Policy packaging can make food system transformation feasible

Abstract

Redesigning food production and consumption is key to limiting global warming, soil erosion and biodiversity loss. Yet, transforming the food system may involve political feasibility problems, as potentially effective policy interventions interfere with citizens’ daily lives. Here, we show that policy packaging—the systematic bundling of different policy measures—can help to mitigate the potential trade-off between political feasibility and problem-solving effectiveness. We use conjoint experiments with citizens from China, Germany and the United States to scrutinize support for different combinations of policies aimed at reducing food systems’ environmental impacts. Our results do not support the widespread claim that costly market-based or push measures per se receive less support than non-market-based or pull measures. Instead, they show that citizens are likely to support even costly policies, but this support varies by country and depends on the specific combination of policy measures, their stringency and revenue earmarking.

Full text

Articles
https://doi.org/10.1038/s43016-020-0047-4
1Department of Humanities, Social and Political Sciences, ETH Zürich, Zürich, Switzerland. 2Institute of Science, Technology and Policy, ETH Zürich, Zürich,
Switzerland. 3Department of Social and Policy Sciences, University of Bath, Bath, UK. ✉e-mail: lukas.fesenfeld@ir.gess.ethz.ch
Practitioners and scientists are calling on governments to regu-
late and incentivize sustainable food consumption and pro-
duction1–5. In addition to supply-side measures that foster
sustainable production methods, many advocate demand-side poli-
cies that target consumption behaviour3,5–7. Demand-side policies
are regarded as paramount for reaching the climate change mitiga-
tion target set in the Paris Agreement, reducing the risks of climate
tipping points and preventing ‘carbon leakage’ (that is, avoiding an
increase in imports of emission-intensive food products as a substi-
tute for reducing the domestic production of these commodities)6–9.
Researchers also suggest packaging (that is, systematically bun-
dling) policy instruments on both the demand and supply side3,5,10–12
to combine multiple marginal gains13 and accelerate the sustain-
ability transformation of the food system1–5.
Arguably, the main hurdle to implementing such measures is not
technical but political, with political feasibility challenges stemming
from insufficient public acceptance of policy interventions6,8,11,14,15.
Environmental policy analysts usually anticipate trade-offs between
problem-solving effectiveness and political feasibility8,11,15,16. While
policy experts consider stringent policies to be more effective at
reducing environmental impacts17–19, public policy specialists often
note that public opposition and political risks are higher when gov-
ernments deploy policies that citizens perceive to be costly and as
interfering with their personal consumption choices8,11,16,20,21. To
identify environmental policies with the potential to be both effec-
tive and politically feasible, it is thus important to investigate public
support for sustainable food policies that are likely to become pub-
licly salient and contested.
To study public support for different policy packages aimed at
reducing the consumption of meat and fish products, we carried
out a survey-embedded conjoint choice experiment in China,
Germany and the United States (total n = 4,874), using quota sam-
pling to ensure representativeness in terms of age, employment
status, gender, income and region (see Methods). These countries
were selected for their importance in global food production and
consumption, as well as their diversity in terms of socioeconomic,
cultural and political systems. Specifically, they are among the
world’s largest producers and consumers of meat and fish products;
hence, they exert a major environmental impact22,23. This experi-
mental approach allowed us to evaluate public support for specific
and more complex policy designs in the form of policy packages11,24.
Thus, our study design mimics realistic choice situations25,26 and
generates new insights about policy packaging that go above and
beyond those of current research that analyses citizens’ preferences
for policy measures in isolation. In what follows, we show how pol-
icy packaging, in contrast with single policy instruments, can help
to mitigate the potential trade-off between political feasibility and
problem-solving effectiveness11.
Designing politically feasible policy packages
Policy support, and thus the political feasibility of policy interven-
tions, depends on citizens’ perceptions of policy-induced costs and
benefits11,16,21,27,28. In line with recent reviews27,28, we acknowledge
that people may misperceive the real cost (and benefit) of poli-
cies. Yet, such (mis)perceptions and their effects on public support
depend on various policy design factors, such as the policy instru-
ment type11,15,28,29, earmarking of policy revenues14,27,28,30 and the level
of policy stringency11,17 (also known as policy intensity18).
Part of the economic literature on policy instruments14,31,32 pro-
poses that market-based instruments are more cost-efficient at
reducing negative environmental externalities but typically receive
less public support than command-and-control instruments (for
example, government standards and regulations)21,32,33. The rea-
son for the gap between expert and public opinion would be that
command-and-control measures conceal personal cost implications
while market-based instruments do not21,34. Others15,28,32 suggest
that citizens usually prefer pull measures perceived to encourage
a desired behaviour (for example, discounts for environmentally
friendly products) over push measures perceived to discourage
undesired behaviour (for example, taxes on environmentally harm-
ful goods or forms of prohibition), as the latter are often considered
more costly and unfair and less effective19,28,29.
Although environmental taxes may be deemed effective despite
their unpopularity, various studies suggest that earmarking revenues
Policy packaging can make food system
transformation feasible
Lukas Paul Fesenfeld 1,2 ✉ , Michael Wicki1,2, Yixian Sun3 and Thomas Bernauer1,2
Redesigning food production and consumption is key to limiting global warming, soil erosion and biodiversity loss. Yet, trans-
forming the food system may involve political feasibility problems, as potentially effective policy interventions interfere with
citizens’ daily lives. Here, we show that policy packaging—the systematic bundling of different policy measures—can help to
mitigate the potential trade-off between political feasibility and problem-solving effectiveness. We use conjoint experiments
with citizens from China, Germany and the United States to scrutinize support for different combinations of policies aimed at
reducing food systems’ environmental impacts. Our results do not support the widespread claim that costly market-based or
push measures perse receive less support than non-market-based or pull measures. Instead, they show that citizens are likely
to support even costly policies, but this support varies by country and depends on the specific combination of policy measures,
their stringency and revenue earmarking.
NATURE FOOD | VOL 1 | MARCH 2020 | 173–182 | www.nature.com/natfood 173

Articles Nature Food
from such taxes could mitigate this dilemma by moderating the
perceived cost–benefit ratio in a direction that enhances public
support14,27,28,30,35. Recent studies of public support for carbon pric-
ing mechanisms14,27,30 show that equal per-capita transfers, ear-
marking revenues for environmental purposes, and transfers to
low-income groups increase public support compared with allocat-
ing carbon tax revenues to general government budgets. In contrast
with per-capita carbon dividends and revenue earmarking for envi-
ronmental purposes, achieving revenue neutrality through cutting
other taxes, such as income taxes, seems to be less popular14,27,30.
Regrettably, no such studies exist for the food sector, particularly
in the context of a concrete behavioural setting such as the daily
consumption of food products.
Policies also differ in terms of their stringency (or intensity)—
here defined as the increase in policy ambitiousness compared
with the status quo17,18. In the current context, policy stringency
captures the degree to which new policies impose additional costs
on polluting activity or imply the creation of additional benefits
for using and providing low-emission alternatives11,17,18. Typically,
stringent measures are expected to be effective at helping meet
environmental policy goals17,19, but can affect public support by
altering perceived policy-induced costs and benefits for citizens at
the same time. We expect that the level of policy stringency influ-
ences public support particularly when policies target consumers
directly, rather than producers6. In other words, the costs and ben-
efits of stringent demand-side measures are directly perceptible to
individuals, whereas those of stringent supply-side measures tend
to be underestimated15.
Apart from notable exceptions9,15, research on these topics has
focused on citizens’ preferences for particular policy instruments in
isolation, ignoring that real-world policy-making and effective envi-
ronmental policies typically take the form of policy packages10,11. In
our experiment, which was focused on how combinations of pol-
icy instruments with different levels of stringency affect political
feasibility, participants chose between two randomly assigned
policy packages with differing characteristics. The policy packages
presented to participants consisted of six different policy instru-
ments (called ‘attributes’ in our analysis), the specific features of
which were experimentally manipulated (see Table 1 and Methods).
Additionally, we included an attribute to earmark tax revenues,
whereby we randomly varied the purpose of earmarking, includ-
ing using revenues for the general budget, low-income groups, and
environmental and climate protection programmes.
Our study builds on the assumption that policy packaging can
help to mitigate the potential trade-off between political feasibility
and problem-solving effectiveness. Reminiscent of a compensation
logic11,36, we propose that the negative political feasibility implica-
tions of policy instruments with high cost visibility can be compen-
sated for by more positively valued policy instruments from which
citizens can expect to extract benefits. Hence, policy packaging may
open up room for political manoeuvre and enable policymakers to
adopt environmental policies that are both strong (and thus effec-
tive) and politically feasible.
Policy design and packaging effects vary across countries
The average marginal component effects (AMCEs)24 differ for each
of the randomly allocated policy design attribute values (that is,
each policy instrument type with a medium or high level of policy
stringency compared with no change in the status quo) according to
the share of respondents supporting a given policy package across
the three countries (Fig. 1). To provide conservative and easy-to-
interpret estimates of politically feasible policy packages (as in
previous studies30,37), we dichotomized the seven-point Likert rat-
ing into a binary variable using the median (4) as the cut-off value
(see Methods for details). Therefore, our measure indicates the
point at which respondents support a policy proposal at ≥5 on the
seven-point rating scale. In line with our compensation argument,
the results show that including instruments perceived as costly
Table 1 | Overview of policy design attributes in the experiment
Policy instrument Policy stringency (high, medium or
no change to status quo) Policy type and primary
target group Perceptions of policy-
induced costs and benefits
by the majority of citizens
Expected policy design
eects on support for
policy packages
Taxes: new tax on meat and
fish products
▪Increase prices by 30%
▪Increase prices by 15%
▪No new tax
▪ Market-based push
instrument targeting
consumer demand
▪ Low benefits and
moderate-to-high costs Moderate-to-strong
negative effects on
support for packages
Regulations: rules about
minimum share of
vegetarian meals in public
cafeterias
▪At least 75% vegetarian meals
▪At least 50% vegetarian meals
▪At least 25% vegetarian meals
▪No such rules
▪ Command-and-control
pull instrument targeting
consumer demand
▪ Low-to-moderate benefits
and moderate-to-high
costs
Campaigns: information
campaigns
▪Frequent campaigns
▪Occasional campaigns
▪No campaigns
▪ Information-based pull
instrument targeting
consumer demand
▪ Moderate benefits and
low costs Moderate-to-strong
positive effects on
support for packages
Discounts: discounts for
low-emission (vegetarian)
alternatives
▪Reducing prices by 30%
▪Reducing prices by 15%
▪No discounts
▪ Market-based pull
instrument targeting
consumer demand
▪ Moderate-to-high benefits
and low costs
Producer subsidies:
reduction of subsidies for
meat and fish producers
▪Eliminating subsidies
▪Halving subsidies
▪ Keeping subsidies at the current
level
▪ Market-based push
instrument targeting
supply side
▪ Moderate-to-high benefits
and low costs
Standards: animal farming
standards
▪Organic standards/free range
▪Higher animal farming standards
▪Standards kept at the current level
▪ Command-and-control
push instrument targeting
producer supply
▪ Moderate-to-high benefits
and low costs
The proposed attribute values for each of the policy design features varied randomly between a low level of policy stringency (that is, no difference from the status quo), a medium level of stringency and
a high level of stringency (see Methods). The policy-induced costs and benefits to citizens thus varied according to the type of instrument and the level of policy stringency. For example, more stringent
demand-side push measures, such as higher taxes on meat and fish products, typically imply moderate-to-high costs and low benefits for consumers, while stringent demand-side pull measures, such
as discounts for low-emission alternatives, imply moderate-to-high benefits and low visible costs. We argue that combining these instruments into policy packages can increase their political feasibility
compared with adopting measures in isolation. For further details, see Methods.
NATURE FOOD | VOL 1 | MARCH 2020 | 173–182 | www.nature.com/natfood
174

Articles
Nature Food
decreases the probability of support for a policy package, while a
combination that includes instruments perceived as beneficial
increases the probability of support.
For example, a new tax that increases meat and fish prod-
uct prices by 30% decreases the share of support by 16 percent-
age points (s.e. = 1.7) in the United States, 12 percentage points in
Germany (s.e. = 1.6) and 8 percentage points in China (s.e. = 1.8),
all else being equal. In the United States, the inclusion of a new tax
decreases the share of support significantly more than in China
or Germany. Unsurprisingly, citizens in all three countries prefer
lower taxes to higher ones. Moreover, including a stringent rule
that requires public cafeterias to offer at least 75% vegetarian meals
reduces the share of support by 8 percentage points (s.e. = 1.3) in the
United States and by 10 percentage points (s.e. = 1.3) in Germany.
However, in China, including this measure does not significantly
decrease the share of support, while a 50% minimum standard even
increases support by 6 percentage points (s.e. = 1.3).
As expected, incorporating policy instruments associated with
lower costs and higher benefits to citizens increases the share of
respondents who support the policy package. Nevertheless, we also
find substantial cross-country variation with these instruments.
While the share of support in Germany increases by around 3 per-
centage points (s.e. = 1.1) when proposals are made to halve pro-
ducer subsidies, we find no such effect for Chinese and US citizens.
In fact, Chinese respondents’ share of support decreases by 3 per-
centage points (s.e. = 1.1) if a policy package includes the elimina-
tion of producer subsidies.
Stricter animal farming standards increase the share of support
in all three countries. In China and the United States, requiring
producers to apply organic farming standards increases the share of
support by 7 percentage points (s.e. = 1.1). In Germany, the effects
are stronger than in the two other countries: requiring organic
farming standards increases the share of support by 19 percentage
points (s.e. = 1.1). Compared with all other policy design features
(policy instruments), stricter producer standards increase support
the most.
Public information campaigns about the negative impacts of
meat and fish consumption on the environment, animal welfare
and human health increase the share of support in China and the
United States, but not in Germany. While the share of support in
China increases by 6 percentage points (s.e. = 1.0) when the govern-
ment frequently informs citizens about the negative impacts of meat
and fish consumption, in the United States this share increases by
around 3 percentage points (s.e. = 1.0).
A policy package’s share of support increases by 2 percentage
points (s.e. = 1.1) in Germany, 3 percentage points (s.e. = 1.1) in
the United States and 7 percentage points (s.e. = 1.1) in China
when the government offers a large discount on low-emission food
Regulations Campaigns Discounts
Taxes Producer subsidies Standards
−0.2 −0.10 0.1 0.2 −0.2 −0.10 0.10.2 −0.2 −0.100.10.2
Organic
standards
Higher
standards
30% lower
prices
15% lower
prices
Eliminating
subsidies
Halving
subsidies
Frequent
campaigns
Occasional
campaigns
30% higher
prices
15% higher
prices
75%
vegetarian meals
50%
vegetarian meals
25%
vegetarian meals
Average marginal component effect on respondents' support share
Attribute value
Attribute value
China
Germany
United States
Fig. 1 | Effects of policy design attributes on the proportion of respondents supporting a policy package by country. Data points with horizontal lines
indicate percentage point estimates for medium and high policy stringency with cluster-robust 95% confidence intervals from linear least squares
regression. The dashed vertical line at 0 on the yaxis denotes the baseline category (that is, no change to the status quo). Robustness checks indicated
that, as expected, the results are internally valid and do not change when incorporating individual-level control variables (for example, age, income, gender
or education) into the analysis (see Supplementary Table 2b). See Table 1 for more details on the policy design attributes.
NATURE FOOD | VOL 1 | MARCH 2020 | 173–182 | www.nature.com/natfood 175

Articles Nature Food
alternatives, such as plant-based protein alternatives. However,
the positive effect of including discounts in policy packages
becomes insignificant in Germany for smaller discounts, while
in China and the United States even small discounts increase the
share of support.
Our analysis of policy packages does not support some of the
findings of previous research examining citizens’ preferences for
different instruments in isolation21,28,32,33. For example, our results
do not support the widespread claim that market-based or push
measures perse receive less support than non-market-based or pull
measures. In contrast, we find that individual support depends on
the country context, specific design features and the specific combi-
nation of instruments.
While earmarking environmental tax revenues for a specific
purpose can increase public support, such an effect is greater for
other policy design features (Fig. 2) and no significant differences
exist between the various earmarking purposes. This suggests
that revenue earmarking may be less effective at mitigating the
support-decreasing effects of environmental taxes in the food sec-
tor—a concrete behavioural setting—than with more general car-
bon taxation14,27,28,30,35. Combining environmental taxes with other
policy instruments perceived as beneficial increases public support
more strongly than any of the earmarking mechanisms. Thus, the
effects of revenue earmarking on public support seem to depend on
whether citizens evaluate policies in isolation or combination.
Overall, these findings show that variation in policy design has
a strong effect on public support. As discussed below (Fig. 3 and
Supplementary Table 1b), the difference in the share of support
between the most and least popular packages can reach 55 percent-
age points. While we find substantial cross-country variation in
the effects of different policy design features, packaging costly and
beneficial measures together can increase support for environmen-
tally effective policies in the food sector in all three countries.
Are effective policy packages politically feasible?
We estimate, per country, how many policy packages from all poten-
tial combinations of policy instruments in our experiment citizens
would support. The grading of policy packages clusters around
the statistical mean, mode and median of 5 (on a scale of 1–7) in
China, and 4 in Germany and the United States (see Supplementary
Fig. 3). In other words, most citizens in Germany and the United
States neither clearly opposed nor supported the vast majority of
policy package proposals our experiments generated, while Chinese
respondents tended to support the majority of policy proposals that
were presented.
Based on the estimated policy design effects presented in Figs. 1
and 2, we now predict the number of packages that would receive
clear support (with a rating of 5 or higher) from a majority of citi-
zens in all three countries. This provides a rough indication of how
liable citizens are to accept policy interventions in the food sector
(irrespective of the intervention’s exact nature). Our prediction
models show that a majority of citizens in all three countries support
at least some policy packages (Fig. 3). While in Germany and the
United States most participants clearly support (with a rating of
5 or higher) <6% of the policy combinations, in China 76% of them
attract strong majority support (see Methods and Supplementary
Table 1a for further details). We also compared a higher-income sub-
population from Germany and the United States with the Chinese
sample, which is biased towards that subpopulation. We found that
the results were not significantly different from those obtained with
the original sample (see Methods and Supplementary Table 2g),
suggesting that the political feasibility of policy interventions
−0.025 00.025 0.050 0.075
Reduce
income taxes
Public programmes for
low-income households
Public environmental and
climate protection programmes
Average marginal component effect on respondents' support share
Attribute value
ChinaGermany United States
Fig. 2 | Effects of tax revenue earmarking on the proportion of respondents supporting a policy proposal by country. Data points with horizontal lines
indicate percentage point estimates with cluster-robust 95% confidence intervals from linear least squares regression. The dashed vertical line at 0 on
the yaxis denotes the reference category (that is, revenues go into the general budget). Robustness checks indicated that, as expected, the results are
internally valid and do not change when incorporating individual-level control variables (for example, age, income, gender or education) into the analysis
(see Supplementary Table 2b).
NATURE FOOD | VOL 1 | MARCH 2020 | 173–182 | www.nature.com/natfood
176

Articles
Nature Food
in the food sector is higher in China than in the two other countries.
This is in line with the Chinese political context, in which the state
takes a strong role in regulating the economy and society38.
The share of respondents supporting different policy packages
that include a particular policy attribute depends on the combina-
tion of instruments (Fig. 3). For example, the proportion of German
respondents supporting policy packages that include tax increase
varies by almost 50 percentage points depending on whether other
policy instruments (for example, higher meat substitute discounts)
are also included. We obtain similar variation in support levels for
policy packages including other instruments, such as reduced pro-
ducer subsidies or rules on the minimum share of vegetarian meals
in public cafeterias. The range of respondents’ support for different
types of packages in China and the United States is smaller than in
Germany. In any case, support levels depend heavily on the specific
combination of policy measures. In China, most combinations of
policy instruments are supported by more than 50% of respondents,
yet in Germany and the United States careful design of particular
packages is pivotal to obtaining majority support.
Given that in the United States and Germany only some packages
receive majority support, we also examined the criteria for designing
such packages (see Supplementary Table 1a). In Germany and the
United States, most citizens currently do not support packages that
include a stringent tax that would increase the price of meat and fish
products by 30% (see Fig. 3). In these two countries, packages that
contain a moderate tax (that is, a 15% price increase) should not
simultaneously regulate the minimum share of vegetarian meals in
public cafeterias, but should be combined with positively evaluated
measures such as stringent animal farming standards, discounts
on vegetarian alternative-to-meat products and information cam-
paigns. For instance, most German respondents support a package
containing a moderate tax increase if subsidies for meat and fish
producers are simultaneously reduced. In the United States and
Germany, majority support for a moderate tax is also dependent on
earmarking the related revenue for public programmes that support
low-income households.
Most respondents in Germany do not support policy packages
that include very stringent rules for public cafeterias, while in the
China GermanyUnited States
Taxes Earmarking Subsidies Regulations StandardsDiscountsCampaigns
0
50
100
0
50
100
0
50
100
30% higher prices
15% higher prices
No new tax
Reduce income taxes
Low-income households
En
v. and climate protection
General budget
Eliminating subsidies
Halving subsidies
Keep at current level
25% vegetarian meals
50% vegetarian meals
75% vegetarian meals
No such rules
Organic standards
Higher standards
Kept at current level
30% lower prices
15% lower prices
No discounts
Frequent campaigns
Occasional campaigns
No campaigns
Distribution of respondents' predicted support share (%)
Fig. 3 | Distribution of respondents’ share of support for policy packages that include particular policy attributes. The box plots within the violin
plots correspond to the first, second and third quartiles (that is, the 25th, 50th and 75th percentiles) of the distribution of support for a policy package
that includes a specific policy attribute. The whiskers of the box plots represent the minimum and maximum share of support. The predicted shares of
support are estimated based on the AMCEs presented in Figs. 1 and 2. We also predict support levels for the least supported, most stringent (that is,
environmentally effective17,19) and most supported packages in each country (see Supplementary Table 1b). The least supported package never receives
majority support and includes high taxes on meat and fish products, stringent rules for public cafeterias, no discounts for vegetarian food alternatives
and no new animal farming standards in all three countries. While in China and the United States the least supported package includes the elimination of
producer subsidies, in Germany it includes keeping subsidies at the current level. The most stringent package receives majority support in China (55%;
s.e.=1.9), while clearly lacking majority support in both Germany (34%; s.e.=1.7) and the United States (32%; s.e.=1.9). In Germany and the United
States, the most strongly supported packages exclude stringent demand-side push measures (for example, taxes) that are perceived to impose high costs
on consumers, while in China the best-supported package includes stringent demand-side push measures. In fact, the best-supported package in China
includes taxes increasing the meat price by 15%, as well as mandating a minimum share of 50% vegetarian meals in public cafeterias. Env., environment.
NATURE FOOD | VOL 1 | MARCH 2020 | 173–182 | www.nature.com/natfood 177

Articles Nature Food
United States some packages including such stringent rules receive
majority support. Again, when introducing such rules, it could be
useful from a political feasibility perspective to also adopt higher
animal farming standards in parallel. In both countries, we also find
that a majority of citizens support a substantial number of policy
packages if discounts for alternative products, reductions in pro-
ducer subsidies or frequent information campaigns are integrated.
Unlike in the two democracies, we find that a majority of Chinese
study participants support policy packages that include high tax
increases on meat and fish products. We identified many packages
that attracted majority support when incorporating a 75, 50 or 25%
minimum share of vegetarian meals in public cafeterias.
While Fig. 3 shows that in Germany and the United States it
is in general difficult to obtain majority support for packages that
include higher taxes, the marginal means plotted in Fig. 4 provide
a more nuanced understanding of how packaging alters support
for new taxes. These results underscore that, in all three countries,
combining higher taxes with more stringent farming standards and
a reduction in subsidies for meat and fish producers can increase the
proportion of respondents who support packages that lead to price
increases. Also, earmarking tax revenues to reduce income taxes
or fund programmes for low-income households can have similar
positive effects on support for packages, including higher tax.
In light of academic14,27–29 and real-world political debates (for
example, recent protests against new consumption taxes in France)
about citizens’ perceptions of the distributive effects of policies, we
investigated to what extent our results depended on the individual
income level in the two democratic country cases. Overall, we find
that income only marginally affects citizens’ preferences for policy
packages (see Supplementary Table 2g,h). While the two lowest
income quintiles in Germany have slightly stronger preferences
for discounts on vegetarian alternatives to meat and fish products
than the two highest income quintiles, higher-income groups in the
United States prefer such discounts and reductions of income taxes
compared with the two lowest income quintiles. Most importantly,
no significant differences between the two income groups were
found in terms of attitudes towards higher taxes in both Germany
and the United States. This suggests that policy packaging—in addi-
tion to discussions about revenue recycling strategies14,27,30—offers
an opportunity to compensate for the perceived costs of instru-
ments such as taxes, and to mitigate political feasibility problems
across different income groups.
We also conducted several subgroup analyses to compare the
effects of policy packaging across voter groups. While in the United
States and Germany we see differences by ideology and party iden-
tification (we could not collect these data in China), the direction
and magnitude of effects are similar between the two countries (see
Supplementary Table 2i–m). This implies that policy packaging can
increase public support across different ideological groups and that
there is political room to design packages that are well supported
across constituencies.
Finally, we also assessed potential heterogeneous packaging
effects across other subgroups, such as those differentiated by previ-
ous meat and fish consumption. Our survey reveals high numbers
of meat and fish consumers in all three countries (that is, 97% in
China, 96% in Germany and 98% in the United States). Therefore,
the results do not change much when comparing the effects of meat
and fish eaters with the full sample findings presented above (see
Supplementary Table 2a–d), although the effects vary depending on
individuals’ frequency and amount of meat and fish product con-
sumption (see Supplementary Table 2e,f). Policy packaging is espe-
cially important to increase support from meat lovers—those most
affected by the policies.
Discussion
To date, research has focused on measuring and describing envi-
ronmental impacts caused by the food sector and assessing how
these impacts could be reduced through changes in agricultural
practices or demand-side measures for consumer behavioural
change1–5. Research on the political feasibility of policy interven-
tions able to reduce the environmental impact of the food system
is still scarce. This study systematically examines how potentially
Earmarking
Reduce
income taxes
Low-income
households
Env. and climate
protection
General
budget
25%
50%
75%
Marginal mean (support share)
Producer subsidies
Eliminating
subsidies
Halving
subsidies
Keep at
current level
25%
50%
75%
Regulations
25%
vegetarian
meals
50%
vegetarian
meals
75%
vegetarian
meals
No
such rules
25%
50%
75%
Standards
Kept at
current level
Higher
standards
Organic
standards
25%
50%
75%
Marginal mean (support share)
Campaigns
Frequent
campaigns
Occasional
campaigns
No
campaigns
25%
50%
75%
Discounts
30% lower
prices
15% lower
prices
No
discounts
25%
50%
75%
Increasing prices by 15% Increasing prices by 30%No new tax China Germany United States
Fig. 4 | Tax-conditional marginal means by country. Tax-conditional marginal means show respondents’ average share of support for all policy packages
when one policy attribute and one of the three tax levels are present. Given that many economists regard environmental taxes on emission-intensive food
products as an efficient and effective policy instrument for reducing the environmental impact of the food system2–4, we were particularly interested in
estimating the effects of packaging on policy support when higher taxes on meat and fish products are combined with other instruments.
NATURE FOOD | VOL 1 | MARCH 2020 | 173–182 | www.nature.com/natfood
178

Articles
Nature Food
effective policy instruments may be designed and combined so as
to reduce public backlash and ensure political feasibility of trans-
forming the food system.
We show that the widely hypothesized dilemma between effec-
tiveness and political feasibility is empirically observable. However,
while policy instruments with costly implications for citizens
reduce public support and increase political feasibility risks, com-
bining costly and beneficial policy instruments into policy packages
can mitigate this dilemma. We identify such compensation effects
across countries and income groups, suggesting that policy packag-
ing might be an opportunity to reduce voters’ distributive concerns
about regulatory interventions in the food sector. Packaging is also a
successful strategy to increase policy support across different ideo-
logical constituencies and of meat and fish eaters.
Our results do not support existing research21,32,33 that finds
costly market-based or push measures perse to receive less sup-
port than non-market-based or pull measures. Although there are
no simplistic decision heuristics for policymakers, careful policy
design can increase support across countries and is pivotal to reduc-
ing political risk. In fact, policymakers interested in ambitious envi-
ronmental policy interventions in the food sector should employ
a holistic policy design strategy3,5,11. They can minimize political
risks by carefully combining different types of policy instruments
with varying degrees of stringency and cost–benefit implications
for citizens. Carefully designed policy packages that are most likely
to pass the political stress test39 include a medium level of policy
stringency for costly demand-side push measures (for example,
higher taxes), strong supply-side push measures (for example,
stricter animal farming standards) and demand-side pull measures
that generate benefits for consumers (for example, larger discounts
for meat alternatives).
We conclude with a cautionary note and suggestions for fur-
ther research. First, even though we only sampled eligible voters
and citizens, one issue common to virtually all survey-embedded
experiments, including ours, is that we cannot fully ensure that
respondents perfectly represent the citizens who turn out to vote or
influence actual political outcomes in other ways.
Second, although to a lesser degree than standard preference
measures, conjoint-based survey experiments may still be affected
by some level of social desirability bias. Further investigation is
necessary to buttress the external validity of our findings and test
policy packaging effects across different issues. Different methods
(for example, field experimental and qualitative) could complement
conjoint experiments to capture potential effects of policy inter-
action and feedback, path dependency and sequencing on public
support17,40–42. In addition, comparative case studies could be used
to investigate the prioritization and coordination in policy design
processes across sectors and within multi-level governance systems.
Third, support for different policy packages may depend on the
specific types of meat and fish products targeted by the respective
proposals, and on the policy frames and source cues provided in the
discussion of a specific policy package. Future studies could explore
whether support varies depending on, for example, the meaning of
organic farming in different countries, or individuals’ perception
of some animal products (for example, fish) as healthy. Survey and
field experimental research could also tease out the effects of differ-
ent arguments (for example, health, environmental and animal wel-
fare effects) and political actors in favour of or against the reduction
of different animal products.
Fourth, food consumption and production are certainly con-
tingent on the sociocultural, political–economic and geographic
conditions in each country. While our results indicate some notable
differences across countries (for example, higher average support
levels in China), we cannot fully control for the possibility that
sociocultural and political factors could lead to different degrees of
social desirability bias across countries. The political context may
also determine individuals’ previous experience with similar policy
proposals (see also Methods and Supplementary Table 5). Field
experiments across countries can better control for such factors and
would allow us to test how citizens’ familiarity with a policy or dif-
ferent types of meat substitutes moderate support levels.
Finally, the costs and benefits of policy changes to citizens are
usually less than those faced by organized interest groups17,40,43. Thus,
ambitious policies for reducing the environmental impact of the
food system must obtain support, or at least acceptance, from food
producers and retailers. Our findings indicate that citizens prefer
policy packages that offer benefits to themselves and impose costs
on producers. However, such packages are likely to encounter oppo-
sition from organized incumbent producer interest groups. This
challenge will force policymakers to consider both the preferences
of organized interests (food producers) and citizens (consumers).
The relative influence of these groups on actual policy outputs is
likely to depend on the public salience of the debate44 and poli-
cymakers’ opportunities to avoid blame45. Future research should
build on the policy feedback, path dependency and sequencing lit-
erature17,40–42 and account for the complex interplay of citizen and
interest group preferences, as well as the interactions of behavioural,
technological and policy changes that can induce political feedback.
Methods
We embedded a conjoint experiment in a survey of three countries—namely,
China, Germany and the United States (n = 4,874)—to investigate citizen support
for sustainable food policy packages.
Experimental design. We asked respondents to evaluate profiles that combine
multiple randomly assigned attributes. We used a conjoint design of fully
randomized paired profiles in which each respondent was shown profiles of two
different hypothetical policy packages displayed side by side (see an example in
Extended Data Fig. 1). Hence, each policy measure constituted an attribute in the
package to which it belonged, and the attribute values were randomly assigned
such that the two policy packages in each pair differed in one or more attribute
values24. This paired-profiles design was chosen because research suggests it
performs well at reducing social desirability bias and replicating real-world
behaviour25. Each policy package contained six types of policy and an additional
attribute related to earmarking for the tax policy.
To help participants understand how the relevant policy measures function,
before showing the pairs of policy packages, we showed a page with brief
descriptions of each policy measure (see the section ‘Description of policy package
attributes for respondents’). The order of policy measures shown on this page was
randomized and the texts for each measure were displayed one after the other as
participants clicked a button. Participants had to spend at least 3 s reading the text
associated with each policy measure before moving to the next one.
Description of policy package attributes for respondents. Various policies are
being considered with respect to meat and fish consumption in the United States.
Please carefully look at the following policy proposals. We will later ask you some
questions about them.
New tax on meat and fish products. The government could introduce a new tax on
meat and fish products in addition to the existing sales tax on all food products.
This tax would increase prices for meat and fish products, and would thus motivate
consumers to buy and eat less meat and fish, and thus more alternatives.
Rules for public cafeterias. The government could introduce minimum amounts or
shares of vegetarian products that must be offered in public cafeterias (for example,
schools, universities, hospitals and government offices). This would obligate
cafeterias to offer fewer meat and fish products and more alternatives, and would
motivate consumers to buy and eat less meat and fish, and thus more alternatives.
Animal farming standards. The government could introduce stricter standards for
animal farming by reducing the use of artificial chemicals, including antibiotics
and growth hormones for animals, as well as pesticides and fertilizers in animal
feed production, and increasing the minimum living space for animals. This
would control water and other environmental contamination, limit resistance to
antibiotics and increase animal welfare. It would result in higher production costs
and higher prices for meat and fish products, and would thus motivate consumers
to buy and eat less meat and fish, and thus more alternatives.
Information campaigns about the impact of meat and fish consumption. The
government could organize and fund nationwide campaigns to inform consumers
about the negative impacts of meat and fish consumption on human health, animal
NATURE FOOD | VOL 1 | MARCH 2020 | 173–182 | www.nature.com/natfood 179

Articles Nature Food
welfare, the climate and the environment. This would motivate consumers to buy
and eat less meat and fish, and thus more alternatives.
Discounts for vegetarian alternatives. The government could put in place financial
discounts for vegetarian alternatives to meat and fish, such as tofu, seitan or lentils.
Incentives such as subsidies and tax breaks would lower the prices of alternatives
to meat and fish. This would motivate consumers to buy and eat more of these
products and less meat and fish.
Reducing subsidies for meat and fish producers. The government could reduce its
current financial support (subsidies) for meat and fish producers. This would result
in higher prices for meat and fish and would motivate consumers to buy and eat
less meat and fish, and thus more alternatives.
Sample. The conjoint experiment described above was embedded in an internet-
based survey conducted in February 2018 in China, Germany and the United
States. It drew on a non-probability, quota-based sample in the three countries
provided by Ipsos. All experiments were internet based and participants were
recruited via the online panels that Ipsos maintains in each country. While these
panels are not probability based, they are non-convenience samples as Ipsos actively
manages and refreshes them to target respondents who match census statistics. For
our survey, Ipsos pre-selected respondents from their panels according to quotas
and constructed samples to match the national voting age population in the three
countries. More specifically, we used hard quotas in our sampling in an attempt
to match distribution by gender, age and region according to each country’s latest
census data (China in 2010, Germany in 2013 and the United States in 2015; see
Supplementary Table 3a–c). Quotas for gender and age were combined to ensure
that each age group was nationally representative in terms of gender distribution.
We also employed soft quotas for education, income, rural–urban population and
occupation to ensure that the samples were not extensively skewed towards certain
sociodemographic groups. The soft quotas worked well in Germany and the United
States, such that samples in these two countries closely followed distribution by
income, education, rural–urban divide and occupation in the national population.
Chinese respondents were recruited from tier I and II cities and the sample
was skewed towards a higher-income and urban population, as rural low-income
populations in China remain under-represented in all existing internet-based
samples. However, due to uneven economic development in China, our sample
represented the most relevant population subgroup of the urban middle class
whose consumption patterns have the most significant environmental impacts in
the country46.
In terms of the consumption-related carbon footprint, that of the urban middle
class in China (>6.4 tCO2 per capita) is comparable to the average consumption
carbon footprint of citizens in industrial countries such as those in the European
Union46. Our samples in all three countries were thus representative of politically
relevant citizens (that is, selectorates47) who represent the voting age population in
the two democratic cases (the United States and Germany), as well as the powerful,
emission-intensive, middle-class, urban population in China46.
We ran robustness check analyses, restricting the German and US samples to
a higher-income subsample of the population, and did not find any substantial
differences from the results for the full population sample (see Supplementary
Table 2e). This suggests that the country-related differences in our results were
indeed explained by differences in the political and cultural context38. Nonetheless,
further research is necessary to better understand cross-country differences
regarding support for environmental policy packages.
Our survey contained different parts addressing several research questions,
including the one on which this paper focuses. The first part covered respondents’
sociodemographic profile, including gender, age, income, education, number of
family members and political ideology. Questions about political ideology could
not be asked in China due to restrictions on such survey activity. The second part
was about previous meat consumption (weekly amount and meat types consumed).
In the third part, respondents were questioned about their food purchasing criteria.
We also measured respondents’ previous awareness of potential sustainability
problems associated with meat consumption and some general policy attitudes
towards reducing the consumption of meat and fish products.
The conjoint experiment was introduced after collecting individual-level data,
which we used in robustness check analyses to confirm that the conjoint experimental
randomization worked as expected. The findings of these robustness checks indicated
that, as expected, the conjoint results were internally valid and did not change when
incorporating individual-level control variables (for example, age, income, gender and
education) into our regression analysis (see Supplementary Table 2b).
We developed the questionnaire in English first. It was then proofread by a
professional translator and then translated by two of the authors who are native
speakers of Chinese and German. The questionnaires are available as part of the
replication material in the public repositor y (see Data availability statement). We
carefully validated and pretested the survey wording by conducting 33 cognitive
interviews with citizens from all three countries.
The median average time for survey completion was 18 min in the United
States, 17 min in Germany and 14 min in China. We invalidated responses that
were submitted within 6 min to ensure that only those respondents who had paid
enough attention to the questionnaire were included in the final sample, which
comprised 4,874 responses from three countries (1,624 in both Germany and the
United States, and 1,626 in China).
In the conjoint experiment, the policy attributes could take the levels shown in
the section ‘Policy package attributes and levels’, and as noted above, the attribute
values were randomly assigned so that respondents had to choose and rate
differently designed policy packages.
Policy package attributes and levels.
New tax on meat and fish products.
(1) Increase prices by 30%
(2) Increase prices by 15%
(3) No new tax
Use of tax revenues.
(1) Public environmental and climate protection programmes (do not display
with 0% tax)
(2) General government budget (do not display with 0% tax)
(3) Public programmes for low-income households (do not display with 0% tax)
(4) Reduce income taxes (do not display with 0% tax)
(5) No tax revenues (only display with 0% tax (that is, tax attribute item 3))
Rules for public cafeterias.
(1) At least 75% vegetarian meals
(2) At least 50% vegetarian meals
(3) At least 25% vegetarian meals
(4) No such rules
Discounts for vegetarian alternatives.
(1) Reduce prices by 30%
(2) Reduce prices by 15%
(3) No discounts
Animal farming standards.
(1) Organic practices (no antibiotics/chemicals) and no cages
(2) Stringent limits on antibiotics/chemicals and large cages
(3) Standards kept at current level
Information campaigns.
(1) Frequent campaigns
(2) Occasional campaigns
(3) No campaigns
Reducing subsidies for meat and fish producers.
(1) Eliminate subsidies
(2) Halve subsidies
(3) Keep subsidies at current level
We chose these policies in consultation with nine food governance experts
in the three countries under examination (see Supplementary Table 4). Most of
these experts agreed that these policies—particularly those with a high level of
policy stringency combined into holistic policy packages—could be effective
at reducing the environmental impact of food systems. Supplementary Table 5
provides an overview of whether and how different policy measures have already
been discussed or adopted at the national level in each of the three countries. This
information provides more context for the reader with respect to how far citizens
have already been exposed to particular policy debates/instruments and could
familiarize with particular food policy measures.
After being shown a pair of two hypothetical policy packages, survey
participants were required to perform two tasks. First, they were asked to indicate
which of the two proposals they preferred. This resulted in a choice outcome
variable coded ‘1’ for the preferred packages and ‘0’ for the rejected ones. Second,
they indicated to what extent they supported or opposed each package using
an ordinal scale of seven degrees ranging from ‘strongly oppose’ to ‘strongly
support’. These ratings resulted in a numerical variable from 1–7 for participants’
level of support for each package. To estimate the share of support, we recoded
the seven-point Likert scale into a binary oppose/support variable. In essence,
the values 1–4 indicate opposition to the proposed policy package, while the
values 5–7 indicate support for the policy package. This coding scheme, in line
with previous studies30,37, provides conservative and hence more externally
valid estimates of feasible policy packages, given that we only consider packages
respondents clearly support (that is, awarded a rating of 5 or higher). The results
presented in Figs. 1 and 2 would slightly change if we used the less conservative,
non-recoded seven-point rating scale or choice outcome as the dependent
NATURE FOOD | VOL 1 | MARCH 2020 | 173–182 | www.nature.com/natfood
180

Articles
Nature Food
variable (see Supplementary Figs. 1a,b and 2a,b). In this study, we are primarily
interested in packages that receive majority support (that is, more than 50% of
respondents indicated a rating of 5 or higher). We identify a considerably higher
number of policy packages that a majority of citizens would probably tolerate, but
do not clearly support (that is, a rating of 4) in each country. Empirical research
buttresses the assumption that policies change when at least a majority of citizens
demand a policy change20,48,49. In light of debates about citizens’ perceptions of the
distributive effects of policies14,27–29, we also explored potential moderating effects
for different income groups (see Supplementary Table 2g,h). We also analysed
policy support among different constituencies and relevant population subgroups.
For Germany and the United States, we studied support among different party
supporters and ideological subgroups (we could not collect these data in China;
see Supplementary Table 2i–m). In all three countries, we also compared effects
contingent on previous food consumption habits (see Supplementary Table 2d–f).
The wording and format of the two outcome questions are shown in Extended
Data Fig. 1. In our experiment, participants were shown four different pairs of
policy packages consecutively and were asked to perform the choice and rating
tasks four times. To avoid ordering effects, we randomized the order of attributes
in each package across all participants but kept this constant per participant in the
following choice tasks to avoid cognitive overload.
Statistical analysis. Each survey participant evaluated two policy packages and
four pairs of profiles, resulting in a total of 38,992 policy package evaluations
(that is, 4,874 respondents × 2 policy packages × 4 choice pairs). Since attribute
values were randomly assigned across all packages and participants, we could
non-parametrically estimate AMCEs, which measure the average causal effect
of each attribute on participants’ support for a policy package24. To do this, we
used least squares regression to regress the choice, rating and binary recoded
rating outcomes on sets of indicator variables that captured the values of each
attribute while omitting one level of each attribute as the reference category and
clustering the standard errors by respondent. The results for the recoded binary
support rating to estimate the effects on respondent share supporting a policy
are displayed in Figs. 1 and 2. We also estimated how the mean of this variable
(equivalent to the proportion who supported the policy) varied in relation to
the policy design features. As a robustness check, we included the results for the
choice and seven-point Likert ratings in the Supplementary Information (see
Supplementary Figs. 1a,b and 2a,b). We also added additional robustness check
analyses (Supplementary Table 2c) based on observations that included only rating
tasks that aligned with the choice tasks. In essence, we dropped those observations
for which the chosen policy package was rated lower than the non-chosen package.
Overall, the robustness check showed that the results did not significantly change
compared with the main results (see Supplementary Table 2c). For our prediction
models, we used the estimated AMCEs from the non-parametrical linear
regression output to calculate which specific combinations of policy instruments
yielded a support level above the majority threshold. In theory, mathematically
speaking, there are 2,916 potentially different combinations of instruments in our
analysis, while our empirically based prediction models showed that in China
>76% (that is, 2,221) of these potential policy packages received majority support.
In the United States and Germany, <6% (United States: 174; Germany: 34) of
the 2,916 theoretically possible packages received a support rating of 5 or higher.
The specific criteria for designing policy packages with majority support (that is,
a rating of 5 or higher) are presented in Supplementary Table 1.
Reporting Summary. Further information on research design is available in the
Nature Research Reporting Summary linked to this article.
Data availability
All data and associated code for replicating the experimental results presented
in Figs. 1–4 of the paper as well as the results presented in the Supplementary
Information are publically available at the Harvard Dataverse: https://doi.
org/10.7910/DVN/FR73RE.
Received: 7 November 2019; Accepted: 13 February 2020;
Published online: 17 March 2020
References
1. Bajželj, B. etal. Importance of food-demand management for climate
mitigation. Nat. Clim. Change 4, 924–929 (2014).
2. Springmann, M. etal. Mitigation potential and global health impacts from
emissions pricing of food commodities. Nat. Clim. Change 7, 69–74 (2017).
3. Springmann, M. etal. Options for keeping the food system within
environmental limits. Nature 562, 519–525 (2018).
4. Godfray, J. etal. Meat consumption, health, and the environment. Science
361, eaam5324 (2018).
5. Poore, J. & Nemecek, T. Reducing food’s environmental impacts through
producers and consumers. Science 992, 987–992 (2018).
6. Creutzig, F. etal. Towards demand-side solutions for mitigating. Nat. Clim.
Change 8, 260–271 (2018).
7. Steg, L. Limiting climate change requires research on climate action. Nat.
Clim. Change 8, 759–761 (2018).
8. Lemken, D., Kraus, K., Nitzko, S. & Spiller, A. Staatliche Eingrie in die
Lebensmittelwahl: Welche klimapolitischen Instrumente unterstützt die
Bevölkerung? GAIA 27, 363–372 (2018).
9. Fesenfeld, L. P., Schmidt, T. S. & Schrode, A. Climate policy for short- and
long-lived pollutants. Nat. Clim. Change 8, 933–936 (2018).
10. Bataille, C., Guivarch, C., Hallegatte, S., Rogelj, J. & Waisman, H. Carbon
prices across countries. Nat. Clim. Change 8, 648–650 (2018).
11. Fesenfeld, L. P. e Eects of Policy Packaging on Public Support for
Transformative Policies (Department of Humanities, Social and Political
Sciences, ETH Zurich, 2018); https://doi.org/10.3929/ethz-b-000397449
12. Fesenfeld, L. Governing urban food systems in the long-run: comparing
best practices in sustainable food procurement regulations. GAIA 25,
260–270 (2016).
13. Alexander, P., Reddy, A., Brown, C., Henry, R. C. & Rounsevell, M. D. A.
Transforming agricultural land use through marginal gains in the food
system. Glob. Environ. Change 57, 101932 (2019).
14. Klenert, D. etal. Making carbon pricing work for citizens. Nat. Clim. Change
8, 669–677 (2018).
15. Wicki, M., Fesenfeld, L. & Bernauer, T. In search of politically feasible
policy-packages for sustainable transport: insights from choice experiments in
China, Germany, and the USA. Environ. Res. Lett. 14, 084048 (2019).
16. Bernauer, T. Climate change politics. Annu. Rev. Polit. Sci. 16, 421–448 (2013).
17. Pahle, M. etal. Sequencing to ratchet up climate policy stringency. Nat. Clim.
Change 8, 861–867 (2018).
18. Scharin, A., Sewerin, S. & Seubert, S. Toward a comparative measure of
climate policy output. Policy Stud. J. 43, 257–282 (2015).
19. Rhodes, E., Axsen, J. & Jaccard, M. Does eective climate policy require
well-informed citizen support? Glob. Environ. Change 29, 92–104 (2014).
20. Stokes, L. C. Electoral backlash against climate policy: a natural experiment
on retrospective voting and local resistance to public policy. Am. J. Pol. Sci.
60, 958–974 (2016).
21. Stadelmann-Steen, I. & Dermont, C. e unpopularity of incentive-based
instruments: what improves the cost–benet ratio? Public Choice 175,
37–62 (2018).
22. OECD–FAO Agricultural Outlook 2018–2027 (OECD, 2018).
23. Ecological Footprint Explorer (Global Footprint Network, 2018); https://data.
footprintnetwork.org/#/
24. Hainmueller, Hopkins, D. J. & Yamamoto, T. Causal inference in conjoint
analysis: understanding multidimensional choices via stated preference
experiments. Polit. Anal. 22, 1–30 (2014).
25. Hainmueller, J., Hangartner, D. & Yamamoto, T. Validating vignette and
conjoint survey experiments against real-world behavior. Proc. Natl Acad. Sci.
USA 112, 2395–2400 (2015).
26. Bansak, K., Hainmueller, J. & Hangartner, D. How economic, humanitarian,
and religious concerns shape European attitudes toward asylum seekers.
Science 354, 217–222 (2016).
27. Carattini, S., Carvalho, M. & Fankhauser, S. Overcoming public resistance to
carbon taxes. Wiley Interdiscip. Rev. Clim. Change 9, e531 (2018).
28. Drews, S. & van den Bergh, J. C. J. M. What explains public support for
climate policies? A review of empirical and experimental studies. Clim. Policy
16, 855–876 (2016).
29. Huber, R., Wicki, M. & Bernauer, T. Public support for environmental policy
depends on beliefs concerning eectiveness, intrusiveness, and fairness.
Env. Polit. https://doi.org/10.1080/09644016.2019.1629171 (2019).
30. Beiser-McGrath, L. F. & Bernauer, T. Could revenue recycling make eective
carbon taxation politically feasible? Sci. Adv. 5, eaax3323 (2019).
31. Hahn, R. W. & Stavins, R. N. Economic incentives for environmental
protection: integrating theory and practice. Am. Econ. Rev. 82, 464–468 (1992).
32. Kallbekken, S. & Sæælen, H. Public acceptance for environmental taxes:
self-interest, environmental and distributional concerns. Energy Policy 39,
2966–2973 (2011).
33. Cherry, T. L., Kallbekken, S. & Kroll, S. e acceptability of eciency-
enhancing environmental taxes, subsidies and regulation: an experimental
investigation. Environ. Sci. Policy 16, 90–96 (2012).
34. Kirchgässner, G. & Schneider, F. On the political economy of environmental
policy. Public Choice 115, 369–396 (2003).
35. Fairbrother, M. When will people pay to pollute? Environmental taxes,
political trust and experimental evidence from Britain. Br. J. Polit. Sci. 46,
661–682 (2019).
36. Häusermann, S., Kurer, T. & Traber, D. e politics of trade-os: studying the
dynamics of welfare state reform with conjoint experiments. Comp. Polit.
Stud. 52, 1059–1095 (2018).
37. Rinscheid, A. & Wüstenhagen, R. Germany’s decision to phase out coal by
2038 lags behind citizens’ timing preferences. Nat. Energy 4, 856–863 (2019).
38. Zhang, L., Xu, Y., Oosterveer, P. & Mol, A. P. J. Consumer trust in dierent
food provisioning schemes: evidence from Beijing, China. J. Clean. Prod. 134,
269–279 (2016).
NATURE FOOD | VOL 1 | MARCH 2020 | 173–182 | www.nature.com/natfood 181

Articles Nature Food
39. Bernauer, T., Prakash, A. & Beiser-McGrath, L. F. Do exemptions undermine
environmental policy support? An experimental stress test on the odd–even
road space rationing policy in India. Regul. Gov. https://doi.org/10.1111/
rego.12225 (2018).
40. Meckling, J., Kelsey, N., Biber, E. & Zysman, J. Winning coalitions for climate
policy. Science 349, 1170–1171 (2015).
41. Schmid, N., Sewerin, S. & Schmidt, T. Explaining advocacy coalition change
with policy feedback. Policy Stud. J. https://doi.org/10.1111/psj.12365 (2019).
42. Schmidt, T. S. & Sewerin, S. Technology as a driver of climate and energy
politics. Nat. Energy 2, 170–184 (2017).
43. Oates, W. E. & Portney, P. R. in Handbook of Environmental Economics
325–354 (Elsevier, 2003).
44. Culpepper, P. D. Quiet Politics and Business Power: Corporate Control in
Europe and Japan (Cambridge Univ. Press, 2010).
45. Weaver, R. K. e politics of blame avoidance. J. Public Policy 6, 371–398 (1986).
46. Wiedenhofer, D. etal. Unequal household carbon footprints in China.
Nat. Clim. Change 7, 75–80 (2017).
47. De Mesquita, B. B., Smith, A., Morrow, J. D. & Siverson, R. M. e Logic of
Political Survival (MIT Press, 2005).
48. Burstein, P. e impact of public opinion on public policy: a review and an
agenda. Polit. Res. Q. 56, 29–40 (2003).
49. Anderson, B., Böhmelt, T. & Ward, H. Public opinion and environmental
policy output: a cross-national analysis of energy policies in Europe. Environ.
Res. Lett. 12, 114011 (2017).
Acknowledgements
The research for this article was funded by the ERC Advanced Grant ‘Sources of
Legitimacy in Global Environmental Governance’ (grant number 295456) and supported
by ETH Zürich. We are grateful for valuable comments by B. Anderson, G. Brückmann,
F. Egli, L. F. Beiser-McGrath, R. Hess, R. Huber, D. Kolcava, V. Koubi, S. Mohrenberg,
Q. Nguyen, D. Presperger, A. Rinscheid, L. Rudolph, A. Schrode and F. Quoss, as well
as the interviewed experts and anonymous reviewers. The ETH ethics commission
provided guidelines for the study procedures and approved the study.
Author contributions
L.P.F., M.W., Y.S. and T.B. developed the study concept and survey design. L.P.F. and Y.S.
conducted the interviews. L.P.F. conducted the analyses and interpreted and prepared the
results. The other authors supported interpretation of the results. L.P.F. wrote the paper
with input from the other authors.
Competing interests
The authors declare no competing interests.
Additional information
Extended data is available for this paper at https://doi.org/10.1038/s43016-020-0047-4.
Supplementary information is available for this paper at https://doi.org/10.1038/
s43016-020-0047-4.
Correspondence and requests for materials should be addressed to L.P.F.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
© The Author(s), under exclusive licence to Springer Nature Limited 2020
NATURE FOOD | VOL 1 | MARCH 2020 | 173–182 | www.nature.com/natfood
182

Articles
Nature Food Articles
Nature Food
Extended Data Fig. 1 | Example of experimental conjoint choice and rating task. We showed study participants four such pairs of random policy packages
and asked them to decide which proposal they preferred within each pair, both in a forced-choice question and on a seven-point rating scale. The different
policy design attributes were described to respondents prior to the four choice tasks.
NATURE FOOD | www.nature.com/natfood

1
nature research | reporting summary October 2018
Corresponding author(s): Lukas Paul Fesenfeld
Last updated by author(s): 18-11-2019
Reporting Summary
Nature Research wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency
in reporting. For further information on Nature Research policies, see Authors & Referees and the Editorial Policy Checklist.
Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.
n/a Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement
A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly
The statistical test(s) used AND whether they are one- or two-sided
Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A description of all covariates tested
A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons
A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient)
AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals)
For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted
Give P values as exact values whenever suitable.
For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings
For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes
Estimates of effect sizes (e.g. Cohen's d, Pearson's r), indicating how they were calculated
Our web collection on statistics for biologists contains articles on many of the points above.
Software and code
Policy information about availability of computer code
Data collection The
data
was
collected
in
collaboration
with
IPSOS.
Ipsos
pre-selected
respondents
from
their
panels
according
to
quotas
and
constructed
samples to match the national voting age population in the three countries. We did not use any own computer code during data
collection.
Data analysis We
used
the
open
source
software
R
package
"cjoint"
(version
2.1.0)
to
nonparametrically
estimate
Average
Marginal
Component
Effects.
For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors/reviewers.
We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information.
Data
Policy information about availability of data
All manuscripts must include a data availability statement. This statement should provide the following information, where applicable:
- Accession codes, unique identifiers, or web links for publicly available datasets
- A list of figures that have associated raw data
- A description of any restrictions on data availability
All
methods,
additional
references,
Nature
Research
reporting
summaries,
source
data,
statements
of
data
availability,
and
associated
accession
codes
will
be
made
publically available at Harvard Dataverse.

2
nature research | reporting summary October 2018
Field-specific reporting
Please select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection.
Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences
For a reference copy of the document with all sections, see nature.com/documents/nr-reporting-summary-flat.pdf
Life sciences study design
All studies must disclose on these points even when the disclosure is negative.
Sample size Describe
how
sample
size
was
determined,
detailing
any
statistical
methods
used
to
predetermine
sample
size
OR
if
no
sample-size
calculation
was performed, describe how sample sizes were chosen and provide a rationale for why these sample sizes are sufficient.
Data exclusions Describe
any
data
exclusions.
If
no
data
were
excluded
from
the
analyses,
state
so
OR
if
data
were
excluded,
describe
the
exclusions
and
the
rationale behind them, indicating whether exclusion criteria were pre-established.
Replication Describe
the
measures
taken
to
verify
the
reproducibility
of
the
experimental
findings.
If
all
attempts
at
replication
were
successful,
confirm
this
OR if there are any findings that were not replicated or cannot be reproduced, note this and describe why.
Randomization Describe
how
samples/organisms/participants
were
allocated
into
experimental
groups.
If
allocation
was
not
random,
describe
how
covariates
were controlled OR if this is not relevant to your study, explain why.
Blinding Describe
whether
the
investigators
were
blinded
to
group
allocation
during
data
collection
and/or
analysis.
If
blinding
was
not
possible,
describe why OR explain why blinding was not relevant to your study.
Behavioural & social sciences study design
All studies must disclose on these points even when the disclosure is negative.
Study description The
study
is
a
quantitative
survey-embedded
conjoint
experiment
with
citizens
across
three
countries
(China,
Germany,
United
States)
and
a total sample size of 4874 respondents. Our experiment resulted in a total (N) of 38,992 observations (i.e., 4,874 respondents x two policy
packages side-by-side x four choice pairs).
Research sample The
conjoint
experiment
was
embedded
in
an
internet-based
survey
conducted
in
February
2018
in
China,
Germany,
and
the
US.
To
increase the external validity of our results we sampled respondents from the population of 18+/eligible voters across the three countries.
The experiment drew on a non-probability, quota-based sample in the three countries provided by Ipsos. All experiments were internet-
based and participants were recruited via the online panels that Ipsos maintains in each country. While these panels are not probability-
based, they are non-convenience samples as Ipsos actively manages and refreshes them to target respondents that match census
statistics. For our survey, Ipsos pre-selected respondents from their panels according to quotas and constructed samples to match the
national voting age population in the three countries. More specifically, we used hard quotas in our sampling in an attempt to match
distribution by gender, age, and region according to each country’s latest census data (China in 2010, Germany in 2013, and the US in
2015; see SI-Tables 3a/b/c). Quotas for gender and age were combined to ensure that each age group was nationally representative in
terms of gender distribution. Additionally, we also employed soft quotas for education, income, rural-urban population, and occupation to
ensure that the samples were not extensively skewed towards certain socio-demographic groups.
Sampling strategy As
described
above,
the
samples
used
in
our
study
are
non-probability-based
and
non-convenience
samples
as
Ipsos
actively
manages
their panels and refreshes them to target respondents that match census statistics. Our experiment resulted in a total (N) of 38,992
observations (i.e., 4,874 respondents x two policy packages side-by-side x four choice pairs). Prior power estimations indicated that this
sample size provides enough statistical power to nonparametrically estimate the average marginal component effects.
Data collection The
data
was
collected
via
the
computer
or
laptop.
The
survey
experiments
were
administered
via
IPSOS,
and
the
researchers
had
no
direct contact with respondents. All treatment conditions were fully randomized.
Timing The
data
was
collected
between
the
15th
of
February
2018
and
the
8th
of
March
2018.
Data exclusions The
average
time
for
survey
completion
was
18
min
in
the
USA,
17min
in
Germany,
and
14min
in
China.
We
dropped
a
total
of
79
respondents who completed the survey in six minutes or less, which we assume is the minimum time to complete the survey in an
attentive manner. The final sample includes 4874 respondents from the three countries (1624 each from Germany and the USA, 1626
from China).
Non-participation The
exact
response
rate
is
unknown
to
the
researchers
given
that
IPSOS
managed
the
data
collection
process.
Unfortunately,
IPSOS
did
not
provide us with the full number of invited individuals to calculate the response rate.
Randomization All
treatment
attribute
values
were
randomly
assigned
across
all
policy
packages
and
participants.

3
nature research | reporting summary October 2018
Ecological, evolutionary & environmental sciences study design
All studies must disclose on these points even when the disclosure is negative.
Study description Briefly
describe
the
study.
For
quantitative
data
include
treatment
factors
and
interactions,
design
structure
(e.g.
factorial,
nested,
hierarchical), nature and number of experimental units and replicates.
Research sample Describe
the
research
sample
(e.g.
a
group
of
tagged
Passer
domesticus,
all
Stenocereus
thurberi
within
Organ
Pipe
Cactus
National
Monument), and provide a rationale for the sample choice. When relevant, describe the organism taxa, source, sex, age range and
any manipulations. State what population the sample is meant to represent when applicable. For studies involving existing datasets,
describe the data and its source.
Sampling strategy Note
the
sampling
procedure.
Describe
the
statistical
methods
that
were
used
to
predetermine
sample
size
OR
if
no
sample-size
calculation was performed, describe how sample sizes were chosen and provide a rationale for why these sample sizes are sufficient.
Data collection Describe
the
data
collection
procedure,
including
who
recorded
the
data
and
how.
Timing and spatial scale Indicate
the
start
and
stop
dates
of
data
collection,
noting
the
frequency
and
periodicity
of
sampling
and
providing
a
rationale
for
these choices. If there is a gap between collection periods, state the dates for each sample cohort. Specify the spatial scale from which
the data are taken
Data exclusions If
no
data
were
excluded
from
the
analyses,
state
so
OR
if
data
were
excluded,
describe
the
exclusions
and
the
rationale
behind
them,
indicating whether exclusion criteria were pre-established.
Reproducibility Describe
the
measures
taken
to
verify
the
reproducibility
of
experimental
findings.
For
each
experiment,
note
whether
any
attempts
to
repeat the experiment failed OR state that all attempts to repeat the experiment were successful.
Randomization Describe
how
samples/organisms/participants
were
allocated
into
groups.
If
allocation
was
not
random,
describe
how
covariates
were
controlled. If this is not relevant to your study, explain why.
Blinding Describe
the
extent
of
blinding
used
during
data
acquisition
and
analysis.
If
blinding
was
not
possible,
describe
why
OR
explain
why
blinding was not relevant to your study.
Did the study involve field work? Yes No
Field work, collection and transport
Field conditions Describe
the
study
conditions
for
field
work,
providing
relevant
parameters
(e.g.
temperature,
rainfall).
Location State
the
location
of
the
sampling
or
experiment,
providing
relevant
parameters
(e.g.
latitude
and
longitude,
elevation,
water
depth).
Access and import/export Describe
the
efforts
you
have
made
to
access
habitats
and
to
collect
and
import/export
your
samples
in
a
responsible
manner
and
in compliance with local, national and international laws, noting any permits that were obtained (give the name of the issuing
authority, the date of issue, and any identifying information).
Disturbance Describe
any
disturbance
caused
by
the
study
and
how
it
was
minimized.
Reporting for specific materials, systems and methods
We require information from authors about some types of materials, experimental systems and methods used in many studies. Here, indicate whether each material,
system or method listed is relevant to your study. If you are not sure if a list item applies to your research, read the appropriate section before selecting a response.
Materials & experimental systems
n/a Involved in the study
Antibodies
Eukaryotic cell lines
Palaeontology
Animals and other organisms
Human research participants
Clinical data
Methods
n/a Involved in the study
ChIP-seq
Flow cytometry
MRI-based neuroimaging

4
nature research | reporting summary October 2018
Antibodies
Antibodies used Describe
all
antibodies
used
in
the
study;
as
applicable,
provide
supplier
name,
catalog
number,
clone
name,
and
lot
number.
Validation Describe
the
validation
of
each
primary
antibody
for
the
species
and
application,
noting
any
validation
statements
on
the
manufacturer’s website, relevant citations, antibody profiles in online databases, or data provided in the manuscript.
Eukaryotic cell lines
Policy information about cell lines
Cell line source(s) State
the
source
of
each
cell
line
used.
Authentication Describe
the
authentication
procedures
for
each
cell
line
used
OR
declare
that
none
of
the
cell
lines
used
were
authenticated.
Mycoplasma contamination Confirm
that
all
cell
lines
tested
negative
for
mycoplasma
contamination
OR
describe
the
results
of
the
testing
for
mycoplasma contamination OR declare that the cell lines were not tested for mycoplasma contamination.
Commonly misidentified lines
(See ICLAC register)
Name
any
commonly
misidentified
cell
lines
used
in
the
study
and
provide
a
rationale
for
their
use.
Palaeontology
Specimen provenance Provide
provenance
information
for
specimens
and
describe
permits
that
were
obtained
for
the
work
(including
the
name
of
the
issuing authority, the date of issue, and any identifying information).
Specimen deposition Indicate
where
the
specimens
have
been
deposited
to
permit
free
access
by
other
researchers.
Dating methods If
new
dates
are
provided,
describe
how
they
were
obtained
(e.g.
collection,
storage,
sample
pretreatment
and
measurement),
where they were obtained (i.e. lab name), the calibration program and the protocol for quality assurance OR state that no new
dates are provided.
Tick this box to confirm that the raw and calibrated dates are available in the paper or in Supplementary Information.
Animals and other organisms
Policy information about studies involving animals; ARRIVE guidelines recommended for reporting animal research
Laboratory animals For
laboratory
animals,
report
species,
strain,
sex
and
age
OR
state
that
the
study
did
not
involve
laboratory
animals.
Wild animals Provide
details
on
animals
observed
in
or
captured
in
the
field;
report
species,
sex
and
age
where
possible.
Describe
how
animals
were caught and transported and what happened to captive animals after the study (if killed, explain why and describe method;
if
released, say where and when) OR state that the study did not involve wild animals.
Field-collected samples For
laboratory
work
with
field-collected
samples,
describe
all
relevant
parameters
such
as
housing,
maintenance,
temperature,
photoperiod and end-of-experiment protocol OR state that the study did not involve samples collected from the field.
Ethics oversight Identify
the
organization(s)
that
approved
or
provided
guidance
on
the
study
protocol,
OR
state
that
no
ethical
approval
or
guidance was required and explain why not.
Note that full information on the approval of the study protocol must also be provided in the manuscript.
Human research participants
Policy information about studies involving human research participants
Population characteristics See
above
Recruitment Participants
were
recruited
via
the
actively
managed
online
panels
of
IPSOS.
IPSOS
invites
respondents
in
line
with
the
quota
criteria and employs multiple standards to minimize self-selection biases. IPSOS is certified under ESOMAR 28.
Ethics oversight ETH
Zurich
Ethics
Commission
Note that full information on the approval of the study protocol must also be provided in the manuscript.

5
nature research | reporting summary October 2018
Clinical data
Policy information about clinical studies
All manuscripts should comply with the ICMJE guidelines for publication of clinical research and a completed CONSORT checklist must be included with all submissions.
Clinical trial registration Provide
the
trial
registration
number
from
ClinicalTrials.gov
or
an
equivalent
agency.
Study protocol Note
where
the
full
trial
protocol
can
be
accessed
OR
if
not
available,
explain
why.
Data collection Describe
the
settings
and
locales
of
data
collection,
noting
the
time
periods
of
recruitment
and
data
collection.
Outcomes Describe
how
you
pre-defined
primary
and
secondary
outcome
measures
and
how
you
assessed
these
measures.
ChIP-seq
Data deposition
Confirm that both raw and final processed data have been deposited in a public database such as GEO.
Confirm that you have deposited or provided access to graph files (e.g. BED files) for the called peaks.
Data access links
May remain private before publication.
For
"Initial
submission"
or
"Revised
version"
documents,
provide
reviewer
access
links.
For
your
"Final
submission"
document,
provide a link to the deposited data.
Files in database submission Provide
a
list
of
all
files
available
in
the
database
submission.
Genome browser session
(e.g. UCSC)
Provide
a
link
to
an
anonymized
genome
browser
session
for
"Initial
submission"
and
"Revised
version"
documents
only,
to
enable peer review. Write "no longer applicable" for "Final submission" documents.
Methodology
Replicates Describe
the
experimental
replicates,
specifying
number,
type
and
replicate
agreement.
Sequencing depth Describe
the
sequencing
depth
for
each
experiment,
providing
the
total
number
of
reads,
uniquely
mapped
reads,
length
of
reads and whether they were paired- or single-end.
Antibodies Describe
the
antibodies
used
for
the
ChIP-seq
experiments;
as
applicable,
provide
supplier
name,
catalog
number,
clone
name, and lot number.
Peak calling parameters Specify
the
command
line
program
and
parameters
used
for
read
mapping
and
peak
calling,
including
the
ChIP,
control
and
index files used.
Data quality Describe
the
methods
used
to
ensure
data
quality
in
full
detail,
including
how
many
peaks
are
at
FDR
5%
and
above
5-fold
enrichment.
Software Describe
the
software
used
to
collect
and
analyze
the
ChIP-seq
data.
For
custom
code
that
has
been
deposited
into
a
community repository, provide accession details.
Flow Cytometry
Plots
Confirm that:
The axis labels state the marker and fluorochrome used (e.g. CD4-FITC).
The axis scales are clearly visible. Include numbers along axes only for bottom left plot of group (a 'group' is an analysis of identical markers).
All plots are contour plots with outliers or pseudocolor plots.
A numerical value for number of cells or percentage (with statistics) is provided.
Methodology
Sample preparation Describe
the
sample
preparation,
detailing
the
biological
source
of
the
cells
and
any
tissue
processing
steps
used.
Instrument Identify
the
instrument
used
for
data
collection,
specifying
make
and
model
number.
Software Describe
the
software
used
to
collect
and
analyze
the
flow
cytometry
data.
For
custom
code
that
has
been
deposited
into
a
community repository, provide accession details.

6
nature research | reporting summary October 2018
Cell population abundance Describe
the
abundance
of
the
relevant
cell
populations
within
post-sort
fractions,
providing
details
on
the
purity
of
the
samples
and how it was determined.
Gating strategy Describe
the
gating
strategy
used
for
all
relevant
experiments,
specifying
the
preliminary
FSC/SSC
gates
of
the
starting
cell
population, indicating where boundaries between "positive" and "negative" staining cell populations are defined.
Tick this box to confirm that a figure exemplifying the gating strategy is provided in the Supplementary Information.
Magnetic resonance imaging
Experimental design
Design type Indicate
task
or
resting
state;
event-related
or
block
design.
Design specifications Specify
the
number
of
blocks,
trials
or
experimental
units
per
session
and/or
subject,
and
specify
the
length
of
each
trial
or block (if trials are blocked) and interval between trials.
Behavioral performance measures State
number
and/or
type
of
variables
recorded
(e.g.
correct
button
press,
response
time)
and
what
statistics
were
used
to establish that the subjects were performing the task as expected (e.g. mean, range, and/or standard deviation across
subjects).
Acquisition
Imaging type(s) Specify:
functional,
structural,
diffusion,
perfusion.
Field strength Specify
in
Tesla
Sequence & imaging parameters Specify
the
pulse
sequence
type
(gradient
echo,
spin
echo,
etc.),
imaging
type
(EPI,
spiral,
etc.),
field
of
view,
matrix
size,
slice thickness, orientation and TE/TR/flip angle.
Area of acquisition State
whether
a
whole
brain
scan
was
used
OR
define
the
area
of
acquisition,
describing
how
the
region
was
determined.
Diffusion MRI Used Not used
Preprocessing
Preprocessing software Provide
detail
on
software
version
and
revision
number
and
on
specific
parameters
(model/functions,
brain
extraction,
segmentation, smoothing kernel size, etc.).
Normalization If
data
were
normalized/standardized,
describe
the
approach(es):
specify
linear
or
non-linear
and
define
image
types
used for transformation OR indicate that data were not normalized and explain rationale for lack of normalization.
Normalization template Describe
the
template
used
for
normalization/transformation,
specifying
subject
space
or
group
standardized
space
(e.g.
original Talairach, MNI305, ICBM152) OR indicate that the data were not normalized.
Noise and artifact removal Describe
your
procedure(s)
for
artifact
and
structured
noise
removal,
specifying
motion
parameters,
tissue
signals
and
physiological signals (heart rate, respiration).
Volume censoring Define
your
software
and/or
method
and
criteria
for
volume
censoring,
and
state
the
extent
of
such
censoring.
Statistical modeling & inference
Model type and settings Specify
type
(mass
univariate,
multivariate,
RSA,
predictive,
etc.)
and
describe
essential
details
of
the
model
at
the
first
and second levels (e.g. fixed, random or mixed effects; drift or auto-correlation).
Effect(s) tested Define
precise
effect
in
terms
of
the
task
or
stimulus
conditions
instead
of
psychological
concepts
and
indicate
whether
ANOVA or factorial designs were used.
Specify type of analysis: Whole brain ROI-based Both
Statistic type for inference
(See Eklund et al. 2016)
Specify
voxel-wise
or
cluster-wise
and
report
all
relevant
parameters
for
cluster-wise
methods.
Correction Describe
the
type
of
correction
and
how
it
is
obtained
for
multiple
comparisons
(e.g.
FWE,
FDR,
permutation
or
Monte
Carlo).

7
nature research | reporting summary October 2018
Models & analysis
n/a Involved in the study
Functional and/or effective connectivity
Graph analysis
Multivariate modeling or predictive analysis
Functional and/or effective connectivity Report
the
measures
of
dependence
used
and
the
model
details
(e.g.
Pearson
correlation,
partial
correlation, mutual information).
Graph analysis Report
the
dependent
variable
and
connectivity
measure,
specifying
weighted
graph
or
binarized
graph,
subject- or group-level, and the global and/or node summaries used (e.g. clustering coefficient, efficiency,
etc.).
Multivariate modeling and predictive analysis Specify
independent
variables,
features
extraction
and
dimension
reduction,
model,
training
and
evaluation
metrics.