In defence of listening: Right-wing populism and the failure of political “science”

Posted on November 13, 2016


Like anyone who is likely to read this, I was horrified by Donald Trump’s election victory. I find the man and what he represents repugnant in ways I struggle to put into words. I don’t see how anything positive can come from his presidency and only hope it will prove to be less disastrous than his toxic campaign suggests it will. And yet the despair I felt on hearing the result was not accompanied by shock. For me that particular emotion has run dry in the last couple of years, as I have listened to political “experts” of various kinds consistently fail to foresee election results and looked on as the bounds of acceptable political discourse have expanded at an exponential rate. If you are shocked at this stage I would suggest it is because you haven’t been paying attention.

The failure of the traditional fonts of political knowledge – pollsters, journalists, political scientists and, of course, establishment parties and candidates – to grasp these events as they were unfolding speaks to a historic crisis in the epistemic basis on which we understand political change. There has been a radical destabilisation of the link between the way public opinion and electoral coalitions shift and the theories and methods we use to track this. The crisis is analogous to the discrediting of neo-classical economics in the wake of the 2008 financial crisis, when dominant models argued that such an event was, for all intents and purposes, impossible. As one analyst of the banks’ financial modelling systems put it:

Under a normal distribution, a -25 sigma event happens with probability once every 10−140 years, which implies that Goldmans (sic.) suffered a number of days (sic.) losses, each of which their models predict occurs approximately once every 14 universes, using the current estimate of the age of the universe of approximately 10−10 years old. (Daníelsson, 2008, Blame the Models, p. 3)

That is to say that the models financial experts based their analyses on were, to put it in the kindest possible terms, junk. The estimated probabilities of a “Remain” vote or a Clinton victory were not quite as high, but these predicted results were similarly treated as foregone conclusions by almost our entire class of political “experts”. The analytical frameworks they used to reach such a false consensus deserves to join neo-classical economics in the intellectual dustbin.

Unfortunately, as with the world of economics, a mix of institutional inertia, vested interests and the apparent lack of any better alternative have meant that dominant approaches to political data collection and analysis remain firmly in place. The power structures that surround these approaches will, inevitably, continue to promote them and to explain away their failures as aberrations. However, I intend to argue here that there is in fact a much better alternative. In simple terms, we must move closer to the social worlds that quantitative and theoretical political science purports, but largely fails, to represent. Not only will pursuing this alternative allow us to better understand the current political crisis, it may actually be the only chance for those of us on the left to influence its trajectory.


Let me first address where I think the problems of existing models lie. It is undeniable that we have a class of journalists and politicians who tend to come from a relatively narrow range of social backgrounds and live in geographically specific regions. By their very definition these groups and their assumptions are unrepresentative of the wider population. To the extent that mainstream politicians and journalists engage with voters beyond these social and geographic parameters, it is increasingly reminiscent of the archetypal colonial anthropologist reporting back authoritatively to their aghast and titillated peers on what the natives want. We can add to this the fact that over the last decade much of our political conversation – among both experts and non-experts – has shifted into social media echo chambers, and the result is a crisis of both communication and representation across social and ideological divides. People are simply not good at understanding and explaining how those who live in and/or get their news from different places think and feel.[1] This is not to say that this was necessarily any better in the past, but there were at least more more common reference points to anchor these relationships.

As important as the growing detachment of the mainstream political and journalistic class and the emergence of social media echo chambers are, however, these are secondary rather than fundamental causes of the epistemic breakdown. They are certainly concerning in and of themselves and have clearly served to amplify the crisis. However, I suspect that with a better underlying understanding of how opinion formation works it would be possible to maintain stronger channels of dialogue even across these highly flawed structures. That is to say the problem is not in its essence social or technological, but epistemological. It is a failure of the very way we understand ourselves as social and political creatures.

What I have heard again and again over recent years from most pollsters, journalists, political scientists and politicians, are analyses built upon three fallacies. Firstly, they make the classic mistake of confusing the concept of risk, which supposes the past can provide an accurate guide to the future, with uncertainty, which does not. Secondly, they believe in the inherent meaningfulness of clearly bounded demographic variables as predictors of political behaviour. Thirdly, they view the notion of a two-dimensional left–right axis of political positioning, with democratic institutions acting as an equilibrating mechanism, as a meaningful analytical construct.

It is to some extent understandable that people should maintain such assumptions because under “close-to-equilibrium” conditions they hold up relatively (though never perfectly) well. However, this does not make them effective analytical tools. As becomes clear during “phase transitions”, or moments of crisis, these always unstable connections quickly break down. During such periods, those who continue to cling to these epistemological assumptions find themselves with sophisticated models that no longer bear any resemblance to reality. As economist Steve Keen asks of the failure of neo-classical economics during financial crises, what is a point of a model that ceases to work at exactly the moment when you most need it?


To explain how such failures occur, let me tackle the first two fallacies together before moving on to the third. If you intend to use the past political behaviour of a particular, pre-defined “demographic” as a predictor of its future political behaviour, you will face a host of problems. For starters, it may no longer make sense to slice the group in the way you traditionally have if the objective conditions and/or subjective motivations of its members have shifted. In terms of its behaviour, this group may have splintered into several smaller ones or merged with another group that was previously separate from it. You will use the same label because you have learned to recognise the group as something stable and it may even largely contain the same people. However, for the purposes of meaningful analysis such a decision no longer makes sense. Then there is the problem that regardless of whether that group has or has not retained a certain coherence, it may have entirely new forces acting upon it. None of this is to say that the past has disappeared without a trace. However, very careful attention is needed to work out what of the past is still causally relevant.

One example of this would be much vaunted “Hispanic vote”, which after Trump’s relentless demonisation of Mexican immigrants was supposedly going to tilt the election in Clinton’s favour. Aside from the fact that the geographic distribution of this vote largely neutralised its potential effect, the expected swing in her favour did not occur. In fact, some 29% of Latinos and a full one in three Latino men voted for Trump – higher than the figure Romney achieved in 2012. There are very plausible reasons for why this occurred. “Latinos” originate from a large and diverse range of countries with different historical relationships to the US, and which are themselves highly stratified by social class, race and ethnicity. They have also been in the US for different lengths of time, ranging from fourth and fifth generation immigrants who don’t speak Spanish to the newly arrived who don’t speak English. Why should wealthier, whiter and/or more established Latinos necessarily be put off by Trump’s stigmatisation of recent and potential Mexican immigrants? This makes sense to anyone with the most rudimentary understanding of this population, and yet it is entirely obscured by the blunt tool of the “Latino” variable.

Another clear example is the fact that white women voted by 53 to 47 per cent for a shamelessly misogynistic male candidate facing numerous sexual harassment allegations, rather than a female candidate who has defended women’s rights relatively consistently throughout her political career. This is only surprising, however, if you assume that the undeniable significance of gender as an objective determinant of one’s social existence translates into it having a direct causal impact on the formation of political subjectivity. This is quite patently not the case for multiple reasons, which it is hardly necessary to list here. Suffice to say that without asking questions like, which women? where? holding which values? subject to which influences?, slicing the electorate using the variable of “female voters” conceals far more than it reveals about their likely electoral choices.


The examples of enormous “demographics” like Latinos and women failing to vote in homogeneous and predictable ways could be dismissed as an operational failure, rather than an epistemological one. It may not be that the approach itself is wrong, but simply that the wrong variables – insufficiently granular, or “intersectional”, ones – have been selected. There is some truth to this, although as I will argue below it does not really get to the heart of the problem. However, there is another aspect of the current political knowledge industry that is even more insidious and which has come under less scrutiny than the failures of the pollsters. This is the idea of the two-dimensional left–right axis as a meaningful construct for interpreting citizens’ engagement with the political system. (I have discussed this in another post on the rise of Jeremy Corbyn as leader of the British Labour Party).

The concept of the political spectrum is sometimes useful as a short-hand for describing, in the very broadest terms, the social and ideological coalitions that representative democratic systems demand in order to make government possible. There are, in most democratic countries, left- and right-wing parties with contrasting and broadly identifiable policy platforms, that seem to betray a certain symmetry as they move from the “centre” to the fringes. When political commentators discuss party-political strategy as though it were a chess game played on two-dimensional plain – where one party can “outflank” the other, or “pitch its tents on the other’s lawn” – they are espousing a particular causal model of how politics works. In this model, the political system and its in-built equilibrating mechanism of requiring parties to compete for voters firmly tethers the political system to society. It is the political science version of rational actor theory, with Homo Politicus at its centre.

In this model, voters of different kinds have relatively self-evident interests and/or preferences, and political parties compete to balance their offer across different parts of the electorate in pursuit of a workable majority. This was the ideology at the heart of the “centrist” governments that dominated the political systems of Europe and North America during the 1990s and 2000s, with their carefully calibrated, focus group-tested retail offers to voters. If a political party failed to meet the demands of a sufficient range of groups, the theory went, it would be punished at the ballot box and be forced to change or face electoral oblivion. Just as neo-classical economics argues that any attempt by government to act over longer timeframes and thus defy the natural, equilibrating mechanism of the market would face disaster, so proponents of the Homo Politicus framework reject the idea of pursuing an ideologically (and morally) coherent platform that, in their view, cannot meet the short-term demands of post-modern voters.

There are multiple problems with such a view. For the current discussion we can leave aside the obvious fallacies that voters pursue their own “interests” and have firmly established preferences. I will also skirt around the frankly absurd idea that governments that oversaw a broad secular trend across Western countries of wage stagnation and welfare state retrenchment, ballooning inequality, and endless foreign wars were “centrist”, except within the narrow confines of formal politics during this particular conjuncture. More important here is the fact that just as the false notion of self-correcting markets concealed and simply delayed a reckoning with the economic contradictions of capitalism, so the focus-group based politics of this period kicked its social consequences down the road. For a while it was able to hold these together by attempting to balance antagonistic pressures, gradually disappointing the majority of voters but not quite enough to dislodge politics from its close-to-equilibrium state. Eventually, however, these were unleashed to produce the centrifugal explosion we have seen in the last couple of years.


There are two important factors for understanding how this process has unfolded. The first is that both identity and opinion formation are gradual and non-linear processes that tend to evade the measures used by pollsters. Polling ossifies the representation of “public opinion” in frameworks that are “etic” rather than “emic”, ie. that reflect the worldview of the political system rather than of citizens. This produces the well-known problem that you can phrase the same basic question in different ways and get dramatically different answers. As this suggests, political subjectivity is not straightforwardly about policies, but is rather bundled up language and the way it constructs positives and negatives, moralities and immoralities.

The consequence of this is that the headline result of a polling question is far less important for predicting future voting behaviour than the interpretive framework on which the individual’s answer was based. Showing that X number of people are concerned about issue Y, does not tell you anything about why they are concerned, or even whether such concern is genuine. The particular issue, for example, might be one that affects a given population directly in very concrete ways. For another population it may be a key component of their worldview, without being one that they experience on a personal level. For another, the issue may have very little or no salience either empirically or morally, but is something that respondents have a vague normative sense that they should be concerned about. Only someone who has a grip on the way that a polling question is being interpreted within the lived reality of a given population is in a position to make a judgement on the meaning of its members’ answers and what it might mean for their future behaviour.

Related to this, and an extremely pertinent point for understanding the march of right-wing populism in 2016, is that polling data is particularly poor at capturing the worldviews of voters who are, in social and subjective terms, more distant from the institutions that gather and analyse the data. Pierre Bourdieu long ago explained that the most interesting respondents to polls are those who abstain altogether. These, he argued, are usually seen as a meaningless residual, but are in fact fundamental to understanding the functioning of representative democracy:

Political science long ago began to register the fact that a large proportion of the persons surveyed ‘abstained’ from answering questions on politics and that these ‘non-responses’ varied signicantly by sex, age, educational level, occupation, place of residence and political tendency. But no conclusions have been drawn, and the psephologists merely deplore this culpable ‘abstention’. As soon as one sees that the inert ‘don’t know’ category is largely recruited from what others call ‘the masses’ or ‘the people’, one begins to suspect the function it performs in the operation of ‘liberal democracy’ and the contribution it makes to maintaining the established order. Abstentionism is perhaps not so much a hiccup in the system as one of the conditions of its functioning as a misrecognized – and therefore recognized – restriction on political participation. (Bourdieu, 1984, Distinction, p. 398)

If we include the “don’t knows” along with those who respond to polls with largely normative, rather than materially- or conviction-based, responses, we start to see how they mislead us into a perception of equilibrium as the natural state of democratic politics. During phase transitions, these diverse “masses” flood both into and out of the system in ways that confound the logic of the Homo Politicus construct. It suddenly becomes clear that for years polls have been asking the wrong questions and therefore drawing the wrong conclusions. Throughout the pre-crisis period, identities and opinions were forming latently in ways the conventional quantitative and theoretical tools of political science were not able to pick up.

This was to some extent true of millennial voters during Obama’s first election (black voters in this case were more clearly detected) and during Bernie Sanders’ and Jeremy Corbyn’s respective leadership campaigns, and it was true of white, blue-collar voters in the Brexit and Trump votes. It was also true in the failure of white women and Latinos to vote in sufficient numbers for Clinton, and of black and young voters to turn out in sufficient numbers. In these contexts, analysts using deductive quantitative and theoretical tools have found themselves completely divorced from the realities they thought their models were capturing.


All of this brings me to my main point. The widespread shock at these election results – and, as I will suggest below, perhaps even the results themselves – are the product of a systemic failure to listen. Crude categories and tools, divorced from the lives of real people, failed to capture dramatic, generational shifts in identities and opinions. I would argue that such a dramatic misreading of these trends would not have been possible if qualitative research, drilling down into the meaning of abstentionism and “normative” participation in the political system, had more successfully influenced the conversation. There should have been armies of qualitative researchers non-judgmentally listening in detail to those people who are furthest from the political system. More importantly still, there should have been armies of quantitative researchers, pollsters, political scientists and journalists reading this work and using it inform their own analyses. At the very least we might then have at our disposal variables that prove to have some explanatory power at the points when we most need them.

This sounds like a fairly trite argument. Indeed, one might be tempted to file it away with the age-old debate over the relative merits of quantitative vs qualitative social science and deductive vs inductive reasoning. Of course, qualitative research has its own clear limitations – it is more difficult to scale up to produce “big-N” conclusions, it is subject to the seemingly less scientific process of inductive interpretation, the individuality of the researcher and her unacknowledged biases more obviously interfere with the data collection process etc. I will not delve too deeply into these well-trodden debates, however I would like to make two important points. Firstly, I am not saying that qualitative research should replace quantitative and theoretical approaches. Rather, I argue that it should inform them far more than it does. The effortless superiority of social science disciplines which make stronger claims to being “scientific” excuses them from engaging meaningfully with qualitative work that would hugely strengthen the basis on which they develop their models and indicators.

My second point is that anyone who has experience of conducting qualitative research will know that it is far less random that it is often given credit for. There is a familiar, somewhat cathartic feeling that qualitative researchers experience once they have reached saturation point in a particular field. Once you speak in depth to enough people within a given population you exhaust the range of possible viewpoints on a given issue that are present within it. These positions will inevitably be diverse, but they will also have certain regularities that relate to the broader way in which the population is inserted into wider society. Mapping these variations allows us to understand the internal differences of populations that from a distance may appear homogeneous, while also helping to identify the bases (material, moral, normative etc.) on which different answers are given. I, myself, spend my days talking to people who are far from the political system, many of whom feel unqualified to discuss political topics. When they do, they often resort to vague platitudes about politicians all being the same and all being corrupt. However, these people also have broader material needs and social subjectivities that determine the space of possibilities in which their engagement with the political system might in the future unfold, whether engagement remains marginal or, at some point, erupts onto the stage in a populist convulsion.

Understanding political subjectivity as a space of possibilities is important because it highlights the fact that the regularities we observe in the world, including in the political behaviour of different populations, have structural bases but are radically undetermined. We have been trained to view different “demographics” as static and homogeneous, with those that defy the norm explained away as anomalies. However, depending on their social reality members of these different groups have a constrained, but nonetheless diverse range of possibilities open to them. White citizens are more susceptible to being seduced by anti-immigrant demagoguery, not least because they and those they know are far less likely to be on the receiving end of its destructive logic. However, some parts of the white population may be more susceptible than others to such an agenda, and no individual, regardless of their social condition, is inevitably so. Furthermore, such an outcome also exists within the space of possibilities for a significant part of the “Latino” population, contrary to what deductive reasoning, shorn of context, might tell us.

The extent to which these different possibilities are actualised is significantly influenced by the effectiveness with which political messages are constructed and solidarities, or at least identification, established across demographic boundaries. Donald Trump was able to draw significant numbers of Latino men and white women into a platform built around a wounded nationalism, while Clinton failed to actualise the potential of drawing sufficient numbers of white working-class men and women into a “progressive” coalition. (Obviously, this refers specifically to the swing voters of the rustbelt. The majority, white, middle-class Trump voters would have had a narrower margin within their possibility space for being drawn into a progressive bloc). This is, of course, partly the result of structural conditions favouring such an outcome – the vagaries of the electoral college system, the widespread disenfranchisement of minority voters, the chronic failure of the media to scrutinise Trump, the psychological effectiveness of scapegoating the powerless as opposed to mobilising positive or rational arguments. However, the Remain and Clinton campaigns also represent historic failures to forge solidarities and develop collective responses to epochal challenges that could cut across many of these demographic boundaries.


The current discussion has primarily focussed on the failures of mainstream political analysts and their use of supposedly neutral quantitative and theoretical tools. However, I would like to end by identifying some important implications of this for those on the left hoping to influence the future direction of politics across the Western world. One of the most depressing aspects of the “Leave” and Trump victories is to see a traumatised left fragment across familiar fault lines. Our social media echo chambers have exploded with accusations and counter-accusations that broadly map onto entrenched and antagonistic economistic and identitarian tendencies. On the one hand, there are those who interpret the Leave and Trump votes as misguided expressions of legitimate grievances by white working-class communities “left behind” by neoliberal globalisation. On the other, there are those who view the vote as an expression of misogynistic and racist revanchism against a diversifying and liberalising society where white male “privilege” has been challenged.

I do not claim to have the answers to these complex questions. However, it seems unnecessary to me to reduce the debate to such a stark and polarised choice. Indeed, I believe the question is probably being framed in the wrong way. I find it striking that while these debates are far more theoretically refined than the banal mainstream conversation, such arguments on both sides frequently end up falling back on the same kinds of categories and etic frameworks used by the pollsters. They are theoretical constructs that contain power structures and variables, but not people. I certainly rarely, if ever, hear these debates providing detailed accounts, let alone directly involving, those people – for Bourdieu, “the people” – who are being defended or judged. I emerge from such debates with no better understanding of what it means to be a “member” of one of these groups. How does their evolution as political subjects unfold, so that they move from the “don’t know” category into voting for Trump, or remain in the “don’t know” category despite being targeted by him? I do not hear satisfactory answers to these questions on the left any more than in the mainstream.

In opposition to such approaches, I would like to make the case for listening. Some view calls to listen to anyone who could vote for a Trump – or, to turn the positions around, anyone who could justify a terrorist attack or participate in an urban riot – as giving into or being passive in the face of hate or violence. However, contrary to common misperception, listening is not a passive activity. When we listen properly we try to make sense of where someone else is coming from. The overcoming of such radical alterity is difficult and perhaps, ultimately, impossible. But in attempting to do so, we change slightly. We humanise the speaker and may, in the process, become more human ourselves. This dialogue also expands the space of possibilities. Those who are listened to have a better chance themselves of learning to listen and of changing, and as a result new subjectivities and solidarities may emerge. These processes are slow, messy and, as often as not, unsuccessful. Not everyone is capable of change. In times of fear many decent people will seek refuge behind the protective walls of class, gender, race and citizenship. If we are honest with ourselves we will recognise that most of us are capable of doing so. Perhaps, to some extent, we already do.

For these reasons, it is more important than ever that we maintain and expand dialogue across the sharpening divides in our troubled societies. We face a moment where structural conditions appear to favour fascism – the politics of division – over both liberalism and socialism. We cannot afford to dismiss potential allies because they do not (yet) fit neatly into theoretically informed political positions that will, inevitably, have selected and excluded different facts from the messiness of the social world. Perhaps most importantly, we can’t expect others to change if we’re not also open to changing ourselves. After the setbacks of this dreadful year, the progressive cause may now need to start again from scratch. Let’s move forward by listening to one another and to those we often ignore, and by trying to grow our space of possibilities together.

[1] This is not a claim of equivalence across the political spectrum. I entirely agree with Simon Wren-Lewis that unchecked mistruths tend to circulate far more in right-wing echo chambers than in liberal and left-wing ones, but that doesn’t mean that the problem is not broader than this. In fact, I would argue that although they may generate more objectively reasonable interpretations, left-wing and centrist bubbles are no better at circulating information that helps them to understand other people’s lives in meaningful ways.
Posted in: Elites, Politics, US