Bringing Values Back In: Values in Europe: A Multiple

Download Report

Transcript Bringing Values Back In: Values in Europe: A Multiple

Bringing Values Back In:
A Multiple Group Comparison with 20
Countries Using the European Social
Survey 2003
Measurement, causes and consequences
To be Presented in Lugano, QMSS, 24.08.06
Eldad Davidov
Together with Peter Schmidt and Shalom
Schwartz (1st study), and with Jaak Billiet
and Peter Schmidt (2nd study)
• Why bringing values?
• Weber;
• Socio demographic variables may affect
values, and values may affect attitudes and
behavior. So values may be the black box in
between.
• This mediation can be different in different
societies.
Outline
• 1) Theory and research questions.
• 2) Data from European Social Survey –ESS
and items.
• 3) Results and conclusions
–
–
–
–
Invariance issues
Possibilities to compare value means
Causes
Consequences
Questions We Want To Answer:
• 1) How many values from the theory do we find in
Europe?
• 2) Can we compare the values across the
countries?
• 3) How are values that we find influenced by
social demographic variables: gender, education
and age?
• 4) Do values affect attitudes towards foreigners, in
particular allowing foreigners into the country and
granting them rights?
1) Theory
• Schwartz‘s measurement theory of values
was first introduced in 1992. The theory
describes universals in the content and the
structure of individual values. It was
measured previously by 10 distinct values
and 40 items. The values are:
The values:
•
•
•
•
•
Achievement (AC)
Hedonism (HE)
Power (PO)
Stimulation (ST)
Security (SEC)
Self-Direction (SD)
Conformity (CO)
Universalism (UN)
Tradition (TR)
Benevolence (BE)
• Some values are closer to other values, and
some values may oppose one another. For
example, tradition may oppose hedonism.
• Close values are expected to correlate
positively and opposing values are expected
to correlate negatively or not at all.
• The 10 values create a continuum, which
can be expressed graphically.
Openness to Change
Self-transcendence
Self-direction
Universalism
Stimulation
Benevolence
Hedonism
Conformity
Tradition
Achievement
Power
Self-enhancement
Security
Conservation
Figure 1: Structural relations among the 10 values and the four higher values (see Devos, Spini, & Schwartz, 2002).
• In empirical studies values from adjacent
types may intermix rather than emerge in
clearly distinct regions. So in empirical
studies it may happen that we will not find
always ten distinct values.
2) The Data
• The data we use is the first round of the
European Social Survey on values,
collected in 2003. It provides for the first
time the opportunity to test Schwartz‘s
value theory with representative and
comparable across countries population
surveys. Previously the theory had been
tested by student surveys, or by
representative data which was not
comparable across countries.
20 Countries (2 Missing)
• 20 countries: 1-AT (Austria), 2-BE (Belgium), 3CH (Switzerland), 4-CZ (Czech Republic), 5-DE
(Germany), 6-DK (Denmark), 7-ES (Spain), 8-FI
(Finland), 9-FR (France), 10-GB (Great Britain),
11-GR (Greece), 12-HU (Hungary), 13-IE
(Ireland), 14-IL (Israel), 15-IT(Italy, missing), 16LU (Luxemburg, missing), 17-NL (Netherlands),
18-NO (Norway), 19-PL (Poland), 20-PT
(Portugal), 21-SE (Sweden), 22-SL (Slovenia).
The 21 ESS Items for Each Value
• 1)
Power (PO):
• Imprich/po1:Important to be rich, have money and
expensive things.
• Iprspot/po2: Important to get respect from others
• 2) Achievement (AC):
• Ipshabt/ac1: Important to show abilities and be
admired.
• Ipsuces/ac2: Important to be successful and that
people recognize achievements
• 3)
Hedonism (HE):
• Ipgdtim/he1: Important to have a good time
• Impfun/he2: Important to seek fun and things that
give pleasure
• 4)
Stimulation (ST):
• Impdiff/st1: Important to try new and different
things in life
• Ipadvnt/st2: Important to seek adventures and
have an exciting life
• 5)
Self-Direction (SD):
• Ipcrtiv/sd1: Important to think new ideas and being
creative
• Impfree/sd2: Important to make own decisions and be
free
• 6)
Universalism (UN):
• Ipeqopt/un1: Important that people are treated equally
and have equal opportunities
• Ipudrst/un2: Important to understand different people
• Impenv/un3: Important to care for nature and
environment
• 7)
Benevolence (BE):
• Iphlppl/be1: Important to help people and care for
others well-being
• Iplylfr/be2: Important to be loyal to friends and
devote to close people
• 8) Tradition (TR):
• Ipmodst/tr1: Important to be humble and modest,
not draw attention
• Imptrad/tr2: Important to follow traditions and
customs
• 9)
Conformity (CO):
• Ipfrule/co1: Important to do what is told and
follow rules
• Ipbhprp/co2: Important to behave properly
• 10) Security (SEC):
• Impsafe/sec1: Important to live in secure and safe
surroundings
• Ipstrgv/sec2: Important that government is strong
and ensures safety
The range of the items
Now I will briefly describe some people. Please
listen to each description and tell me how much
each person is or is not like you.
• 1 Very much like me
• 2 Like me
• 3 Somewhat like me
• 4 A little like me
• 5 Not like me
• 6 Not like me at all
• 7 Refusal
• 8 Don't know
• 9 No answer
3) Descriptive Results of Items
• The range of items across countries is not
very large, but there are nevertheless
differences.
• In practice, social scientists often compare
on the item level. Therefore, let’s look at
some countries.
Value Items
Ipudrst
Ipsuces
Ipstrgv
Ipshabt
Iprspot
Ipmodst
Iplylfr
Iphlppl
Ipgdtim
Ipfrule
Ipeqopt
Ipcrtiv
Ipbhprp
Ipadvnt
Imptrad
Impsafe
Imprich
Impfun
Impfree
Impenv
impdiff
Israel and Germany
5
4,5
4
3,5
3
2,5
Germany
2
Israel
1,5
1
0,5
0
• Greece for example gives a clear picture: it
has the highest scores for the values
achievement, security, tradition, stimulation,
universalism, power.
• How do other Mediterranean countries do?
• Israel for example scores most highly in
Europe only for two values- power and
stimulation.
• Spain for example scores most highly in
Europe in three values- universalism,
benevolence and tradition. So geography
does not tell us the whole story.
• How is Germany doing?
• In the middle golden way. Values tend to
score around the average and there are no
extreme items.
• And Switzerland?
• Switzerland is strongest in self-direction,
hedonism and universalism, and weakest in
conformity.
• However, Scandinavia tells us a different story.
• Sweden for example has the lowest scores for
universalism, benevolence, security and conformity.
Maybe people know that the state takes care of the
people so they do not feel the need to do it
themselves.
• Norway has the lowest scores for universalism,
security and also self-direction.
• So at least for some Scandinavian countries
geography and social system have a similar story to
tell.
• We would like to compare countries also on the
value level, and not only on the item level as we
are doing here. In such a way we can control for
measurement error.
• In order to be able to compare the means of the
values (which are the constructs here), we first
have to make sure the values mean the same
thing all over Europe.
• Ensuring that values mean the same can be done
by showing measurement invariance, that the
indicators are related to the values equally in all
the countries.
3) Data Analysis
1) Twenty separate analyses for each country.
2) A multiple sample analysis of all 20
countries together.
1
• At first we computed 20 correlation matrices for
each country separately using pairwise deletion for
missing values (see Browne 1994 and Schafer and
Graham 2002, which demonstrate why pairwise is
better than listwise and adequate if there is no more
than 5% missing values).
• The correlations ranged from negative values for
indicators belonging to constructs, which are
theoretically apart in the map of indicators, to highly
positive values for adjacent value constructs and for
indicators belonging to the same construct.
1
• Then we tested the theory for each country
separately. In all countries some constructs
correlated too highly. In order to solve the
problem of non positive definite matrices
caused, we had to unify such constructs.
• As a result we identified 5-8 values in the
20 countries
Country
Number of Values
Unified Values
1. AT
8
MALE, KOTR
2. BE
6
MALE, KOTR, WOUN,
STSE
3. CH
7
KOTR, MALE, WOUN
4. CZ
7
MALE, WOUN, KOTR
5. DE
7
MALE, WOUN, KOTR
6. DK
8
KOTR, MALE
7. ES
8
KOTR, MALE
8. FI
8
KOTR, MALE
9. FR
7
KOTR, MALE, WOUN
10. GB
8
KOTR, MALE
11. GR
5
MALE, KOTR, WOUN,
HEST, STSE
12. HU
5
WOUN, KOTR, MALE,
SIUN, HESE
13. IE
6
MALE, KOTR, WOUN,
HEST
14. IL
7
WOUN, MALE, STSE
15. NL
8
KOTR, MALE
16. NO
8
MALE, KOTR
17. PL
6
WOUN, KOTR, HEST,
MALE
18. PT
7
KOTR, WOUN, HEST
19. SE
8
KOTR, MALE
20. SL
5
KOTR, WOUN, HEST,
MALE, STSE
2
• Then we ran the simultanuous analysis for
20 countries
d3
d4
1
d5
1
ac1
d6
1
ac2
1
he1
1
he2
1
AC
HE
1
d2
1
po2
1
1
d1
po1
PO
ST
1
st1
st2
1
1
d21
sec2
1
1
d20
SD
SEC
sec1
1
1
d19
1
co1
CO
UN
BE
1
1
tr2
1
d17
tr1
1
d16
be2
1
d15
be1
1
d14
1
sd1
d9
sd2
d10
1
un1
d11
1
un2
d12
1
un3
TR
d8
1
co2
1
d18
d7
1
d13
• Again we had to unify constructs
correlating too highly causing non positive
definite matrices. We ended up with
identifying 7 values
• The constructs unified were Power and
Achievement, Conformity and Tradition,
and Universalism and Benevolence
d3
d4
1
ac1
d5
1
d6
1
ac2
1
he1
he2
1
HE
1
d2
1
po2
1
1
d1
po1
POAC
ST
1
st1
st2
1
1
d21
sec2
1
1
d20
SD
SEC
sec1
1
1
d19
1
co1
COTR
UNBE
1
d17
tr1
1
d16
be2
1
d15
be1
1
d14
1
sd1
d9
sd2
d10
1
un1
d11
1
un2
d12
1
un3
tr2
d8
1
co2
1
d18
d7
1
d13
• Finally, according to modification indices,
in order to improve the model five items
intended to measure particular value
constructs also had significant, negative,
secondary loadings on motivationally
opposed value constructs
d3
d4
1
ac1
d5
1
d6
1
ac2
1
he1
he2
1
HE
1
d2
1
po2
1
1
d1
po1
POAC
ST
1
st1
st2
1
1
d21
sec2
1
1
d20
SD
SEC
sec1
1
1
d19
1
co1
COTR
UNBE
1
d17
tr1
1
d16
be2
1
d15
be1
1
d14
1
sd1
d9
sd2
d10
1
un1
d11
1
un2
d12
1
un3
tr2
d8
1
co2
1
d18
d7
1
d13
Answer to first Question
• In simple words- we found a model which
works for all the 20 European countries
(configural invariance).
• But- we have a model which has only 7
values and not 10.
• In order to answer the second question on
differences in values between countries, we
have to test for metric (measurement)
invariance. Metric invariance will guarantee
that the values mean the same over the 20
European countries
Measurement Invariance:
Equal factor loadings across groups
Group A
dA11
dA22
Item a
Item b
lA11=1
lA21
lA31
dA33
Group B
dB11
f 11
k A1
A
A1
dB22
Item a
Item b
lB11=1
lB21
lB31
dB33
Item c
 B1
fB11
k B1
Item c
fB21
fA21
dA44
d
A
55
dA66
Item d
Item e
Item f
dB44
lA42=1
lA52
lA62

dB55
A
2
fA22
k A2
dB66
Item d
Item e
lB42=1
lB52
lB62
Item f
 B2
fB22
k B2
Steps in testing for Measurement Invariance
• Configural Invariance
• Metric Invariance
• Scalar Invariance
• Invariance of Factor Variances
• Invariance of Factor Covariances
• Invariance of latent Means
• Invariance of Unique Variances
Steps in testing for Measurement Invariance
• Configural Invariance
• Metric Invariance
• Equal factor loadings
• Same scale units in both groups
• Presumption for the comparison of latent means
• Scalar Invariance
• Invariance of Factor Variances
• Invariance of Factor Covariances
• Invariance of latent Means
• Invariance of Unique Variances
Full vs. Partial Invariance
• Concept of ‘partial invariance’ introduced by
Byrne, Shavelson & Muthén (1989)
• Procedure
• Constrain complete matrix
• Use modification indices to find non-invariant
parameters and then relax the constraint
• Compare with the unrestricted model
• Steenkamp & Baumgartner (1998): Two indicators
with invariant loadings and intercepts are
sufficient for mean comparisons
• One of them can be the marker + one further
invariant item
• We constrained the loadings of all items on
the seven factors to be the same in each of
the 20 countries
• Fit indices suggest a reasonable fit for this
model too, a fit sufficient not to reject the
model (RMR=0.08, NFI=0.89, CFI=0.91,
RMSEA=0.01 and PCLOSE=1.0)
• To conclude: we found also metric invariance:
items are related to values equally in the different
countries.
• Therefore at least statistically comparing the
means of the values across countries is
substantially meaningful (to be sure we should do
cognitive pretests in different countries, but we do
not have them)
• According to results of the invariance test, factor
covariances vary considerably across countries
• The next test is scalar invariance. To
guarantee scalar invariance, we have to set
the intercepts to be equal across groups.
• The global fit measures suggest we should
reject this model.
• Implication: Means of values cannot be
compared meaningfully across groups.
• Prospects for future possibilities to compare
latent means (Little et al. 2006).
In a new study (work in progress) we test effects
of Gender, education and age on values
According to Kohn/Schoenbach (1993) :
• people with higher education  more self directed
• people with higher education  less conformist
According to Steinmetz, Schmidt, Tina-Booh and
Wieczorek (in progress)
• men  less universalist, and score higher on power in
Germany
According to Heyder, 2003 and dissertation (in progress)
• Higher age  more conformist
Gender
7 Values:
Power and achievement
Education
Security
Conformity and tradition
Age
Universalism and benevolence
Self-Direction
Stimulation
Hedonism
Gndr
educ
d3
birth
year
d4
d5
d6
1
1
1
1
ipshabt
ipsuces
ipgdtim
impfun
1
HE
1
d2
d1
1
1
dhe
iprspot
1
imprich
1
MALE
1
1
dml
d21
d20
1
ipstrgv
1
1
SI
impsafe
ST
impdiff
ipadvnt
1
1
d7
d8
dst
dse
1
1
SE
ipcrtiv
impfree
1
dsi
dkt
d19
d18
1
1
1
ipbhprp
ipfrule
1
duw
KOTR
1
1
UNWO
ipeqopt
ipudrst
impenv
imptrad
ipmodst
1
1
d17
islam
d16
iplylfr
iphlppl
1
1
d15
d14
1
1
1
d11
d12
d13
1
d9
1
d10
Results
Men
Power and
Achievemen
t
Security
Conformity
and Tradition
Universalism
and
Benevolence
SelfDirection
Stimulation
Hedonism
Higher
Education
Power and
Achievement
Security
Conformity
and Tradition
Universalism
and
Benevolence
SelfDirection
Stimulation
Hedonism
Older Age
Power and
Achievemen
t
Security
Conformity
and Tradition
Universalism
and
Benevolence
SelfDirection
Stimulation
Hedonism
Muslim
Power and
Achievement
Security
Conformity
and Tradition
Universalism
and
Benevolence
SelfDirection
Stimulation
Hedonism
Dark blue: for all countries higher, light blue:for most countries higher
Dark gray: for all countries lower, light italic gray: for most lower
Green: effects in different directions in differernt countries.
In a new study…(work in progress)
• We argue that values are more stable than attitudes
(Ajzen/Fishbein, Eagly/Chaiken 1993)
• This justifies using values to explain attitudes and
opinions
• Our intention is to explain two latent variables
from the ESS 2003: Allowing immigrants into the
country and Conditions to allow immigrants into
the country
Indicators
• Allow into country is measured by 4 indicators:
– D5: Allow many/few immigrants of different
race/ethnic group from majority
– D7: Allow many/few immigrants from poorer countries
in Europe
– D8: Allow many/few immigrants from richer countries
outside Europe
– D9: Allow many/few immigrants from poorer countries
outside Europe
– Scale: 1=allow many, 4=allow none
Indicators 2
• Conditions to allow was measured by two
indicators:
– D10: Qualification for immigration: good
educational qualifications
– D16: Qualification for immigration: work skills
needed in country
– Scale: 0=extremely unimportant, 10=extremely
important
The problem
• There is not much theory about these relations.
• Ajzen Fishbein postulated for example a causal
relation between conformism and attitudes
towards immigrants.
• Billiet postulated this relation too, and also the
effect of security needs on attitudes to immigrants.
• Theory is needed to further explain such relations.
• We expect:
• People scoring high on Tradition,
conformity and security to allow less
immigrants in.
• People scoring high on universalism and
benevolence to allow more immigrants in.
Results
• We guarnteed invariance across 21 countries to
allow comparison of the effect of values on
opinions
• people scoring high on Hedonism, Universalism
and benevolence, power and achievement want to
allow more immigrants into the country.
• People scoring high on stimulation and self
direction, conformity and tradition, and on
security want to allow less immigrants into
country, and set more conditions for allowing
them.
Conclusions
• What did we learn?
• The model works well in Europe but for 7 values.
• Maybe more items will solve this problem, and we may
find out we can identify 10 values, but we cannot be sure.
• We find meaningful relations between socio-demographic
characteristics , opinions on immigration and values.
Effects of gender and education postulated in previous
studies were confirmed in many countries. Effects of
confomity on attitudes towards immigrants as
operationalized here was confirmed.
What next?
• In the next steps we would like to:
• 1) Conducting a full model simultanuously with socio dem. Variables, values and
opinions to find direct and indirect relations.
Soc.dem.
Charact.
Attitudes
opinions
Values
Behavior
• 2) Doing it for several countries simultanuously to compare the structural effects
• 3) Compare the means with the new method which does not require scalar
invariance, and try to give meaningful explanations for differences, such as
geographical, political and historical differences between countries
• What we conducted here was a preliminary test for such comparisons
• Thank you very much for your attention!!!!