Revealing Hidden Secrets

Knowing that people don't wear their hearts on their sleeves, psychologists have longed for a "pipeline to the heart."  One strategy, developed nearly a half century ago by Edward Jones and Harold Sigall, created a “bogus pipeline.” Researchers would convince people that a machine could use their physiological responses to measure their private attitudes.  Then they would ask them to predict the machine's reading, thus revealing attitudes which often were less socially desirable than their verbalized attitudes. 

 More recently, psychologists have devised clever strategies for revealing “implicit attitudes,” by using reaction times to assess automatic associations between attitude objects and evaluative words. (In contrast to consciously held “explicit attitudes,” implicit attitudes are like unconscious habits.)

 A new working paper (abstract; PDF) by Katherine Coffman and fellow economists demonstrates a third strategy for getting people to reveal sensitive information—the “Item Count Technique” (ICT). 

One group was given four simple statements, such as “I spent a lot of time playing video games as a kid,” and then was asked how many of the four statements “apply to you.”  A second group was given the same four statements plus a fifth:  “I consider myself to be heterosexual,” and then was asked how many of the five statements “apply to you.”

Although no individual is asked to reveal which specific statements are true of them, a comparison of the two groups’ answers reveals—for the sampled population—the percent agreeing with the fifth statement.

Thus, without revealing anyone’s sexual orientation, the ICT aggregate data showed a 65 percent higher rate of non-heterosexual identity than was self-reported among people who were asked straight-out: “Do you consider yourself to be heterosexual?”   

But then let’s not discount public surveys.  Nate Silver’s digest of presidential polling data correctly predicted not only the 2012 U.S. national presidential outcome, but also the outcome in all 50 U.S. states.  Specific, explicit attitudes can predict behavior.