May 2016

The Death Of Polling ? Editorial: Bobby Duffy Aalia Khan Gideon Skinner

Information www.ipsos-mori.com [email protected]

Follow www.twitter.com/IpsosMORI

01. 12. 22. Foreword by Bobby Duffy How Britain voted The digital election A comprehensive analysis of A new kind of debate? the 2015 election campaign Steve Ginnis, Head of Digital Gideon Skinner, Head of 02. Political Research The Death of polling? An international perspective 26. Gideon Skinner, Head of 18. We need to talk Political Research about Scotland Julia Clark, Senior Vice Cracks in the foundation How Scotland voted and what President, Ipsos Public Affairs The impact of UKIP, the it tells us about the future of Greens and May 2015 British politics

Dr Rob Ford, Senior Lecturer in Mark Diffley, Head of Ipsos 08. Politics Manchester University MORI Scotland Poll position An interview with Professor Samuel Wang, founder of the 20. Princeton Election Consortium On the campaign trail Bobby Duffy Ann gives us her thoughts on the election and beyond

Ann Treneman, The Times Foreword Bobby Duffy

Welcome to this international on to the BBC’s Broadcasting House. And finally, you can’t talk about edition of Ipsos MORI Social Research Lord Ashdown (a previous contributor to last year’s election without the Institute’s Understanding Society. ‘The this journal) declared that if the numbers momentous shift in Scotland. Mark Death of Polling?’ may be a deliberately were right, he would eat his hat. Of Diffley, head of our Edinburgh office, dramatic title to mark one year on from course, the exit poll was accurate - it recalls Ipsos MORI's poll following the the UK General Election, but it’s fair to was the eve-of-election polls that were Scottish Independence referendum in say it has not only changed the British out, none predicting the Conservative October 2014, which was the first to political landscape, but has also shaken majority. Gideon reflects on how we show Labour’s collapse in Scotland. the polling industry. But are we alone in ended up with that surprising final result Mark gives us his reflections on the facing these challenges? in May 2015, and how it was the “war of referendum, as well as the factors In this edition, we examine the state the weak”, a term coined by The Evening behind Nicola Sturgeon’s continued of polling across the world, using case Standard’s Political Editor Joe Murphy. election success. studies from some of the 30 countries Our polls for this newspaper showed We hope you enjoy reading about and five continents Ipsos MORI has that in the run up to the election, both this General Election special, one year polled in since 2007. Our political Labour and Conservatives had as many on. The answer we offer to the question experts, Gideon Skinner and Julia Clark weaknesses as strengths – it’s just that posed by this publication is that polling in the London and Washington offices Labour’s were more decisive. is not mortally wounded, and will take us through the modern challenges Ann Treneman, journalist and author continue to play an important role in of polling. This includes universal themes of All in this together: My five years political insights. Sam Wang says it best: like the rise of insurgent parties and as a political stalker, takes us on the “change is not death – it is life. Like technology changing the way we reach campaign trail and gives us a candid Sarah Palin said, only dead fish go with people, as well as challenges that are view of how Westminster politics played the flow! Polling is as interesting as ever.” specific to each country. out. We are also delighted to include Dr Ipsos MORI remains committed to Lessons need to be learned, but Rob Ford’s analysis - one of the most understanding society from our broad should political polling be reduced to prominent thinkers of this election. Ann range of social and political research, in simple ‘horse race’ predictions? We Treneman is sure that David Cameron the belief that this leads to better politics, are delighted to have discussed this had no idea that he was still going to be policy and practice. If you would like to with Professor Samuel Wang, eminent in Downing Street while Dr Ford views discuss any of the research here, please neuroscientist by trade who somehow the result as an “unexpected resumption get in touch. finds time to also be a ground-breaking of normal service”, and dissects the rise polling analyst. Sam founded the of the insurgent parties. Princeton Election Consortium website, Social media played a significant role featured in the New York Times and the in this election. Ipsos MORI is known for Wall Street Journal, and is adamant its political polling, but it makes up less that polling is not dead, but very much than one percent of what we do. Our Bobby Duffy alive and evolving. He argues that the range of research is much broader and Managing Director “breathless” reporting of the ‘horse race’ increasingly focuses on applying social Ipsos MORI in the American press misses the richness research methods to the digital space. Social Research Institute of polling data and the insights it can offer Our head of Ipsos MORI Digital Steve us. Sam also discusses why we need Ginnis, in collaboration with Demos, the @BobbyIpsosMORI both political forecasters and pollsters Centre for Analysis of Social Media and and how the popularity of data journalism University of Sussex, recall, among other can improve the reporting of elections. things, how John Major and Ed Miliband Returning to the UK election, it was (for a very short time, admittedly) got this time last year when our (and many more mentions on Twitter than Kim other partners’) exit poll was projected Kardashian and One Direction! 1. Ipsos MORI - Understanding Society May 2016

Julia Clark The Death of polling? Washington Gideon Skinner An international perspective. London

One year on from the UK general 5stelle was underestimated in the 2013 as an opportunity to step back from election and the British polling industry Italian elections,2 while in Israel, the the day-to-day and review the state is still dealing with the consequences. polls were predicting a dead heat, which of polling across the world. Ipsos, The British Polling Council Inquiry has Binyamin Netanyahu overturned in the having conducted polls in almost 30 released its report into the failures of final vote.3 countries over 5 continents since 2007, the polls,1 and while many pollsters Of course, as pollsters, it is in our is uniquely positioned to undertake are still introducing changes to their interest to point out the outstanding this assessment, given that our experts methodologies the upcoming referendum accuracy of other recent polls carried are the public face of polling in many on Britain’s membership of the European out in South Africa, Canada, during of these countries. For this article, Union is keeping the spotlight firmly on the US primaries, and most recently we've spoken to our experts from Italy, the latest measures of public opinion. for London’s Mayoral Election. But Sweden, Ireland, Canada, South Africa, But it’s not just in Britain where the as the largest polling company in the the Netherlands, Australia, India, the UK science of political polling has received world, it is also our responsibility to and the US – to answer the question: is some body blows in recent years. scrutinise failures and seek to remedy polling really dead? Spoiler alert: Not To take just a few examples, the surge them. So, in the spirit of ‘glass half full’, even close. towards Beppe Grillo’s MoVimento we take these recent polling failures

Figure ONE.

The performance of the polls, from the good to the could-do-better.

Italy 2013 GB 2015 Sweden 2014 Ireland 2016 Canada 2015 South Africa 2014

Poll Result Poll Result Poll Result Poll Result Poll Result Poll Result

Riv C Con Mod FF Lib ANC 2.3% 37.7% 23.3% 25% 40% 62.2% 3.3% 36% 21.7% 23% 38% 63%

Centre Left Lab Lib FG Con DA 29.5% 31.2% 5.4% 26% 32% 22.2% 34.4% 35% 6.4% 28% 31% 22%

Centre UKIP Cen Lab NDP EFF 10.6% 12.9% 6.1% 7% 20% 6.4% 11.9% 11% 5.7% 6% 22% 5%

Centre Right LibDem KD SF BQ Other 29.2% 8.1% 4.6% 14% 5% 9.3% 28.3% 8% 5.7% 15% 4% 10%

5* Other Soc Dem Other Green 25.6% 10.2% 31% 29% 4% 20% 10% 28.6% 28% 4%

Other Van 2.8% 5.7% 2.1% 7.6%

MP 6.9% 10.3%

SD 12.9% 9.4%

F! 3.1% 3.6% Average 2.4 1.5 1.6 1.4 1.2 0.8 error 2. The Death of Polling?

So what is the state of polling across in interviewing a representative sample, the world? Our data suggests it’s a and how to best account for turnout. real mix. Given the range of countries First of all, the social context is We are only we cover, it’s not surprising that the more complex, technology is changing, nature of the challenges faced varies and politics is fragmenting in many as good as our substantially. For example, in countries countries, which has implications for such as India and South Africa, there polling as much as the societies whose last poll are real practical difficulties given views we are trying to understand. their demographic profile: very large Countries struggling to recover from the populations and geographies, rural and economic crisis, where there is growing urban differences, with many different disaffection with politics, decreasing ethnicities and languages to cover. turnout, with increasingly fractured general election there. Similarly, in And there are further challenges in politics and rising insurgent parties all Italy, the Five Star Movement was getting accurate responses, such as create difficulties for polling that didn’t underrepresented (although the overall the dominance of the caste system in exist in the early years of the industry. increase was picked up), and in South India, the question of how to get past To look at just one of those, the rise Africa polls have tended to overestimate the influence of the village head in of radical or insurgent parties is often the ANC, the biggest party. some areas of South Africa, or ensuring difficult for polls to pick up, especially Having said that, the impact of rising that women can express their choices where there is a lack of historical data insurgents doesn’t always have the freely in more patriarchal societies. to provide the empirical basis for our same implications for the polls. In the These are all instances when there assumptions. For example, as shown Netherlands the left wing Socialist party is no substitute for local expertise on in Figure Three, Sweden’s electorate was over-estimated, and in the UK the the ground, but there are also some has transformed from two political polls generally performed better than common methodological issues that blocks to three in just ten years. And expected in predicting the vote shares our experts are grappling with in every in an otherwise good performance by for the minor parties. Finally, Ireland is country – in particular, how to respond the polls, the Swedish Democrats were an interesting case study of how social to changing social factors, challenges underestimated in the most recent change can affect poll estimates for different parties. A decade ago the bias was towards the biggest brand, Fianna Figure TWO. Fail, but since the economic crisis for which Fianna Fail received much of The challenges for the polls. the blame, if anything there is a now a reluctance to admit to voting for them. Social and political change Sampling is another key concern, for all survey researchers. In the Makes predicting Makes it harder to UK, the challenges of getting a turnout more get a representative representative sample (in particular, important sample getting enough people who are not Does the sample contain engaged in politics) was the key issue the right mix of voters and identified by the British Polling Council Turnout non-voters? Sampling inquiry. In Sweden, while the polls Can we identify voters and non-voters correctly? were mostly accurate, our analysis suggested they underrepresented areas that voted Sweden Democrats, Individual country factors and overrepresented areas with high support for the Greens (intriguingly, one 3. Ipsos MORI - Understanding Society May 2016

so can lead to fewer polls overall, and Figure THREE. the demand for faster, cheaper, and more regular polling in other countries The rise of insurgents – how Sweden moved from two political blocks to has led to a greater diversity of three in ten years. methods and sampling approaches. This in turn creates further 60% Centre-Left Coalition Centre-Right Coalition Sweden Democrats challenges for pollsters, for example difficulties getting young people to answer calls on their mobiles from unknown numbers, the lack of geographic identifiers on many mobile numbers making it harder to sample 30% across regions, or the difficulties online panels can have in covering the very oldest groups in the population. At other times, though, mode seems to make little difference – in both the UK and Italy, for example, despite a range

0% of methods, no mode was obviously Left-Right Parties Election 2006 Left-Right Parties February 2016 superior to the others. Turnout is another crucial factor factor here seemed to be an interviewer landline penetration in countries such that affects many countries, especially effect – where interviewers had a as South Africa means telephone is not given the general decline in turnout foreign-sounding name, or a different a viable option at this stage. throughout the world (as shown in the accent, response rates were lower in This method is expensive though, and chart below). As usual, there are some high Sweden Democrat-supporting areas). And the general problems of low response rates affect many around Figure FOUR. the world. Difficulties in getting representative Voter turnout in parliamentary elections. (%) samples are often linked with changes 95 in the way polling is carried out (the 90 ‘mode’), especially where there has been a move towards newer methods 85 of interviewing, such as greater use of 80 mobile phone interviewing, or online panels. Again, though, our international 75 perspective reminds us that this is not 70 the case everywhere. In countries 65 such as India, South Africa and Ireland, traditional face-to-face polling is still the 60 norm – and often very successful too, 55 such as in the 2016 elections in Ireland (where telephone sampling can over- 1950 1960 1970 1980 1990 2000 2010 represent more middle-class and public sector voters), and indeed the lack of Global average Commonwealth Europe Northern America 4. The Death of Polling?

exceptions to the rule – in countries models based on the constituency need to do properly), and by better such as Sweden and Australia turnout is patterns of voting in 2015, and the weighting procedures. In the UK, for high, and so makes less of a difference results in our polls showed dramatic example, downweighting the proportion (helped of course by compulsory voting changes in predicted levels of turnout of readers of quality newspapers in Australia, although even then there is with no change in sampling procedures. in our polls is already improving our a need to deal with the small proportion In the US, the level of turnout can make coverage of non-voters. There is a of spoilt ballot papers). all the difference between a Republican particularly interesting response in The impact differential turnout has victory and a Democrat one, as shown in Canada, where our successful pre- on results can be significant (although the chart below. election survey of 2015 (with an average again not in every country – for example error of just 1.2%) employed a mix- Canada ran a sophisticated likely mode methodology. Our analysis there turnout model in the 2015 elections, How to respond? suggested that different modes had and found that different turnout levels biases towards certain types of voter made no difference at all to the final Given the range of individual attitude-types, even when correcting prediction). In Italy, part of the cause country factors, there is unlikely to for basic demographics – online modes of the underestimation of the 5Star be a single silver bullet response to tended to overestimate progressive vote was an assumption that the more these challenges – indeed, the variety voters, while telephone polls tended to disillusioned voters would not vote, as of methods provides a rich variety of overestimate conservative voters (again, they had in the past – but instead they lessons to learn from. Achieving more it is worthwhile pointing out that this were motivated to vote for this new party. representative polls can be helped is not always the pattern seen in other In the UK, although not identified as a by improving sampling coverage from countries). A mixed mode approach, prime cause by the BPC inquiry, better first principles (although a move to then, was developed to cancel out the turnout predictions are likely to be a key pure random probability sampling is different biases in the different methods, aspect going forwards – some pollsters unlikely to be the answer, not least involving 1,000 interviews online (both have developed sophisticated turnout because of the time and expense they from panel and non-panel sources) and Figure FIVE.

Turnout levels make a big difference in the US. 2014 Generic Congressional Ballot by Turnout Levels.

51 Republican Vote 50 49 49 48 48 47 47 47 46 46 45 44 44 Democratic Vote

Actual 2014 Turnout 2012 Turnout

Generic Congressional Vote Share 30% 35% 40% 45% 50% 55% 60% 65%

Turnout Level (by Likely Voter Model) 5. Ipsos MORI - Understanding Society May 2016

1. Already touched on above is the move towards new/mixed mode methods, in response to the desire for more, faster, and cheaper polling, and to correct biases in the old methods. Again, the practical implications of this may differ from country to country. In South Africa, for example, the relative preponderance of mobile phones may mean that polling via SMS is the next step after face-to-face polling, while in Canada as mentioned above we are already exploring a four-way mixed sample split. And there are other techniques too taking advantage of new digital methods to understand public opinion in greater depth, such as our analysis of social listening during the British general election, and the online community we ran for the BBC, which allowed us 1,000 interviews by telephone (both broader attitudes and motivations to build up a close relationship with landline and mobile phone). towards voting. In the 2014 general individual panel members week-by- The accuracy of our prediction based election, for example, we estimated that week as the campaign evolved. on this method suggests it is well worth the ANC would receive 61% - 65% of the exploring further. vote, depending on the level of turnout 2. Related to this, the challenges There are also some very thought- (for example, we predicted 63% if turnout of diminishing client budgets provoking responses to dealing with was moderate). In the end turnout was – especially in the media – but turnout, with South Africa a good 74%, in line with our ‘moderate’ turnout rising expectations for rapid and example. First of all, our polling there category, and the ANC received 62%, comprehensive insight on the state includes rigorous tests that go well very close to our prediction. of public opinion is something that beyond simply asking respondents their is faced in every country in which likelihood to vote, involving sanity checks we poll, and helps drive the need against official registration figures, and The future for constant innovation in the way physical checks during the interview we do polling. itself to ensure respondents have their ID of polling books proving they are registered to vote 3. One trend, in response to growing (obviously only possible thanks to their So what changes might we see awareness of the limitations of face-to-face methodology). Secondly, ahead? Some might laugh at the idea polling, is to see more attention given the importance of turnout to the of pollsters trying to predict what might beginning to be paid to figures results, we publish predictions based on happen in the future, but when it comes beyond the simple ‘horse-race’ high, medium and low turnout scenarios, to developments in our own industry results. This is exacerbated by modelling responses from a number of there are some common the difficulty in many countries of different questions taking into account patterns developing. translating national vote shares into both expressed likelihood to vote and seat forecasts (for example, in the 6. The Death of Polling?

UK even if the polls had shown the correct vote shares, traditional seat Figure SIX. calculators would still have struggled to predict the exact final outcome). How big data and aggregation allows us to model for the impact The most famous example of this is of incumbency. Gallup in the US, who have stopped producing vote intention figures

and now focus on monitoring public opinion on other variables such as 4 leader approval and issue ratings. (2012)

4. Another response to the limitations of any individual poll is greater use of Incumbents modelling/big data approaches. The US is taking the lead here, perhaps not surprising in a country that Successors Odds of Winning has seen the total number of polls 65% 60% 55% 50% 45% 40% 35% explode to over 17,000 published in the 2012 presidential election Government Approval Rating alone, and at Ipsos we are now polling continuously 24/7, totalling Public Opinion Research (CAPOR) to mention the alternatives, such as the 11,000 interviews a month. This in 2015, and is a charter member of recent Oldham West by-election in the hugely increases the opportunities to the AAPOR Transparency Initiative UK,7 which in the absence of polls still aggregate different polls and develop in the US.6 The common thread recovering from their general election models based on the full set of data throughout these is a commitment failure, was called wrong by both available to us, rather than expecting to transparency in order to promote pundits and betting markets. any single poll to provide a perfect better research standards, and a Most importantly, as we try to show in prediction. For example, this allows better understanding among the the rest of this edition, even if we are us to demonstrate that it is much media and wider public on how to unlikely to see the end of ‘horse-race’ more likely for an incumbent to win an interpret polls and make use of their polling any time soon, the true value of election based on a given approval results. our research is much broader, in what rating than it is for a new candidate. it can tell us about the wider attitudes, So where does this leave polling – values, and motivations of the public, 5. The final trend we are seeing is an apart, perhaps, from a feeling that it is to truly help us better understand the increasing focus on transparency surprising that anyone should try in the societies in which we live. Polling isn’t initiatives, as pollsters take on more face of this challenges? It seems clear dead yet, especially in information- responsibility for improving public that the future will bring more changes hungry societies where comment understanding of our research. – some no doubt will fail, but others will and conjecture would quickly replace An early example of this is the advance the quality and accuracy of data. But it is incumbent upon us as National Council for Public Polls polling (as Professor Sam Wang says professionals to ensure that our polling (NCPP) in the US, set up in 1969,5 in our interview later, “change is not is transparent, reliable and accurate to which provided the model for the death, it’s life – polling is as interesting keep it alive and well. British Polling Council, established as ever”). And the more doom-laden in 2004. At the same time, Ipsos prophecies ignore the often less- has been heavily involved in setting heralded successes pollsters have in up the Canadian Association for many countries around the world, not 7. Ipsos MORI - Understanding Society May 2016

Poll position Bobby Duffy Bobby Duffy interviews Professor Samuel Wang. London

forecasters like Nate Silver are doing? I view my calculations as presenting a fresh snapshot of current conditions. In a general election campaign, public opinion doesn't move that fast, so it is possible to have some clue about what will happen months into the future by looking at polls only. In the US, midterm elections are harder to predict, so it’s really only the immediate view that polls give. At FiveThirtyEight, they like to use econometric factors which have predictive value at times far from the election. It’s a separate activity, which is good for making predictions before polls are available, and for predicting states or districts where data isn't available.

Polling has had some well publicised lapses of late – Israel and the UK for Sam is a professor of neuroscience way Florida would go. So if I could see example. But it has also had some and molecular biology at Princeton that, certainly some high paid consultant successes – Canada, Sweden and University. In 2004, he founded on the campaign was telling Gore the South Africa. What’s your view on the Princeton Election Consortium same thing – that Florida was critical. whether the accuracy and value of website which takes a meta- I became fascinated with that polling is changing – is it getting analysis approach to polling. Sam’s because it was a very close and unusual worse or not? election analysis has featured in race. I realised that it came down to titles like the New York Times and Florida, Pennsylvania and Ohio and As far as I can tell, every decade or the Wall Street Journal. whoever won two of those would so, there is some crisis that makes probably win the presidency. I got into people think that polling is dead. Polling You are a professor of neuroscience arguments about it with a colleague who failures have occurred frequently and have authored several books on is a string theorist. When he thought throughout history, and are usually linked the subject. What made you want to about it, he agreed - I was just so with changes in how pollsters reach create your own model of political delighted to win an argument with a string people. The Literary Digest once said polling? theorist that I thought I could do a fancier with great assurance that Franklin version of that. I came up with a tracking Roosevelt would be defeated in a I was watching that movie, index that turned all the states into landslide. Of course, they were wrong, Fahrenheit 9/11 and it begins with Al probabilities, took the analysis further and and it turns out they were using a mailing Gore greeting the voters of Florida. I posted it on a blog. My analysis became list from a previous election. People had wanted to know why he chose that state, popular that year among financial traders moved in the intervening time, especially how he knew to be there– was it just by and social scientists, and was on the those hit by the Great Depression, chance? Gore obviously has high levels front page of the Wall Street Journal.8 making the list of voters they could reach of talent available to him, and I, by So I was hooked. biased towards Republicans. reading the polls and careful analysis My point is that there will always be published in The New Republic could What is your approach to predictive new technology that makes it easier to see that the race depended on which analysis - is it different to what reach people. Look at the US - in the 8. Interview: Professor Samuel Wang

last five years, we have had a major that general elections polls are pretty transition from landline to cell phones. accurate in the US, primary elections So it would be a good idea for pollsters When you aren’t so accurate and multi-party to think about how to keep up - our elections are extremely challenging. changing population requires new look into Both readers and journalists should think technology for reaching them. about what to expect from polls. Given So I don't think polling is getting what pollsters the difficulties of reaching people, I think worse. Pollsters are trying out new it’s a miracle we can get within five methods all the time. In decades to have taken points. In both the recent Iowa and New follow, landline phones became a better Hampshire elections, Trump came in method – and now internet sampling. the trouble with two to four points - and that’s an Call me contrarian, but I think polling is open primary. It’s pretty impressive. more interesting than ever. to ask, there When you look into what pollsters have taken the trouble to ask, there are Has the recent rise in multi-party are all kinds all kinds of interesting material inside elections, challenger parties and polls. I think journalists do use that firebrand candidates made polling of interesting information but that’s the huge part of more challenging? the value of polling – not just the top material inside line number. Multiparty elections like the UK and Israeli elections, where strategic voting polls – not just In the aftermath of the May 2015 occurs, are hard to poll. In the US, polls election, the British Polling Council’s in two-party races are as accurate as the top line findings show that the challenge of ever. So far in the US Presidential race, sampling was a key factor in the UK.9 polling errors have been between two number. You already mentioned that the and five percentage points. Opinion can method of reaching voters keeps move fast in primaries, and that's not so needing to evolve. What else do you different from past years. But I also don’t think that polling should just be about the two horse race question. Pollsters are asking so many other questions that are super interesting. They are asking why young voters like Bernie Sanders and older voters prefer Hilary Clinton. There’s all kinds of richness in polling data.

That’s very interesting. People don’t always make the connection between polls as a way of predicting a horse race and giving a wider insight. Do you think expectations of accuracy on that narrow question are too high?

I think it’s important to know how accurate polls are going to be - we know 9. Ipsos MORI - Understanding Society May 2016

think are the challenges facing polling separate? Your modelling approach is in the US? fairly clean and straight. So you don’t I had the want pollsters to be modelling or In my view, the biggest challenge is ‘messing’ with their base data? low turnout in midterm elections, which idealistic happens in the US when there is no Yes, I think poll analysis and Presidential race. In 2014, which was a hope that forecasting should be kept separate, for midterm year, turnout was the lowest it two reasons. First, people often want to had been in over 72 years.10 In such a statistical know what the polls say, and that is a situation, it is hard to identify likely-voter statistically well-defined problem. populations with accuracy. methods Second, I think the forecasting question Consistent with that, midterm is equally interesting. When the two are elections have the largest polling errors. could reduce in tension - that is important news. For As most voters become less engaged, example, polls currently strongly favour this becomes an important issue - not the breathless Donald Trump for the Republican just for pollsters, but for democracy. nomination but predictive factors like Another big challenge is how to survey horserace endorsements point in a very different movable populations. States like Nevada direction. That’s a fascinating story and and Alaska have many residents who commentary focussing too much on the nerdy horse just arrived, or are about to leave. In race predictions misses that tension. those cases, eligible voters often have that followed Data journalists are in a unique position out-of-state area codes. In these cases, to tell that story, as long as they keep voter lists, cell phone sampling, and the release of the two separate. internet sampling are important. a single poll. On that, we have seen a huge rise in In the last decade or so, there has data journalism as an important part been a rise in poll aggregators, of media outlets. Do you think there is forecasters and new models both in an over reliance on polling in political the US – your website, The Upshot/ debates these days or does it still FiveThirtyEight/RealClearPolitics – horserace commentary that followed the provide an important function in and in the UK – Polling Observatory/ release of a single poll. I thought that telling voters what others think? (Rob Ford/Chris Hanretty/Will this would open up more airtime for There is discussion in the UK about Jennings). With the availability of analysis of issues – that hasn't quite restricting the publication of polling aggregation tools and technology, happened! Instead, we have a genre of in the run-up to elections. can everyone and anyone be a commentary on polls and the horserace, forecaster now? divorced from issues. It's become Polls play a central role in shaping popular, but I wonder whether we are coverage. However, I have noticed that I think it’s an area that’s maturing – really any better informed for it. I think journalists are perfectly capable of Nate Silver and people like him have those sites can contribute by analyzing weaving a story that is completely at really blazed the trail. His success in the the data by itself, as a separate object; odds with what polling data tell them! So New York Times inspired them to found and then the data in the context of I am not entirely sure what the net effect The Upshot, that now features some issues. That is very hard, and requires [of restricting polls] would be. pretty interesting voices. both data people and journalists. It’s important for writers to know what When I got into this activity in 2004, I polls can't tell them. A single poll is had the idealistic hope that statistical never to be trusted, especially when it methods could reduce the breathless Should polling and forecasting stay gives a result that is different from other 10. Interview: Professor Samuel Wang

polls. That's called an outlier and usually declined because it’s not in the media’s Supreme Court has had difficulty in to be discarded. Broadly, I think the interest to avoid it - outliers are the establishing a clear standard and I think details of polls often have rich unusual and interesting event. If Donald the right way to approach it is statistical. information about the voting public. Trump talks about building a wall or Now judges and lawyers are not known There are many stories to be found in Marco Rubio gets stuck on a script, for their love of statistical analysis and, as the cross-tabs of polls. that’s unusual and journalists will report far as I know, nobody chooses to study The reporting of outlier polls hasn’t it. But if they are discussing a statistical law because they love math. But here is outlier -a data point that is different from an example of a discipline that could use other data points taken at the same time this powerful method of reasoning, but it’s - that’s not the kind of ‘unusual’ they not a common tool to them – the legal In the case should be reporting. profession is filled with very smart people who don’t always turn to math. So there of polling, What ways do you think we can is something missing here. improve media reporting and public As for the popularity of statistics, I change is not understand of how to interpret polls? guess there is a nerdy glamour to people like Mark Zuckerberg and Nate Silver, death - it is Journalists are storytellers and their which helps... stories inevitably have an emotional life. Like Sarah tone, but I think there needs to be a What changes do you see for balance there with factual content. It forecasters and polling in the future? Palin said, would be cool if they were trained to ask What will accurate and successful whether their stories were consistent forecasting/polling looking like five or only dead fish with numerical data like polls. Also, it ten years from now? would be great if they could tell when go with the they were being swept up in a wave of As Yogi Berra said "It’s tough to make enthusiasm. I think that journalists predictions, especially about the future". flow. Polling is follow, or at least trail, public opinion Forecasting in other domains, like the more than they realise, and are more economy, seem to get slightly better with as interesting aware of one another’s attitude than time. As the track record gets longer, we they acknowledge. They are social are going to develop a better idea of what as ever. animals like the rest of us but have an we don’t know. extra responsibility to be aware of unknowing biases. Data based analysis Our title for our journal is the “Death has the potential to provide a way for of Polling?” What’s your view of its writers to check these biases. state of health?

Do you think the increased popularity In the case of polling, change is not of books on the application of death - it is life. Like Sarah Palin said, statistics is a good thing - are only dead fish go with the flow. Polling is statistics finally sexy? as interesting as ever.

This is super important. People who are good with numbers have many career Sam's analysis can be found here: advantages. I’ve been spending a lot of http://election.princeton.edu/ time with legal scholars lately, working on the problem of gerrymandering. The 11. Ipsos MORI - Understanding Society May 2016

How Britain voted Gideon Skinner A comprehensive analysis of the 2015 election campaign. London

The death of polling maybe a next time – particularly in picking up figures. Many of us working in polling grand description of what happened differential levels of turnout among would argue that that undersells the full last May, but it is clear that it does different party supporters. Nor should value of public opinion polling. Certainly need some resuscitation. The Exit Poll it be forgotten that the polls did make it’s not what we at Ipsos MORI find most broadcast, just after Big Ben struck several correct calls, some of which interesting in our research. 10pm on Thursday 7th May, pointed to a we were expecting to be harder to get It also ignores all the other types of Conservative majority, when many had right – for example the collapse of the research that we are utilising to gain a been expecting a hung parliament. Liberal Democrats, the SNP landslide, richer understanding of the motivations There was a variation in the vote and the rise of UKIP to third place (not behind voters’ behaviour, such as online shares predicted in the eve-of-election to mention the Exit Poll itself in which and traditional focus groups, social media polls, but they all pointed to broadly the we were heavily involved). analysis, in-depth digital communities, same outcome, and none picked up the It is though worth taking a wider look longitudinal tracking studies, learning from extent of the final Conservative lead. at what the polls had to say. Much of the neuroscience and behavioural techniques, We have already covered some daily focus on opinion polls in the media and so on. Our mission is to make sense of the reasons that might explain and the Twittersphere was on small, of society, so what can we learn from the performance of the polls, and often statistically insignificant changes all the research we conducted over the the areas where we need to improve in the horse-race voting intention election that explains the final result? 12. How Britain voted

The Conservatives had real The rise of UKIP, although taking problems. Labour’s votes from all parties, was also particularly attracting past Conservative They struggled “despites” voters. Our final election aggregate analysis estimates that 14% of 2010 to reach out Let's begin by examining Labour's Conservative supporters switched to strengths and why it thought it had UKIP in 2015.13 to certain a chance of winning. Labour were Furthermore, Ed Miliband had consistently much more liked than the clearly touched a nerve with his groups, in Conservatives – the party of the heart campaign on the cost of living. Eight (if not of the head, to which we will in ten agreed with him that there was a particular come). In our party image questions, cost of living crisis. Personal economic Labour did particularly strongly on optimism was half that of country younger “understanding the problems facing economic optimism and only one in Britain”, and “looking after the interests five thought their family would benefit women, and of people like me”.11 They were also from the growth in the economy. The the only major party to be more liked public thought he "got" the problems in much of than disliked – 52% said they liked the they faced – his best score on his Labour party, compared with just 33% own detailed image ratings was for the country liking the Conservatives.12 understanding the issues facing Britain. The Conservatives had real problems. Finally, there was Labour’s trump outside of the They struggled to reach out to certain card – most trusted on public services, groups, in particular younger women, and particularly on the totemic issue of South East and in much of the country outside of the NHS. The 2014 winter crisis boosted the South East and South West. They the NHS to the top of voters’ concerns, and South were strongly associated with being and although David Cameron had had closer to the rich, businessmen, and the some success in nullifying Labour’s West. City – not the most popular groups in a strength on the NHS prior to the 2010 country still recovering from the 2008 election, this time Labour were going in financial crisis. Meanwhile, the “Rose with their customary lead. Garden” sheen of the UK’s first coalition for many years had quickly worn off, and Britons reverted to their traditional dislike The Conservatives for this form of government. The Conservatives themselves were make their facing structural barriers to achieving their first outright victory for 23 years. strengths count The so-called bias in the electoral system meant that they needed around So how does our polling help to twice as big a lead as Labour to achieve explain why the Conservatives were a majority. History was also against them able to win despite these challenges? – regardless of what the opinion polls It was The Evening Standard’s said, only two governments since 1900 Political Editor Joe Murphy who first had increased their vote share after characterised the 2015 election as the more than two years in office, and none “war of the weak” when using our polls. in the last sixty years. In fact, for several years it was clear that 13. Ipsos MORI - Understanding Society May 2016

Labour's biggest problem was the image Figure SEVEN. of its brand. Despite a better campaign for Ed Miliband, Cameron owned key 'PM' Despite being qualities throughout the Parliament. seen as the party of the Good in a crisis Capable leader Out of touch with heart, the odinary people Has got a lot Looks after some of personality Has a clear sections of society Conservatives vision for Britain more than others Has sound More style had regained judgement than substance

their Understands the problems facing Britain reputation as

the party of ¡ Cameron ¡ Miliband ¡ Clegg ¡ Farage the head.

Labour had as many weaknesses as for Conservative supporters as the strengths, and neither party could count prospect of UKIP in government was for on their advantages. Labour supporters. Labour had similar structural Labour's biggest problem was the and historical challenges to the image of its brand. Despite Labour Conservatives and their own precedent being seen as the party of the heart, to overturn. No opposition party for over the Conservatives had regained their 80 years had managed to win a majority reputation as the party of the head – a after a single term out of office.14 And perception they lost to New Labour in they also had their own challenger the 1990s and early 2000s. Labour also in the wings. Polling consistently failed to paint itself as a One Nation showed – to widespread incredulity – party. It was seen as the party of its that the SNP had managed to snatch traditional client groups (the working victory from the jaws of independence classes, the North, trade unions, benefit referendum defeat, and were on course claimants and immigrants), but not as to achieve an astonishing swing in representing the South, middle-classes, Scotland, putting many of Labour’s homeowners or businesses. safest seats at risk. This held a double The difference was even starker when whammy for Labour – the prospect of the voters compared the two leaders. the SNP having influence over the next Leader image counts to voters – they are government was as much a turn-off making an emotional decision as well as 14. How Britain voted

Despite Labour’s campaign a purely rational one – and our historical leader since the start of the campaign, data shows two things. Firstly, that no but in the same survey Cameron was still on the cost Prime Minister had ever been successful twice as likely to be seen as capable).15 with ratings as low as Ed Miliband’s so In the end, Labour didn’t make their of living, close to an election. Secondly, the traits strengths count. They were unable to that make most difference to voters are convince voters that they would make a it was the not personality or even being out of difference to their lives – particularly on touch, but classic competency issues the economy. Conservatives such as being a capable leader, good in Despite Labour’s campaign on the a crisis, and having sound judgement. cost of living, it was the Conservatives who convinced And they were the very issues that David who convinced voters they were to be Cameron had captured in the public’s trusted on the economy, recovering from voters they mind, in some cases leading Miliband by the 2012 “omnishambles” budget16 to over 25 percentage points. retake the lead on this key issue. were to be Our tracking data during the Our post-election polling showed campaign showed that Miliband did that when it came to the crunch, it was trusted on the improve public perceptions of him, but the economy, not the NHS, that was from a low base, and not enough on the most important factor for voters. economy. the key Prime Ministerial attributes (for Qualitative research suggests that example, more people said their opinions while the Conservatives were unable to had improved of him than any other completely allay fears that they would Figure EIGHT.

Q: Before - Looking ahead to the next general election, which if any, issues do you think will very important to you in helping you decide which party to vote for? After - Thinking about the general election, which single issue was most important to you in helping you decide which party to vote for?

Managing the economy 25 / economic situation 41 Healthcare/ 15 NHS/hospitals 60 9 Benefits 48 8 Poverty/Inequality 31 7 Asylum and Immigration 41 5 Europe/EU 38 3 Taxation 31 3 Education/schools 31 Care for older and 2 disabled people 34 2 Housing 33 0 60

¡ % After the election (week 7) ¡ % Before the election (week 4) 15. Ipsos MORI - Understanding Society May 2016

bring in privatisation of the NHS, people also suspected Labour of using it as Figure NINE. a political football. The issue even reinforced perceptions of Labour as Labour increased the lead it already had among young people. But the inefficient and wasteful. party actively lost ground among the oldest group – those most likely to vote. As a result, while Conservatives 33% were almost as likely as Labour voters 42% 18-24 26% +3.5% to say the NHS was an important issue 28% to them, it was Conservative policies on 30% 36% the NHS that they saw as credible, not 25-34 32% +2.5% 33% Labour’s. 31% During the election campaign, Ipsos 35% 35-44 35% +2% MORI hosted an online community with 35% 28% the BBC and found similar patterns 32% 45-54 34% +0.5% in the follow-up interviews. People 36% wanted to send Labour a message that 27% 31% it needed to change, and it was caught 55-64 40% +3.5% 37% both coming and going. On the one 33% hand people felt that it was still tainted 22% 65+ 44% -7% by the previous Labour government, 47% with too many of the same faces, failing Vote share by age 2010-15 Con-Lab Swing to learn the lessons of the crash, and ¡ Labour 2010 ¡ Labour 2015 ¡ Conservative 2010 ¡ Conservative 2015 losing touch with its core support. But on the other, Labour had lost one of its most successful features of the landslide years – the sense that it a broad coalition of support outside the Summer Budget was more positive stood for “aspiration but with fairness”. their heartlands of young people, than negative, and George Osborne is Meanwhile, the Conservatives offered renters, London and the North. holding on to the best ratings for any reassurance and credibility, especially The SNP remain well ahead in Conservative Chancellor since Nigel while the economy was still recovering. Scotland, while Labour’s task of Lawson (although it has fallen since the winning seats in England is not made March Budget 2016). any easier by a reversal of electoral Ipsos MORI's latest polling on public The road to 2020 bias in the system. Thanks in part to services also suggests that people may the Conservatives picking up many be changing their expectations in the The 2015 general election saw the seats from the Liberal Democrats for a light of ongoing austerity. Fewer people smallest swing between Conservatives relatively small increase in vote share, feel affected by the cuts now than in and Labour since the Labour party was an equal number of votes between 2012, even though clearly many more founded, and the worry for Labour (and the two main parties would now give cuts have been carried out since then, hope for the Conservatives) must be that the Conservatives nearly 50 more the balance of opinion still thinks cuts so far there is little sign of any major seats. And while the impact of the have been necessary, and perceptions shift in the parties’ positions. Boundary Review is still unknown, most of quality across many public services The Conservatives consolidated their estimates suggest it will increase the have held up (with one notable position among the groups most likely to Conservatives’ net lead over Labour. exception of the NHS, where concern is vote – the old, and the middle classes – Meanwhile, the Conservatives at record levels). and won 19 seats across the South East continued to solidify their message of And what of Labour’s leader, Jeremy and South West.17 Labour failed to build a long-term economic plan. Reaction to Corbyn (whose victory, incidentally, 16. How Britain voted

was picked up first by the polls)? He is certainly seen as representative of Figure TEN. a new type of politics – a focus group watching his conference speech live Q: Are you satisfied or dissatisfied with the way… doing his job as Prime told us he seems “one of us”, “he’s Minister/Leader of XX Party? believable, he’s passionate”,18 and our analysis of social media gave him a Foot - Nov '80 21 23 +2 resounding thumbs-up. Kinnock - Oct '83 13 33 +20

However, it’s not just supporters on Major - Dec '90 22 37 +15 Twitter he has to win over. Corbyn is Smith - July '92 12 30 +18 polarising opinion to a rare extent so Blair - Aug '94 16 34 +18 early in his reign, but concerningly for Hague - Jun '97 13 12 Labour age remains one of the dividing -1 factors, with younger people much more Duncan Smith - Sep '01 15 15 0 positive, while older voters are sceptical. Howard - Nov '03 17 26 +9 He speaks more to those who already Cameron - Jan '06 17 31 +14 describe themselves as left-wing, than Brown - July '07 20 36 +16 those who put themselves nearer the Miliband - Oct '10 22 41 +19 centre. And the events of the leadership Corbyn - Sep '15 36 33 -3 campaign itself, and its aftermath, Net+% have seen dramatic increases in those describing Labour as extreme or divided. ¡ % Dissatisfied ¡ % Satisfied Overall, Labour still has to answer the question it failed to do so in 2015, as one swing voter told us: “I’ve got a mortgage and two young kids and I feel secure right now… if Labour come in would they rock the boat?”19 Of course, that does not mean 2020 is in the bag for the Conservatives. The Labour still has to answer the interconnected issues of Europe and immigration, concern about housing, tax question it failed to do so in credits and a prospective leadership battle all have the potential to take away 2015, as one swing voter told their upper hand. But despite all their faults in predicting the final outcome, the us: “I’ve got a mortgage and underlying reasons for the Conservatives’ victory were identified by the polls a long two young kids and I feel secure way in advance of the election itself. The polls need to improve the right now…if Labour come in accuracy of their final predictions in the future, but they will play a crucial role in would they rock the boat?”. making sense of the next election result too. 17. Ipsos MORI - Understanding Society May 2016

Cracks in the foundation The impact of UKIP, the Greens and May 2015.

dominant as the only plausible alike had decayed, and local candidates alternative government. The two main unaccustomed to genuine competition parties won 86% of the seats in the found themselves unable to respond to House of Commons, about the same as the SNP dual appeal to nationalism and their share in the previous election. The traditional left wing values. Liberal Democrats, the Conservatives' There was also a great insurgent coalition partners, collapsed and the wave in England and Wales. Nigel idea of multi-party coalition politics Farage's UKIP won 3.9 million votes, seemed to collapse with them. 3 million more than in 2010, the best Closer examination suggests a more performance by an upstart party with complicated picture. An electoral tidal no links to the existing Westminster wave has transformed Scotland from a establishment since the emergence of Labour heartland into an SNP one party Labour over a century ago. The Greens state. Nicola Sturgeon's nationalists now won another 1.2 million votes, more Rob Ford is a politics lecturer at the control 56 of the 59 available seats, than quadrupling their 2010 total. Yet University of Manchester and has after achieving swings so large that they both parties found themselves nearly co-authored Revolt on the Right and nearly broke the BBC's election night shut out of Parliament thanks to the Sex, Lies & the Ballot Box. swingometer. The main reason for this workings of the British electoral system is the realignment of Scottish politics on parties with evenly spread support - In some respects, the British election around the independence question - despite winning nearly one in six of the of May 2015 was an unexpected more voters now back independence, total votes the two parties only have a resumption of normal service. The and almost all of those who do now back single MP each among the 650 who sit Conservative party won enough seats the SNP. However, the SNP's success on the Commons benches. Under the to govern Britain with an absolute also reflected Labour's long neglect of proportional representation electoral majority, and the Labour party remained Scotland; voter loyalties and local parties systems employed in many European countries, UKIP could have 80 or more MPs, and the Greens as many as 25. While the electoral system Nigel Farage's UKIP won suppressed its political impact, this was an unprecedented rejection 3.9 million votes, 3 million of politics as usual by the British electorate. What sorts of voters turned more than in 2010 - the best to the two insurgents, and what impact will their surging support have on performance by an upstart party British politics in this Parliament and beyond? The surge in UKIP support with no links to the existing was driven by "left behind" voters - older white British voters of modest means Westminster establishment and few educational qualifications. UKIP's best performances came since the emergence of Labour in economically declining towns, often ports and manufacturing areas over a century ago. where industry collapsed years ago and nothing has come to replace it since. Farage's voters express deep disaffection with traditional politics, 18. Cracks in the foundation

The old social divisions of class and status, and the which has delivered nothing for them, old political in which the new parties have grown. but strong attachment to British, or What next? UKIP and the Greens more usually English, nationalism and arguments have given voice to powerful concerns, identity. Immigration and the EU have but at present they lack systematic become the twin lightning rods for these about the role representation in Parliament thanks voters. Immigrants as both a scapegoat to the effects of the electoral to blame for their marginalised and of the state, system. Without a strong presence declining economic and social position in Westminster, both parties will and a threat to the traditional identity no longer struggle to win attention and resources and values they treasure. The EU as commensurate to their electoral a symbol of the distant, alien and motivate support. They could buckle under the unaccountable political elites who run weight of frustrated expectations - their lives yet ignore their concerns. voters as they many Green voters could be attracted Green voters are, in every respect by the idealism of Jeremy Corbyn's but one, the polar opposite to Ukippers. once did. Labour, UKIP voters could decide While UKIP's support is strongest that they are better off aligning with among the oldest and those with a Conservative majority government no qualifications, the youngest and than a single UKIP MP. The strength of most highly qualified voters are the these voters' concerns, and the depth Greenest. While UKIP voters are strongly class and status, and the old political of their disaffection, suggest this will nationalist, hostile to the EU, and arguments about the role of the state, only delay a reckoning. The partisan socially Conservative, Green voters are no longer motivate voters as they once and class loyalties that bound voters cosmopolitan, committed Europeans did. New values and new arguments to the traditional parties are long gone, and socially liberal in every respect. have taken their place, in particular and unlikely to return. Voters no longer Ukippers oppose immigration as a deep divisions over identity, social want to fall into line behind two large cultural and economic threat and dislike values and social change which pits parties, who can longer represent the immigrants, Green voters embrace progressive young graduates against range of their concerns and values. The it as economically beneficially and their conservative grandparents. This likelier outcome in Britain, as elsewhere enriching to society, and are the most divide poses problems for the traditional in Europe, is thus for more volatile and welcoming towards immigrants. While British parties of government as it cuts pluralistic politics. Britain's old political UKIP voters concentrate in declining and across the traditional conflicts which elites need to figure out how to respond homogenously white former industrial have defined them, and splits both of to this change. As Scotland shows, if towns and ports, Green voters are them down the middle. Labour has the cracks in their foundations grow too heaviest on the ground in booming voters who proudly wave St. George wide, even the mightiest party fortresses university towns and cities and the cross flags, and others who see the can come crashing to the ground. diverse and rapidly growing boroughs same flag as a symbol of oppression. of central London. There is only one The Tory tent encompasses social thing Green and UKIP voters agree on liberals who embrace diversity and - "politics as usual" in Britain is broken. immigration, and social conservatives Greens, like Ukippers, have very negative who find mass migration very views of mainstream politics and very threatening. These internal divisions low trust in the political system. make it hard for either party to muster The simultaneous surge in support a coherent response - any strong for two parties with polar opposite views position risks alienating as many voters reflects a deep change in our political as it attracts. This inability to respond system. The old social divisions of effectively has helped create the space 19. Ipsos MORI - Understanding Society May 2016

On the campaign trail Journalist and author Ann Treneman gives us her thoughts on the election and beyond.

man who had been stung by a deadly the first few years, the issue of Scottish jellyfish. And Nick Clegg? I think that, independence was virtually invisible. until 10.03 on election night, he rather It was only when the referendum was fancied his chances as staying on as upon us, that Westminster got excited, Deputy Prime Minister. moving en masse north of the Border So where did we go wrong? How and then, after the result, just as quickly did we miss the train coming down the back. But the fact is the SNP may have tracks, heading straight for us? Part lost the battle but, buoyed by a tsunami of the reason is that we were looking of energy from their supporters, they elsewhere, mostly inwards towards the are now winning all sorts of wars. Westminster Bubble, talking to ourselves about ourselves. Devolution may be top The Liberal Democrats of the political agenda but the press and When I looked back through the the commentariat remains SW1-centric. sketches it was clear that the Lib Dems It is amazing how quickly we forget Here, then, are a few factors that we started losing the country (though not when it comes to politics. During the should have paid more attention to: the commentariat) almost immediately. election, as a special treat to myself, The truth is that the press forgives easily, I kept a dictionary of the words that Scotland drawing a line and moving on, but the defined GE2015. Words like “kitchen”, Westminster has consistently voters are not as kind. It wasn’t tuition and “Russell Brand” and “’Ed Stone”. underestimated the SNP and the power fees, it was the hypocrisy of the about- Words like “Sturgeon” and “pumped and allure of passionate politics. It was turn that did for the Lib Dems. What if, I up” and “green surge” (and, yes, “brain interesting for me, when I looked back ask myself, the Lib Dems had handled fade”). Do you see how, even reading over my sketches during the coalition coalition government differently, refusing those words now, they feel as faded as years for my new book All in this the lure of tuition fee increase? What charity shop rejects? together, how little Scotland featured. In if they hadn’t taken on the Sisyphean It was billed as the biggest election task of Lords reform? What if they had in a generation. It was going to be so focused instead on core vote issues close that all of us press at Westminster of helping the poorest in terms of tax, cleared our diaries until the end of Whatever schools, health, transport? The election May, assuming that we’d all be at work, could have looked very different. covering whatever painful version of the they say hung parliament that had been elected. Labour But then, because voters will do what now, David I interviewed Ed Miliband during they like, despite what the press says, his leadership election and listened it wasn’t really so close after all. The Cameron to his belief that Labour, by moving a election that was going to change the bit to the left, would attract back those world didn’t even change who was in had no idea voters it had lost to the Lib Dems. All of Number 10. his efforts over the next five years, and But we shouldn’t be too fooled by that he was during the election, were aimed at that, that. The fact is that the result was creating a world of moral capitalism a shocker. Whatever they say now, going to stay where Lib Dems would feel at home. But David Cameron had no idea that he in the process, he forgot to pay attention was going to stay at Downing Street. in Downing to the angry dispossessed who were Indeed, I rather think that Samantha was drifting towards UKIP and the furious already planning their next life phase. Street. Scots who were sick of being taken Ed Miliband reacted to his defeat like a for granted. Ed Miliband concentrated 20. On the campaign trail

on non-doms instead. He fought HIS to amuse (us) but George Osborne is We have yet election but this turned out to be the not so easily derailed. And, yes, I did say wrong one and he paid the price. Osborne. He’s the future leader of the to understand Tory party. I don’t need a crystal ball to The Tories see that. fully how It was striking for me how, during the coalition years, one man changed Ann Treneman’s new book, All the allure of almost everything: his persona, his In This Together, is published by appearance, his policies. George Robson Press at £14.99. She was the passionate Osborne began as a rather unpopular parliamentary sketchwriter for The Times (remember the Olympic booing?) from 2003 until September 2015. politics is artisto-Chancellor. He ended it as a man who invented the Northern Powerhouse, going to stealing Caesar’s hair-cut and Labour’s clothes. When the Tories announced that reshape the they were now the party of the workers, I admired the chutzpah but laughed. But political I had underestimated George Osborne who is determined not only to reinvent landscape, but himself but his party too. It strikes me, looking back, that underestimate David Cameron may remain in Number 10 but almost everything else has it at your peril. changed. All of our main parties are in the throes of change brought about by the coalition and the election. We have yet to understand fully how the allure of passionate politics is going to reshape the political landscape, but underestimate it at your peril, as the Labour Party has found out. No one ever imagined that the one with the beard was going to win. So what are my predictions for the future? The Labour party is, like a wonky shopping trolley, always turning in on itself. The Corbynistas are going to, at some point, have to deal with the world as it really exists. That should be fun. UKIP hope Europe will give them a surge. I wouldn’t be so sure. As for the SNP, Nicola Sturgeon may just win her war of independence: now that really will be a crisis (for the SNP). As for the Tories, events such as pig-gate continue 21. Ipsos MORI - Understanding Society May 2016

The digital election Steve Ginnis A new kind of debate? London

Figure ELEVEN.

Q: To what extent do you agree or disagree that social media platforms such as Facebook and Twitter…

…are giving a voice to people who would not User 79% 11% normally take part in Non-user 42% 25% political debate …are breaking down the barriers between voters User 56% 27% and politicians and Non-user 32% 33% political parties

…are making political User 57% 23% debate more divisive than it used to be Non-user 37% 23%

…are making political User 50% 29% debate more superficial than it used to be Non-user 50% 19%

¡ % Agree ¡ % Disagree

The jury is still out on the value of four million viewers sent 3,000 Tweets were reserved for Nigel Farage’s claim social media to the political debate. per minute into the Twittersphere. Over that the studio audience was biased.21 On the one hand, we believe that the course of the second debate, Whilst the televised debates social media platforms are giving a 239,000 Tweets either mentioned provided the biggest concentration voice to those who would not normally one of the party leaders or used the of social media activity during the take part, and are breaking down the debate hashtag to pass comment. campaign, there was also evidence barriers between voters, politicians The biggest ‘boos’ and ‘cheers’ in of a prolonged social media dialogue and parties. On the other, we are this new dual-screening commentary which was breaking down the barriers concerned that the same vehicles for seemed to map the peaks and troughs between MPs and voters. Our partners discussion are making political debate collected through the real-time ‘worm’ at Demos counted 479 MPs and 794 more divisive and more superficial Ipsos conducted on behalf of the prospective candidates with accounts than it used to be. So which is it? BBC.20 Whereas the worm asked a on Twitter, with 1.88 million Tweets Wisdom of the Crowd, a research carefully-selected few (50 participants exchanged between politicians and project by Ipsos, Demos, University per debate) to constantly rate what was voters in the weeks leading up to the of Sussex and CASM LLP, analysed being said during the programme using election. Perhaps surprisingly, 23% of social media engagement during the a keypad, social media gave a platform the 59,179 messages sent by MPs were 2015 general election and found that to thousands of voices who otherwise counted as ‘replies’ to messages from the sheer volume of discussion during would not have been heard. Across both other people, rather than just re-tweets the campaign was evidence enough formats of research, the biggest cheers or posts aimed at broadcasting policy.22 of a new type of political dialogue. in the second debate arrived when party Over the course of the first televised leaders joined together to criticise David debate, hundreds of thousands of its Cameron’s no-show, the biggest boos 22. The digital election

Ed Miliband and David Cameron received more Figure TWELVE. mentions per Comparison of coverage of events 20 – 26th April 2015. week than % coverage as proportion of total articles collected One Direction Entertainment 72% Entertainment 39% - even John Politics 28% Politics 61% 100% 6% 10% Major's speech 90% 5% 80% 11% warning of a 70% 10% 40% 7% Labour-SNP 60% 50% 25% deal got more 40% 21% mentions 30% 6% 5% 20% 3% 19% than Kim 10% 10% 9% 11% 0% Kardashian Election Unspun - UK online media Ipsos MORI - social media monitoring

¡ SNP manifesto launch ¡ Avengers movie premier in that time ¡ John Major speech on LAB/SNP ¡ Kim Kardashian ¡ Cameron's Aston Villa mix-up ¡ Robert Downey Jr Storms out of interview period, such ¡ Shapps' Wikipedia profile edits ¡ Clarkson's boss quits Top Gear was the social media Such was the social media a greater share of coverage to politics momentum, that over the last three over popular entertainment stories. momentum. weeks of the campaign our analysis However despite the magnitude showed that Ed Miliband and David of engagement, there are also signs Cameron each received more mentions that the role of social media, as an per week than One Direction; even John agent of political debate, was limited. Major’s speech warning of a Labour- Firstly, to what extent was social SNP deal got more mentions than all media really giving voice to those who discussions relating to Kim Kardashian otherwise wouldn’t have a platform? The in the same week. Compared with accounts mentioned the most in our data from coverage of the election by collection of conversation about David traditional news sources, collected by Cameron and Ed Miliband were the the Media Standards Trust in ‘Election accounts of the Daily Mail, Telegraph, Unspun’,23 social media appeared to give the i100, political party accounts, and 23. Ipsos MORI - Understanding Society May 2016

Figure THIRTEEN.

Response to television debates.

journalists. This suggests that social media was partly just an extension of the traditional political debate. Secondly, a review of the networks in which conversations took place on social media also suggests less evidence of true political debate between people with opposing views. In a thorough mapping of the Twittersphere, Demos were able to identify distinct galaxies of conversation between political ideologies. Although the Labour-leaning galaxy was by far the biggest in terms of both number of accounts and number of contributions, discussion 250 Cheers was confined between those who held 200 22:30 the same political views and rarely 150 took place at the pan-galactic level.24 100 In addition to the ‘who’ and the 'who- 50 between’, sceptics might also point to 0 the quality of content generated through -50 social media. Despite the vast number -100 of people Tweeting their views during -150 the live televised debates, we classified -200 just 11% of these comments as relating -250 directly to policy issues such as education, the economy or immigration. -300 Boos The vast majority of comments related -350 19:30 20:00 20:30 21:00 21:30 22:00 instead to the personality of the leaders, including comments about their ¡ Cameron ¡ Sturgeon ¡ Farage ¡ Clegg ¡ Bennett ¡ Miliband appearance and unprompted sledging. What did voters make of it all? At the height of the campaign, we asked members of the BBC’s 2,000 strong Election Uncut digital community for could be shared and believed that were concerned that conversation on their views on the role of social media social media content was subject to less social media was being dominated by a in the election. The active supporters bias and more honesty than the spin few loud voices, and that the tendency of social media in political debate portrayed in traditional media. However, towards humorous content made it had had positive experiences of using whilst it appeared to benefit some difficult to sustain serious debate. social media during the campaign. voters, others were disappointed by the They enjoyed the speed at which news evidence they had seen. These voters 24. The digital election

During the first televised debate, hundreds of thousands of its four million viewers sent 3,000 Tweets per minute into the Twitterspherse.

Active supporter: Those who had little or no direct Opponents in principle: experience of social media for political debate fell into two camps. Passive I have seen posts from friends supporters believed that social Social media is full of and colleagues from all sides media was the future and a key to egotistical, self-seeking of the political spectrum giving a voice to voters who might people who bend their party where there has been lively not otherwise share their opinions. In line to meet their followers/ "debate" and where video links contrast, others were opponents in friends. It is not a true have been posted to back up principle, and believed that political representation of what the discussions. It has been debate through social media will always people thing. It is extremely quite informative. be fundamentally unrepresentative, superficial. unhelpful and inappropriate.

Passive supporter:

Disappointed by the Many would not stand up in a It is clear that social media brought evidence: public meeting to air a point of great benefit to some voters who used it view, but would be able to do either to find information about policies it via social media. and candidates, keep informed with I think it 'dumbs' down the developments in the campaign, or debate when it becomes so engage directly with politicians. But it casual: it's very 'normal' for is also clear that these benefits were politicians to be slated, our not shared by all. Our analysis cannot country to be complained give definitive answers to the exam about etc, and social media question of how social media changes really provides the fuel for debate, but it does conclude that there this to happen. is some truth in all our perceptions of social media’s impact on political debate, both for good and ill.

25. Ipsos MORI - Understanding Society May 2016

We need to talk about Scotland Mark Diffley How Scotland voted and what it tells us about the future of British politics. Edinburgh

Support for Labour in Scotland has fallen in every UK General Election, bar one, since 1997.

Unlike in the rest of the UK, the question of Scotland’s constitutional In the end, Scotland returned story of the election in Scotland future would remain at the forefront 56 SNP MPs from our 59 seats, a was clear cut; the surprise was not of political debate here in Scotland. triumph that nobody could have that the SNP won but that it took Our poll in October 201425 gave predicted as the referendum polls so many seats in its landslide. the first indication of something closed on September 18th. The starting gun on Scotland’s extraordinary happening in Scottish On one level, the SNP victory was general election campaign was fired politics; that the referendum’s unsurprising; the 45% who voted ‘Yes’ in the early hours of September 19th losers, the SNP, were likely to be were always most likely to cast their 2014, when David Cameron stood on winners in the next big electoral vote just months later for the party the steps of 10 Downing Street and test, the 2015 general election. most identified with independence. announced a plan for ‘English Votes And from October 2014 all the way The electoral preferences of most ‘No’ for English Laws’ to run alongside through to the election the picture voters were split between the three the process of enhanced devolved remained stable and consistent. The main ‘unionist’ parties giving the SNP powers for the Scottish Parliament. question was not whether the SNP an inherent advantage. The fact that our Coming in the immediate aftermath would be the largest party in terms referendum polling regularly highlighted of the independence referendum, the of votes cast, but how many seats around 1 in 10 'No’ voters backing Prime Minister’s statement set the tone would turn yellow in the landslide. the SNP only enhanced the party’s for the election and highlighted that the advantage as the election approached. 26. We need to talk about Scotland

Our poll in October 2014 was the first to give Figure FOURTEEN. indication of Q: How would you vote if there were a General Election tomorrow? something

1% 6% extraordinary 3% 6% 2% 6% happening 10% 10% October 2014 April 2015 in Scottish (Certain to vote) (Certain to vote) 52% 54% politics; the

23% 20% referendum's losers - the SNP lead +31 SNP lead +34 SNP - were ¡ SNP ¡ Labour ¡ Conservative ¡ Lib Dem ¡ Green ¡ Other likely to be Voting intention in Scotland, October 2014 and April 2015 winners in the next big Figure FIFTEEN. electoral test. Q: Which party do you think has the best policies for Scotland?

48% 18% 9% Health and the NHS 4% 3% 1% 15% 49% 16% 10% Education 4% 2% 1% 17% 40% 14% Crime and anti 15% 4% social behaviour 2% 2% 23% 28% 9% 6% The environment 4% 36% 1% 16% 0% 50%

¡ SNP ¡ Scottish Labour ¡ Scottish Conservatives ¡ Scottish Lib Dems ¡ Scottish Greens ¡ Other ¡ Don't know 27. Ipsos MORI - Understanding Society May 2016

Figure SIXTEEN.

Q: How satisfied or dissatisfied are you with the way… is doing her job as On one level, first Minister / his/her job as leader of the XXXX party? the SNP +100 victory was +48% unsurprising; NET +31% NET +9% the 45% who +1% NET 0 NET -5% voted ‘Yes’ NET were always most likely to -100 cast their vote Nicola Sturgeon Kezia Dugdale Ruth Davidson Willie Rennie Patrick Harvie Satisfied 71% Satisfied 32% Satisfied 43% Satisfied 29% Satisfied 47% just months Disatisfied 23% Disatisfied 31% Disatisfied 34% Disatisfied 34% Disatisfied 16% later for the However, while the reasons for the she towers over her rival party leaders, party most SNP success in May were reinforced attracts plaudits from across the and heightened by the independence political spectrum and looks likely to identified with referendum, they have actually been remain Scotland’s leading political established over a much longer period. figure for some time to come. independence. Part of the story lies in the long- The roots of this decline are held advantages the SNP has in varied and complex. Having been public perceptions of key policies the dominant party in the coalition and the party leaders. Our most administrations in the first seven recent poll for STV highlighted the years of the new Scottish Parliament, advantage the SNP holds over its rivals the party now trails the SNP on all across each of the key policy areas important voter measures, resulting in in deciding voters’ preferences. its collapse at the general election. At the same time, the public And there is no sign of the decline continues to be highly satisfied with being halted in the near future. The the performance of First Minister challenges facing both the Scottish Nicola Sturgeon, further reinforcing leader Kezia Dugdale and UK leader the overwhelming dominance of her Jeremy Corbyn in attempting to party. With at least 70% of voters return lost voters to Labour’s fold regularly happy with her performance, are numerous and significant. 28. We need to talk about Scotland

As forecast, the 2016 Holyrood election27 produced a clear victory Figure SEVENTEEN. for the SNP, continuing the party’s dominance of Scottish politics. However, Labour's falling vote share in Scotland. the failure to achieve a second overall majority in the parliament means 50% that the party will head a minority 46 administration for the next five years. 45% 43 It is the Conservatives, rather than 42 40 Labour, who made the biggest dent 40% 39 in the SNP majority, winning a record

31 seats at the election and making 35% 34 it the dominant opposition force. All 32 of which means that the SNP will 30% 29 need to cooperate, negotiate and compromise with the other parties to 25% 24 ensure it achieves its manifesto goals. 20% It also means that the road ahead to 1997 1999 2001 2003 2005 2007 2010 2011 2015 a second independence referendum Date of Election may be a little less straightforward. ¡ % UK General Election ¡ % Holyrood Election

Support for Labour in Scotland has fallen in every UK General Election, bar one, since 1997.

29. Ipsos MORI - Understanding Society May 2016

References

1. http://www.britishpollingcouncil.org/bpc-inquiry-report/ 17. http://www.bbc.co.uk/news/election-2015-32633099

2. http://www.voxeurop.eu/en/content/news-brief/3461591- 18. http://www.newstatesman.com/politics/staggers/2015/10/ boost-grillo-italy-ungovernable what-do-focus-groups-make-jeremy-corbyn

3. http://www.israelelection2015.org/polls/ 19. Ipsos MORI Election Uncut community, commissioned by the BBC: https://www.ipsos-mori.com/election2015/ 4. http://www.theatlantic.com/politics/archive/2015/10/ electionuncut.aspx gallup-poll-2016-presidential-primary- fivethirtyeight/409531/ 20. http://www.bbc.co.uk/news/election-2015-32171041

5. http://www.ncpp.org/?q=node/1 21. http://www.telegraph.co.uk/news/general-election-2015/ politics-blog/11512203/Election-debates-Who-does- 6. http://transparency.aapor.org/index.php/transparency Twitter-think-is-winning.html

7. http://www.theguardian.com/politics/2015/dec/04/labour- 22. http://www.bbc.co.uk/news/blogs-trending-31766165 sweep-to-conclusive-victory-in-oldham-byelection 23. https://www.ipsos-mori.com/researchpublications/ 8. http://synapse.princeton.edu/~sam/wsj_4nov04_pollcalc- researcharchive/3571/Dave-and-Ed-vs-Zayn-and-Kim-K- followup.pdf who-are-the-social-media-winners.aspx

9. http://on.ft.com/1KeDdFx 24. https://www.youtube.com/ watch?v=5A9c6Yyw47k&app=desktop 10. http://www.nytimes.com/2014/11/12/opinion/the-worst- voter-turnout-in-72-years.html 25. http://www.slideshare.net/IpsosMORI/scottish-public- opinion-monitor-october-2014 11. Ipsos MORI Election Uncut community, commissioned by the BBC: https://www.ipsos-mori.com/election2015/ 26. http://www.bbc.co.uk/news/election-2016- electionuncut.aspx scotland-36251471

12. Ipsos MORI Political Monitor 2015

13. www.ipsos-mori.com/researchpublications/ researcharchive/3575/How-Britain-voted-in-2015.aspx

14. http://www.gallup.com/

15. https://www.ipsos-mori.com/researchpublications/ researcharchive/3563/Ed-Miliband-and-Nicola-Sturgeon- felt-to-have-most-improved-perceptions-over-the- campaign.aspx

16. http://www.standard.co.uk/news/politics/david-cameron- blasted-over-budget-omnishambles-7661054.html 30. References

FIGURE References 15. Source: Ipsos MORI Scottish Public Opinion Monitor. Base: c.1000 Scottish adults aged 18+, April 2015.

1. Source: Ipsos Italy, Ipsos MORI UK, Ipsos Sweden, 16. Source: Ipsos MORI Scottish Public Opinion Monitor. Ipsos Ireland, Ipsos Canada, Ipsos South Africa. Base: c.1000 Scottish adults aged 18+, April 2015.

2. Source: Ipsos MORI, 2016 17. Source: Ipsos MORI Scottish Public Opinion Monitor, 1997-2015. 3. Source: Ipsos Sweden, 2006-2016

4. Source: www.idea.int/vt/, April 2016

5. Source: Reuters / Ipsos Poll, September – October 2014

6. Source: Reuters / Ipsos Poll

7. Source: Ipsos MORI. Base: 1,000 British adults 18+, 12-15 April 2015. Split sample for Nick Clegg (507) and Nigel Farage (493).

8. Source: Ipsos MORI. Base: 10,227 GB adults (of which 5,992 were “absolutely certain to vote” or said they had already voted), interviewed 10 April – 6 May 2015. Data is weighted to final result.

9. Source: Ipsos MORI Political Monitor. Base: c1,000 British adults aged 18+ each month.

10. Source: Ipsos MORI Political Monitor, March 2015.

11. Source: Ipsos MORI Political Monitor/King’s College London. Base 1,010 British adults aged 18+, 8th – 10th February 2015.

12. Source: Ipsos MORI / Election Unspun. Comparison of coverage of events 20th – 26th April 2015.

13. Source: Wisdom of the Crowd project with following partners: Demos, University of Sussex, Centre for Analysis of Social Media, Ipsos MORI, The Telegraph. Analysis of Twitter during ITV’s Leaders’ debate, 2nd April 2015.

14. Source: Ipsos MORI Political Monitor, October 2014 and April 2015. Base: 1,002 adults aged 18+ across Great Britain. 31. Ipsos MORI - Understanding Society May 2016

32.

Information

Ipsos MORI 3 Thomas More Square London E1W 1YW t: +44 (0)20 3054 5000 www.ipsos-mori.com www.twitter.com/IpsosMORI

About Ipsos MORI's Social Research Institute

The Social Research Institute works closely with national government, local public services and the not-for-profit sector. Its 200 research staff focus on public service and policy issues. Each has expertise in a particular part of the public sector, ensuring we have a detailed understanding of specific sectors and policy challenges. This, combined with our methodological and communications expertise, ensures that our research makes a difference for decision makers and communities. Understanding Society is part of the Ipsos Views publications programme. For more information see www.ipsos.com/ipsos-views.