – and somehow extract from that to the wider electorate. How reliable is that as a method?
IC: Sampling, as a method, is reliable so long as the sampling is carried out effectively. A sample does not have to be particularly large, and the size of the sample is not the critical issue, so long as the sample itself is truly representative – so long as it is a true cross-section of the population for which it stands. The problem with sampling is that it’s very difficult to guarantee that the sample is perfectly representative, mainly because there will always be some people who are on the list to be interviewed who – for whatever reason – are not interviewed. They may refuse, they may not be located, they may be out of town. Where electoral studies have a difficulty, and where all social science based on surveys have a difficulty, is in estimating the degree to which the sample is unrepresentative and then compensating for that by applying statistical weights to the responses they get from those who do answer the survey.
NW: Now, unlike much social science, it’s very obvious what the practical use of some of this research is – particularly in the political sphere. But that potentially brings its own problems because there’s a question of where the funding comes from and what the motivation of the funders is.
IC: Electoral surveys are funded either by the Economic and Social Research Council, which has found the money for a national election survey at every general election going back to 1964 – 50 years, and I think it’s commendable that they have done that; I think that this was a visionary and longsighted thing to do. The other source of funding for surveys are typically newspapers, or sometimes private organisations. Newspapers will commission polls because it will always provide them with a story, and sometimes it will provide them with a front-page story.
NW: But the interesting thing is that sometimes the poll results actually affect the next poll result.
IC: I know it’s said that opinion polls themselves can affect the way people vote. I think that that happens pretty rarely, and when it does happen it happens only marginally. The one impact that I think opinion polls have had – more than rarely – on voting is their impact on turnout. I think it is the case that if people believe that the vote is going to be very close they are more likely to turn out, and their perception that the vote is going to be pretty close – of course – comes from the polls. Similarly, if the polls are suggesting it’s going to be a walkover for one of the parties, I think that does depress turnout to some degree.
NW: Well, that’s one myth in this area. I wonder if there are any other popular myths about polls that need dispelling.
IC: I mean there are a number of myths about opinion polls, aren’t there. One is that it’s possible to manipulate the results by asking questions in a particular way. That’s not entirely a myth; occasionally there are some very badly designed polls which do ask a question in a tendentious or slanted way and then produce the response that the sponsor wants. But, you know, there are an awful lot of experts on opinion polls around the country, and organisations that try and do something like that very soon get found out.
There is a myth that people deliberately lie to the polls and, as a result of deliberately lying to the polls, the polls are inaccurate. I don’t think there are more than a tiny number of people who deliberately lie to the polls. I do think that there are always a small number of people who will say “don’t know” when they do know how they’re going to vote. And they say “don’t know” possibly because they don’t want to admit yet to themselves that they’re going to vote in the way that they do. And I think occasionally they say “don’t know” because they support a minority party or a minority position: they have unfashionable views and they’ve kept fairly quiet about it in their family, or amongst their friends at work, and so they go on being fairly quiet about it if they are approached by a pollster.
NW: What about the weather? Some people think the weather affects how elections turnout.
IC: I don’t think the weather has any impact whatsoever. It’s assumed that the country’s a kind of nation of wimps, and, you know, if it’s a little bit wet then people don’t come out to vote. And it’s also assumed that if it’s sunny people feel better about life and are more optimistic, which may be true, but quite how that has an impact on the party they support, I’m not sure. I can give you examples of elections which had very high turnout in the most dreadful you could imagine: for example, in February 1974, turnout was one of the highest turnouts recorded since World War ll. and the weather was truly dreadful.
NW: Your work in this area is very much in the area of social science. What is it that makes it social science and not just pure science?
IC: The observations that social scientists make are observations about human beings. Human beings are not atoms or enzymes or particles. Human beings have wills of their own that act in unpredictable ways. Indeed, the very business of studying human beings itself affects the way that they behave. And so, there’s always an element of unpredictability that cannot be captured by social science research, however cleverly it is designed, and however robustly it is undertaken. It is true that, in some areas, social scientists can measure the probability of an unpredictable event occurring. It’s a little bit like the weather, although the weather has got more of a scientific basis to it. Weather forecasts deal in probabilities not in certainties; social scientists deal in probabilities not in certainties, although it is more difficult for social scientists to attach a precise probability – if that doesn’t sound a contradiction in terms – to a particular event happening. What I think the social sciences are better at doing is busting myths about why particular events occurred, or the conditions in which people live. For example – going beyond psephology now, I’m just talking about social sciences in general – surveys, particularly panel surveys, have done a great deal to improve our understanding and therefore undermine all kinds of prejudices about very important social phenomena.
Let me just give you one example: for a very very long time, it’s been assumed that in Britain, there is an underclass of people who are poor, who have a low education, who are disadvantaged in all kinds of ways. And the assumption is that there is a permanent class of such people and very often their children and their children’s children belong to the same class. We know that that’s simply not the case, and we know that from good social science. What good social science tells us is that at any one point in time, there will be a certain – what you might call a stock – of poverty in a society. But they also show that if you look at change over time, and you’re looking at the same people changing over time, there’s a considerable flow in and out of poverty. That’s where I think social science really scores, and where it helps to improve policy-making by the government and other public agencies.
NW: I wondered if any research in your own or other people’s in the area of electoral studies has really surprised you in the results it came up with.
IC: I can think of two results that have surprised me recently, in the last two or three years. First of all, I was very struck – and I think it’s fair to say surprised – at the very large gender gap in the American presidential elections that elected Obama twice. There is now a really significant difference between women – particularly younger women and single women – and men. Women are much more likely to vote for the Democrats than men, and those women who are single – a lot of single women in the United States – by which I mean not married, were even more likely to be voting Democrat rather than Republican. I think that reflects something about the culture wars in the United States on abortion, on women’s rights and so forth.
And then, in the United Kingdom, I think the bit of public opinion that has most surprised me in the last two or three years has been the substantial volume of support for gay marriage. I had always assumed that gay marriage was one of a number of liberal causes which found support amongst the well-educated, and perhaps the young, rather than the old – but still only a minority support. And I really have been surprised that ever since questions have been asked about gay marriage, there has been a consistent and really substantial majority in favour of it, and that is a reflection of cultural changes that have clearly been occurring – had been accelerating – and I think had been underestimated by – what you might call – more traditionalist elements in British society.
NW: Ivor Crewe, thank you very much.
IC: Well, thank you very much.