Happiness Science

1
(No Transcript)
2
Happiness Science Popular Topic
3

• What is happiness?
• An evaluation of a life. A happy life is a good
  life.
• How is happiness measured?
• Standard economics (Utility/)
• Welfare economics (Capabilities/HDI)
• Subjective Measures (Quality of Life, Subjective
  Well-Being)

4

• Happiness Science
• Book Well-Being The Foundations of Hedonic
  Psychology (Kahneman, Diener, Schwarz, 1999).
• policy relevance
• development of valid indicators
• existing economic indicators are limited
• focus on traded goods
• flawed assumptions (behavioral economics)

5

• Happiness Science
• Focus on subjective measures
• Subjective Well-Being (SWB)
• Affective Component (AWB)
• Amount of Positive Affect / Negative Affect
• Cognitive Component (CWB)
• Life Satisfaction
• Average Domain Satisfaction

6

• Cognitive Well-Being (CWB)
• Life satisfaction judgments
• A global assessment of ones life
• Widely used in happiness surveys
• The majority of empirical findings in happiness
  science are based on these measures.

7
Example World Value Survey Taking all things
together, would you say you are 1 Very happy2
Rather happy3 Not very happy4 Not at all
happy All things considered, how satisfied are
you with your life as a whole these days? Using a
scale on which 1 means you are completely
dissatisfied and 10 means you are completely
satisfied where would you put your satisfaction
with your life as a whole? Completely
dissatisfied Completely
satisfied1 2 3 4 5 6 7 8 9 10
8

• Promises
• subjective / evaluation based on individuals
  own point of view (not paternalistic)
• comprehensive
• Problems
• requires willingness to participate
• requires cognitive abilities
• insensitive to environmental influences
  (set-point, adaptation)
• may rely on inappropriate comparison
  standards(satisfaction treadmill, relative vs.
  absolute judgments)

9

• Participation Problems
• National representative surveys routinely
  include life satisfaction questions.
• Few respondents do not answer these questions.
• Responses are not random (high correlation
  between two independent questions).
• Conclusion
• A general problem of survey-based indicators,
  but not specific to happiness science.

10

• (Lack of) Cognitive AbilitiesHeuristics and
  Biases
• Traditionally studied by social psychologists
  and behavioral economists (Kahneman, Schwarz,
  etc.)
• The heuristics and bias research program is
  itself biased and has focused on demonstrating
  biases in human judgments (Giegerenzer, Funder).
• This has lead to a biased perception of humans
  abilities.
• Individual bias may often cancel out in
  aggregated measures of life satisfaction (e.g.,
  national averages).

11

• Example Context-Effects
• In a well-known example, Strack, Martin, and
  Schwarz (1988) presented the following two
  questions consecutively in a survey administered
  to students How happy are you? and, How many
  dates did you have last month The correlation
  was .12 when the general happiness question came
  first, but when the dating question came first,
  the correlation rose to .66 (Kahneman, 1999, p.
  22).
• difference between two correlations, effect
  size q .67

12

• Example Context-Effects
• Two important conclusions can be drawn from
  this finding, WHICH HAS BEEN REPLICATED MANY
  TIMES WITH DIVERSE POPULATIONS AND IN A VARIETY
  OF LIFE DOMAINS (Schwarz Strack, 1999, this
  volume).
• First, people EVIDENTLY compute an answer to
  the subjective happiness question on the fly,
  instead of retrieving a prepared answer from
  memory.
• Second, respondents APPEAR TO anchor their
  report of well-being on their satisfaction with
  any significant life domain to which attention
  has been drawn. (Kahneman, 1999, p. 22).

13
Kahneman et al. (2006) Would you be happier if
you were richer? A focusing illusion SCIENCE,
312, 1908-1910. Same example the dating
question EVIDENTLY caused that aspect of life to
become salient and its importance to be
exaggerated when the respondents encountered the
more general question about their happiness (p.
1908).
14

• Schimmack and Oishi (2005)
• Meta-analysis of all studies that manipulated
  item-order (no priming r .32, priming r .40,
  effect size q .09).
• Replication of Strack and Schwarz (1988) dating
  study (no priming r .39, priming r .49,
  effect size q .12).
• Correlation with average domain
  satisfaction(priming r .71, no priming r
  .78, effect size q .16).

15

• Conclusions
• Priming effects are weak
• Satisfaction in important life domains that were
  not primed is a strong predictor of global life
  satisfaction judgments.
• Chronically accessible information is more
  important than temporarily accessible
  information.
• You get a noble price for pushing a paradigm, not
  for accurate reporting of empirical evidence.

16

• Stability and Change(Adaptation/Set Point)
• Genetic dispositions may produce stable
  differences between individuals.
• Environmental influences may have short-lasting
  effects due to adaptation.
• Policy implication Even if it could be
  measured, it could not be changed.

17

• Empirical Evidence
• Meta-analyses and longitudinal panel studies
  provide evidence for stability and change.
• Veenhoven (1994) meta-analysis
• Ehrhardt et al. (2000) SOEP
• Fujita and Diener (2005) - SOEP
• Schimmack and Oishi (2005) meta-analysis
• Schilling (2006) SOEP
• Schimmack and Lucas (2007) SOEP
• Anusic and Schimmack (in prep.) Meta

18

• Modeling Stability and Change
• Trait
• State
• Error / Fluctuation
• Stability of State Variance
• High slow adaptation
• Low fast adaptation

19
Trait State Error Plot
20
Error Free Trait State Plot
21
Greymultiple itemsBlacksingle items
22

• Schimmack and Lucas (2007)
• A dyadic study of stability and change of
  married couples.
• Spousal similarity in trait variance
• Assortative mating (genetic similarity)
• Stable environmental factors
• Spousal similarity in state variance
• Mutual social influence
• Shared environmental factors

23
(No Transcript)
24

• Conclusion
• Evidence for a stable trait component,
  presumably due to genetic dispositions.
• Evidence for a slowly change state component. No
  evidence for quick adaptation.
• Both components contribute about equally to the
  error free variance in life satisfaction.
• Evidence for spousal similarity in both
  components.
• Change may be due to changing circumstances
  rather than simple adaptation to stable
  circumstances.

25

• Environmental factors that produce change in life
  satisfaction?
• Unemployment (down, up after reemployment)
• Disability (down, adaptation evidence mixed)
• Widowhood (down, slow adaptation)
• Divorce (down, then up in new relationship)
• Marriage (up and down, no adaptation)
• Having children (on average up, adaptation
  unknown)
• Bigger house (up, adaptation unknown)
• Source. Several articles by Rich Lucas, review
  article by Diener et al. 2006) children effect
  based on poster German Sociological Society
  2007house effect based on preliminary
  unpublished results of SOEP data.

26

• Relative versus AbsoluteNational Differences in
  Happiness
• Studies of individuals within a nation fail to
  reveal causes that produce differences across
  nations.
• Changes within nation may be caused by absolute
  or relative judgments of well-being.
• Large survey studies of national representative
  samples show marked differences between nations.
• Last year, researchers published a world map of
  happiness.

27
(No Transcript)
28

• Theoretically Important Questions
• What is the correlation between per capita GDP
  in Purchasing Power Parity and happiness?
• Is the relation linear or non-linear
  (log-function, diminishing marginal utility)?
• What predicts discrepancies between these two
  measures of nations well-being (welfare)?
• standard economics (error in happiness measures)
• happiness economics (false assumptions of
  standard economics)

29

• Schimmack, Oishi, Diener (in preparation)
• used two WVS items (N 80 nations)
• avoid computation of average
• estimate correlations separately for frequencies
  of different response categories
• modeling shows that indicators are not
  unidimensional.
• one dimension shows high loadings of categories
  7,8, and 9, other dimension has high loading of
  10s.
• GDP predicts frequencies of 7s, 8s, and 9s.
• Latin America predicts frequencies of 10s.

30
Top 10 Happy Nations 1. Finland2.
Netherlands3. Iceland4. Luxembourg5.
Sweden6. Australia7. Norway8. Canada9.
Ireland10. USA
Top 10 Bias Nations 1. Puerto Rico2.
Colombia3. Venezuela4. Brazil5. El
Salvador6. Malta ?7. Switzerland ?8.
Denmark ?9. Mexico10. Austria ?
31
Happiness and Wealth (PPP)
32

• Results
• Linear correlation with PPP, r .83
• Correlation with Log-PPPP, r .82
• Multiple correlation, r .85
• unique linear, beta .51
• unique log, beta .35

33
Lowest 10 Nations ResidualsUnhappier than PPP
predicts 1. Zimbabwe2. Luxembourg3.
Ukraine4. Russia5. Tanzania6. Belarus7.
Moldova8. Armenia9. Pakistan10. Georgia
34
Top 10 ResidualsHappier than PPP Predicts 1.
Indonesia2. Colombia !3. China4. El
Salvador5. Mexico !6. Dominican Republic7.
Nigeria8. Finland9. Malta10. Philippines !
35

• Human Development Index(Education, Longevity,
  Log (PPP)
• Correlation with happiness, r .73
• Controlling for PPP, beta .17, n.s.
• Gini(Income Inequality)
• Correlation with happiness, r -.24
• Controlling for PPP, beta .13
• Correlation with bias, r .55
• Controlling for Latin America, beta .27

36

• CO2 Emissions
• Correlation with happiness, r .57
• Controlling for PPP, beta -.16, n.s.
• Electricity Consumption
• Correlation with happiness, r .66
• Controlling for PPP, beta -.03

37

• Conclusion
• Life satisfaction judgments are at least
  partially based on absolute information.
• PPP predicts life satisfaction beyond the
  fulfillment of basic needs (proxy for utility).
• Other national indicators do not explain
  discrepancies between happiness and PPP.
• Measurement error in PPP may account for some of
  the discrepancies?

38

• Hedonic indicators (AWB)?
• Less empirical evidence, but often highly
  correlated with CWB.
• Life satisfaction more responsive to
  unemployment than affective well-being
  (Schimmack, Schupp, Wagner, in press) (hedonic
  treadmill, bread and circuses).

39

• Happiness Science
• Important research area
• Wealth of data
• Remaining problems
• cardinality
• bounded measure (problem?)
• More empirical (positive happiness science) work
  needed before it can be used in public policy
  (normative happiness science).