Philip Cowley and Robert Ford
Elections are important. But that isn’t the only reason some people spend their lives studying them.
Elections are also, at least some of the time, interesting, fun, shocking and exciting. Yet that’s not why you should study them either.
Elections are always revealing. Those willing to dig into the detail of elections and voting will find human nature, in all its dizzying variety, contained within. That is why they are worth studying.
This is a book about elections not as a means of choosing governments but as a means to learn about the human condition, and about what makes us tick. ‘If we would learn what the human race really is at bottom’, claimed Mark Twain, ‘we need only observe it at election time’ – and the only thing that has changed since Twain wrote that in 1885 is that the more we learn about elections, the more we realise how accurate he was.
Like all things involving human beings, the reality of elections is often a long way from the myth – and it isn’t always uplifting. It can, for example, be a bit depressing to learn that candidates with surnames that begin with a letter that comes early on in the alphabet have an advantage over those lower down. Or that better-looking candidates will out-perform the ugly ones. Or that both Leave and Remain voters think Postman Pat is one of their own. All this is true, and is discussed in what follows.
But at the same time it is uplifting to discover that, for all the criticism voters can get (and give), they haven’t completely lost faith in politics despite all its flaws; they respond fairly coherently to what governments do, even if they don’t know the details; most of them will get off the couch and vote if only someone asks them nicely; and they really do try their best, most of the time, to make sense of a difficult and confusing political world and vote in accordance with what they value most.
The genesis of this book lies in two earlier volumes, Sex, Lies and the Ballot Box and its imaginatively named follow-up, More Sex, Lies and the Ballot Box. Both books were well received and sold well enough. (By this we mean they sold well for books on how people vote, not that J. K. Rowling became unduly concerned for her sales by the appearance of a rival.) This volume contains a combination of chapters from the two earlier books – revised, where appropriate – along with many entirely new chapters, largely, but not exclusively, dealing with Brexit. Indeed, so much has changed since the first book came out in 2014 that even some of the revised chapters have been so thoroughly altered that they are essentially new. Whatever one’s views on Brexit, both the referendum and events since have been full of what educators call ‘teachable moments’, things that perfectly illustrate how people, on all sides, are not quite as rational or all-knowing as they like to think they are.
The chapters are written by members of the Political Studies Association’s specialist group on Elections, Public Opinion and Parties, known as EPOP, which has been running for over twenty years and is one of the PSA’s most active groups. The book is written by people who love elections so much that the electoral cycle is part of the natural rhythm of their lives. We know that many of our readers will be similarly enthused, but there is also another goal, and another audience, for the book: to try to reach people who had perhaps not realised how interesting elections and voting can be.
The study of voting – psephology, as it is known – draws on a range of materials, from electoral results to focus groups with voters to opinion polling. The chapters that follow make use of all of these types of sources. Britain, in particular, is lucky in the breadth and depth of data available to researchers seeking to understand the public mood.
Yet recent years have provided plenty of fodder for polling sceptics. Neither of us can say anything about polls on social media these days without some smart alec popping up to announce that ‘you can’t trust polls any more’, as a result of Trump or Brexit or Corbyn (or whatever is their obsession). The first problem with this argument is that it often isn’t true. American pollsters, for example, correctly predicted a popular vote win for Hillary Clinton (it’s not their fault the voters lived in the wrong places) and, as one of the chapters in this book makes clear, polls in recent elections across the globe have performed as well as they ever have, if not better.
Polls are sometimes wrong, but they are always uncertain, and the bigger problem with polling is that this uncertainty gets lost in the stories people build from them. The idea that Trump couldn’t win or that Brexit was impossible was a narrative crafted by pundits and campaigners – the polls were never so certain. We know polls will go wrong sometimes, but we also know that people are very prone to leap to strong conclusions based on limited evidence. Some argue that the errors in the polls are so frequent, or our ability to judge them properly so limited, that they do more harm than good to our election campaigns, or even to our broader political conversation. Perhaps, such critics say, it is time to do away with the distraction of unreliable polls altogether. We think the opposite; with so much up in the air, it is more important than ever to keep ourselves grounded with the best evidence about the public mood that we can gather. Polls and analysts get things wrong sometimes. But get rid of them and the vacuum will only be filled with bad punditry and biased speculation, which will get more things more wrong, more often. Ignorance, as Barack Obama once said, is never a virtue; it’s just not knowing stuff.
That said, no one who does it seriously thinks that measuring public opinion is without difficulties. As many of the chapters in this book demonstrate, the public can give different responses to almost identical questions; they have inconsistent attitudes; voters can support or oppose policies that do not exist; voters often think they voted when they didn’t (and think they didn’t when they did). In Yes, Minister, there is an exchange between the Permanent Secretary Sir Humphrey Appleby and the more junior Bernard Woolley after Bernard presents some unwelcome opinion poll data and is told to go away and do another poll producing the opposite response. Bernard protests that the public can’t be both for and against something. Sir Humphrey’s response: ‘Of course they can, Bernard.’ And Sir Humphrey was right. But polling is still better than the alternative – which is to make it up or to assume that you or your friends have some profound insight into the general population (spoiler alert: you don’t). The answer to poor polling is not no polling but better polling, along with a better understanding of what the polls can (and can’t) tell us.
Here are fourteen rules to help understand polls and polling:
- Be sceptical about genuine polls but ignore the voodoo ones entirely. The first question to ask is simple: is this a genuine poll? Is it carried out by a reputable company, which has attempted to ensure that the people taking part are broadly representative of the public? Even good companies get things wrong occasionally, so some scepticism is always justified, and there are legitimate debates over the best way to tap into the public mood. But it’s fairly easy to spot the ones that aren’t pukka. Is it a self-selecting poll on a newspaper website or a phone-in poll on TV? Is it someone using Twitter to push for a ‘big sample’? The views of 10,000 activists on Twitter are no more informative or representative than the views of ten of your mates around a pub table (that is: not informative at all). If it’s any of these so-called voodoo polls, don’t waste your time being sceptical. Just ignore it altogether.
- If a poll looks especially unusual or interesting, it is probably wrong. This is known as ‘Twyman’s Law’ after Tony Twyman, a media research analyst. Eye-catching polling results are often the product of some blunder by the pollster, such as a poorly designed question, or due to the vagaries of sampling. Plus, chance alone will throw up the occasional outlier with eye-catching results from time to time. This wouldn’t be much of a problem, except…
- If a poll looks unusual or interesting, it will get more headlines. ‘Poll shows same as the poll we did last week’ isn’t a headline. ‘Poll shows dramatic and unexpected shift’ is a headline. As a result, what people read about polling tends to be skewed towards the errors and outliers, and towards change over stability (even when the change is within the margin of error). The blame for this doesn’t rest solely with journalists or editors; anyone following polls on social media can see it is the outliers that rack up the likes and retweets. We are all suckers for surprises.
- Polls aren’t as precise as people sometimes pretend. Given all this, it is worth remembering that even an absolutely kosher poll has the potential for error of around +/- 3 percentage points on any single finding. (On a combination of findings, such as the lead of one party over another, the potential random error is even larger.) It is therefore always better to focus on trends over time – is the lead getting larger or smaller? – than obsessing over any one poll finding. And be especially cautious about analysing parts of any poll – what are called the cross-breaks – such as the figures just for men, or those in London, or working-class people. Here, the numbers involved will be smaller and the random error larger still.
- The public isn’t stupid, they’re just not all that interested. Most people don’t pay attention to politics most of the time. This is hard for those whose job it is to follow politics 24/7 to understand or take into account, and it can lead to problems. Sometimes people are asked questions in polls which assume a level of knowledge about politicians or policies that most people simply don’t have. Even then, most voters will gamely try their best using whatever hints are provided in the question to guide them on unfamiliar terrain. Ambitious campaigners can exploit this tendency to pump up support for an obscure politician or proposal, by linking them to something better known or popular. But don’t underestimate the public – they can pick up the drift of developments even if they seem to know nothing about the detail. As a result…
- How you ask the question matters. Everyone involved in polling knows that question wording matters – and so reputable companies take real pains over it. There is a good reason for this: most people know what they think in broad-brush terms but know little about detailed or obscure issues, so they will look for hints in the question to help them make sense of it. As a result, even small changes in how a question is asked can produce a big shift in responses, if the change alters what voters think the question means (as discussed in Chapter 1). This effect also highlights that what the question writers think their question is about is often very different from what a typical voter thinks it is about.
- People see the world through partisan lenses. Most of us are not neutral. We have a team we cheer for and a team we oppose. When new information appears, these allegiances skew how we understand it. Many of the chapters below show how partisan filters affect the way people respond to polling questions (even when they apply to fictional characters, marital infidelity or cats). You can also watch this process unfold in real time on social media whenever a poll is published – partisans will sing the merits of strong polls for their party, which they are eager to share, while criticising or ignoring less-flattering polling. Reverse the poll results, and the critics and cheerleaders switch sides.
- People don’t always tell the truth. People routinely claim to be much better citizens than they are – the numbers claiming they watch political coverage regularly, or who will turn out to vote, are far above the true figures revealed by TV ratings and election tallies. People are also so unwilling to admit their ignorance that most will happily attempt to answer meaningless or illogical questions. These fibs are still revealing, though – because they show how people want to be thought of as good citizens and don’t like admitting ignorance.
- People aren’t good at predicting (or remembering) what influences them. If you ask people, ‘Would x influence your vote?’, they often say ‘Yes’. But if people were that easily influenced, elections would be much more volatile and less predictable than they are. People simply aren’t very good at understanding their own decision processes and systematically overestimate the importance of whatever notion a pollster puts in front of them. For exactly this reason, ‘Would x influence your vote?’ questions are very popular with campaign groups, who will tout the answers as evidence of the burning importance of their pet issue.
- Individuals are unique, but crowds are more predictable. Everyone’s heard about a relative who lived to a ripe old age despite cheerfully inhaling sixty a day. Few people think that means smoking isn’t harmful for your health. The elderly relative is just an outlier. People understand this at one level: they know that the same forces impact on everyone exposed to them, moving them all together. But everyone also thinks they are the exception to the rule: ‘Moving in crowds is what other people do; I take my own path.’ Both views are right. Studying public opinion is often like watching pebbles on a beach – every pebble looks unique, starts in a unique place, and moves in a unique way. But all the pebbles are pushed by the same wave, and they tend to move in the same general direction.
- Crowds may be more predictable, but they’re not uniform. At any election or referendum, it is common to hear commentators declare that that ‘the people’ think something or other. This is always nonsense, because ‘the people’ never think just one thing. The only certainty in life is that, even if many, or most, people think x, there are still plenty of people who think y, along with some who think both x and y, some who don’t know, some who won’t say, some who think they think x but really think y, some who say they think y but go and vote for x anyway, and so on.
- Accusations of pollster bias usually say more about the accuser than the pollster. Political polling is a high-stakes public relations exercise for pollsters; it is their opportunity to demonstrate how good they are at gauging public opinion. They spend a lot of time and effort trying to be as accurate and unbiased as possible. Those who attack pollsters as biased are usually the most biased of all – partisans who do not have any incentive to be accurate and every incentive to attack the bearers of bad news about their team.
- Public opinion is part of politics, but it is not the same as politics. A well-conducted poll can tell you what people are thinking; it does not tell politicians what to do. Public opinion can change, not least because politicians can try to change it. When Bob Hawke, then Prime Minister of Australia, was told that he needed to be careful how he handled a racially sensitive issue, because some of his supporters held racist attitudes, he replied: ‘Then tell me what I need to say to turn them around.’ Too much analysis of polling presents it as the end-product of politics rather than just part of the process.
- Most people think everything in politics has an obvious explanation. They are wrong, as the rest of this book will demonstrate.
Like its predecessors, this book isn’t meant to be an introductory textbook. This volume offers an eclectic series of sketches, each introducing an aspect of elections and political behaviour. Each of the chapters offers a 1,000-word essay. These are not monographs, and most of them summarise years, in some cases decades, of research. Each chapter ends with a short account of further reading, and there is a detailed bibliography in case any of the subject matter stirs you to dig deeper. We make no claim for comprehensiveness, but between them the following fifty chapters incorporate: polling, ignorance, political geography, targeting, gender, sex, race, racism, Scotland, partisanship, Wales, young people, trust, turnout, apathy, alienation, death, volatility, religion, issue ownership and salience, Northern Ireland, manifestos, candidates, class and Brexit. And then more sex.
As with the two earlier books, our colleagues were enthusiastic about the book and a pleasure to work with. Editing a book like this was not always straightforward, but they have made it easier by always (well, almost always) responding to our often detailed and insistent editorial requests swiftly and in good spirit. We are also grateful to all the staff at Biteback for their support. We think the end result is worth it. We hope you do too.