Why do girls' STEM aspirations differ between countries?

Why do girls' STEM aspirations differ between countries?
How cultural norms and institutional constraints shape young
women's occupational aspirations
Marcel Helbig, Wissenschaftszentrum Berlin für Sozialforschung
Kathrin Leuze, Leibniz Universität Hannover
Womens‘ share in tertiary education 1970/80
Below 30 %
52 to 60 %
30 to 40 %
60 to 70 %
40 to 48 %
70 % and more
48 to 52 %
Quelle: Unesco: Global Education Digest 2009
2
Seite 2
Womens‘ share in tertiary education 2005
Below 30 %
52 to 60 %
30 to 40 %
60 to 70 %
40 to 48 %
70 % and more
48 to 52 %
Quelle: Unesco: Global Education Digest 2009
3
Seite 3
.3
.2
.1
0
Share of girls and boys with STEM preferences
.4
STEM aspirations of 15-year-old girls and boys (PISA 2006)
NL SE FR FI LU LV CH BE DK AT IS US SK NO DE UK IE EE PL IT CZ HU LT ES SI CA PT GR BG JP
Girls' preferences
Source: PISA 2006, authors‘ calculations
Boys' preferences
Seite 4
Research question
How can we explain cross-national variations
in girls‘ preferences for STEM occupations?
STEM occupations:
- Science: e.g. physicists, chemists, biologists
- Technology: e.g. computing professionals, computer
programmers
- Engineering: e.g. electrical engineers, mechanical
engineers, chemical engineers
- Mathematics: mathematicians, statisticians
Seite 5
5
Rational choice explanations for STEM aspirations
(Jonsson 1999, Eccles 1994)
• Probabilities of success / expectancy-value:
• Higher competencies in and valuation of mathematics
and science
 Higher probabilities of success for entering STEM
occupations
• H1: Countries with high female-to-male competence
advantage in reading and mathematics
• Better chances for girls in general to succeed in the
education system
• Better confidence of girls for successfully entering STEM
occupations
 more STEM aspirations of girls
Seite 6
Rational choice explanations for STEM aspirations
(Jonsson 1999)
• Benefits:
• Different valuation of income and status (boys) vs.
working with others and communicating (girls)
 gender-differentiated valuation of returns to STEM
occupations
• H2: Countries with a large service sector
• good employment chances (benefits) for women in
„female-typcical“ occupations
• Better options for girls to achieve their work-related benefit
expectations
 less STEM aspirations of girls
Seite 7
Socialization explanations for STEM aspirations
(Hannover 2008, Ruble et al. 2006, Eccles 1987)
• Normative gender-role expectations:
• gender-typical parental expectations in general
• parents with higher socio-economic status
 more progressive gender role expectations, also
regarding occupational aspirations
• H3: Countries with progressive gender ideology:
• less gender-typical role expectations of the wider social
environment (teachers, peers, media etc.) regarding
occupational preferences
 more STEM aspirations of girls
Seite 8
Socialization explanations for STEM aspirations
(Hannover 2008, Ruble et al. 2006, Eccles 1987)
• Same-sex role models:
• Girls orient themselves towards mothers and boys
towards fathers as role models to learn from
• Mothers working in STEM occupations
 ‘same-sex role models‘ to learn from
• H4: Countries with less occupational sex segregation:
• more same-sex role models in the wider social
environment (teachers, acquaintances, media, etc.)
working in gender-atypical occupations
 more STEM aspirations of girls
Seite 9
Data and methods
 Programme for International Student Assessment 2006
 Focus on 30 EU and OECD countries:
AT, BE, BG, CA, CZ, DK, EE, FI, FR, DE, GR, HU, IC, IE, IT, JP, LT,
LV, LU, NL, NO, PL, PT, SK, SI, ES, SE, CH, UK, US
 Dependent variable: STEM occupational aspiration of 15year-old pupils at the age of 30
 Sample restriction: Aspiration of academic occupation
(ISCO major groups 1 to 3)
61,394 girls, 49,835 boys
 Method: Multilevel random intercept logistic regression
models with countries as second level (AMEs)
 all metric variables are mean centered and standardised
Seite 10
10
Individuallevel control variables
 General: age, grade, migration background
 Rational choice:
 probability of success / expectancy:
objective: mathematics and science competencies
subjective: science self-concept
 value of science: personal value of science
 benefits: instrumental science motivation
 Socialization:
 gender norms: socioeconomic status of household
(ESCS index of economic, social and cultural status)
 same-sex role model:
STEM occupation of mother and father
Seite 11
11
Countrylevel independent variables
Rational choice:
• Female-to-male overall competence advantage:
Differences between mean reading and mathematics scores for
girls and boys for the year 2006 (PISA 2006)
• Size of the service sector:
Service sector employment as share of total employment 2006
Socialisation:
• Progressive gender ideology:
Country mean of gender role attitudes towards women‘s and
mothers‘ employment (ISSP 2002, WVS 2000)
• Occupational sex segregation:
Index of Dissimilarity (ILO 2000)
Countrylevel control variable:
Female to male tertiary graduation rate 2006 (Eurostat, OECD)
Seite 12
12
Results of the multilevel random intercept logistic
regression models – countrylevel variables
Per cent women in
occupational category
H1: High female-to-male
competence advantage
H2: Large size of the service
sector
H3: Progressive gender
ideology
H4: More occupational sex
segregation
Female-to-male graduation
rate
Variance country level
intercept only model %
% explained individual vars.
% explained full model
Girls
Boys
Expected
Observed
Expected
Observed
+
0.155*
?
-0.043
-
-0.350***
?
-0.336***
+
-0.014
-
-0.027
-
-0.382***
+
-0.103
- 0.098
-0.034
11.35
5.39
-10.41
46.73
-13.96
43.23
Significant * p<0.05, ** p<0.01, *** p<0.001, std. coeff., average marginal effects Seite 13
Girls‘ competence advantage: interaction with sex
-3.5
-3
Linear Prediction
-2.5
-2
-1.5
-1
Predictive Margins with 95% CIs
2
NL
3
4
5
6
7
8
9
10 11 12 13
Girls' overall competence advantage
Boys
Girls
14
15
16
BG
Seite 14
Size of service sector: interaction with sex
-4
-3
Linear Prediction
-2
-1
0
Predictive Margins with 95% CIs
50
PL
55
60
65
70
Size of service sector
Boys
Girls
75
80
LU
Seite 15
Progressive gender norms: interaction with sex
-3.5
-3
Linear Prediction
-2.5
-2
-1.5
-1
Predictive Margins with 95% CIs
.4
JP
.45
.5
.55
.6
Progressive gender norms
Boys
Girls
.65
.7
DK
Seite 16
Index of dissimilarity: interaction with sex
-5
Linear Prediction
-4
-3
-2
-1
Predictive Margins with 95% CIs
40
US
45
50
55
60
Index of dissimilarity
Boys
Girls
65
70
CZ
Seite 17
Summary of main findings
 Cross-national variations in STEM aspirations matter, but
not as much as individuallevel differences
 more longitudinal analysis of STEM aspirations on the
individual level
 On the country level, girls‘ STEM aspirations are higher
 the better girls perform in school relative to boys
 the smaller the size of the service sector
 the less progressive the prevailing gender norms
 the lower the general occupational sex segregation
 Question: What to do?
 Are gender-typical occupational aspirations problematic?
Seite 18
THANK YOU FOR YOUR
ATTENTION!
Seite 19
Results of the multilevel random intercept logistic
regression models – individuallevel variables
STEM occupational
aspiration
Girls
Boys
Expected
Observed
Expected
Observed
Math competencies
+
0.045
+
0.137***
Science competencies
+
0.193***
+
0.107**
Science self-concept
+
0.145***
+
0.094***
Personal value of science
+
0.157***
+
0.084***
Instrumental science
motivation
+
0.650***
+
0.265***
High parental SES
+
-0.111***
-
-0.155***
Mother STEM occupation
+
0.810***
0
0.331**
Father STEM occupation
0
0.549***
+
0.573***
% Variance explained
21.62
15.00
Significant * p<0.05, ** p<0.01, *** p<0.001, std. coeff., average marginal effects Seite 20
Index of Dissimilarity of 30 OECD and EU countries
80
70
60
50
40
30
20
10
0
KO US JP GR IT CA PO NL ES LU PT SI UK FR BE BU CH DE AT IR DK LT HU NO LV IC SE FI EE SK CZ
Source: ILO, authors‘ estimations
Seite 21
21
Further robustness checks
 Models without Japan: same results
 Girls‘ (dis-)advantage in math: similar results, but not as strong
 Male to female science self-concept (country mean)  stronger
gender-essentialist beliefs of male- and female-typical science
self-concepts as part of the gendered selves
 General value of science (country mean)  increased value of
working in male-typical occupations also for girls
 no effect
 Gender pay gap  higher benefits in male-typical occupations
 no effect
 Share women in management  increased probability of
success of girls for working in a male-typial occupational field
 no effect
Seite 22
Girls‘ (dis-)advantage in math: interaction with sex
-3.5
-3
Linear Prediction
-2.5
-2
-1.5
-1
Predictive Margins with 95% CIs
-4
-3
-2
-1
Female (dis-)advantage math
girl=0
0
1
girl=1
Seite 23