what do moral attitudes tell us about public preferences
TRANSCRIPT
What do moral attitudes tell us
about public preferences for
lockdown measures in the UK?
Mesfin Genie, Mandy Ryan, Luis Enrique Loria-Rebolledo,
Verity Watson (HERU, Aberdeen)
Ruben A. Sakowsky (Medical Ethics and History of Medicine,
Gottingen)
Daniel Powell (Health Psychology, Aberdeen)
Shantini Paranjothy (Public Health Consultant, Aberdeen)
Motivation
• The COVID-19 pandemic has infected over 160M people resulting in over 3M deaths globally.
• The majority of European and high-income countries have focused on reducing the R number to < 1.
• Keeping R < requires several government-mandated interventions:
• Social distancing
• Closure of schools and non-essential businesses
• Other social restrictions (Lockdown, wearing face coverings, hand-washing etc)
• Success depends on compliance with lockdown/other restrictions.
• Governments have offered guidelines that would help “flatten the curve”
Anti-lockdown protests in the UK
Some government officials themselves have violated the
regulations they proposed
Why do some people decide to comply or ignore the regulations designed to protect public health?
• One framework to understand this involves exploring one’s moral intuitions.
• Moral Foundation Theory (MFT, Graham et al., 2009) – how people make judgments about “right vs wrong.”
• MFT proposes 5 basic moral foundations (caring, fairness, loyalty, authority, purity):• Caring/harm – intuitions that prevent harm and caring for others who are in need of protection • Fairness/equality – intuitions involving fair practices and equality • Ingroup/loyalty – sacrificing for one’s ingroup (protecting the interests of one’s own group)
• Authority/respect – respect for those who are higher in authority, obedience to authority figures • Purity/sanctity – a motivation to be pure both physically and spiritually
• People who care about others may be more likely to adopt behaviours that protect public health.
• People who value respecting authority may be more willing to comply with rules that governments promote.
Previous studies
• Individuals who place greater value on the sanctity foundation are more hesitant to use vaccines for children (Rossen et al., 2019).
• Promoting vaccine use using messages stressing caring, fairness, and sanctity might encourage vaccine roll-out (Rossen et al., 2019).
• Individuals who place value on the caring and fairness foundations are more likely to donate money to charity (Nilsson et al., 2016).
• Individuals who consider fairness to be an important value when making a moral judgement are less inclined to show speeding or rushing behaviour while driving (van den Berg et al., 2019).
• Caring, fairness, and sanctity concerns predict complying with wearing face masks and social distancing, while authority did not predict behavioural compliance (Chan, 2021).
Research questions
• Is there a link between MFT (caring, fairness, loyalty, authority, and purity) and preferences for various features of government lockdowns in the context of COVID-19?
• To what extent the five moral foundations predict the public’s willingness to accept/comply with more restrictive lockdowns to save lives?
DCE and MFQ20
• Discrete Choice Experiment (DCE) – to understand public preferences for pandemic responses in the UK
• Moral Foundation Questionnaire (MFQ20) – to understand the potential link between moral foundations and preferences for features of lockdown policies
• 22 statements covering 5 moral foundations – 4 items per foundation
• 2 survey formats/sections – 2 items for each foundation
Attributes and levels used to form policy scenarios
* Number of infections were linked to the excess death feature using an Infection Fatality Rate of 0.7%.
Feature Description Levels
Type of Lockdown
(Severity of restrictions)
How restrictive the lockdown (based on a colour/tier
system).
Green, Yellow, Amber,
Red
Length How long the lockdown is in place (in weeks) 3, 6, 10, 16 weeks
Postponement of usual
non-pandemic care
Whether non-pandemic medical care is postponed. No procedures are
postponed; Some
procedures are postponed;
All procedures are
postponed
Excess deaths Number of excess deaths (expressed as a fraction of
10,000).
1, 4, 9, 13
Infections* The number of infections (expressed as a fraction of
10,000).
100; 600; 1,300; 2,000
Ability to buy things How much of the goods that respondents are able to buy
today will they be able to buy in a year’s time.
100% of their shopping
trolley; 90% of their
shopping trolley; 80% of
their shopping trolley; 70%
of their shopping trolley
Job losses How many people lose their job (expressed as a fraction of
100).
0; 4; 15; 25
Lockdown restrictions
Design
• D-efficient design (Rose & Bliemer, 2013) with non-informative priors
• 24 choice tasks
• 3 blocks
• 8 tasks per respondent
• Virtual think-aloud interviews
• Participants recruited via Facebook advertisement (n=23)
• Stakeholder Advisory Group (n=4) – four countries of the UK
Example choice task
Moral Foundation Questionnaire – MFQ20
• Short form MFQ20 (www.moralfoundations.org) – extensively validated measure of 5 moral foundations (2 sections, 11 items in each section)
• From MFQ20 to the 5 moral foundations – examples:
• Harm/care (4 items from two response formats):
• Whether or not someone cared for weak or vulnerable (S1) (1 = not very relevant to 6 = extremely relevant)
• Compassion for those who are suffering is the most crucial virtue (S2) (1 = strongly disagree to 6 = strongly agree)
Care = (mfq_a1 + mfq_a7 + mfq_b7 + mfq_b1)/4
• Fairness/equality (4 items from 2 sections):
• Whether or not someone acted unfairly (S1)
• When the government makes laws, the number one principle should be ensuring that everyone is treated fairly (S2)
Recruitment and sample
• Qualtrics
• A representative sample of adults >18 years from the 4 nations of the UK (n = 4021)
• England (n=1112)• Scotland (n = 1143)• Northern Ireland (n = 848)• Wales (n=1098)
• Respondents screened by Qualtrics using sex and age using quotas to ensure a balance in each nation.
• Piloted early October 2020 using sample of 200 (50 per nation).
Results: MIXL model (WTSL-space, number of lives saved)
(1) (2)
MIXL - Mean SD
1. Model parameters Coeff (St.err) Coeff (St.err)
Alternative specific constant -1.06 (0.09) *** -
Yellow lockdown (ref: green lockdown) 0.17 (0.21) 0.41 (0.24) *
Amber lockdown (ref: green lockdown) -0.57 (0.19) *** 3.47 (0.19) ***
Red lockdown (ref: green lockdown) -2.53 (0.22) *** 7.01 (0.20) ***
Duration of lockdown (1-week increase) -0.23 (0.02) *** 0.32 (0.02) ***
All healthcare postponed (ref: none) -2.39 (0.21) *** 4.82 (0.18) ***
Some healthcare postponed (ref: none) -1.07 (0.19) *** 1.47 (0.15) ***
Ability to spend (10% increase) 0.84 (0.09) *** 0.92 (0.06) ***
Job loss (1 out of 100 increase) -0.28 (0.01) *** 0.25 (0.01) ***
Excess deaths (1 out of 10,000 increase) -1.17 (0.06) *** 2.24 (0.11)
2. Model diagnostics
Log-likelihood -17250.86
McFadden's pseudo-R² 0.26
Number of observations 33608
Number of respondents 4201
Trade-offs (Willingness to save lives - WTSL): Excess Deaths
In the UK, how many lives would have to be saved to accept …
WTSLred = 2.53 out of 10,000
UKPop 2019: 66M
-2.53 per 10,000 or 16,698 lives
would have to be saved to accept
red lockdown
All planned procedures postponed
WTSLpostponed = 2.39 out of 10,000
Heterogeneity in WTSL – the role of the five core moral foundations
MIXL-Mean
Model parameters Coeff (St.err) Fairness Care Authority Loyalty Purity
Alternative specific constant -0.91 (0.09) *** 0.26 *** -0.08 -0.11 -0.13 -0.02
Yellow lockdown (ref: green lockdown) 0.41 (0.20) ** 0.18 -0.26 0.29 0.13 0.06
Amber lockdown (ref: green lockdown) -0.90 (0.19) *** 0.13 0.47 ** 0.04 0.35 0.16
Red lockdown (ref: green lockdown) -2.62 (0.23) *** 0.74 *** 0.33 0.68 ** 0.40 * -0.29
Duration of lockdown (1-week increase) -0.21 (0.02) *** 0.01 0.04 ** -0.03 * -0.03 * 0.02
All healthcare postponed (ref: none) -2.36 (0.22) *** 0.29 0.25 0.06 0.10 -0.44 **
Some healthcare postponed (ref: none) -0.56 (0.17) *** 0.09 -0.14 0.20 -0.06 -0.02
Ability to spend (10% increase) 0.74 (0.08) *** -0.001 -0.20 *** 0.06 -0.03 0.12 *
Job loss (1 out of 100 increase) -0.28 (0.01) *** 0.02 * 0.02 ** 0.02 -0.02 ** -0.001
Excess deaths (1 out of 10,000 increase) -1.17 (0.06) *** 0.67 *** 0.18 *** -0.18 ** -0.29 *** -0.03
Excess lives required to be saved to accept/comply with red lockdown… do moral attitudes play a role?
2.62 per 10,000 (approximately 17,292lives) need to be saved to compensate or accept the most restrictive lockdown
For those who place a greater value on fairness
foundation …
-2.62+0.74 = 1.88 per 10,000 (approximately 12,408 lives) need to be saved to compensate or to accept the most restrictive lockdown
For those who value respecting authority …
-2.62+0.68 = 1.94 per 10,000 (approximately12,804 lives) need to be saved to compensateor to accept the most restrictive lockdown
Excess lives required to be saved to accept government’s decision of postponing all planned procedures… do moral attitudes play a role?
2.36 per 10,000 (approximately 15,576lives) need to be saved to compensate or accept government’s decision of postponing all planned procedures
For those who place greater value on
purity/sanctity foundation …
-2.36-0.44 = 2.8 per 10,000 (approximately 18,480
lives) need to be saved to compensate or to accept
government’s decision of postponing all planned
procedures
Summary of main findings
• Findings provide insight to trade-offs general public are willing to make.
• Preference heterogeneity in the trade-offs people are willing to make.
• Part of the heterogeneity could be explained by people’s moral intuitions.
• Fairness and authority dimensions of the MFT seem to explain some of the heterogeneity in terms of WTSL for the most restrictive lockdown.
• Government communications could be framed more on the relevant moral dimensions to increase behavioural compliance (e.g., please follow gov’t rules to protect your fellow citizens, we are all in this together, etc)
21
References • Chan, E. Y. (2021). Moral foundations underlying behavioural compliance during the COVID-19 pandemic.
Personality and Individual Differences, 171. https://doi.org/10.1016/j.paid.2020.110463
• Graham, J., Haidt, J., & Nosek, B. A. (2009). Liberals and Conservatives Rely on Different Sets of Moral Foundations. Journal of Personality and Social Psychology. https://doi.org/10.1037/a0015141
• Graham, J., Nosek, B. A., Haidt, J., Iyer, R., Koleva, S., & Ditto, P. H. (2011). Mapping the Moral Domain. Journal of Personality and Social Psychology. https://doi.org/10.1037/a0021847
• Nilsson, A., Erlandsson, A., & Västfjäll, D. (2016). The congruency between moral foundations and intentions to donate, self-reported donations, and actual donations to charity. Journal of Research in Personality, 65. https://doi.org/10.1016/j.jrp.2016.07.001
• Rose, J. M., & Bliemer, M. C. J. (2013). Sample size requirements for stated choice experiments. Transportation, 40(5). https://doi.org/10.1007/s11116-013-9451-z
• Rossen, I., Hurlstone, M. J., Dunlop, P. D., & Lawrence, C. (2019). Accepters, fence sitters, or rejecters: Moral profiles of vaccination attitudes. Social Science and Medicine, 224. https://doi.org/10.1016/j.socscimed.2019.01.038
• van den Berg, T. G. C., Kroesen, M., & Chorus, C. G. (2020). Does morality predict aggressive driving? A conceptual analysis and exploratory empirical investigation. Transportation Research Part F: Traffic Psychology and Behaviour, 74. https://doi.org/10.1016/j.trf.2020.08.017
Descriptive statistics of the five moral foundations
Moral foundation Mean SD Reliability (Cronbach’s α)
Harm/Care 4.701 0.935 0.76
Fairness 4.718 0.887 0.73
Loyalty 3.846 1.028 0.66
Authority 3.987 1.035 0.68
Purity 4.184 1.106 0.74
Baseline results (MNL and MXL models in preference space)
(1) (2) (3)
MNL MXL SD
1. Model parameters Coeff (St.err) Coeff (St.err) Coeff (St.err)
ASC1 -0.17 (0.013) *** -0.22 (0.017) *** -
Yellow lockdown (ref: green lockdown) 0.09 (0.023) *** 0.18 (0.033) *** 0.13 (0.169)
Amber lockdown (ref: green lockdown) -0.01 (0.022) 0.02 (0.034) 0.42 (0.080) ***
Red lockdown (ref: green lockdown) -0.28 (0.022) *** -0.37 (0.040) *** 1.41 (0.053) ***
Duration of lockdown (1-week increase) -0.03 (0.002) *** -0.04 (0.003) *** 0.06 (0.006) ***
All healthcare postponed (ref: none) -0.07 (0.019) *** -0.38 (0.029) *** 0.60 (0.048) ***
Some healthcare postponed (ref: none) -0.05 (0.019) *** -0.09 (0.024) *** 0.02 (0.071)
Ability to spend (10% increase) 0.10 (0.007) *** 0.09 (0.010) *** 0.10 (0.046) **
Job loss (1 out of 100 increase) -0.03 (0.001) *** -0.04 (0.002) *** 0.04 (0.002) ***
Excess deaths (1 out of 10,000 increase) -0.13 (0.002) *** -1.93 (0.040) *** 1.61 (0.052) ***
2. Model diagnostics
Log-likelihood -18887.58 -17325
McFadden’s pseudo-R² 0.19 0.25
Number of observations 33608 33608
Number of respondents 4201 4201
Heterogeneity in WTSL – the role of individualising vs binding foundations
MXL SD
1. Model parameters Coeff (St.error) Coeff (St.error) Individualising Binding
ASC -1.08 (0.10) *** - 0.11 (0.09) -0.20 (0.08) **
Yellow lockdown (ref: green lockdown) 0.03 (0.21) 0.24 (0.17) -0.27 (0.22) 0.45 (0.20) **
Amber lockdown (ref: green lockdown) -0.93 (0.20) *** 3.50 (0.18) *** 0.53 (0.20) *** 0.36 (0.17) **
Red lockdown (ref: green lockdown) -2.46 (0.24) *** 6.92 (0.20) *** 0.81 (0.24) *** 0.69 (0.20) ***
Duration of lockdown (1-week increase) -0.22 (0.02) *** 0.30 (0.01) *** 0.06 (0.02) *** -0.03 (0.01) **
All healthcare postponed (ref: none) -2.36 (0.22) *** 5.03 (0.18) *** 0.49 (0.19) ** -0.29 (0.16) *
Some healthcare postponed (ref: none) -1.00 (0.19) *** 1.61 (0.17) *** -0.10 (0.15) 0.04 (0.12)
Ability to spend (10% increase) 0.73 (0.08) *** 0.63 (0.07) *** -0.15 (0.06) ** 0.11 (0.05) **
Job loss (1 out of 100 increase) -0.29 (0.01) *** 0.26 (0.01) *** 0.04 (0.01) *** -0.01 (0.01)
Excess deaths (1 out of 10,000
increase) -1.19 (0.06) *** 2.09 (0.11) *** 0.80 (0.06) *** -0.43 (0.05) ***
2. Model diagnostics
Log-likelihood -17115.27
McFadden's pseudo-R² 0.26
Number of observations 33608
Number of respondents 4201
From MFQ20 to the 5 moral foundations …
• Summing scores and averaging across 4 items each (Graham et al., 2011).
• Harm/care (4 items from two response formats): • Whether or not someone suffered emotionally (S1)• Whether or not someone cared for weak or vulnerable (S1)• Compassion for those who are suffering is the most crucial virtue (S2)• One of the worst things a person could do is hurt a defenceless animal (S2)
• Fairness/equality (4 items from 2 sections):• Whether or not some people were treated differently than others (S1)• Whether or not someone acted unfairly (S1)• When the government makes laws, the number one principle should be ensuring
that everyone is treated fairly (S2)• Justice is the most important requirement for a society (S2)
From MFQ20 to the 5 moral foundations …
• Ingroup/loyalty (4 items from two formats):• Whether or not someone’s action showed love for his or her country (S1)• Whether or not someone did something to betray his or her group (S1)• I am proud of my country’s history (S2)• People should be loyal to their family members, even when they have done
something wrong (S2)
• Authority/respect (4 items from two formats):• Whether or not someone showed a lack of respect for authority (S1)• Whether or not someone conformed to the traditions of society (S1)• Respect for authority is something all children need to learn (S2)• Men and women each have different roles to play in society (S2)
From MFQ20 to the 5 moral foundations …
• Purity/sanctity (4 items from two response formats):• Whether or not someone violated standard of standards of purity and
decency (S1)
• Whether or not someone did something disgusting (S1)
• People should not do things that are disgusting, even if no one is harmed (S2)
• I would call some acts wrong on the grounds that they are unnatural (S2)
• Care + fairness individualising foundation
• Loyalty + authority + purity binding foundations