Post on 28-Dec-2015
transcript
Why don’t women do as well as men in Chemistry Finals at
Oxford ?Jane mellanby
Department of Experimental Psychology
Chemistry at Oxford
FHS Chemistry 1996-2013: % Firsts by gender (MChem only)
0.0%
10.0%
20.0%
30.0%
40.0%
50.0%
60.0%
Male 32.0% 25.8% 33.6% 29.6% 32.8% 44.5% 46.0% 37.3% 40.7% 38.8% 39.8% 46.5% 46.4% 45.1% 50.0% 45% 45% 52%
Female 19.6% 34.0% 22.2% 37.5% 36.0% 14.3% 23.6% 28.3% 31.4% 35.9% 31.1% 36.2% 41.3% 34.8% 30.0% 46% 29% 30%
1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
• If it is a problem, it is mainly an Oxford (and Cambridge) problem not specifically a Chemistry problem
‘Gender gap’ 2012
26% of women and 32% of men obtained firsts
In any given FHS either men and women do the same or men do better than women
Few differences are actually statistically significant within an honour school
‘Gender gap’ 2013
• Overall figures not yet available• Men did better in:
HistoryEnglish language and literatureM. PhysicsMedical sciencesBiochemistryEngineering scienceBiological Sciences
Factors in the University structure
• Proportion of women professors and lecturers in a subject ?
• Proportion of women undergraduates in a subject ? Critical mass requirement ?
Proportion of women lecturers and professors ?
• MPLS : 13.2% women
• English: 62.7% women
• But both show a ‘gender gap’ in Finals results
Proportion of women undergraduates ?
• MPLS: ca. 29% women
• English: ca. 61% women
• Both show the ‘gender gap’ in firsts.
• So presence of a ‘critical mass’ of females reading a subject does not eliminate ‘gap’.
The studies
• Study 1. 1996-7. Questionnaire before Finals. Finals class
and marks. (Mellanby, Martin and O’Doherty, 2000)
• Study 2. 2002-2008. Oxford Admissions Study. (Mellanby and Zimdars, 2011; Mellanby, Zimdars and Cortina-borja, 2013)
Questionnaire at admissions and before Finals.
First public exam and finals class and marks
Strategy: Study 1
• Measure everything we can think of
• See which have sex differences
• See which predict Finals marks and class
• Only those which have sex difference AND predict Finals can explain gender gap
Conclusions (Study 1)
Measure Sex difference Predicts ?Verbal ability NO YESNon-verbalability
YES NO
Academicefficacy
YES NO
Academicmotivation
NO NO
Risk-taking NO NOHours of work NO NO
Conclusions (Study 1): Emotional state
Measure Sex difference Predicts ?Depression YES NOAnxiety YES NOExam stress YES NOLoneliness YES NOSelf-esteem YES NOHappiness YES NO
The only factor in Study 1...
That had BOTH a sex difference
AND predicted degree class was
expectation of a first:
Males expected to do better, and their expectation correlated with their actual results.
The correlation between actual and expected result for females was much lower than for males
Strategy: Study 2
• Measures taken at admissions and before finals
• Looked at changes over time
• Included First Public Exam (FPE) results
• Statistical modelling of probability of getting a first
Sex differences in Study 2
At admissions Before Finals• Self-esteem: M>F M>F• Depression: F>M not measured• Trait anxiety: F>M F>M• Happiness: M>F M>F• Expectation M>F M>F (no correlation
of a first with score at admissions)
Hours of revision F>M
Results: What predicts getting a first?
(Logistic regression for probability of getting a first)• Being male (+)• Verbal IQ (+; men only) • Expectation of a first (+; both)• Trait anxiety (+; women only)• FPE results (+; both)• Hours of revision (+; men only)• Deep learning approach(+; men only)• General self-esteem (negative; both)
High trait anxiety in women predicts first class degree
(Mellanby and Zimdars, 2010)
0 20 40 60 80 100
0.00
00.
010
0.02
0
female & 1st
acc4b$anxiousEnd[b2 & b3]
0 20 40 60 80 100
0.00
00.
010
0.02
0
male & 1st
acc4b$anxiousEnd[!b2 & b3]
0 20 40 60 80 100
0.00
00.
010
0.02
0
female & not 1st
acc4b$anxiousEnd[b2 & !b3]
0 20 40 60 80 100
0.00
00.
010
0.02
0
male & not 1st
acc4b$anxiousEnd[!b2 & !b3]
Performance decile in finals: comparison between males and females (Mellanby, Zimdars and Cortina)
0.0 0.2 0.4 0.6 0.8 1.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
Finals decile
dens
ity e
stim
ate
FemaleMale
Effect of performance in prelims/mods and expectation of a first on finals performance
40 50 60 70 80 90 100
0.0
0.2
0.4
0.6
0.8
1.0
prelims mark
prob
abili
ty o
f ge
ttin
g a
1st
female exp 1stfemale exp 2.1
male exp 1stmale exp 2.1
Interaction of factors that predict getting a first
• Expectation of a first predicts it
• Males have higher expectation
• For a given mark in FPE males had a higher expectation of getting a first.
• Expectation plus FPE accounts STATISTICALLY for the ‘gender gap’
• Why is the better performance by men at Oxford (and Cambridge) not seen at most other universities ? What is different about Oxford and Cambridge ?
What is different about Oxford and Cambridge ?
• Very high previous achievement of undergraduates
• Internationally acknowledged ‘top’ universities• Tutorials• Exams mainly timed unseen papers not
coursework-based• Publicity about the gender gap
• What sort of a psychological construct is this ‘expectation of getting a first’ ?
• We can equate it with ‘academic self-concept’
Academic self-concept versus general self-esteem
• These are separable
• Academic self-concept is context-specific
• It is related to academic achievement
• General self-esteem has social, emotional and physical aspects.
• It is related to well-being(Shavelson, 1982; Hansford and Hattie, 1982; Rosenberg et al, 1995)
• So why should academic self-concept be lower in women at Oxford ?
• What happens to them during their time at Oxford ?
Change during time at Oxford
Change over 3 years
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
self esteem happiness anxiety
psychological measure
chan
ge
male
female
Hypotheses re why females’ academic self-concept is lower
• Background sex differences in academic expectations and self-belief (Correll, 2004)
• Stereotype threat (student newspaper effect; prizes/awards for FPE success) ?(Steele, 1997)
• Big fish little pond/little fish big pond ? (Marsh, Kong and
Hau, 2000)
• Differential effect of negative feedback in tutorials (Jackson, 2003)
• Male-oriented environment.
Effect of lowered academic self-concept before Finals
• “ I am stupid therefore I need to ‘learn’ more”
• Work longer hours than men
• Inefficient revision strategy. Learning, learning, learning
• Lowered exploitation of deep learning approach.
Intelligence ?
• The males are not more ‘intelligent’
• But their academic outcome is related to their verbal IQ while women’s is not
So What can or should be done?
• Emphasize that Oxford women are not stupider than Oxford men. This may help to remove the stereotype threat.
• We need to be aware of the detrimental effect of negative criticism in women on academic self-concept
• Female attitude to revision strategy: this may improve if academic self-concept is raised.
• And we need to keep on monitoring the gender gap
Effect of expectation of a first and of self-esteem on Finals performance
5 10 15 20
0.2
0.3
0.4
0.5
0.6
0.7
0.8
prob
abili
ty o
f get
ting
a 1s
t
self esteem at the end
F expect 1st F expect 2.1 M expect 1st M expect 2.1