Bad group logic

The day before the UK general election, and there’s a lot of political communication flying about on social media. There are positive persuasions (“Vote Labour!”), negative persuasions (“Don’t vote Tory”), and a curious kind of message which interests me now: “Tories are evil”. Here’s an example:

The Conservative Party is an eternally irritating force for wrong that appeals exclusively to bigots, toffs, money-minded machine men, faded entertainers and selfish, grasping simpletons who were born with some essential part of their soul missing”. Charlie Brooker

The first curious thing about this quote is the logical content, which is either ridiculous or sinister. The Conservative Party got 13.6 million votes in the last election, more than any other party. Brooker’s claim is weirdly specific, and if we assume the number of “toffs, money-minded machine men and faded entertainers” is small, we are forced to conclude that Conservative voters are either “bigots” or “selfish, grasping, simpletons”. Either Brooker has made a mistake about what Conservative voters are like, or millions of ordinary people are bigots or simpletons. This view leads two places: First, to the conclusion that people who aren’t as insightful as Brooker shouldn’t be allowed to vote. Second, that there is no point trying to persuade most people honestly – bigots and simpletons can’t understand the evidence and arguments.

The second curious thing about this quote is the intent. The kind of contempt seems unlikely to work on anyone voting Conservative. Nobody would recognise themselves in Brooker’s list of villains, or – if they did – be motivated to weigh his values highly. It’s persuasive currency for the political opposition is zero. It might even be worse than nothing, since any Conservative voter is likely to be incensed, rather than dissuaded.

Messages like this, from simple slurs to more complex arguments that assume the irrationality or stupidity of people who vote differently, don’t exist to persuade. They are cheerleading for political partisanship, playing to the home crowd. Social media is rocket fuel to this kind of performative tribalism, which has it deep roots in our psychological capacity for groupness.

By ‘groupness’ I mean our tendency to identify social groups and use these to navigate our informational worlds. It means that we make decisions about who to believe and who to help based on which group we think they belong too first, using this group membership to filter all other information about them, including what they say or do.

This tendency is usually on on a high trigger, as demonstrated by a phenomenon called the ‘minimal group paradigm‘. I saw a nice demonstrate of the phenomenon recently from David Hauser on twitter:

Let’s just walk through this, since a seemingly trivial classroom demonstration of the minimal group paradigm has important implications for understanding political tribalism.

First, Dave asks the class “is a hotdog a sandwich?”. Question one can be any question, as long as the there are two different answers which are both chosen by some of the class, and as long as the those answers are of no relevance to question two. Some classic research used people’s art preferences. Dave asks if a hotdog is a sandwich. 69% of the class think Yes, 31% think No.

Next, Dave asks people to pick from two options

A: Give $3 to everyone who agrees with me and give $4 to everyone who disagrees with me.

B: Give $2 to everyone who agrees with me and give $1 to everyone who disagrees with me.

Most of the class, 71%, pick B.

Take that in for a moment. Most people are choosing less money for themselves ($2 rather than $3), and less money for many other people in the class – which is bizarre, or at least at odds with economic rationality in itself – and they are doing it because of social groups which didn’t exist until moments before, and have been conjoured out of the air by the instructor asking a stupid question about hotdogs.

Based on this arbitrary grouping, people would rather those who disagreed with them get less ($1 rather than $4) even to their own cost and the general cost of everyone who agrees with them! The are controls you could run to confirm that the effect is driven by the grouping induced by the first question, but the general conclusion must be that tribalism is a hell of a drug, and dangerously easy to invoke.

This helps us understand the attraction of “They Are Evil” messaging. These messages are not persuasive. Thinking like this is the opposite of persuasive – it is polarising, driving people further apart in their views and making communication across the divide harder. We do it because it feels good, awarding us and our tribe the moral equivalent of $2 when we could have had £3, but at the vindictive gain of awarding the opposing tribe $1 rather than $4. Collectively it diminishes us, and encourages a view that people who disagree are biased, selfish or otherwise beyond redemption and persuasion.

Please don’t give in to this kind of thinking. And please be careful of creating new tribes by asking unnecessary questions with binary answers.

2 replies on “Bad group logic”

An interesting, thought-provoking post, but there is a giant ambiguity (at least in the way you have described it here) in Dave’s thought-experiment.

“Give….” without clarification could mean you (the person answering the question) will give. This would mean option B leaves the person answering hugely better off as they will be giving away $1 or $2 to multiple individuals rather than $3 or $4. I assume this is an ambiguity created by the desire for brevity here and not a flaw with the experiment as conducted?

(And by the way, if Dave is to be the provider of the money and the answerer is to be among those receiving the money – which is also unclear here! – then there may be at play a desire not to impoverish Dave or the unknown money provider!)


I hear you, and that is definitely an interpretation some people make (judging by my twitter replies). However, the basic effect (although maybe not its strength in this scenario) has been demonstrated in a wide range of circumstances (including ones which remove for the ambiguity), which gives me confidence in the basic take-away: people can be induced by surprisingly minimal markers of group affiliation to engage in spiteful resource allocation

Leave a Reply

Your email address will not be published. Required fields are marked *