Vice and a number of other news and academic outlets have recently reported that Trump and the Brexit campaign used a psychographic analytics firm to help win their respective elections.
“Psychologist Michal Kosinski developed a method to analyze people in minute detail based on their Facebook activity. Did a similar tool help propel Donald Trump to victory? Two reporters from Zurich-based Das Magazin (where an earlier version of this story appeared in December in German) went data-gathering.” – Vice.com
Since I build psychographic and demographic profiles of customers for a living, I was particularly interested in this story. The audiences I build for clients perform well on Facebook and other platforms. The most recent Facebook Saved Audiences I created for clients performed better than the Lookalike Audiences Facebook generates from their customer lists. My abilities in audience building are good. But I wondered if this could this be a new methodology I should be using for Follow Us clients.
My personal Facebook profile has a ton of data – hundreds of Friends, likes, and daily activity. I decided to take one for the team and uploaded my profile onto the site discussed in the piece here. The journalists writing about The Cambridge Psychometrics Center described the analytics tool as a super-accurate personality analyzer and behavior predictor. I got ready to be blown away by their algorithms.
Full disclosure: I am a knee-banging, bleeding heart liberal member of the much reviled East Coast Liberal Elite. I prepared for the Cambridge Psychometrics tool to tell me about my love for Hillary and all things liberal.
That’s not what happened. Almost every single metric was wrong:
- Analytica prediction of gender: likely male
- Actual gender: female
- Analytica prediction of sexual orientation: likely gay male
- Actual sexual orientation: straight woman
- Analytica prediction of personality type: INTP
- Actual personality type: ENFP
- Analytica prediction of political views: 60% of my profile predicted I was liberal, but I likely also held strong conservative views at 40%
- Actual political views: see left of Che Guevera above
- Analytica prediction of occupation: psychologist or one of the “helping professions”
- Actual profession: owner of digital marketing consultancy, which (I hope) helps people and uses psychological principals. I’ll give them partial accuracy on this one.
- Analytica prediction of intelligence: 47th percentile (the analysis cited my likes for E! News and The Washington Post as predictors of my mediocre intelligence).
- Actual intelligence: As a child, I took an IQ test and scored a few points away from genius level. I will concede that too much time watching The Simple Life on E! in the 90’s may have brought my score down significantly.
- Analytica prediction of marital status: single, no kids
- Actual marital status: married for over 10 years to a one man with whom I have two biological children.
I am a little nervous about sharing my political and personal info online. This is the world we live in. Our data is everywhere and the personal is political. This experiment was a worthy exercise to see if Big Data is as smart as it thinks it is.
Yes, Orwell’s 1984 is feeling prescient to many of us in 2017. If this sample size n of 1 can be used to predict how accurate this particular tool is overall, they have far to go before I start deleting my Facebook page in fear. And I will not be contacting the creators to use their tools in Follow Us Digital’s audience building. My audiences convert well with my current techniques.
At least I understand now why I saw so many Trump banner ads and ‘You May Know’ suggestions for gay chat lines on my feed last year 😉