Would it tell Mark Zuckerberg your deepest insecurities, prejudices, and desires? You might be shocked. We certainly were, when we psychoanalysed all of the personal data given to us by our subject.
Platforms like Meta have powerful predictive analytics that can reach deep down inside your mind and find things perhaps unbeknownst even to yourself. Myriad studies have revealed how your digital footprints can be used to predict your personality – your funky Spotify plays make you extraverted, your philosophical GoodReads likes make you open, and your financial transactions on legals and investments make you disagreeable. Even footage of the way you walk has been used in a recent study (I always suspected I had a neurotic shuffle).
You probably know all this (that’s why you reject cookies). But you might not be aware of just how invasive it can get.
In a famous study, researchers from Cambridge University collected personality on Facebook using a quiz, then analysed the traits’ relationships with Facebook likes. Extraverts ‘liked’ beer pong and disorganised people ‘liked’ comedy, as you’d expect; and another study found that, on average, the algorithm could predict personality better than could a family member, friend, workmate or cohabitant.
But the researchers could predict other things, more personal than personality.
Sexuality, drug use and relationship status, for example – via ‘liking’ things like Adam Lambert, Big Momma’s House and Weight Watchers, respectively. Other studies have been able to predict sexuality based on facial photographs alone, and famously the US retailer Target used purchase data predict that a teenage girl was pregnant before she’d told her own father.
Just how far into your mind can these algorithms reach?
Freud argued that the subconscious always forces its way to the surface through some means or another. It’s like a plastic ball in a bathtub – if you push a subconscious urge back under the suds, it will just pop up somewhere else. As the fertility rate in America falls, for example, the ownership of small dogs increases – the parental instinct must find its outlet.
In the same way, these invisible forces surely manifest in digital traces. Your Google searches, your biometrics, the tone of your Twitter posts, all revealing the ever-churning currents of your subconscious.
At Kubik, we set out to answer this question by combining technology with psychology - data with psychoanalysis - and becoming forensic psychologists with only digital fingerprints as our clues.
We pioneered a methodology we call data ethnography, qualitatively combing through the digital DNA of a single subject to see what it could tell us about their psychological make-up.
If we were their therapist, what could we learn from the behavioural traces they leave online?
We collaborated with a single subject who gave us over a year’s worth of their personal data, and their consent to pore through it and analyse it. They downloaded their data from Instagram, Facebook, Amazon, Netflix, Apple Health and their bank statements, and gave it all to us. We explore the data in three ways: at a topline level, across time series, and through correlational analyses matching variables by date.
Looking at a word cloud of their Amazon purchase data, the word ‘women’ occurred a lot. Immediately, we knew our subject was probably female. She was also in a relationship (‘wedding’), and she was self-conscious about her looks (‘body’). There were early signs of personality, too: the most common word was ‘black’, which may suggest she’s neurotic (since research consistently shows neurotic people prefer darker colours).
There were other signs of neuroticism in the data. Our subject was a regular watcher of crime dramas and true crime documentaries, a favourite of neurotics.
She was also concerned with mental health, having bought a gratitude journal as well as books like Maybe You Should Talk To Someone: A therapist, her therapist, and our lives revealed. In fact, her bank statements revealed she visited a therapist regularly. One book – Imperfect: The gifts of imperfect – might be revealing about how she feels about herself.
This neuroticism sometimes manifested itself in a need for distraction and escape. Sometimes she would binge on Netflix – on Sunday 29th April 2018, she watched 9.5 hours of Netflix including 7 episodes of Gossip Girl in a row. On another weekend, she watched 5.5 hours of Seinfeld and nothing else (the day before, she’d spent significantly more than usual on alcohol and went to bed at 1am).
Her Netflix hours were significantly higher during the winter months, reaching four and a quarter hours a day during November to January, compared to one and a quarter the rest of the year. She would also watch some shows obsessively: for Too Hot to Handle season three, once she starting watching it, she watched more and more every day, even watching it as soon as she woke up at 8am, and nothing else, until it was finished.
This also suggests a high need for completion – a consistent and conscientious personality that sees things through. She watched all three and half hours of The Irishman in one sitting, for example. She also put money into savings on a regular basis, and most of her expenditure was on essentials rather than frivolities. As well as financial budgeting, there was even evidence of health budgeting: on many of the days she spent money on drinking and dining, she would go running during the day, as if planning her calorific expenses in advance.
Our subject also seemed to be smart and analytical. Her Instagram comments revealed that she commonly used abstract metacognition words like “thought” and “guess”. Her Amazon purchases included many functional terms like ‘wireless’; she also liked to watch Netflix shows like Explained, Unsolved Mysteries, and 100 Humans: Life’s questions answered. One day, she watched brainless David Spade comedy The Wrong Missy for just four and a half minutes before giving up and watching a movie called The Book Club instead. Similarly, she appeared to be open to new experiences – for example, her watched Instagram reels mostly covered topics like photography, fashion, and travel.
Socially, our subject showed signs of extraversion. The third biggest category of expenditure was wining and dining, her Instagram comments were full of positive affect (e.g., ‘haha’, ‘😘’, ‘❤️’), her Amazon purchases included topics like ‘party’, and she enjoyed reality Netflix shows like Queer Eye, Glow Up, and Real Housewives.
On the other hand, her desire for social interaction many not totally satisfied. On Facebook, she gave almost twice as many comments as she received, and her Instagram comments were polite (e.g., ‘thank you’), suggesting a formal distance. The ‘uses and gratifications’ theory of media consumption argues that people watch content to satisfy needs they have: the fact that our subject watched so many sociable shows and sitcoms (a word cloud of titles revealing words like ‘friends’, ‘lover’, ‘stranger’) might reflect her unfulfilled needs.
She additionally seems to be a bit cynical – that is, disagreeable. In the entire period of study, he spent only £2 on donations. She also bought products like a Grumpy Cat mug which read, “I don’t like morning people. Or mornings. Or people.”
Overall, we deduced that our subject is intelligent, curious, and adventurous – she’s someone likes solving puzzles. She suffers from some insecurities and anxieties, which can develop into an obsessive-compulsive desire for distraction; she also keeps people at arm's length and has an unfulfilled yearning for social intimacy. However, she is protected from her neuroses by high conscientiousness and future planning.
We showed this profile to our subject, who exclaimed, “This is so, so, so good. I am so impressed. It was scarily accurate!! I am actually shocked how accurate it is. It feels like my therapist wrote it.”
On a larger scale (privacy issues notwithstanding), these psychological insights could be used for targeted comms and user experiences. On a far more basic level, a study from Cambridge found that personality-targeted Facebook ads increased conversion rates by up to 40%.
Yet it could also be used for good, sending self-help nudges and boosts. For our subject, we found a statistically significant, negative correlation between hours of sleep and hours of Netflix watched – but positive correlations between sleep and both Instagram likes received and steps. In theory, she could be nudged into healthier sleeping by encouraging her to watch less Netflix, walk more, and have a more fulfilling day worth sharing online – and all this could be done in a tone to match her thoughtful, organised personality while meeting her need for reassurance and social connection.
There is limitless potential, it seems, in decoding our digital DNA.