Methodologies to integrate
subjective and objective
information to build
well-being indicators
Filomena Maggino – Elena Ruviglioni
e-mail:
filomena.maggino@unifi.it
First of all
I would like to thank
Stefano Tarantola for his invitation and
for having encouraged me to present this
work on which we are working together.
the Joint Research Centre group for its
co-operation not only in preparing this
work: the idea to work on this topic was
– Relationships between the two components
Methodological issues in developing indicators
– Hierarchical design
Applied approach to integration
– Multi-stage multi-technique approach
Particular aggregation issues
– Construction of complex indicators
– Definition of macro-units
– Analytical approaches to integration
– Problems in selecting indicators
An Example in four stage
Final remarks
Conceptual framework
Three philosophical approaches
– The ability to select goods and
services that one desires
Æ economic indices
– Normative ideals
Æ social indicators
– Subjective experiences
Conceptual framework
Conceptual framework
└►
Observation approaches
└
►
Measures
└►
Operative models
└►
Methodological
approaches for
managing the
complexity
Societal well-being should be assessed
through a multidimensional and
integrated approch
Quality of Life approach (
QoL
)
of Quality of Life (
QoL
)
Various definitions of QoL share a clear
definition between
• Objective components
• Subjective components
└►
Objective components at micro level
Objective components at macro level
Subjective components
Objective = descriptive
Subjective = evaluative
Social structure
Living conditions
Evaluations of living conditions
Subjective QoL, in terms of well-being
More
objective
More
subjective
Conceptual framework
Conceptual framework
It is impossible and undesirable to
consider one perspective separated from
the others
Integration represents the MOST valid
and complete approach in order to study
QoL
Interrelating and combining individual
living conditions and subjective well-being
components
└►
Two perspectives:
1. objective QoL at macro level can be considered an
antecedent with respect to subjective QoL
(subjective well-being).
¾ In this case, objective indicators (input) can be
interpreted in terms of contextual conditions that can
explain the subjective indicators (output)
2. objective QoL conditions at macro-level and
subjective QoL (perceptions) are independent;
perceptions are influenced by individual
characteristics and not by the objective living
conditions.
¾ In this case, subjective indicators (input) can be
considered as an important component driving the
improvement of objective conditions.
Zapf’s model:
→
↓deprivation
adaptation
low
dissonance
well-being
high
Objective
living conditions
low
high
Subjective
well-being
Level of
→
↓
Conceptual framework
Costanza’s model:
→
↓Conceptual framework
– Social Capital (SC)
• networks and norms that facilitate cooperative action
– Human Capital (HC)
• the knowledge and information stored in our brains, as
well as our labour
– Built Capital (BC)
• manufactured goods such as tools, equipment, buildings
– Natural Capital (NC)
• the renewable and non-renewable goods and services
provided by ecosystems
– Time (T)
Social epidemiology approach:
Epidemiology + behavioral sciences
in order to investigate
social determinants of
population distributions of
health, disease, and well-being
Methodological issues
Hierarchical design:
ELEMENTARY INDICATORS
Ð
LATENT VARIABLES
Ð
AREAS TO BE INVESTIGATED
Ð
CONCEPTUAL MODEL
COMPONENTS
Methodological issues
Hierarchical design is completed by
identifying the relations between:
– Latent variables
– Latent variables and the
corresponding indicators
– Elementary indicators
objective and subjective information
relies on
└►
Applied approach to integration
Definition of conceptual framework
Organizational context (system of indicators)
– Levels:
• micro
• macro
Perspectives of analysis:
– Aggregation of
• indicators (reflective or formative approach)
• units
– Integration of objective and subjective
characteristics
Applied approach to integration
Multi-stages multi-techniques approach
└►
Conceptual framework
▼
Definition of objective and subjective
components
▼
Conceptual perspective of integration
(CPI)
▼
→
↓
Stage I : Indicators aggregation
Perspective:
Creation of complex indicators by synthesizing
elementary indicators
Level of analysis:
From elementary indicators to synthetic indicators
Analytical issues:
Reflective indicators Æ scaling models
Formative indicators Æ composite indicators construction
→
↓
Stage II : Integration
Perspective:
Understanding relationships between objective
and subjective characteristics
Level of analysis:
Micro level
Analytical issues:
Different solutions (consistently with CPI)
→
↓
Stage III : Units aggregation
Perspective:
Creation of macro-units by synthesizing
elementary units
Level of analysis:
From micro units to macro units
Analytical issues:
Following homogeneity/functionality criteria
→
↓
Applied approach to integration
Stage IV : Integration
Perspective:
Understanding relationships between objective
and subjective characteristics
Level of analysis:
Macro level
Analytical issues:
Different solutions (consistently with CPI)
→
↓
(a)
construction of complex indicators
Observational units aggregation
(b)
definition of macro-units
Particular aggregation issues
(a)
Construction of complex indicators:
Reflective criterion
(Homogeneity)
Æ Synthetic indicator
Formative criterion
(Heterogeneity)
Æ Composite indicator
Æ Comprehensive/Summary indicator
Condensation Æ New synthetic values
Particular aggregation issues
(b)
Definition of macro units by
condensation:
Information Æ same level
Micro level Æ aggregation Æ proper scale
This problem involves both objective and subjective
indicators with different solutions.
not observable
subjective well-being
subjective
(i) population information
(ii) territory information
(i) individual living
conditions
objective
information
Macro
Micro
Level of observation
Particular aggregation issues
Aggregation of objective information
criteria
Compositional
– Information refers to population (observed at
individual level
Contextual
Particularly delicate (characteristics
non-cumulative) Æ ad-hoc aggregating criteria
Homogeneity
Functionality
Particular aggregation issues
Homogeneity:
Segmentation analysis
Partitioning analysis
Tandem analysis
Particular aggregation issues
Functionality:
Groups
Areas
Time periods
Particular aggregation issues
Analytical approaches to integration
Structural model approach
Multi-level approach
Life-course perspective
Composite indicators
1. Settlement/aggregation area sizes
2. Time frames
3. Population composition
4. Domains of life composition
5. Objective vs subjective indicators
6. Positive vs negative indicators
7. Input vs output indicators
8. Benefits and costs
9.
Measurement scales
10. Report writers
11. Report readers
12. Quality-of-life model
13. Distributions
14. Distance impacts
15. Causal relations
An example
Goal :
to illustrate the multi-technique
multi-stage characterization of the
proposed approach
By using:
subjective and objective data
Provided by:
– European Social Survey project
– Joint Research Centre (JRC – European
Commission)
An example
First stage:
– synthesis of basic indicators at individual
level
Second stage:
– understanding relationships between objective
and subjective characteristics at micro level
Third stage:
– synthesis of micro units
Fourth stage:
– understanding relationships between objective
and subjective characteristics at macro level
An example : first stage
B31 the national government
B30 present state of economy in country
0 (left) 10 (right) B28
placement on left-right scale
Self-placement
B12 the United Nations
B11 the European Parliament
B10 politicians
B9 the police
B8 the legal system
reflective 0 (no trust at all)
10 (complete trust) B7 country’s parliament Trust in Politics Model of measurement Scaling technique Item Nr Items Variable Area
E19 voluntary organizations E18 religion E17 work E16 politics E15 leisure time E14 friends formative 0 (extremely unimportant) 10 (extremely important) E13 family Values: important in life 0 (extremely dissatisfied) 10 (extremely satisfied) B29
how satisfied with life as a whole Life satisfaction 0 (extremely unhappy) 10 (extremely happy) C1
how happy are you Happiness
Subjective aspects
An example : first stage
1. living comfortably 2. coping 3. difficult 4. very difficult on present income F31
feeling about household’s income nowadays Income Socio-demographic profile D9 many/few immigrants from poorer countries outside Europe
D8 many/few immigrants from richer countries outside Europe
D7 many/few immigrants from poorer countries in Europe
D6 many/few immigrants from richer countries in Europe
D5 many/few immigrants of different race/ethnic group from majority reflective 1. allow many 2. allow some 3. allow a few 4. allow none to come and live here D4
many/few immigrants of same race/ethnic group as majority
Acceptance of immigration: allow Immigration and asylum issues Model of measurement Scaling technique Item Nr Items Variable Area
An example : first stage
0.6
IMP_W Work dimension0.4
IMP_PL Private life dimension0.6
TRUST_II International institutions COMPOSITE312
Personal life principles0.4
IMP_C Caring dimension0.7
SAT_NSS Satisfaction for national socialservices
0.4
IMP_PLPrivate life dimension Welfare dimension
15
COMPOSITE20.8
TRUST_NS National security
0.8
SAT_NF Satisfaction for national foundations0.6
IMP_ALActive life dimension Public & political life
18
COMPOSITE10.8
TRUST_NP National politics Aggregated score Variance explained (%) Obtained component Item loadingSynthetic indicators
An example : second stage
0.95 3.13 -0.29 -3.19
Public & political life
COMPOSITE1 0.79 2.17 -0.47 -1.96
Non-acceptance of immigration
IMMIGR 0.98 2.24 -0.34 -2.30Self-placement on left-right scale
B28
0.85 2.46 1.10 -1.14
Feeling about household’s income
nowadays
F31 0.93 1.34 -0.37 -3.74Happiness
C1 0.97 1.31 -0.58 -3.10Life satisfaction
B29 CLUSTER1
(n=7369) SD max. mean min.INDICATOR
0.91 3.15 0.10 -5.03
Personal life principles
COMPOSITE3 0.86 2.90 0.12 -4.32
Welfare dimension
COMPOSITE2 0.76 4.08 0.60 -2.50Public & political life
COMPOSITE1 0.76 2.17 -0.64 -1.96
Non-acceptance of immigration
IMMIGR 0.92 2.24 0.26 -2.30Self-placement on left-right scale
B28
0.63 2.46 -0.61 -1.14
Feeling about household’s income
nowadays
F31 0.59 1.34 0.48 -3.74Happiness
C1 0.54 1.31 0.54 -3.10Life satisfaction
B29 CLUSTER2
(n=14855)An example : second stage
0.97 3.07 -0.24 -5.71
Personal life principles
COMPOSITE3 0.94 3.85 0.48 -3.83
Welfare dimension
COMPOSITE2 0.90 2.36 -0.49 -3.85Public & political life
COMPOSITE1 0.78 2.17 0.48 -1.96
Non-acceptance of immigration
IMMIGR 0.90 2.24 -0.46 -2.30Self-placement on left-right scale
B28
0.68 2.46 -0.40 -1.14
Feeling about household’s income
nowadays
F31 0.58 1.34 0.54 -3.23Happiness
C1 0.60 1.31 0.53 -3.10Life satisfaction
B29 CLUSTER3
(n=9703) SD max. mean min.INDICATOR
An example : second stage
1.11 3.22 -0.11 -5.59
Personal life principles
COMPOSITE3 0.99 3.29 -0.54 -4.34
Welfare dimension
COMPOSITE2 0.99 3.61 -0.26 -3.47Public & political life
COMPOSITE1 0.79 2.17 0.81 -1.96
Non-acceptance of immigration
IMMIGR 0.99 2.24 0.30 -2.30Self-placement on left-right scale
B28
0.89 2.46 0.47 -1.14
Feeling about household’s income
nowadays
F31 1.04 1.34 -0.93 -3.74Happiness
C1 1.00 1.31 -0.86 -3.10Life satisfaction
B29 CLUSTER4
(n=10418) SD max. mean min.INDICATOR
An example : second stage
CLUSTER PLOTS
Cluster Parallel Coordinate Plots
1 COMPOSITE3 B28 COMPOSITE2 COMPOSITE1 C1 IMMIGR B29 F31 In de x of C as e -10 -5 0 5 2 COMPOSITE3 B28 COMPOSITE2 COMPOSITE1 C1 IMMIGR B29 F31 In de x of C as e -10 -5 0 5 3 B29 F31 4 B29 F31
Cluster Profile Plots
1 COMPOSITE3 B28 COMPOSITE2 COMPOSITE1 C1 IMMIGR B29 F31 2 COMPOSITE3 B28 COMPOSITE2 COMPOSITE1 C1 IMMIGR B29 F31 3 F31 4 F31
Medium-low Low
Medium-high High
Personal life principles COMPOSITE3 Low High Medium-high Medium-low Welfare dimension COMPOSITE2 Medium-low Low High Medium-low
Public & political life COMPOSITE1 High Medium-high Low Medium-low Non-acceptance of immigration IMMIGR Right Left Centre-right Centre-left self-placement on left-right scale B28 Some difficulties Comfortable Very comfortable Many difficulties Feeling about household’s income nowadays F31 Low High Medium-high Medium-low happiness C1 Low Medium-high Medium-high Medium-low life satisfaction B29
4
3
2
1
An example : second stage
-2
-1
0
1
Dim(1)
-1.5
-1.0
-0.5
0.0
0.5
1.0
D
i
m
(
2
)
50-64
voted yes
65 and more
cluster 2
male
31-49
cluster 3
cluster 4
female
cluster 1
less than 31
voted no
An example : third stage
42345 10418 9703 14855 7369 N 100.0 24.6 22.9 35.1 17.4 Total 1519 100.0 30.0 21.5 31.4 17.1 Slovenia SI 1999 100.0 8.2 17.5 63.0 11.3 Sweden SE 1511 100.0 47.5 11.8 12.9 27.9 Portugal PT 2109 100.0 32.6 11.5 17.1 38.8 Poland PL 2036 100.0 12.4 26.6 51.4 9.6 Norway NO 2364 100.0 17.0 25.0 50.7 7.4 Netherlands NL 1552 100.0 18.4 27.5 45.5 8.6 Luxembourg LU 1206 100.0 28.0 15.1 37.5 19.4 Italy IT 2497 100.0 22.3 19.0 26.1 32.6 Israel IL 2046 100.0 15.3 18.3 49.9 16.4 Ireland IE 1685 100.0 56.3 11.9 10.5 21.2 Hungary HU 2566 100.0 51.1 12.5 11.4 25.0 Greece GR 2051 100.0 23.0 32.3 32.5 12.2 United Kingdom GB 1503 100.0 33.3 28.9 25.4 12.4 France FR 2000 100.0 14.7 35.5 39.4 10.5 Finland FI 1728 100.0 27.5 20.9 31.1 20.5 Spain ES 1500 100.0 7.1 26.6 60.1 6.2 Denmark DK 2919 100.0 23.8 28.7 30.9 16.5 Germany DE 1360 100.0 35.1 13.8 23.8 27.4 Czech Rep. CZ 2040 100.0 8.8 22.9 57.5 10.9 Switzerland CH 1897 100.0 15.6 26.8 43.1 14.5 Belgium BE 2257 100.0 21.8 41.2 23.4 13.6 Austria AT Cluster 4 Cluster 3 Cluster 2 Cluster 1 Total NAn example : third stage
-0.2
-0.1
0.0
0.1
0.2
0.3
0.4
0.5
D
i
m
(
2
)
DK
CH
cluster 2
NO
NL
LU
FI
BE
IE
cluster 3
GB
AT
HU
GR
PT
cluster 4
cluster 1
CZ
IL
FR
SI
ES
IT
DE
0 10 20 30 40 50 60 70 % cluster 1 0 2 4 i n e q u alit y of i nc o me d DK NL LU NO FI SE FR AT CZ HU SI DE IE BE 0 10 20 30 40 50 60 70 % cluster 2 0 2 4 i n e q u alit y of i nc o me d HU AT CZ FR DE SI NODK SE NL IE LU BE FI r = 0.509 r = -0.437 0 10 20 30 40 50 60 70 % cluster 3 0 2 4 6 8 i n e q u alit y of i nc o me dsi tr i b uti o n PL PT HU GR CZ IT SE IEES SI AT FI GB FR DE LU BE NO DK NL 0 10 20 30 40 50 60 70 % cluster 4 0 2 4 6 8 i n e q u alit y of i nc o me dsi tr i b uti o n DK SE NO FI IE BE NLATLU HU GR PT CZ FR PL SI IT ES DE GB r = -0.350 r = 0.418