Transcript Professor Nigel Harvey, University College

Trust and social progress:
Discussion
Nigel Harvey
University College London
Data types favoured by different
social science disciplines
• Sociologists tend to focus on verbal data and
statements (responses to surveys and polls,
narratives, discourse analysis, etc).
• Economists, particularly micro-economists,
prefer to use non-verbal behavioural data
(revealed preferences, purchasing behaviour).
• Psychologists are interested in the relations
between these two types of data (implicit versus
explicit learning, automatic versus controlled
processing).
Trust and happiness
• Stated community trust (ratings of agreement
with ‘In general, people can be trusted’) predicts
stated happiness (ratings of life satisfaction).
• Does the same relation hold between revealed
trust (eg willingness to act on someone’s advice)
and revealed happiness (eg prescribing levels of
anti-depressants)?
• If it does not, the relation between stated
variables may reflect people’s lay theories of how
a third variable (eg economic conditions) affects
both their community trust and their happiness.
Lay theories
• To say that the relation reflects lay theories is not
to belittle its importance. Such theories can
influence not only the content of verbal behaviour
(responses to polls, interviews, conversations) but
also verbally mediated non-verbal behaviour.
• The problem for psychologists is to determine the
extent to which non-verbal behaviours (voting,
large purchases) are verbally mediated and
influenced by lay theories of how society works.
• What is the status of ‘rational expectations’?
Stated versus revealed trust
• Onora O’Neill, in her 2002 Reith lectures, says:
“We may end up claiming not to trust, and yet,
for practical purposes place trust in the very
sources we claim not to trust”. There is a
dissociation between stated and revealed trust.
• RL points out that, across countries, stated trust
correlates 0.65 with the proportion of dropped
wallets containing owners’ addresses that are
returned. But the latter is a measure not of
revealed trust but of revealed trustworthiness.
Determinants of stated trust
• Earle & Cvetkovich (1999) argued that trust
placement is non-rational (ie not based on
evidence of trustworthiness).
• They found people’s judgments of trust in an
agency correlated 0.66 with how much they shared
values with it. They assume assessment of shared
values is based on ‘value-bearing narratives’
produced by the agency
• We (Twyman et al, 2005) have replicated this with
risks associated with sports, jobs, drugs and travel.
Determinants of revealed trust
• “The greatest trust between man and man is the
trust of giving counsel” (Francis Bacon): the
extent to which we use advice from one source
rather than another reveals our relative trust in
those sources.
• Revealed trust has a rational basis. We are
more likely to use advice from sources who have
given better advice in the past (Harvey &
Fischer, 1997). Lack of trust in government
advice about MMR can be seen in the light of
the failure of that advice in the context of BSE.
Relation between stated and
revealed trust
• Different factors appear to determine levels of
stated and revealed trust. The extent to which
they coincide is an empirical question.
• We compared stated and revealed trust in
advice about risks received from government
agencies and consumer support services.
• When the government agency gave better
advice, revealed trust in it was higher than
stated trust in it. When it gave worse advice,
revealed trust was lower than stated trust in it.
Do people know what will make
them happy?
• Various studies (eg Kahneman & Snell, 1990)
suggest that people do not know what will make
them happy. People assessing how their liking of
plain yogurt would change after eight daily servings
thought it would decrease. In fact, it increased.
• People asked what they would like to eat for seven
days in advance select variety whereas those
asked on each day do not.
• These results suggest people have lay theories of
happiness that are not always appropriate.
Inappropriate lay theories of
happiness
• Schkade & Kahneman (1998) asked two groups
(Midwesterners & Californians) to assess how
satisfied they were with their lives in general and
with certain aspects in particular (eg job prospects,
climate). Two other groups did the same for people
in the other place.
• Happiness of Californians and Midwesterners did
not differ. But Midwesterners thought Californians
would be happier. They overrated the importance
Californians gave to climate and culture.
Inappropriate lay theories may influence migration.
Context in happiness judgments:
The focussing illusion
• Wilson et al (2000) asked people to judge how
happy they would be if their favourite football team
won a future fixture. After their team had won, they
asked them how happy they were.
• People had overestimated how happy they would
be and how long their elation would last.
• Asking them to fill in a diary before the game
detailing what they would be doing after it reduced
the effect. They realised other events would
influence their happiness besides the game result.
• Focussing occurs with retrospective judgments too.
Discussion points
• Value of different types of data and some
knowledge of the relation between them.
• Onora O’Neill’s claims about dissociations
between revealed and stated trust.
• Effects of context and lay theories on
validity of survey results.
• Cross-cultural differences in ways of
dealing with lack of trust and their relation
to happiness.