Does Science Promote Women? Evidence from Academia: 1973-2001 for presentation at: Science and Engineering Workforce Project at NBER October 20, 2005 Donna.

Does Science Promote Does Science Promote Women? Evidence from Women? Evidence from

Academia: 1973-2001Academia: 1973-2001for presentation at:for presentation at:

Science and Engineering Workforce Project at NBERScience and Engineering Workforce Project at NBEROctober 20, 2005October 20, 2005

Donna K. GintherDonna K. GintherUniversity of KansasUniversity of Kansas

Shulamit KahnShulamit KahnBoston UniversityBoston University

IntroductionIntroduction

Federal agencies and others Federal agencies and others monitor the status of women in monitor the status of women in science and find they are under-science and find they are under-represented without evaluating represented without evaluating likely causes.likely causes.

CAWMSET (2000)CAWMSET (2000) GAO (2004)GAO (2004) Nelson and Rogers (2005)Nelson and Rogers (2005)

IntroductionIntroduction

January 16, 2004 January 16, 2004 Chronicle of Chronicle of Higher EducationHigher Education reports: reports: “Women are Underrepresented “Women are Underrepresented in Sciences at Top Research in Sciences at Top Research Universities”Universities”

Is under-representation caused Is under-representation caused by the promotion process?by the promotion process?

Research QuestionResearch Question

Does Science Promote Women?Does Science Promote Women? We examine gender differences in:We examine gender differences in:

Tenure Track jobsTenure Track jobsPromotion to TenurePromotion to TenurePromotion to Tenured, Full Professor Promotion to Tenured, Full Professor

We find negligible gender We find negligible gender differences in promotion.differences in promotion.

Literature ReviewLiterature Review

From Scarcity to Visibility, From Scarcity to Visibility, Long Long et. al. (2001)et. al. (2001) Shows that women in science, Shows that women in science,

broadly defined, have made broadly defined, have made progressprogressRepresentation, Salary, PromotionRepresentation, Salary, Promotion

Aggregates data such that it’s Aggregates data such that it’s difficult to observe problemsdifficult to observe problems

Literature ReviewLiterature Review

Women in Science, Women in Science, Xie and Xie and Shauman (2003)Shauman (2003) Life course approach to science careersLife course approach to science careers Sex differences in research productivity Sex differences in research productivity

have declined, explained by observable have declined, explained by observable characteristicscharacteristics

Find small pay differences, some Find small pay differences, some promotion differences for nonacademicspromotion differences for nonacademics

Literature ReviewLiterature Review

Promotion in SciencePromotion in Science NSF (2004)NSF (2004)

Finds significant gender gap in Finds significant gender gap in promotion in sciencepromotion in science

Gap is reduced after controlling for Gap is reduced after controlling for family characteristicsfamily characteristics

Combine science and social science Combine science and social science in the analysisin the analysis

Literature ReviewLiterature Review

Academic Labor MarketsAcademic Labor Markets Promotion differentials: Long, Allison Promotion differentials: Long, Allison

and McGinnis (1993) Kahn (1993, and McGinnis (1993) Kahn (1993, 1995) Ginther and Kahn (2004)1995) Ginther and Kahn (2004)

Ginther and Hayes 1999, 2003; Ginther Ginther and Hayes 1999, 2003; Ginther 2004, 2003, 20012004, 2003, 2001

There is no, single academic labor There is no, single academic labor market. market.

DataData

Use 1973 - 2001 Survey of Use 1973 - 2001 Survey of Doctorate Recipients (SDR)Doctorate Recipients (SDR) Biennial, Longitudinal Survey of Biennial, Longitudinal Survey of

U.S. DoctoratesU.S. Doctorates Used by NSF to analyze scientific Used by NSF to analyze scientific

labor forcelabor force

DataData

Longitudinal SampleLongitudinal Sample:: Individuals who received their Individuals who received their Ph.D. between 1972 and 1991 Ph.D. between 1972 and 1991 observed between 1973 and observed between 1973 and 2001.2001. Tenure-track sub-sample: Those Tenure-track sub-sample: Those

who report ever having a tenure-who report ever having a tenure-track job.track job.

DataData

Academics in the Sciences: Academics in the Sciences: Life SciencesLife Sciences

Agriculture and Food ScienceAgriculture and Food Science Biology and Life SciencesBiology and Life Sciences

Physical SciencesPhysical Sciences ChemistryChemistry Earth ScienceEarth Science Physics Physics Computer Science / MathematicsComputer Science / Mathematics

EngineeringEngineering

DataData Dependent variables: Dependent variables:

Probability of Tenure Track jobProbability of Tenure Track job Probability of Promotion to tenure Probability of Promotion to tenure

and full professorand full professor Duration between Ph.D. and Duration between Ph.D. and

promotion to tenure and full promotion to tenure and full professorprofessor

DataData Independent variables: Independent variables:

GenderGender Age Ph.D.Age Ph.D. Year Ph.D.Year Ph.D. RaceRace Academic fieldAcademic field Degree institution characteristicsDegree institution characteristics

DataData Time-varying Independent Time-varying Independent

variables: variables: University/College employer University/College employer

characteristics characteristics Rank and Tenure status Rank and Tenure status Primary / Secondary work Primary / Secondary work

activitiesactivities Government Support of ResearchGovernment Support of Research Publications****Publications****

Data DifficultiesData Difficulties

Biennial SurveyBiennial Survey Changes in the sampling frameChanges in the sampling frame Numerous missing observations, Numerous missing observations,

required a lot of imputationrequired a lot of imputation Imputed productivity from three Imputed productivity from three

years of observed publicationsyears of observed publications

Empirical MethodsEmpirical Methods

Probit models (dependent Probit models (dependent variable):variable):

Tenure track within 5 years of Tenure track within 5 years of Ph.D.Ph.D.

Tenured at 11 years after Tenured at 11 years after Ph.D.Ph.D.

Tenured, Full Professor at 15 Tenured, Full Professor at 15 years after Ph.D.years after Ph.D.

Empirical MethodsEmpirical Methods

Hazard of PromotionHazard of Promotion Proportional Hazards Proportional Hazards

Model with time-varying Model with time-varying covariatescovariates

Hazard model is preferred Hazard model is preferred specificationspecification

Stylized FactsStylized Facts

Women’s representation in Women’s representation in science depends upon the science depends upon the fieldfieldLife Science—ProgressLife Science—ProgressPhysical Science, Engineering,Physical Science, Engineering,

—Anemic representation—Anemic representation

1970 1975 1980 1985 1990 1995 2000

Year

0

10

20

30

40

50

Per

cent

age

Fem

ale

ScienceLife SciencePhysical ScienceEngineering

Figure 1: Percentage of Doctorates Granted to Females, 1974-2000 Suvey of Earned Doctorates

Source: 1974-2000 Survey of Earned Doctorates

Probability of Tenure Track Probability of Tenure Track JobJob

Model 1 Model 2 Model 3 Model 4

Science -0.038 -0.044 -0.031 -0.033

(0.009) (0.009) (0.010) (0.010)

Life Science -0.041 -0.059 -0.075 -0.077

(0.012) (0.013) (0.013) (0.013)

Physical Science -0.002 0.003 -0.010 -0.015

(0.016) (0.016) (0.017) (0.017)

Engineering 0.000 0.014 0.009 0.013

(0.033) (0.034) (0.035) (0.035)

Demographics No Yes Yes Yes

Degree Characteristics No No No Yes

Fields No No Yes Yes

Probability of Tenure Track JobProbability of Tenure Track Job—Including family variables—Including family variables

Science Life Phys. Eng.

Female 0.156 0.108 0.206 0.072

(0.018) (0.025) (0.029) (0.064)

Female*Married -0.171 -0.149 -0.236 0.009

(0.024) (0.033) (0.041) (0.092)

Female*Total Children -0.029 -0.022 -0.055 -0.053

(0.013) (0.017) (0.022) (0.045)

Female*Young Children -0.059 -0.068 -0.021 0.000

(0.028) (0.038) (0.050) (0.100)

1975 1980 1985 1990 1995 2000

Year

0

10

20

30

40

Per

cent

age

Fem

ale

A) Science

1975 1980 1985 1990 1995 2000Year

0

10

20

30

40

Per

cent

age

Fem

ale

B) Life Science

AssistantAssociateFull

Figure 2: Percentage Female by Academic Rank, Science Disciplines

1975 1980 1985 1990 1995 2000

Year

0

10

20

30

40

Per

cent

age

Fem

ale

C) Physical Science

1975 1980 1985 1990 1995 2000

Year

0

10

20

30

40

Per

cent

age

Fem

ale

D) Engineering

Source: 1973-2001 Survey of Doctorate Recipients

1975 1980 1985 1990 1995 2000

Year

0

10

20

30

Perc

enta

ge F

emal

e

Source: 1973-2001 Survey of Doctorate Recipients

ScienceLife SciencePhysical ScienceEngineering

Figure 3: Percentage of Tenured Faculty who are Female, by Discipline

Gender Promotion GapGender Promotion Gap

Previous Research (Ginther and Previous Research (Ginther and Kahn 2004; Ginther and Hayes 1999, Kahn 2004; Ginther and Hayes 1999, 2003) has shown a significant gender 2003) has shown a significant gender promotion gap inpromotion gap in EconomicsEconomics HumanitiesHumanities

What about science?What about science?

Promotion to TenurePromotion to TenureFull

SampleLife

SciencePhysical Science

EngineerIng

Female Probit Coefficient 0.00 -0.03 0.01 0.02Promoted 11 Yrs Ph.D. (0.88) (0.19) (0.73) (0.75)

Survival Curve 1.26 0.35 0.00 0.49HomogeneityHomogeneity (0.26) (0.55) (0.95) (0.49)

Female Risk Ratio 0.97 1.02 1.00 1.06(No Covariates) (0.33) (0.60) (0.96) (0.56)Model 1 Female Risk Ratio 0.95 0.89 0.93 1.00(Covariates ex. Productivity) (0.14) (0.02) (0.22) (0.97)Model 2 Female Risk Ratio 0.97 0.92 0.94 1.03(Including Productivity) (0.29) (0.07) (0.28) (0.82)

0 5 10 15 20Years after Ph.D.

0.0

0.2

0.4

0.6

0.8

1.0P

red

icte

d S

urv

ival F

unct

ion

Predicted Survival

A. Science

MaleFemale

Figure 4: Predicted Survival Without Tenure Functions, by Gender and Discipline

0 5 10 15 20Years after Ph.D.

0.0

0.2

0.4

0.6

0.8

1.0

Pre

dic

ted S

urv

ival F

unct

ion

B. Life Science

0 5 10 15 20Years after Ph.D.

0.0

0.2

0.4

0.6

0.8

1.0

Pre

dic

ted S

urv

ival F

unct

ion

C. Physical Science

Source: 1973-2001 Survey of Doctorate Recipients

0 5 10 15 20Years after Ph.D.

0.0

0.2

0.4

0.6

0.8

1.0

Pre

dic

ted S

urv

ival F

unct

ion

D. Engineering

Promotion to Full ProfessorPromotion to Full ProfessorFull

SampleLife

SciencePhysical Science

Engineer-ing

Female Probit Coefficient -0.05 -0.09 -0.02 0.09Promoted 15 Yrs Past Ph.D. (0.02) (0.00) (0.51) (0.37)Survival Curve 7.57 0.61 11.59 0.14 Homogeneity (0.01) (0.44) (0.00) (0.71)Female Risk Ratio 0.90 0.96 0.79 0.95(No Covariates) (0.01) (0.48) (0.00) (0.74)Model 1 Female Risk Ratio 0.95 0.93 0.87 1.09(Covariates ex. Productivity) (0.34) (0.37) (0.11) (0.89)Model 2 Female Risk Ratio

0.97 0.96 0.891.04

(Including Productivity Covariates)

(0.54) (0.61) (0.19) (0.82)

0 5 10 15Years after Tenure

0.0

0.2

0.4

0.6

0.8

1.0P

red

icte

d S

urv

ival F

unct

ion

Predicted Survival

A. Science

MaleFemale

Figure 5: Predicted Survival Without Full Professor Functions, by Gender and Discipline

0 5 10 15Years after Tenure

0.0

0.2

0.4

0.6

0.8

1.0

Pre

dic

ted S

urv

ival F

unct

ion

B. Life Science

0 5 10 15Years after Tenure

0.0

0.2

0.4

0.6

0.8

1.0

Pre

dic

ted S

urv

ival F

unct

ion

C. Physical Science

Source: 1973-2001 Survey of Doctorate Recipients

Tenure at Research ITenure at Research IFull

SampleLife

SciencePhysical Science

EngineerIng

Female Probit Coefficient 0.01 -0.03 -0.01 0.00Promoted 15 Yrs Past Ph.D. (0.84) (0.40) (0.88) (0.97)Survival Curve 3.48 1.11 0.27 2.97 Homogeneity (0.06) (0.29) (0.61) (0.08)Female Risk Ratio 0.92 0.94 0.96 1.24(No Covariates) (0.10) (0.35) (0.65) (0.14)Model 1 Female Risk Ratio 1.00 0.91 0.97 1.17(Covariates ex. Productivity) (0.99) (0.22) (0.72) (0.33)Model 2 Female Risk Ratio 1.02 0.93 0.98 1.23(Including Productivity Covariates)

(0.75) (0.31) (0.86) (0.20)

Full Professor at Research IFull Professor at Research IFull

SampleLife

SciencePhysical Science

Engineer-ing

Female Probit Coefficient -0.04 -0.06 -0.04 0.27Promoted 15 Yrs Past Ph.D. (0.24) (0.11) (0.56) (0.09)Survival Curve 15.69 6.79 9.52 0.00 Homogeneity (0.00) (0.01) (0.00) (0.95)Female Risk Ratio 0.77 0.80 0.68 0.99(No Covariates) (0.00) (0.02) (0.01) (0.95)Model 1 Female Risk Ratio 0.89 1.02 0.68 1.02(Covariates ex. Productivity) (0.18) (0.87) (0.01) (0.95)Model 2 Female Risk Ratio 0.91 1.01 0.71 1.07(Including Productivity

Covariates)(0.23) (0.90) (0.03) (0.78)

ConclusionsConclusions

Does Science Promote Women?Does Science Promote Women? YESYES

Gender differences in tenure trackGender differences in tenure track Explained by marriage and family.Explained by marriage and family.

Little evidence of gender Little evidence of gender promotion gap to tenure, full promotion gap to tenure, full professor.professor.

ConclusionsConclusions

One exception:One exception: Full Professors in Physical Science at Full Professors in Physical Science at

Research I UniversitiesResearch I Universities Each academic field presents Each academic field presents

different hurdles for women in terms different hurdles for women in terms of pay and promotion.of pay and promotion. Science: 12% gender gap in salaries at Science: 12% gender gap in salaries at

Full Professor (Ginther 2004)Full Professor (Ginther 2004)