0

The Impact of Implicit Bias on Women and Underrepresented Minorities in STEMLEAD 2008

Dr. Sapna CheryanDepartment of PsychologyUniversity of WashingtonJuly 21, 2008

1

1

Increasing emphasis on diversity…

Yet women and minorities are still underrepresented in many domains.

2

0

10,000

20,000

30,000

40,000

50,000

N o n -Min o rity 4 1 ,9 4 0 1 3 ,8 2 8 2 3 ,7 8 6 1 6 ,3 9 0

As ia n /P a c. Is l. 5 ,8 4 9 3 ,2 6 9 2 ,9 2 5 4 ,3 7 8

U R M 3 ,2 4 7 1 ,0 8 2 1 ,6 6 7 1 ,7 5 8

B io lo g ica l a n d a g ricu ltu ra l s c ie n ce s

Ma th e m a tics a n d co m p u te r s c ie n ce s

P h ys ica l a n d e a rth s c ie n ce s

E n g in e e rin g

Number of Doctoral-Degreed Faculty (All Tenure Statuses) by Discipline and Ethnicity, 2003

(Source: Survey of Doctorate Recipients, 2003)

http://www.cpst.org Data provided by CPST and Lisa Frehill

3

Current Institution

URMAsian/Pacific

Islander Non-Minority

R1 < R1 R1 < R1 R1 < R1

PhDInst.

R1 79.2% 63.7% 81.7% 68.4% 86.9% 65.4%

<R1 19.8% 34.7% 16.7% 31.5% 10.2% 33.2%

Unk. 1.1% 1.6% 1.6% 0.1% 2.9% 1.4%

http://www.cpst.org Data provided by CPST and Lisa Frehill

A majority of faculty at R1 schools earned Ph.D.s at R1s . . .URM and Asian/Pacific Islanders at R1s are more likely to have come from non-R1 schools than non-minority faculty. (Source: National Study of Postsecondary Faculty, 2004)

4

The URM pipeline to doctoral degrees is far from “full”. . . Not at “parity” with representation in the population.(Degree awards, 2005)

14.1%12.4% 13.3% 12.3%

20.5% 19.9%18.7%

12.0%8.0%

11.0%

20.6%22.5%

9.7%10.1%

5.9%8.3%

6.6%

12.5% 13.1%

8.1%8.4%

0%

5%

10%

15%

20%

25%

30%

35%

Life Sciences Mathematics ComputerScience

Engineering PhysicalScience

Psychology SocialSciences

Per

cent

UR

MBachelor's Master's Doctorate

Source: CPST analysis of IPEDS data using NSF's WebCASPAR system. Life Sciences includes biological and agricultural sciences; Physical sciences includes the earth, atmospheric and ocean sciences disciplines. URM = Under-represented minority and includes African American, American Indian and Hispanics.

Parity line: 31% of U.S. 18-24 year olds are URM

Data provided by CPST and Lisa Frehill

The URM pipeline to doctoral degrees is far from “full” . . . Not at “parity” with representation in the population.(Degree awards, 2005)

5

Percent Women in STEM Fields (2004 CPST)

0

10

20

30

40

50

60

Bachelor's Degrees Master's Degrees Ph.D. Degrees Faculty

Perc

ent W

omen

6

Women in the Engineering "Pipeline"

19.4%22.6% 22.2%

17.4%

12.3%

5.2%

11.1%

0%

5%

10%

15%

20%

25%

30%

35%

40%

45%

50%

Bachelors,2006

Masters, 2006

Doctorates,2006

AssistantProfessors,

2004

AssociateProfessors,

2004

FullProfessors,

2004

EmployedEngineers,

2007

Source: CPST analysis of NSF's WebCASPAR database (degree data), American Society for Engineering Education (faculty data), and Bureau of Labor Statistics (overall employment).

7

Women as a Percent of Ph.D.s Employedin Universities & 4-Year Colleges by STEM Field and Rank, 2003

0%

10%

20%

30%

40%

50%

60%

70%

80%

C omputer Sciences

Life Sciences

Mathematics

Physica l Sciences

Psychology

Socia l Sciences

Engineering

Other Fac ultyA s s is tant Prof es s orA s s oc iate Prof ess orProf es s or

Source: Analys is of or iginal data from : National Sc ienc e F oundation. W omen, M inor i ties , and Persons with D is ab i l i ties in Sc ienc e and Engineer ing, 2007 .

http://w w w .cps t.org

8

8

Why we care1) People are missing out on well-respected, influential,

and flexible careers (Kalwarski, Mosher, Paskin, & Rosato, 2007)

2) STEM fields are missing out on potential talent (National Academy of the Sciences, 2003)

3) STEM fields are missing other perspectives (Margolis & Fisher, 2002)

4) Strength of diverse groups (Sommers, 2006)

9

Influence of racial composition onjury decision making (Sommers, 2006)

N = 29 juries (6 people each)

10

Diverse juries:• More information exchange

• Took longer• Discussed more case facts• Discussed more missing evidence

• More accurate• Fewer inaccurate statements• Fewer uncorrected inaccuracies

• More openness to discussing race• Discussed more race-related topics• Fewer objections to considering race

Almost all driven by Whites!

Influence of racial composition onjury decision making (Sommers, 2006)

11

Whites’ in diverse groups less likely to vote guilty, even before deliberations began

In diverse juries, Whites are:• Reminded to not be prejudiced• Processing trial info more closely • More receptive to discussing racism

Benefits of diversity

Influence of racial composition onjury decision making (Sommers, 2006)

cut

12

12

Why are we not there?

the role of bias

13

13

Common understanding:bias = conscious, intentional, to inflict harm

But what the research shows: bias = automatic, outside of our awareness,

unintentional, conflicts with our conscious beliefs

Bias is not what most people think it is

14

14

15

15

16

16

Unconscious bias: The Implicit Association Test

A demo

https://implicit.harvard.edu/

17

17

Resume study (Neumark, 1996)

18

18

Example of Gender Bias• Study of Swedish

Medical Research Council review

• Women needed to produce more than 99“impact factors” to be perceived as competent as men with only 20 impact factors.

Nature 387:341-343.

2

2.1

2.2

2.3

2.4

2.5

2.6

2.7

2.8

2.9

0-19 20-39 40-59 60-99 >99

Total Impact"C

ompet

ence

Sco

re"

Men

Women

Post-doc applications (Wennerås & Wold, 1997)

19

19

0

10

20

30

40

50

60

Train

ing

Teac

hing

Applica

tion

Rese

arch

Skills

& A

bilities

Care

er

FemaleMale

0

5

10

15

20

25

Person

al Lif

e

Publica

tions CV

Patie

nts

Colle

ague

s

Female

Male

“Exploring the color of Glass: Letters of Recommendation for Female and Male Medical Faculty”Discourse & Society 14: 191-220.

Example of Gender Bias

Recommendation letters (Trix & Psenka, 2003)

•Analysis of 300 letters of recommendation for medical faculty.

•Descriptions of women by letter writers emphasized teaching.

•Descriptors of men by letter writers emphasized their role as researchers and professionals.

•Fewer superlatives used to describe women.

20

20

Responding to Leaders (Butler & Geis, 1990)

21

21

Asking favors (Flynn, 2007)

service

22

22

Impact of stereotypes on career aspirations(Cheryan, Plaut, Davies, & Steele, under review)

Computer science majors are…

“Nerdy, techie, stay up late coding and drinking energy drinks, no social life.”

“Pale, sometimes socially frustrated, inquisitive, skilled, focused.”

“They are usually guys, very intense, very intelligent, intuitive, and quick. They don't frequently take showers.”

23

23

Room in Gates CS building decorated with stereotypical or non-stereotypical objects

Signaling belonging (Cheryan et al., under review)

24

24

Interaction: F(1, 35) = 10.22, p < .01Cheryan, Plaut, Davies & Steele (under review)

Signaling belonging (Cheryan et al., under review)

25

25

Interaction: F(1, 35) = 10.22, p < .01Cheryan, Plaut, Davies & Steele (under review)

Signaling belonging (Cheryan et al., under review)

26

26

Method Method (Spencer, Steele, & Quinn, 1999)

• Male (N = 24) and female (N = 30) students with college math experience

• Administered a 30 minute GRE math subject test, divided:“gender differences” vs. “no gender differences”

stereotype threat - fear of confirming a negative stereotype about your group (Steele, 1997)

Stereotype threat (Steele, 1997)

27

27

ResultsResults

Stereotype threat

0

5

10

15

20

25

30

35

40

45

"Genderdifference"

"No genderdifference"

Sco

re c

orr

ect

ed f

or

guess

ing

MenWomen

Spencer, Steele, & Quinn (1999)

28

28

0

5

10

15

20

25

30

35

40

45

"Genderdifference"

"No genderdifference"

Sco

re c

orr

ect

ed f

or

guess

ing

MenWomen

ResultsResults

Stereotype threat

F(1, 50) = 5.66, p < .05Spencer, Steele, & Quinn (1999)

29

29

Stereotype threat (Dar-Nimrod & Heine, Science, 2006)

MethodMethod• Female students with college math experience• Administered two math sections separated by a

verbal comprehension section1. “No gender differences”2. Gender differences because of genes3. Gender differences because of experience

Gender diff vs. no gender diff manip remind you of anything

30

30

Stereotype threat (Dar-Nimrod & Heine, Science, 2006)

31

31

Bias can manifest in different ways- deliberate- unconscious- “in the air”

Impact behaviors

Finding solutions (stay tuned)

Takeaways

Lessons from jury studyIndiv & structural & cultural…linking it to the rest of the workshopDavid’s fall email

32

32

• Acknowledge that diversity can be a competitive advantage

• Establish clear written procedures that minimize cognitive errors

• Promote diversity and ensure an equitable workplace at every level of the institution

• Construct welcoming environments

A few solutions

33

33

Thank you!

34

Pre-deliberation opinions

Main effect of race

Main effect of prime

Whites in diverse groups had fewer guilty votes than Whites in all-White groups

Sommers (2006)

Influence of racial composition onjury decision making (Sommers, 2006)

end