Scientific American 2019-04

ADVANCES

12 Scientific American, April 2019

SOURCE: “PATTERNS OF IMPLICIT AND EXPLICIT ATTITUDES, VOL. I: LONG-TERM CHANGE AND STABILITY FROM 2007 TO 2016,”

BY TESSA E. S. CHARLESWORTH AND MAHZARIN R. BANAJI, IN

PSYCHOLOGICAL SCIENCE,

VOL. 30, NO. 2; FEBRUARY 2019

Graphic by Amanda Montañez

0

0.

0.

0.

0.

0.

0

0.

0.

0.

0.

0.

0.

0.

0.

0.

1.

Sexuality

Race

Skin tone

Age

Disability

Body weight

2008 2010 2012 2014 2016 2008 2010 2012 2014 2016

Neutral

More biased
(toward typically
preferred group)

Implicit Association Test Score

Explicit Preference Score

Test showed

images of faces

Test showed

images of bodies

Researchers analyzed data from

4.4 million implicit bias tests

completed by U.S. participants, controlling

for factors such as the time of year the test was taken.

Higher test scores indicate stronger implicit preferences

for straight, white, light-skinned, young, nondisabled

or thin people. Scores for the body weight test shifted

around 2011, when researchers started showing

silhouettes of bodies rather than faces.

Participants’ explicit biases were

assessed by asking them to select

statements expressing how much they favored

one group over another, such as “I strongly prefer

young people to old people.”

Implicit Bias Explicit Bias

Average monthly scores

(light lines)

Trends with seasonal adjustments

(bold lines)

PSYCHOLOGY

Biases Aren’t

Forever

Implicit prejudice against

certain groups is declining

Psychologists have lots of evidence that

implicit social biases—our unconscious,

knee-jerk attitudes associated with specific

races, sexes and other categories—are

widespread, and many assumed they do

not evolve. The feelings are just too deep.

But a new study finds that over roughly the

past decade, both implicit and explicit, or

conscious, attitudes toward several social

groups have grown warmer.

The study used data from a standard

test of implicit attitudes collected via a Web

site called Project Implicit. Participants

were asked to quickly press a certain com-

puter key in response to positive words,

such as “happy,” and a different key in re -

sponse to negative words, such as “tragic,”

that appeared on a screen. These words

were interspersed with images or words

that represented two categories of people,

such as blacks and whites, and participants

were asked to flag these using the same

keys. Faster reactions when, for example,

black rather than white faces shared a key

with negative words suggested a racial bias.

Tessa Charlesworth and Mahzarin

Ba naji, psychologists at Harvard Universi-

ty, analyzed more than four million results

collected over a 10-year period from U.S.

adults who had taken implicit association

tests for sexuality, race, skin tone (in which

faces differ in color but not shape), age,

disability and body weight. Respondents

also answered questions on the screen

asking them to explicitly rate how much

they liked people in each of the categories.

In line with previous findings, explicit

bias decreased in all six categories from

2007 through 2016; the drop ranged from

49 percent (for sexuality) to 15 percent (for

body weight). But more surprisingly,

implicit bias also decreased—by 33 percent

for sexuality, 17 percent for race and 15 per-

cent for skin tone ( graphic ). Most of the

reductions occurred in all generations, the

researchers reported in a study published

online in January in Psychological Science.

“It’s a really cool paper,” says Keith Payne,

a psychologist at the University of North

Carolina at Chapel Hill, who has found sim-

ilar bias reductions in his own work. “I think

it’s going to start a lot of conversations.”

Charlesworth and Banaji also found,

however, that implicit biases about age and

disability did not change over time, and

those against overweight people nudged

up by 5 percent.

Several factors might explain the dis-

crepancies among categories, the research-

ers say. In their data set, implicit biases for

race, skin tone and sexuality were lower to

begin with than those for age, disability and

body weight. And the types of implicit biases

that decreased the most are also the biases

that have received more societal attention.

Meanwhile the stigma associated with

obesity may have increased in recent years.

Next, the team plans to explore implicit

and explicit attitude change across demo-

graphics and geographical regions, as well as

whether trends have changed since the 2016

U.S. presidential election. — Matthew Hutson