Can one tweet change a person’s mind? No, but a barrage of them might

To save time, I need to build a bot that finds the appropriate xkcd for a post.
There’s lots of annoying responses going around to the issue of social media bots, but one of the most annoying to me is the canard of ‘it’s all stupid, they can’t be influential because who’s going to change their mind on how to vote after seeing one tweet or Facebook post?’ It particularly irritates me because it’s conflating two bad ideas together. First, the idea that our political opinions are fixed and immutable things, near impossible to change, and secondly, ignoring that the sheer quantity of content produced by bots is important in itself.

The first issue is something I wrote about in more detail a while ago but in short if your view of people’s political opinions is that they’re entirely fixed, you’re generally wrong. Part of the problem here is that people with strongly held and developed political opinions of their own tend to assume everyone else is similar to them in their access to and desire to use information on which to develop their opinions. It’s people with high information levels about a subject assuming that everyone else has the same level of information, when in fact most people’s opinion towards most things is what Converse called a ‘nonattitude’, having so little knowledge of an area that any opinion they’re asked to give it on it is effectively random. Almost everyone has some issues on which they’re high information individuals – for some of us it’s politics, for others it’s sport, some know lots about art or cooking or construction or any number of issues where high information individuals might have huge disputes about something, while the rest of us would probably stare blankly at anyone who asked us for our opinion on it.

So while people might have an opinion built up over time about which party they support in political matters, they tend to gloss over the detail of policy, which means that when they’re asked to decide on a specific policy in an environment where their traditional party cues are meaningless – through the medium of a referendum, say – positions are going to be a lot less strongly held and much more subject to change. Our opinions on most issues aren’t a neatly organised filing cabinet, neatly organised and fully cross-referenced, they’re much more like piles of paper strewn across numerous shelves from which we tend to grab whichever looks best at the time. We can have multitudes of different considerations hanging around there, waiting for us to find the right context to consult them in and weigh them against each other.

This is why it is silly to say that one tweet can change the way someone thinks on an issue, in just the same way that it’s unlike that any one leaflet, poster, TV ad, newspaper article, or even blog post might decisively change someone’s mind for good. The aim of anything like that isn’t to change your mind there and then, but to get into your consciousness enough to form a consideration that you will later form an opinion on. Again, this isn’t about seeing something, going to your mental filing cabinet, pulling out a folder and rewriting it, but rather seeing something and sticking it on a shelf for future reference. One consideration alone might not be enough to make you act in a certain way, but a bunch of them developed over time and coming from a number of different sources could well be. It’s why companies don’t just broadcast their adverts once and figure everyone will see it, why billboards remain in place for weeks not hours, and why political parties deliver rainforests worth of paper during election campaigns. It’s not about changing minds so much as it is about getting people to have a balance of considerations that’s favourable to you in their head when they’re asked to make a decision – and once they’ve made that decision, human nature often means they’ll tell themselves that was their opinion all along, and there wasn’t any way they could have changed it…

No, polling companies aren’t trying to turn us into fascists

There was a minor social media storm yesterday evening when some people shared a question that YouGov are currently asking in one of their surveys. People were asked on a scale from strongly disagree to strongly agree how much they agreed with the statement “the best way to run the country would be to have a strong leader who does not have to bother with parliament or elections”.

Now, seen on its own that question might seem odd and somewhat scary, but its not uncommon to see questions like that asked in political science research. Indeed, as Chris Hanretty reminded me in a Twitter discussion, the specific wording of that question is taken directly from the World Values Survey. The WVS is a project that’s been running since the 1980s, asking people in many different countries their views on a lot of different issue,s including politics, which then gives social scientists (including political scientists like me, as I’ve used WVS data in my PhD research) a useful data set of comparative information about opinions in different countries. By asking the same questions of people in different countries, and asking those questions repeatedly in different waves of the survey over the decades, we can find out a whole lot of things about how people’s attitudes are similar or different over the world. The WVS website has details of all the questions asked in the different waves, and also has an online analysis tool where you can look at the data yourself. (Or, if you’re the sort of person who likes to analyse data in even more depth and has a stats package on your laptop, you can download the data and analyse it yourself. More of that later.)

The ‘strong leader’ question is one of four asked about people’s opinion on the political system, amidst a wider section on political beliefs and actions. People are asked to indicate if they strongly agree, agree, disagree, or strongly disagree with them (unlike YouGov, there’s no neither option). Here’s the text in full (from the Wave 6 (2010-14) questionnaire):

I’m going to describe various types of political systems and ask what you think about each as a way of governing this country. For each one, would you say it is a very good, fairly good, fairly bad or very bad way of governing this country?
V127. Having a strong leader who does not have to bother with parliament and elections
V128. Having experts, not government, make decisions according to what they think is best for the country
V129. Having the army rule
V130. Having a democratic political system

(I’d be interested to know if YouGov asked the other three questions – the formatting of the question in the image suggests it was one of several)

As I’m one of those people with a stats package on their laptop, I’ve processed the answers to those questions and sorted each one by the countries in the WVS (from Algeria to Yemen) so you can see the results for yourselves. Not to sound too clickbaity, but some of the results may surprise you. One of the purposes of this sort of research is to look at the political norms in each country, giving us a chance to compare them and see what’s actually going on beneath the surface. These questions aren’t just being asked in a bunch of similar democracies, but across a range of different types of government, so they let us see how the type of system you live in affects your opinion on different ways of running the country.

The point of the World Values Survey (and any other competently done survey or opinion poll) is to attempt to get to people’s real opinions, not just the ones they publicly express because they’re socially acceptable. Framing of the question matters too. If you ask someone ‘are you a fascist?’ they’ll most likely say no because fascism is generally seen as a bad thing, but ask them the ‘strong leader’ question and they may give you a different answer that comes closer to their real opinion. (With the disclaimer that describing something as a ‘real opinion’ is making a bunch of assumptions – see this post I wrote a while ago for more on how people form opinions) If you want to find out how many people would support dictatorship, you need to frame the question and the survey in a way that gets them to give that honest opinion. Ask people if they’d electrocute a stranger because someone in authority told them to and they’ll probably say they wouldn’t, put them in a situation where they have to do that and they might.

This is not YouGov trying to prepare the ground for a fascist takeover, it’s researchers (and I don’t know who commissioned it – it may be part of WVS Wave 7, it may be someone borrowing their question format for something else) trying to find out what people genuinely think about an important political issue. We know that there’s an authoritarian trend in many different countries, and if you want to counter it, surely it’s important to see how widespread genuine support for it is? The question about a strong leader isn’t what should be worrying you, it’s how people answer it.

Worth Reading 187: Californian murder

Workplace coercion – “Why are people who profess to be classical liberals apparently so indifferent to workplace tyranny?” asks Chris Dillow.
Why the 2016 Election Will Be One of the Most Pivotal Moments of Our Time – It’s fun to laugh at the clown car the US election resembles at the moment, but there are important things at stake.
Claims of the increasing irrelevance of universities are ideology masquerading as evidence – Three political science professors present “the difference between lazy journalism and quality social science research: the former pontificates, the latter take the time and trouble to create and test the evidence.”
In support of a universal basic income: introducing the RSA basic income model – The RSA has come out in favour of a basic income and presents a very interesting model of how it could be implemented.
Do candidates dream of electric sheep? – This is the age of the American uncampaign, where candidates routinely evade election laws through ridiculously funded Super PACs that are ostensibly not part of their official campaign.

Closing the Overton Window

A window, which has been shown to actually work
A window, which has been shown to actually work
It’s become quite common for political pundits and online commentators to talk sagely about the “Overton Window” when they’re talking about public opinion. (Click here for a sampling of recent uses of it) It’s presented as an immutable law of political science – there’s a range of possible policies available on any given issue or set of issues, but only a certain amount (those within the Overton Window) are possible/politically deliverable at any given time. Thus, to make something politically possible that’s currently not, one must shift the window somehow.

There’s always a general vagueness about just how it works, but that the the Overton Window exists is taken as read by a lot of people, and I’m sure most people who cite it believe it’s an idea that’s been discussed by academics over many years and has been rigorously tested and proven to be correct. You might think that, but you’d be wrong. Here’s a search for it with Google Scholar, which indexes millions of pieces of academic research and writing, which brings back no serious academic references to it. It does get mentioned in a few Glenn Beck books, some obscure libertarian pamphlets and in publications from the think tank it emerged from, but no serious academic attention has been paid to it.

Why is that? I’d propose that the most obvious explanation is that the Overton Window isn’t actually a theory, it’s just a vague assertion. See, for instance, this article about it (pdf) which starts from the assumption that it exists and goes from there. Any actual theory of political science has to be testable, and if the Overton Window was a theory it would come with a set of testable predictions about public opinion and how it shifts over time. It would also come with justification for some of the simplifying assumptions it appears to be based on, most notably that all the policy options for a certain issue can be placed on a straight line. None of that exists for the Overton Window, and none of its advocates appear to have put any effort into trying to prove it, preferring to assert its truth without testing.

I’ve written before about Zaller’s Receive-Accept-Sample model of public opinion, and I think it provides an interesting counterexample to the Overton Window in that it’s a genuine theory which has been thoroughly tested and shown to make accurate predictions about how public opinion works. However, as it’s slightly more complex than ‘you can do some things, you can’t do others, but if you want to do the others then sizeable donations to our think tank might help make them possible’ it doesn’t get mentioned much by supposed experts on public opinion in the media.

This is a shame, because I think Zaller’s work on how elite opinion leads public opinion (in The Nature and Origins of Mass Opinion) is a much more useful concept than talk of windows. Zaller shows that when elites (politicians, media and other opinion formers) are generally agreed on an issue, public opinion follows suit. Public opinion on an issue only begins to shift when elite opinion begins to divide and differ. To use Zaller’s example (though there are many others in the wider literature), American opinion on the Vietnam War was mostly in favour of it until newspapers and politicians started raising questions about it, and then it began to fade. (That’s a bit of a simplification, but the key point is that public opinion tends to follow elite opinion, not vice versa) If there’s not an elite consensus, then public opinion will tend to cover a much wider spectrum.

Public opinion is a much more vague and nebulous thing than an idea like the Overton Window gives credit for. In truth, most people have what Converse called ‘nonattitudes’ to most issues: they’re capable of giving an opinion on something if asked, but that opinion is often completely random based on what they’ve seen and heard recently rather than a fixed and carefully considered outcome of rational thought. Trying to analyse something so amorphous using concepts that are fundamentally weak is almost a guarantee of misunderstanding what’s actually happening. Public opinion can be shifted much more quickly and much more widely than people think but the key to doing that is finding messages that reach people and become part of their considerations, not obsessing over the location of mythical windows.

2015 General Election Day 29: My intentions are getting firmer

Let’s start with a couple of tales from the doorsteps, and see which one you find the most amusing. Both of them are from Devon, so maybe they do elections differently down there. First, we have the Greens of Exeter with a complaint:

Maybe I find this more than amusing than most, given that the Greens called on me back when I was first a candidate and also called on me the other week, but if you’re ‘appalled’ at someone knocking on a door, it feels to me that someone’s outrage-o-meter has been set a little too sensitively.

Meanwhile, in Torbay, the local Conservatives found all that door-knocking such a bother when they last did it in 2011 that they’re not going to bother doing it again this year. Not doing any canvassing is a model adopted by many candidates, few of whom are successful and even fewer of whom tell their electorate that they’re not doing it. We shall have to wait and see what success this approach bears for them.

On a wider note, I do think the model of canvassing that most parties use is badly broken, especially the further out from an election it’s used. It’s always good for politicians to get out on the doorsteps and talk to people, but expecting people to have firm political identities that an be recorded and treated as fixed is a mistake, in my opinion. The idea that you can define large chunks of the population as being definite supporters of any party isn’t backed up by any of the current data on how voters see themselves (see Elections and Voters in Britain for a lot more on this). As with so many things in our politics, a lot of canvassing rests on assumptions made in the mid-20th century that aren’t reflective of how people are now.

One other thought on voter intentions that might be of interest. A few weeks before the election, I went to a presentation by Chris Hanretty (one of the people behind Election Forecast) explianing their model. It’s assumptions follow polling trends from previous elections where for a long period in the run up to polling day, past electoral performance is as important as current opinion polling. It’s not quite as simple as taking an average between the two and calculating a swing – there are lots of weightings and demographic data in their model that are important – but one point he made is important: from around ten days out, current polling becomes a much more important part of the prediction than past performance. If that holds then we would expect to see increases in the Tory and Lib Dem shares in polls over the next ten days, while Labour and UKIP fade off while their overall prediction stays roughly the same. However, if the polling stays around its current level, then we’ll likely see opposite changes in the prediction.

However, for those of you wondering about the accuracy of the different focusing models, I must arn you of a potential flaw in May 2015‘s. As I’ve mentioned before, my department are having an election prediction competition and if May 2015’s current prediction is the final result, then I’d win the contest (and £200). This implied accuracy of my predicting skills is something you might want to take into account while assessing different forecasting models and websites. (For comparison, I’m 11th of 37 based on Election Forecast and 5th against Elections Etc’s current numbers)

Today’s minor party focus breaks from the order of the list in response to a request from Therese on Twitter who wants to know more about the Hoi Polloi who are not so much a party as one person’s description of themselves. That person is Geoff Moseley, a cinematographer and he’s standing for Parliament as the vanguard of a peaceful revolution, wanting to stop politics being just “the current dichotomy of power rocking back and forth between the left and right wings of the same deranged bird”. Beyond that webpage, though, there’s very little about him which implies that the revolution will be a very peaceful one.

Today on Election Leaflets, we have the first sign of an organised anti-SNP tactical voting campaign on the ground with this leaflet from Scotland In Union. How many of them there are being delivered, I don’t know, and their recommendations for who to vote for on their website seem to be based more on bookies’ odds than polling data, but it will be interesting to see if there is tactical voting in Scotland as part of the wider shift in voting intention there that seems likely to show up there next week.

There are just ten days left. I’m starting to feel that I might just find something to write about every day until then…

Receive-Accept-Sample: How people form opinions

the-nature-and-origins-of-mass-opinionAs we’re now into the election campaign, the entire purpose of which is to get people to form a certain set of opinions and then act on them on May 7th, I thought it was about time I went back to doing another post on a concept from political science that seeks to explain how opinions are formed.

What I’ll be looking at in this post is John Zaller’s 1992 book The Nature and Origins of Mass Opinion, and the model for public opinion he sets out in, which he terms the Receive-Accept-Sample or RAS model. This is a widely accepted model of how public opinion formed, but not universally agreed upon, and it’s also worth noting that it was published in 1992, so before any widespread use of the internet.

I’m going to try and explain the model as simply as I can, but remember that this is a 300+ page book with lots of charts and tables so Zaller’s arguments are a lot more complex and subtle than the precis of them I’ll provide here. The book is worth reading, or you can also try Drew Westen’s The Political Brain, which takes a more psychological approach to public opinion, but is more up to date and discusses Zaller along with other approaches. (I wrote about it here)
Continue reading Receive-Accept-Sample: How people form opinions

Worth Reading 157: Carthago Delenda Est

What ISIS Really Wants – Long but very good piece about the history, ideology and theology of the Islamic State.
The national interest demands it, let’s ban golf courses – If you think solar farms are a waste of productive land, why do you not protest about an even bigger waste?
The Austerity Con – A good explainer of the situation by Simon Wren-Lewis in the LRB.
How I became an erratic Marxist – Having just been writing an essay on Marx for my MA, I found this piece by Yanis Varoufakis fascinating, but I think it’ll be of general interest too.
Two Polarities of Anti-Politics: why trying to be friends with both Ukip and Green supporters won’t work for the mainstream parties – Interesting research from Southampton University on what drives support for UKIP and the Greens.

Worth Reading 139: Hadrian’s tomb

Russell Brand and our political culture – Chris Dillow argues that Brand gets publicity because our political culture as a whole is anti-intellectual.
Stuffing envelopes and getting stuffed – An alternative take on Liberal Democrat campaigning by Alex Harrowell.
The Laborers Who Keep Dick Pics and Beheadings Out of Your Facebook Feed – the realities of social media content monitoring.
The world will change around 2020 – According to David Boyle, that’s what the trends are pointing to.
Profs Bumble Into Big Legal Trouble After Election Experiment Goes Way Wrong – This is why conducting political science research is hard. However, I do hope the researchers involved are adding up all the news stories about them as ‘instances of our research methodology being cited in public discussion’.

Worth Reading 127: Still with both arms

How political science conquered Washington – Relevant to my interests and things I’ve talked about before: how political commentary in the US is taking more notice of academic research.
Victim-blaming: an all-pervading curse – How a culture of blaming the victim lets the real culprits off the hook.
How Jim fixed it: the strange, dark life of Jimmy Savile – Rachel Cooke’s New Statesman review of Dan Davies’ book on Savile’s life.
Chicken – Flying Rodent on how ‘human rights’ and ‘political correctness’ are handy shields to hide behind when you’ve failed at doing your job and want people to not blame you.
No Name – A superb piece of writing from Jack Graham on the Ripper murders and how coverage of them has ignored the women.

And as a more general recommendation, Justin McKeating is back writing on the web again. Go read.