Do cultural differences in moral psychology explain political conflict on climate change?
Plan:
Work through Feinberg & Willer, 2013 ‘The Moral Roots of Environmental Attitudes’
What are their background assumptions, and what is the evidence for them?
What is their theoretical framework?
cultural variation in ethical foundations: context
We are the last suriving human species. And a relatively young one,
only around 300ky old (compared to Neanderthals who survived for
perhaps 500ky, and homo erectus who lived for maybe 1.8my).
For most of our species’ existence we were a marginal hominin
with a small population in Africa.
There were around 30,000 of us alive at any time.
Although some of our ancestors
ventured out of Africa into Europe and Asia around 100kya,
those early explorers seem to have died out while other hominins,
particularly Neanderthals and Denisovians, thrived.
We were one of many hominin species,
and one with a small population and fragile existence
threatend by drought and other environmental pressures.
Around 90–60kya we were reduced to a global population
of just a few thousand.
But something like 50kya we underwent a cultural transformation.
There was an explosion in art, ritual, trade and tool-making.
This coincided with the spread of homo sapiens from Africa to
every corner of Earth in a very short time.
The transformation from one of many hominin species to
the sole survivors involved rapid and enormous cultural change.
Modern ethical abilities likely emerged around this time
and played a role in enabling homo sapiens to flourish in
a wide variety of environments.
For most of the last 50ky we were roughly 100,000
widely dispersed hunter-gatherers living in bands of perhaps 30.
Our ethical abilities enabled food sharing, cooperative breeding
and provided a defence against pathogens.
Trust was likely important for building large-scale trade networks.
And our ethical abilities must have contributed to our ability to live
in cohesive groups with effective, but not too dominant leaders.
(Respect for authority and its flip side: reverse hierarchical dominance.)
This phase of our existence, as roughly 100,000
widely dispersed hunter-gatherers living in bands of perhaps 30,
lasted for around 50ky and
only came to an end for most people around 5000 years ago.
That’s 2000 generations as hunter–gatherers and
just 200 generations or so of living in farming and pastoral
communities.
City living is even more recent phenomena, stretching back
fewer than 8 generations for nearly all of us.
(Around 8 generations ago (circa 1800) , only about 3–5% of the global population lived in cities;
but 1950, 30% of humans lived in cities. It wasn’t until 2007 that
more than half of humans became city dwellers.)
So what we’re saying is that for 2000 generations we were hunter-gatherers,
and for the 200 or so generations that we have been farming,
mostly we were operating on a fairly small scale.
It’s only for the last 8 generations or so, the most recent 0.4% of the time
of our homo spaien major cultural revolution, that any significant
number of us have interacted with larger groups.
Why am I stressing this?
Because cultural variation in moral foundations is not an issue if
you are living in a small, cohesive group. It is not an issue if
you do not have wealth or power.
The need to work closely, and share everything, with people who have
profoundly different ethical views is a recent one.
Something few of us humans have more than eight generations of experience
of.
And something that we have, until now, barely even been aware of as a challenge.
The tragedy of anthropogenic climiate change that is so easy to mitigate
most vividly illustrates the challenge.
How are social liberalism and conservatism measured?
These are extremists, so not part of what we are tracking.
(Have to mention this because to me it seems obvious)
How are social liberalism and conservatism measured?
How much in favour of ...
1 capital punishment
2 making abortion illegal
3 less strict gun control
4 more socialized health care
5 legalization of same sex marriage
6 not punishing illegal immigration
7 democrats
8 republicans
... are you
‘First, participants completed an eight-item assessment of political
orientation (Nail, McGregor, Drinkwater, Steele, & Thompson, 2009), in
which they were instructed to indicate the extent to which they were in
favor of or against “each of the following eight policies, practices,
and political groups,” on a scale ranging from 1 strongly against to 7
strongly in favor: (1) capital punishment, (2) making abortion illegal,
(3) less strict gun control, (4) more socialized health care, (5)
legalization of same sex marriage, (6) not punishing illegal
immigration, (7) democrats, and (8) republicans.
(Wolsko, 2017)
Feinberg & Willer (2015)
Instructions (to Liberals in Study 1): ‘Now, we would like you to write
a persuasive argument (4-5 sentences) aimed at convincing conservative
Americans who oppose same-sex marriage of why they should be in favor of
same-sex marriage.
Note: In a follow-up study, we will actually present your argument to
conservative Americans who oppose same-sex marriage to see if they are
persuaded by your argument. Participants who are able to effectively
persuade these future study participants will be entered into a drawing
for a $50 bonus.’
Conservatives asked to write argument for Liberals aimed at supporting Official English.
‘Feinberg and Willer (2015) found that when participants were asked to identify whether a morally reframed argument or a more typical moral argument would be more persuasive to those on the other side of the political spectrum, 64% of liberal participants and 85% of conservative participants correctly selected the morally reframed argument.’
(Feinberg & Willer, 2019, p. 7)
Note that some participants attack their targets’ morality.
start of lecture
Question
Do cultural differences in moral psychology explain political conflict on climate change?
Plan:
Work through Feinberg & Willer, 2013 ‘The Moral Roots of Environmental Attitudes’
What are their background assumptions, and what is the evidence for them?
What is their theoretical framework?
The story so far ...
1. ‘Moral convictions and the emotions they evoke shape political attitudes’
2. There are at least two foundational domains of human morality, including harm and purity.
3. ‘liberals and conservatives possess different moral profiles’
attitudes do not matter, moral values do.
‘believers [in climate change] are more likely than skeptics to report
that they intend to behave in climate-friendly ways (r = .32) but on
actual behaviors the difference is modest (r = .17).’
In other words, knowing whether people are skeptics or believers tells you surprisingly little about their willingness to engage in actions that matter’
‘For example, the difference between believers and skeptics in terms of their willingness to put a price on carbon is relatively small (r = .20)’
(Hornsey & Fielding, 2020, p. 21).
This is repeated just to stress that we have data for the assertion.
There are effects, but they are small-to-medium only.
And the smaller effects are for specific policies vs general things.
(Remember this is people who *do* vs people who *do not* believe in anthropogenic climate change.)
Figure caption: ‘Correlations between climate change belief and outcome variables.
Error bars represent 95% confidence intervals.’
I did look but I do not fully understand the methods. I think the key point here is that we are
asking whether, when there is a difference in climate scepticism, there is a corresponding difference
in support for various policies.
Hornsey, Harris, Bain, & Fielding (2016, p. figure 3)
Doran, Böhm, Pfister, Steentjes, & Pidgeon (2019, p. figure 3)
‘moral concerns turned out to be substantially more important
than consequence evaluations, explaining about twice as much of the variance.’
The story so far ...
1. ‘Moral convictions and the emotions they evoke shape political attitudes’
2. There are at least two foundational domains of human morality, including harm and purity.
3. ‘liberals and conservatives possess different moral profiles’
3
‘liberals and conservatives possess different moral profiles regarding the five moral foundations’
Feinberg & Willer (2013, p. 2)
van Leeuwen & Park (2009, p. figure 1a)
Subjects are Dutch students
van Leeuwen & Park (2009, p. figure 1b)
Implicit measure: IAT test of conservative and liberal concepts; which are implicitly
associated with good things?
This is not to say that there is no exceptions ...
Exceptions
New Zealand: Harm and Fairness not linked to socially conservative/liberal
(Davies, Sibley, & Liu, 2014)
Using participants in New Zealand,
Davies et al. (2014, p. 434) found that ‘Although Harm/care and
Fairness/reciprocity showed significant negative correlations with
conservatism, these relationships were weak, indicating that these
foundations are not related to ideology. [...]
the individualizing foundation results
are surprising, and different to those found by Graham et al. (2011).’
This matters because it means we have another thing to explain:
why is harm/care sometimes linked to political ideology (Graham et al’s US sample)
and sometimes not (Davies et al New Zealand sample)?
‘Latvia is the only country [...] with a
positive care-conservatism,
and negative right-authority and right-sanctity associations.’
(Kivikangas, Fernández-Castilla, Järvelä, Ravaja, & Lönnqvist, 2021, p. 71)
Important because if this is true, an explanation of the
association between politics and moral foundations needs to explain
why purity is sometimes right and sometimes left.
Graham, Haidt, & Nosek (2009, p. figure~1)
Also works with a web sample collected in USA
Graham et al. (2009, p. figure~3)
Graham et al, 2009 figure 2
(Graham et al., 2009, p. figure 2)
‘We tested whether the effects of political identity persisted after
partialing out variation in moral relevance ratings for other demographic
variables. We created a model representing the five foundations as latent
factors measured by three manifest variables each, simultaneously predicted
by political identity and four covariates: age, gender, education level,
and income. [...] Including the
covariates, political identity still predicted all five foundations in the
predicted direction [...]. Political identity was the key
explanatory variable: It was the only consistent significant predictor
[...] for all five foundations’
(Graham et al., 2009, p. 1032)
There was a problem about invariance
invariance ?!
Just a reminder from unit `moral_foundations_operationalised`—not for discussion here
MFQ-2 : from 2023 (Atari et al., 2023; Dogruyol et al., 2024)
MFQ-1: from 2007 (Haidt & Graham, 2007)
Davis et al. (2016): no scalar invariance for Black people vs White people
(Davis et al., 2016) found metric but not scalar invariance
Atari, Graham, & Dehghani (2020): scalar non-invariance for US vs Iranian participants
Doğruyol, Alper, & Yilmaz (2019): metric non-invariance for WEIRD/non-WEIRD samples
Nilsson (2023): non-invariance for class within Sweeden
Iurino & Saucier (2020): problems with fit
We found
care and equality
to be
negatively correlated with political conservatism,
while
proportionality, loyalty, authority, and purity
were positively associated with
right-wing ideology.
(Atari et al., 2023, p. 1171)
Graham et al, 2009 figure 2
invariance ?!
First reply to the invariance challenge is that a related
result was found with MFQ-2, which did exhibit measurement invariance
(except for purity).
But there is also a second reply ...
metanalysis
‘care and fairness are generally negatively, and loyalty, authority, and sanctity,
generally positively related to conservative political orientation’
(Kivikangas et al., 2021, p. 77)
example: purity
(Kivikangas et al., 2021, p. figure 6 (part))
There is a similar figure for each foundation. This one is for purity (which shows the largest effects).
effect sizes are small to medium (purity is the strongest)
BACKGROUND ‘Cohen suggests that r values of 0.1, 0.3, and 0.5 represent small, medium, and large effect sizes respectively.’
https://www.statmethods.net/stats/power.html#:~:text=We%20use%20the%20population%20correlation,and%20large%20effect%20sizes%20respectively.
(Other sources say .25 is medium)
Some studies do not find a non-zero estimate of effect size (cross 0.0),
but none find evidence for an effect in the opposite direction ...
Graham et al, 2009 figure 2
invariance ?!
If we have measurement invariance, that should make it harder to find and to replicate effects.
But the meta-analysis indicates the opposite.
So there is some hope that this is true
repeated for summary of main point
Graham et al. (2009, p. figure~3)
3
‘liberals and conservatives possess different moral profiles regarding the five moral foundations’
