By Patrick Fagan
Mark Zuckerberg knows what you like in the bedroom.
Metaphorically, of course – but also, perhaps, literally. In 2013, a Cambridge University study found that not only could Facebook likes predict personality, but they could also predict intimate traits like depression, drug use, and sexuality. Research since then has similarly suggested, for example, that even profile pictures can predict personality as well as sexuality.
This shouldn’t come as a surprise. Since the Cambridge Analytica affair, it is well understood that innocuous social media data can predict political preferences – with, for example, Democrats tweeting more about feelings and Republicans about group membership. If you want to see the principle in action, TIME worked with Sam Gosling (author of Snoop: What Your Stuff Says About You) to create a quiz that can predict your politics based on just ten seemingly unrelated questions. It’s just a short hop from the polling booth to the bedroom, as the odd auto-asphyxiating conservative politician reminds us, and so too does research by a survey finding, among other things, that liberals prefer threesomes and conservatives prefer spanking.
The fact is, it’s all connected. Our behaviour is driven by latent psychological constructs which predict how we’ll act across different contexts in fairly consistent ways. Someone who votes for liberal political candidates will also be open to nonmonogamy, as well as read philosophy and listen to jazz because there is an underlying trait driving all of those behaviours – in this case, openness to novelty and ambiguity.
Psychological Principle
In psychology, this relates to a principle known as thin slicing. If you were given a very thin slice of cake, you could look at it and accurately predict what the rest of the cake looks like; likewise, given a very thin slice of a person, it’s possible to make quite accurate predictions about them based on inferred underlying characteristics.
This is why first impressions as so powerful: one study showed participants photographs of faces for just one-tenth of a second and found that ratings of perceived personality had a high level of accuracy (which did not change even given more time). Yet, it is not just a person’s face that gives away their character. Research has found, for example, that accurate personality judgements can be made using cues like a target’s office or bedroom, their social media profile, or even their email address.
Predictability of Personality
There are two practical implications here.
The first is that people can be read rather well from their digital footprints. A review, Human and Computer Personality Prediction from Digital Footprints, summarised how data points like social media posts, smartphone logs and linguistic features have been used by algorithms to predict personality and concluded that the computer is generally more accurate than human judgements. For example, an analysis of Spotify data found that personality could be predicted not only by genre preferences but also by derived metrics, such that conscientiousness was related to country, soul and funk, as well as having a premium account and being less likely to skip tracks. Elsewhere, personality has been predicted by things like purchases (for example, those conscientious people are more likely to spend on savings accounts and less likely to spend on takeaways) and reading (conscientiousness is correlated positively to history and negatively to comedy).
The second implication is that adverts and user experiences can be targeted based on inferred traits, and these are more likely to be more effective. Early research found that attitudinal evaluations were higher for messages that matched the reviewer’s personality (e.g., “You’ll always be where the excitement is” for extraverts), and more recent research found that personality-matched Facebook ads had up to 40% more clicks and 50% more purchases ‘in the wild’. Likewise, in the last year or so, personality-targeted adverts and behavioural interventions have received significant attention and promising results – with researchers finding, for example, that citizens are more strongly persuaded by personality-matched political ads, ads are more engaging when tailored to individual differences in persuasive susceptibilities, and that a personality-targeted dieting intervention significantly reduced calorie consumption.
Personality Traits and Customisations
Taking the Big Five personality traits as a common framework, here is just a flavour of how messaging can be customised according to the audience.
People who are open to experience will respond better to surrealist, abstract, complex, and asymmetrical aesthetics; they prefer wording to be more verbose and unique; they prefer messaging which encourages them to reflect and come to their own conclusion rather than being instructed directly, and which focuses on hopes and benefits rather than duties and losses.
Conscientious people like art to be representational (e.g., landscapes and portraits) and simple; they prefer copy to be factual and newsy, and they are more likely to be nudged by appeals to conformity and achievement.
Extraverts prefer aesthetics to be loud, bright, and warm; they respond better to messaging which is positive and social; and they are more likely to be nudged by future discounting (i.e., having things in the here-and-now), sex appeals, and humour.
Those scoring high in agreeableness prefer aesthetics that are curved, calm, and orange; they respond well to messaging which encourages them to take other people’s perspectives; and they are more likely to be nudged by appeals to avoid confrontation, and stick to morals, and avoiding regret.
Finally, neurotics respond best to aesthetics that are structured and romantic, though they also like dark colours and sad paintings; they respond to wording that emphasises feelings of personal distress; and they are nudged by attractive messengers, among other things.
Implications for Nudges
Ultimately, the point is that yes, nudges work (generally), but different nudges will work for different people. This point was illustrated nicely by a study in which over 50,000 letters were posted in South Africa to advertise loans. The researchers found that a picture of a smiling, attractive woman increased loan uptake – but only among men.
As Richard Thaler once said, “Just as no building lacks an architecture, so no choice lacks a context.”
This article was edited by Ankit Shanker
