Generics and Stereotypes Andrei Cimpian, NYU andrei.cimpian@nyu.edu - - PowerPoint PPT Presentation

Generics and Stereotypes

Andrei Cimpian, NYU

andrei.cimpian@nyu.edu http://CimpianLab.com Twitter: @AndreiCimpian

What are stereotypes1,2?

• Despite variability across researchers, 2

common elements:

(1) Stereotypes are beliefs or generalizations about groups (2) that guide our interactions with the social world

(e.g., Ashmore & Del Boca, 1981; Judd & Park, 1993; Jussim, 2012)

1 explicit (vs. implicit) stereotypes; 2 cognitive (vs. cultural) products

Also…

• A shared assumption that stereotyping is

continuous with general processes involved in reasoning about categories

• Differences in content, not structure or process

(e.g., Ashmore & Del Boca, 1981; Judd & Park, 1993; Jussim, 2012)

The problem

• “Stereotypes = beliefs/generalizations about

groups” is underspecified

• Beliefs about groups – and about categories

more generally – can be generic or statistical (quantified)

politicians are liars some/all/64% of politicians are liars

(e.g., Carlson, 1977; Carlson & Pelletier, 1995; Gelman, 2003; Gelman et al., 1998, 2004; Leslie, 2007, 2008; Nickel, 2016; Sterken, 2015)

This matters because…

• Generic and statistical beliefs about categories

differ considerably

• in how they map onto the world
• in their cognitive ease
• in their developmental trajectory

(e.g., Carlson, 1977; Cimpian, 2016; Leslie & Gelman, 2012; Gelman, 2004; Leslie, 2007, 2008; Nickel, 2016; Sterken, 2015)

The question arises…

• Well, what are stereotypes? Are they generic

beliefs about groups, or are they statistical beliefs?

• Not necessarily either/or
• And how could we find out?

Our criterion

• Despite variability across researchers, 2

common elements:

(1) Stereotypes are beliefs or generalizations about groups (2) that guide our interactions with the social world

Functional criterion: Which type of belief is more closely related to social judgments?

Prediction

• Social judgments will rely more on generic

than statistical beliefs

• because generic beliefs are cognitively easy and

developmentally primitive

Overview

• Four studies comparing generic and

statistical beliefs as predictors of social judgments

• Broader implications
• Can stereotypes said to be accurate?

Study 1

Matt Hammond Lecturer (Assistant Professor) Psychology, Victoria University of Wellington (NZ)

Data and syntax on OSF: https://osf.io/s6zab/

Study 1: Materials

• Finding stereotypes:
• searched Google Scholar with relevant key words

(e.g., “stereotypes”) and downloaded papers on first 3 pages of results

• 45 unique stereotypes
• Participants on MTurk (n = 65) were asked if people

believed each stereotype

• 27 stereotypes were retained (with average scores

above the midpoint)

Study 1: Materials

• Examples (in generic form):
• African Americans are athletic
• African Americans are musical
• Asians are intelligent
• Doctors are intelligent
• Lawyers are greedy
• Men are logical
• Women are compassionate

Study 1: Participants

• 97 participants from MTurk
• 55% women
• ages: 20–83 (M = 36.54, SD = 13.18)

Study 1: Measures

• For each of the 27 stereotypes:
• Generic belief:

Ps rated agreement with generic statements: e.g., “African Americans are athletic” −3 = strongly disagree to 3 = strongly agree

• Statistical belief:

Ps estimated the prevalence of the trait: e.g., “What percentage of African Americans are athletic?” 0–100 sliding scale AS SEPARATE BLOCKS

Study 1: Measures

• The social judgment DV:
• Does an unfamiliar member of a group possess a

stereotypical trait? “Suppose that Person Y is African American. Is Person Y athletic?” −3 = very unlikely to 3 = very likely

Study 1: Results

Generic Beliefs Social Judgments (expectations) Statistical Beliefs

multilevel model with cross‐classified random effects for subjects and items

β = .315*** [.277, .354] β = .514*** [.476, .552] b = −.199 [−.269, −.129]

*** p < .001 ***

Study 1: Additional results

• Same results for the subsets of stereotypes about

race/ethnicity, gender, and profession, bs < −.208, ps < .001

• No moderation by social desirability (Crowne &

Marlowe, 1960; e.g., “I never hesitate to go out of my way to help someone in trouble”)

Study 1: Conclusion

• Initial evidence that generic beliefs provide a stronger

guide for social judgments than statistical beliefs do

• Generic beliefs might be more central to the structure of

stereotypes

Study 2

• Why do social judgments rely more on generic (vs.

statistical) beliefs?

• Possible answer: because reasoning with generic

beliefs in easier (e.g., Cimpian, 2016; Hampton, 2012)

Study 2

• If so, we might expect stronger reliance on generic

beliefs (and weaker reliance on statistical beliefs) for certain participants:

(1) those with intuitive (vs. analytic) cognitive styles: The Cognitive Reflection Test (Frederick, 2005):

e.g., “Soup and salad cost $5.50 in total. The soup costs a dollar more than the salad. How much does the salad cost?”

Study 2

• If so, we might expect stronger reliance on generic

beliefs (and weaker reliance on statistical beliefs) for certain participants:

(1) those with intuitive (vs. analytic) cognitive styles (2) those with more authoritarian tendencies: Right‐Wing Authoritarianism (Altemeyer, 1996)

e.g., “It is always better to trust the judgment of the proper authorities in government…”

Social Dominance Orientation (Sidanius & Pratto, 1999)

e.g., “Inferior groups should stay in their place”

Study 2

• If so, we might expect stronger reliance on generic

beliefs (and weaker reliance on statistical beliefs) for certain participants:

(1) those with intuitive (vs. analytic) cognitive styles (2) those with more authoritarian tendencies (3) those who are older (?) e.g., working memory and fluid reasoning abilities typically decline with age (e.g., Salthouse, 2015).

Some evidence from Study 1

Generic Beliefs Social Judgments (expectations) Statistical Beliefs

β = .315*** [.277, .354] β = .514*** [.476, .552] b = −.199 [−.269, −.129]

*** p < .001 *** Older (+1 SD) participants: b = .641*** [.583, .699] Younger (–1 SD) participants: b = .474*** [.420, .527]

Study 2

• New set of stereotypes:
• Participants on MTurk (n = 33) were asked to list common

stereotypes, regardless of whether they believed them to be true

• Only stereotypes mentioned by 3 or more participants

were retained

• Final set: 30 stereotypes

Study 2

• Examples (in generic form):
• Asian people are bad drivers
• Black people are lazy
• Blondes are dumb
• Muslims are terrorists
• Native Americans are alcoholics
• Southerners are ignorant
• Teenagers are irresponsible
• White people are racist

Study 2

• Changed how we measured generic beliefs:
• same sliding‐scale format used for statistical beliefs (except

still Strongly Disagree to Strongly Agree)

Study 2: Main Results

Generic Beliefs Social Judgments (expectations) Statistical Beliefs

β = .329*** [.304, .354] β = .458*** [.433, .484] b = −.129 [−.175, −.084]

*** p < .001 *** N = 185, MTurk 68% women ages 18–78, M = 37.13

Study 2: Cognitive Style

Generic Beliefs Social Judgments (expectations) Statistical Beliefs

β = .329*** [.304, .354] β = .458*** [.433, .484] b = −.129 [−.175, −.084]

*** p < .001 *** Analytic thinkers (+1 SD): b = .366*** [.324, .408] Intuitive thinkers (–1 SD): b = .296*** [.257, .335]

Study 2: Authoritarianism

Generic Beliefs Social Judgments (expectations) Statistical Beliefs

β = .329*** [.304, .354] β = .458*** [.433, .484] b = −.129 [−.175, −.084]

*** p < .001 *** High‐authoritarian subjects (+1 SD): b = .283*** [.242, .324] Low‐authoritarian subjects (–1 SD): b = .379*** [.339, .419] High‐authoritarian subjects (+1 SD): b = .521*** [.478, .565] Low‐authoritarian subjects (–1 SD): b = .428*** [.389, .467]

Study 2: Age

Generic Beliefs Social Judgments (expectations) Statistical Beliefs

β = .329*** [.304, .354] β = .458*** [.433, .484] b = −.129 [−.175, −.084]

*** p < .001 *** Older (+1 SD) participants: b = .500*** [.456, .544] Younger (–1 SD) participants: b = .425*** [.383, .466]

Study 2: Conclusion

• Additional evidence that generic beliefs are more

strongly related to social judgments than statistical beliefs are

• Social judgments may rely on generic beliefs more in

part because they are low‐effort

Study 3

• Would we find the same results with respect to non‐

social categories (animals, artifacts, etc.)?

• We compared social and non‐social stimuli
• 30 non‐social stimuli were borrowed from Khemlani, Leslie,

and Glucksberg (2012; e.g., ducks lays eggs; barns are red; sharks attack swimmers)

• Same 30 social stimuli as in Study 2
• Identical procedure for the two sets of stimuli

Study 3

• We simplified the statistical belief question:

“Out of all of the [category members], how many [have feature]?” sliding scale: none to all

Study 3: Social stimuli

Generic Beliefs Social Judgments (expectations) Statistical Beliefs

β = .330*** [.299, .362] β = .451*** [.420, .483] b = −.121 [−.176, −.065]

*** p < .001 *** N = 98, MTurk

Study 3: Non‐social stimuli

Generic Beliefs Non‐social Judgments (expectations) Statistical Beliefs

β = .605*** [.574, .636] β = .322*** [.292, .353] b = .283 [.227, .338]

*** p < .001 *** N = 96, MTurk

Generic Beliefs Social Judgments (expectations) Statistical Beliefs

β = .330*** [.299, .362] β = .451*** [.420, .483] b = −.121 [−.176, −.065]

*** p < .001 *** Older Younger Analytic Intuitive High‐authoritarians Low‐authoritarians

Study 3: Social stimuli – moderators

Study 3: Non‐social stimuli – moderators

*** p < .001 Analytic Intuitive Analytic Intuitive

Generic Beliefs Non‐social Judgments (expectations) Statistical Beliefs

β = .605*** [.574, .636] β = .322*** [.292, .353] b = .283 [.227, .338]

***

Study 3: Conclusion

• Contrast between judgments in the social and non‐

social domains:

• Non‐social judgments exhibited a stronger link with

statistical (vs. generic) beliefs

• Potentially, unique role of generic beliefs in the social

domain

• Interesting to speculate about why: greater ambiguity and

complexity in the social domain?

Study 4

• Expand our investigation to a new social judgment:

explanations for stereotypical features of individuals

• In particular, biological (essentialist) explanations:

“Person Y is Asian and is smart. To what extent is Person Y’s being smart due to their biology?” −3 = Not at All to 3 = A Great Deal

Dar‐Nimrod & Heine, 2011; Gelman, 2003

Study 4: Main results

Generic Beliefs Statistical Beliefs

β = .086*** [.044, .127] β = .174*** [.133, .215] b = −.088 [−.164, −.013]

* p < .05 *** p < .001 * N = 177, MTurk

Social Judgments (biological explanations)

(see also Cimpian & Markman, 2011; Cimpian & Erickson, 2012)

Study 4: Moderators

Generic Beliefs Social Judgments (biological explanations) Statistical Beliefs

β = .086*** [.044, .127] β = .174*** [.133, .215] b = −.088 [−.164, −.013]

* p < .05 *** p < .001 * Analytic Intuitive High‐authoritarians Low‐authoritarians

Study 4: Conclusion

• Generic (vs. statistical) beliefs are also more predictive
• f biological explanations for individuals’ stereotypical

traits

Take‐away

• Our initial question:

What are stereotypes? Are they generic beliefs about groups, or are they statistical beliefs?

• Functional criterion:

Which type of belief is more closely related to social judgments?

• Our evidence suggested an answer: generic beliefs

Take‐away

• Generic beliefs are more central to stereotype

structure—the more powerful and active component

• f stereotypes

Advantages of this view

• Incorporates important insights from the literature on

concepts and categorization

• Resolves a tension in the literature:
• Stereotypes are overgeneralizations and oversimplifications
• Stereotypes can also reflect statistics in the environment
• Is consistent with prominent views of stereotypes as

low‐effort cognitive shortcuts that people adopt to cope with the overwhelming complexity of the social environment (e.g., Fiske, 1998)

Implications of this view: Accuracy

• If stereotypes are (mostly) generic beliefs, what does

this mean for their potential accuracy?

• Reasons to be skeptical – generic beliefs…
• were preferred by low‐effort, prejudiced thinkers
• have a loose mapping to statistical facts (e.g., Leslie, 2008)
• are resistant to counterevidence (e.g., Chambers et al., 2008)
• give the impression of strong statistical support (e.g., Cimpian et

al., 2010)

• accompanied by misleading explanatory intuitions (e.g., Cimpian

& Markman, 2011)

Thank you!