Tag: Brexit

Cognitive ability and voting ‘leave’ on Brexit

In a new study published in the journal PLoS ONE on November 22, a pair of researchers from the University of Bath in the UK have reported that “higher cognitive ability” is “linked to higher chance of having voted against Brexit” in the June 2016 referendum. The authors have reported this based on ‘Understanding Society’, a “nationally representative annual longitudinal survey of approximately 40,000 households, funded by the UK Economic and Social Research Council”, conducted in 12 waves between 2009 and 2020. The researchers assessed people’s cognitive ability as a combination of five tests:

Word recall: “… participants were read a series of 10 words and were then asked to recall (immediately afterwards and then again later in the interview) as many words as possible, in any order. The scores from the immediate and delayed word recall task are then summed together”

Verbal fluency: “… participants were given one minute to name as many animals as possible. The final score on this item is based upon the number of unique correct responses”

Subtraction test: “… participants were asked to give the correct answer to a series of subtraction questions. There is a sequence of five subtractions, which started with the interviewer asking the respondent to subtract 7 from 100. The respondent is then asked to subtract 7 again, and so on. The number of correct responses out of a maximum of five was recorded”

Fluid reasoning: “… participants were asked to write down a number sequence—as read by the interviewer—which consists of several numbers with a blank number in the series. The respondent is asked which number goes in the blank. Participants were given two sets of three number sequences, where performance in the first set dictated the difficulty of the second set. The final score is based on the correct responses from the two sets of questions—whilst accounting for the difficulty level of the second set of problems”

Numerical reasoning: “Participants were asked up to five questions that were graded in complexity.The type of questions asked included: “In a sale, a shop is selling all items at half price. Before the sale, a sofa costs £300. How much will it cost in the sale?” and “Let’s say you have £200 in asavings account. The account earns ten percent interest each year. How much would you havein the account at the end of two years?”. Based on performance on the first three items, partici-pants are then asked either two additional (more difficult) questions or one additional (simpler) question”

On the face of it, the study’s principle finding, rooting the way people decided on ‘Brexit’ in cognitive ability, seems objectionable because it’s a small step away from casting an otherwise legitimate political outcome – i.e. the UK leaving the European Union – as the product of some kind of mental deficiency. Then again, in their paper, the authors have reasoned that this correlation is mediated by individuals’ susceptibility to misinformation, that people with “higher” cognitive ability are better able to cut through mis- or dis-information. This seems plausible, and in fact the objectionability is also mitigated by awareness of the Indian experience, where lynch mobs and troll armies have been set in motion by fake news, with deadly results.

This said, we must still guard against two fallacies. First: correlation isn’t causation. That higher cognitive ability could be correlated with voting ‘remain’ doesn’t mean higher cognitive ability caused people to vote ‘remain’. Second, the fallacy of the inverse: while there is reportedly a correlation between the cognitive abilities of people and their decision in the ‘Brexit’ referendum, it doesn’t mean that pro-Brexit votes couldn’t have been cast for any reason other than cognitive deficiencies. One Q&A from an interview that PLoS conducted with one of the authors, Chris Dawson, and published together with the paper makes a similar note:

Some people might assume that if Remain voters had on average higher cognitive abilities, this implies that voting Remain was the more intelligent decision. Can you explain why your research does not show this, and what misinformation has to do with it?

It is important to understand that our findings are based on average differences: there exists a huge amount of overlap between the distributions of Remain and Leave cognitive abilities. We calculated that approximately 36% of Leave voters had higher cognitive ability than the average (mean) Remain voter. So, for any Remain voters who were planning on boasting and engaging in one-upmanship, our results say very little about what cognitive ability differences may or may not exist between two random Leave and Remain voters. But what our results do imply is that misinformation about the referendum could have complicated decision making, especially for people with low cognitive ability.

The five tests that the researchers used to estimate cognitive ability (at least in a relative sense) are also potentially problematic. I only have an anecdotal counter-example, but I suspect many readers will be able to relate to it: I have an uncle who is well-educated (up to the graduate level) and has had a well-paying job for many years now, and he is a staunch bhakt – i.e. an uncritical supporter of India’s BJP government and its various policies, including (but not limited to) the CAA, the farm laws, anti-minority discrimination, etc. He routinely buys into disinformation and sometimes spreads some of his own, but I don’t see him doing badly on any of the five tests. Instead, his actions and views are better explained by his political ideology, which is equal parts conservative and cynical. There are millions of such uncles in India, and the same thing could be true of the people who voted ‘leave’ in the 2016 referendum: that it wasn’t their cognitive abilities so much as their ideological positions, and those of the people to whom they paid attention, that influenced the vote.

(The reported correlation itself can be explained away by the fact that most of those who voted ‘leave’ were older people, but the study does correct for age-related cognitive decline.)

The two researchers also have a big paragraph at the end where they delineate what they perceive to be the major issues with their work:

Most noticeably, the positive correlation between cognitive ability and voting to Remain in the referendum could, as always, be explained by omitted variable bias. Although we control for political beliefs and alliances, personality traits, a barrage of other socioeconomics factors and in our preferred model, house-hold fixed-effects, the variation of cognitive ability within households could be correlated with other unobservable traits, attitudes and behaviours. The example which comes to mind is an individual’s trust in politicians and government. Then Prime Minister of the UK David Cameron publicly declared his support for remining in the EU, as did the Chancellor of the Exchequer. The UK Treasury published an analysis to warn voters that the UK would be permanently poorer if it left the EU [63]. In addition to this were the 10 Nobel-prize winning economists making the case in the days leading up to the referendum. Whilst cognitive ability has been linked with thinking like an economist [64,65], Carl [51] also finds evidence of a moderate positive correlation between trust in experts and IQ. Moreover, work on political attitudes and the referendum have shown that a lack of trust in politicians and the government is associated with a vote to Leave the EU [56]. Therefore, the positive relationship between cognitive ability and voting Remain could be attributable for those higher in cognitive function to place a greater weight on the opinion of experts. A final note is that our dependent variable is self-reported which may induce bias, for instance, social desirability bias. Against that, the majority (75.6%) of responses were recorded through a self-completion online survey and we do control for interview mode, which produces no statistically significant effects.

It’s important to consider all these alternative possibilities to the fullest before we assume, say, that improving cognitive ability will also lead to some political outcomes over others – or in fact before we entertain ideas about whether people whose cognitive abilities have declined, according to some tests and to below a particular threshold, should be precluded from participating in referendums. If nothing else, problems of discretisation quickly arise: i.e. where do we draw the line? For example, while people with Alzheimer’s disease can be kept from voting, should those who are mathematically illiterate, and would thus probably fail the fluid reasoning and numerical reasoning tests? Similarly, and expanding the remit from referendums to elections (which isn’t without problems), which test should potential voters be expected to pass before voting in different polls – say, from the panchayat to the Lok Sabha elections?

Consider also the debates at the time Haryana passed the Haryana Panchayati Raj (Amendment) Act in 2015, which stipulated among other things that to contest in panchayat polls, candidates had to have completed class 10 or its equivalent (plus adjustments if the candidates are from an SC community, women, etc.). Obviously those contesting the polls would be well past their youth and unlikely to return to school, so the Act effectively permanently disqualified them from contesting. As such, while the answers to the questions above may be clearer in less unequal societies like those of the UK, they are not so in India, where cognitively well-equipped people have been criminally selfish and public-spiritedness has been more strongly correlated with good-faith politics than education or literacy.

At the same time, the study and its findings also reiterate the significant role that mis/disinformation has come to play in influencing the way people vote, for example, which makes individuals’ cognitive abilities – and all the factors that influence them – another avenue through which to control, for better or for worse, the opportunities we have for healthy governance.

Cognitive flexibility and nationalism 2.0

Remember that paper about cognitive flexibility and nationalism? The one that said people who are more nationalistic in their politics tend to have lower cognitive flexibility? I’d blogged about it here. I hadn’t read the study’s paper, published in the Proceedings of the National Academy of Sciences, because I didn’t think I had to to be able to call the study’s conclusions into question. An excerpt from my previous post:

… ideological divisions, imagined in the form of political polarisation, are bad enough as it is without people on one side of the aisle being able to accuse those on the other side of having “low cognitive flexibility”. The nuance can be worded as prosaically as the neuroscientists would prefer but this won’t – can’t – stop the less-nationalistic from accusing the more-nationalistic of simply being stupid, now with a purported scientific basis.

This is why I believe something has to be off about the study. The people on the right, as it were on the political spectrum, are not stupid. They’re smart just the way those of us on the left imagine ourselves to be. Now, one defence of the study may be that it attempts to map a hallmark feature of the global political right, sort of a rampant anti-intellectualism and irrationality, to its neurological underpinnings – but nationalism is more than its endorsement of traditions or traditional values.

As it turned out: if I had, the paper would have revealed more problems with the study and made a stronger case than I was able that it is quite likely a product of the “publish or perish” kind of thinking. The reason I revisit this study now is an interesting conversation I had with Shruti Muralidhar, a cortical and hippocampal neuroscientist, currently a postdoc at the Massachusetts Institute of Technology. Before I’d written my post, I’d asked Shruti if she could read the paper and possibly critique it.

My primary concern was basically about assigning a kind of “hierarchy of cognitive abilities” to the political spectrum – that sounds dangerous. By saying the political right has less cognitive flexibility, I’d felt like the study was reaching the conclusion that there might be a purely biological explanation for why people behave the way they do. This kind of reductionism is eminently dangerous.

According to Shruti, “This understanding of the paper is not far from what they want the reader to take away – but sadly, they have little or no backing to actually prove or disprove this claim.” She summed her observation up in a few points (quoted verbatim):

  1. Cognitive flexibility is simply just that. It doesn’t mean more or less intelligence, smarts or anything like that. In fact, it might not even be a “positive” trait depending on the situation at hand.
  2. The study’s authors have administered only two cognitive tests, and one of them clearly gives counterintuitive, unexpected or, one might even say, “wrong” results, as in goes against the study’s primary hypothesis.
  3. These are correlation studies, which usually are to be taken with a bagful of salt.

The first question that arises then is why the authors – or PNAS – decided to publish their study when 50% of their tests turned in results that opposed their hypothesis: that the more nationalistic are less flexible, cognitively speaking.

She also pointed out many issues with the language in the paper, especially lines that could be misinterpreted easily. Some in particular stuck out because they revealed a deeper epistemological issue with the study.

Shruti said, “The authors clearly admit that cognitive flexibility is a multi-dimensional beast and that it is difficult to understand it completely,” and often suggest that they don’t understand it completely themselves. One giveaway is that they keep saying variations of “We need more and better tests”.

A bigger giveaway is a line on page 6: “However, it is also conceivable that immersing oneself in strongly ideological environments may encourage psychological inflexibility and promote a preference for routines and traditions.” In other words, if A stood for “more nationalistic” and B for “less cognitive flexibility”, then the authors were saying that A therefore B while also admitting that B therefore A. In other other words, their correlation was in doubt, leave alone causation. This portion concludes thus:

Nevertheless, more research is necessary to understand the nature of cognitive flexibility and the various ways in which it manifests in relation to ideological thinking.

The authors haven’t defined cognitive flexibility explicitly in their paper, instead referencing older studies on the subject. Even so, Shruti said that those papers might not be able to provide the final word either because, as one of her peers had pointed out, “Since this study is EU/Britain-specific, their idea of what ideological inflexibility is might also be different from, say, India’s or the rest of the world’s. Europe thrived on systems and thinking-within-the-box for centuries.”

All together, the paper appears to describe a study of the “low-hanging fruit” variety. Its central hypothesis has been neither proved nor disproved, the reader is left in doubt about whether the tests were properly chosen (and why more tests weren’t performed), and the paper is strewn with admissions that the authors don’t claim to understand what one of the more important keywords in the study really means.

Worst of all (to me) is that the paper has been published with a misleading headline and the university press release, with an incredibly misleading one that should take all the responsibility for fake news born as a result (and strengthening the case that they shouldn’t be trusted). And there’s quite a bit of it:

  • PsychCentral has an article that only quotes Leor Zmigrod, the lead author of the atudy and a psychologist at the University of Cambridge
  • The same is true of an article in The Guardian by Nicola Davis. The headline goes ‘Brexiters tend to dislike uncertainty and love routine, study says’ – more of the reductionism at work.
  • Andrew Brown of The Guardian takes the study’s conclusions at face value, writing in his column:

… some kinds of political argument are going to be literally interminable. Obviously this isn’t true of any particular issue. Even the question of our relations with Europe will be settled some time before the heat death of the universe. But it may be replaced by something else which arouses the same passions and splits the population in the same way, because the cognitive traits [Zmigrod] is analysing are all part of the normal variation of humanity.

In fact, it seems no prominent coverage of the paper has invited an independent researcher to comment on its findings. I concede that I myself didn’t speak to a psychologist – Shruti is a neuroscientist – but all of Shruti’s observations are hard to ignore.

Finally, if I were looking to publish a paper right now, I’d hypothesise that flattering, non-critical coverage of scientific papers – peer-reviewed or otherwise – is more common among news publishers if each paper makes it easier for the publication to maintain its political position.

Featured image credit: mwewering/pixabay.