On Labels

Asian-American women took a maths test. Prime their ethnicity first and they scored higher. Prime their gender and they scored lower. Same women, same test, and the only variable was which half of themselves had been mentioned (Shih, Pittinsky & Ambady, 1999).

On the AP Calculus exam, women scored better when the gender box came after the test rather than before. The reanalysis estimated 4,763 more women a year would earn the credit if the question moved to the end of the page (Danaher & Crandall, 2008).

Tell a woman that men score five percentile points higher at maths for genetic reasons and she does worse on the test. Tell her the same gap comes from how teachers treated boys and she does better. Same gap, same test, two hundred plus women, published in Science (Dar-Nimrod & Heine, 2006).

Roughly 600 adolescents. Being on a team didn't predict delinquency. How often they played didn't predict delinquency. Calling themselves a jock predicted significantly more of it, across both gender and race (Miller et al., 2007).

791 undergraduates. Athlete identity went with lower depression and lower odds of a suicide attempt in the past year. Jock identity went with higher odds of one. Same sport, same body, two nouns (Miller & Hoffman, 2009).

900 teenagers were asked in Grade 10 which Breakfast Club character they were, then tracked for years. The label they picked at 15 predicted their substance use, their education and their psychological adjustment into their twenties (Barber, Eccles & Stone, 2001).

Tell teachers at random which children are about to bloom and those children gain real, measurable IQ points by year end. The label wasn't true. It didn't need to be (Rosenthal & Jacobson, 1968).

Fail a test, then get offered the tutoring. If you think ability is a fixed thing you have, you turn the tutoring down. The label decided the ceiling before the class started (Hong, Chiu & Dweck, 1999).

After failing a test, some students compared their scores with people who'd done worse and some compared with people who'd done better. Which one they picked was decided before the test, by what they believed ability was (Nussbaum & Dweck, 2008).

Kahan gave 1,100 people a hard statistics problem. Presented as a skin cream trial, the numerate crushed it. Presented as gun control data, the same problem, the same numbers, and everyone polarized and got less accurate. The polarization didn't shrink among the best mathematicians. It grew (Kahan et al., 2017).

If you openly prove you are a good and fair person on one test, you're more likely to discriminate and be a biased actor on the next (Monin & Miller, 2001).

Over two hundred psychologists, psychiatrists and social workers read the same patients. When the case was framed biologically, their empathy dropped. The authors called it alarming, because clinician empathy predicts whether people actually get better (Lebowitz & Ahn, 2014).

If you praise someone for being smart they are 40% more likely to lie about their score to a stranger (Mueller & Dweck, 1998).

Ask four year olds to "be helpers" instead of "to help" and they help more, until the moment they spill the milk. Then the helpers quit and the helping children keep going (Foster-Hanson et al., 2020). This is the whole letter in one experiment. The noun works right up until you fail, and then it turns on you.

Researchers asked Spanish people if they'd throw themselves in front of a runaway train to save five others. If the five were Spanish, most said yes. If they were European, most said yes. If they were American or any "other" group, most said no (Swann et al., 2010).

One class of fifth graders was told repeatedly that they were neat and tidy people. Another was told repeatedly that they should be. Being told what they were beat being told what to do, and it was still holding two weeks later (Miller, Brickman & Bolen, 1975).

People with low self-worth leave partners who think too well of them, because a label wants agreement more than it wants your happiness (Swann, self-verification theory).

Men on a phone call with a woman they'd been told was attractive were warmer, and she became warmer back, on a recording where the raters could hear only her (Snyder, Tanke & Berscheid, 1977). The label was in his head and came out of her mouth.

Give trained clinicians an identical case file and swap "substance abuser" for "person with a substance use disorder," and the abuser gets punishment while the person gets treatment (Kelly & Westerhoff, 2010).

Show a chewed eraser and say to someone "This is a dog's chew toy" and they will not touch it. Say "This could be a dog's chew toy" and 40% will use it as an eraser (Langer & Piper, 1987).

452 German 15-year-olds. Identifying as "Alternative" correlated with self-harm at around r = 0.3, and 45.5% of them reported self-injury. The subculture didn't just describe them, it came with a behavioural repertoire.

People who see themselves as many separate things handle bad news better than people who see themselves as one thing. If you're only a footballer/lawyer/doctor/parent, a bad game or a lost job is a bad life (Linville, 1987).

Eighty four hotel maids. Half were told the work they already do counts as exercise. Nobody changed what they did. Four weeks later the informed group had dropped weight, blood pressure, body fat and BMI (Crum & Langer, 2007).

Ask people at 50 how they feel about growing old, then check the death records up to 23 years later. The ones who held positive views lived 7.5 years longer, after controlling for age, gender, income, loneliness and actual health (Levy et al., 2002).

If you never experiment with another identity before you're 18 then you are likely to score higher on rigidity and authoritarianism a decade later (Marcia, 1966).

Say the label out loud and you'll defend positions you've never actually thought about. Once it's public, being wrong costs less than being inconsistent (Cialdini).