Gender attitudes: acceptance or choice
Ko Oudhof
Statistics Netherlands
Steps
•
•
•
•
•
•
Attitudes and measurement
Theory of Hakim
What’s the problem?
Research design
Results
Conclusions
UNECE/CES-work session gender statistics 2004
Attitudes (common elements in most
definitions)
• Oriented on object, person, institution
or event
• Evaluative component
• Cognitive component
• Affective component
• Stable condition or construct
• Intermediary between object stimulus
and behavioural response:
consistency
UNECE/CES-work session gender statistics 2004
Hakim - preference theory
• Publication: Work-Lifestyle Choices in 21st Century
(2000) etc.
• Heterogeneity of lifestyles of women possible
because of new scenario
– Contraceptive revolution
– Equal opportunities revolution
– Expansion white collar occupations
– More jobs for secondary earners
– Subjective elements in lifestyle choices more important
• Three groups of preferences (of women)
– home-centred
– work-centred
– adaptive
UNECE/CES-work session gender statistics 2004
UNECE/CES-work session gender statistics 2004
Hakim on attitude measurement
•
•
•
•
Preferences versus attitudes
Subjective factor too often ignored
More holistic approach (?)
Researchers should measure more often
values and attitudes by fewer questions
• Low predictability of behavior from attitudes
• Main reason: in attitude measurement
public beliefs are confused with
personal goals
UNECE/CES-work session gender statistics 2004
Public moral versus personal preferences
• One should not discriminate xxx
• Which neighbour will you prefer…xxx or yyy
• People should use more often public transport
• From now on I will go to my work by public transport
• Women and men should divide paid work equally
• I/ my wife will go looking for a job tomorrow
• Men should do more housework
• From now on I/my husband will do ……
UNECE/CES-work session gender statistics 2004
Testing public versus personal
• Both man and woman should contribute to the family
income
• Man should earn the money, the woman should take
care of household and family
• Men should do a larger part of the household chores
than nowadays
• How we should divide our tasks at home has been
subject of discussion
• I prefer to be the one who is working outside the
home and will be responsible for the household
income
• My partner thinks it's important that we divide paid
work and caring tasks equally among ourselves
UNECE/CES-work session gender statistics 2004
Testing criterion
• Assuming that persons in different situations
will differ in relevant attitudes concerning
that situation: which items will distinguish
(explain, predict) better?
• Paid work
• Housework
• Child care
• All three: today + tomorrow
UNECE/CES-work session gender statistics 2004
Earner types
Typology of division of
household work
Typology of division of child
care
Permanent earnerless
Permanently doing everything
Permanently doing everything
Earnerless but change wanted
Permanently doing greater part
Permanently doing greater part
Permanent single earner
Doing everything but wanting
less
Doing everything but wanting
less
Doing greater part but wanting
Permanent one and a half earners
less
Doing greater part but wanting
less
Permanent half and half earners
Permanently equally divided
Permanently equally divided
Permanent double earners
Permanently doing less
Permanently doing less
Single earner but more earners
wanted
Doing less but wanting more
Doing less but wanting more
One and a half earners but half
and half earners wanted Doing nothing but wanting more Doing nothing but wanting more
Double earners but less wanted
Other
UNECE/CES-work
session gender statistics 2004
Pseudo R-squared for various groups of variables in analysis
0,800
0,700
0,600
0,500
0,400
0,300
0,200
0,100
0,000
F
M
earner type
only background variables
F
M
household work
background + public attitudinal variables
F
M
child care
background + personal attitudinal variables
all variables
Multinomial logistic regression
• In logit analysis dependent is binary
– having job (yes/no)  odd (yes)=p/p-1
• When independent is binary (m=1/f=0)
– Odd (men)=p1/1-p1 and odd(women)=p2/1-p2 
– Odd ratio (men/women) = (p1/1-p1)/(p2/1-p2)
• When independent is multinomial
– Dummies: odd ratio is relative to reference category
• When dependent is multinomial
– Dummies: odd ratio is relative to reference category
• When independent is interval
– Coefficient is change in odds by change per unit
• So, odds ratio = 1 means (relatively) no effect
UNECE/CES-work session gender statistics 2004
MALE
situation
pair 25-44 without children
pair 25-44 with children
pair 45-64 without children
pair 45-64 with children (=ref.group)
educational level
social-economic statust
housing tenure (rent =reference)
public
men should do contribute more to the care for
children than nowadays
Women are better fit to raise small children
than men
personal
in my work men will care child
in my circle men will for care for child
discussion on division of tasks at home
I prefer working and earning
my partner wants division paid work and care
It's difficult for my partner to leave the care
for the child(ren) to me
I value highly to be at least one working day at
home with my child(ren)
number of valid observations
doing less or
nothing
doing everything permanently equally
or doing more
divided
(=ref.group)
n.a.
0,67
n.a.
n.a.
0,88
0,84
1,33
n.a.
0,16
n.a.
n.a.
1,06
1,00
0,79
n.a.
n.a.
n.a.
n.a.
n.a.
n.a.
n.a.
2,38
1,26
n.a.
0,90
0,75
0,74
1,53
0,55
0,94
1,23
1,03
0,98
0,87
0,74
1,26
n.a.
n.a.
n.a.
n.a.
n.a.
0,51
0,78
n.a.
1,63
1,34
n.a.
7
110
196
high score on variable = (strongly) agree/(completely) applying
p<.01
p<.05
p<.10(italics)
n.a.=not applicable as reference category
Analysis of items
• No assumptions on level of categories (as in
PC or factoranalysis)
• HOMALS-analysis indicates
– Ordinality in scale in public items
– contrasts between neutral categories and one or
two extreme categories in personal items
– Interpretation difficult: a) different situations in
which issues are not very relevant b) preference
less important than arrangements by mutual
agreement
UNECE/CES-work session gender statistics 2004
Characterisation of dimension (low versus
high)
paid work to be divided versus paid work
might be men's task (pub.op.work-dim1)
pw1
division paid work is more or less an issue
versus statements not applicable
(pers.op.work-dim.1)
iw1
0,19
experiencing hardly pressure on division paid
work versus experiencing pressure
(pers.op.work-dim.2)
iw2
-0,20
0,00
women should do housework versus equal
division tasks (pub.op.hhwork-dim.1)
ph1
-0,85
-0,18
0,16
few hesitations in opinion versus maybe child
care should be women's task (pub.op.child
care-dim.1)
pc1
0,35
-0,05
-0,12
-0,34
child care is women's task versus not
agreeing (pub.op.child care-dim2)
pc2
-0,28
-0,09
0,01
0,33
0,00
division child care no clear issue versus
statements on division child care not
applicable (pers.op.child care-dim.1)
ic1
0,13
0,98
0,12
-0,13
-0,08
-0,07
experiencing hardly any pressure on division
tasks versus experiencing pressure
(pers.op.child care-dim.2)
ic2
-0,23
-0,13
0,97
0,19
-0,13
0,03
X
X
X
X
p<.01
X
X
X
pw1
overlap of majority of items
X
iw1
iw2
ph1
pc1
pc2
0,00
ic1
ic2
Conclusions
• Effect public as well as personal items
• Shortcomings in quality of items and
answering categories: attitudes?
• Comparibility of public and personal items?
• Other relevant factors in personal choices?
• Conceptual issue: Choices on paid work and
homework as one-dimensional problem?
UNECE/CES-work session gender statistics 2004
Hakim’s model disaggregated in two dimensions
Care
High
35%
Medium
40%
Low
25%
High
25%
????
Mixed (Workcentred/adaptive)
typical workcentered
Medium
40%
Mixed (homecentred/adaptive
)
adaptive type
Mixed (Workcentred/adaptiv
e)
Low
35%
typical homecentered
Mixed (homecentred/adaptive)
????
Career
UNECE/CES-work session gender statistics 2004