Influence
Part 9  The Marketplace of Attention
Chapter 260 of 360

Psychographics for Sale: Cambridge Analytica and the OCEAN Model

Cambridge Analytica's claim about itself was that it held psychological profiles on tens of millions of American voters and could target them with messages matched to their personality traits rather than to their stated politics.

The data came from an app called thisisyourdigitallife, a personality quiz built by the Cambridge researcher Aleksandr Kogan. Around three hundred thousand people took it and consented to give their data. Facebook's platform at the time also permitted an app to collect data on the taker's friends, who had consented to nothing, and the total reached — on Facebook's own later estimate — around eighty-seven million people.

The scientific basis is real and predates the company. Michal Kosinski and colleagues at Cambridge had published work showing that Facebook likes predict personality traits, intelligence, sexual orientation and political views with substantial accuracy — in some domains better than a person's friends could.

The targeting claim built on this is that a message can be tuned to the recipient's Big Five profile: a high-neuroticism voter receives a fear-based framing of the same policy that a high-openness voter receives framed as innovation. Some experimental work supports message-personality matching producing modest gains in persuasion, and the mechanism is fluency — a message that matches how you process is easier to process and generates less counter-argument.

The honest assessment of the company's actual electoral impact is that it is unproven and probably far smaller than either its own marketing or the subsequent alarm suggested. Cambridge Analytica was, among other things, selling itself.

The structural consequence, though, does not depend on how well the targeting worked, and it is the reason this chapter matters.

When political advertising is delivered individually and privately, no shared public record of what was claimed exists. In a broadcast environment, a campaign that says one thing in Florida and the opposite in Michigan can be caught, because both statements are public. In a targeted environment, each voter sees a message calibrated to them, nobody sees the set, and there is no artifact for a journalist or an opponent to hold up.

Accountability requires a common frame. The technique removes it, and the removal is not a side effect.

The affair broke in March 2018 through Christopher Wylie and reporting by the Observer, the Guardian and the New York Times. Facebook was fined five billion dollars by the FTC in 2019. Cambridge Analytica closed.

The case

The Facebook-Cambridge Analytica affair exposed in March 2018 by The Guardian, Observer and New York Times: Aleksandr Kogan’s app harvested data on up to 87 million Facebook users, which Cambridge Analytica used for Big Five (OCEAN) psychographic political targeting.

The mechanism

The method rests on Michal Kosinski’s finding that digital footprints predict personality traits, allowing message variants tuned to a target’s neuroticism or openness rather than to their stated beliefs. Matching persuasive appeals to trait profiles increases processing fluency and reduces counterargument, the message-matching effect documented in persuasion research. Because each voter sees only their own tailored frame, no shared public record of claims exists — accountability collapses along with the common frame.

What this chapter covers

  1. A personality quiz becomes a data pipeline
  2. Trait-matched messages lower counterargument
  3. 2014-2015 harvest, March 2018 exposure
  4. Voters receive incompatible private realities
  5. Whistleblower and press reveal the harvest

Cross-ref: Attack — Speak to the Trait