Evidence from 33 countries challenges the assumption of unlimited wants

Abstract

Humans have unlimited wants. This foundational economic principle and widely accepted assumption about human nature poses considerable challenges to addressing sustainability because pursuing wealth and economic growth to meet unlimited wants increases resource use and pollution. Here we show evidence that this principle is not universal, and actually applies only to a minority of people. Across 42 community samples (N = 7,860) from 33 countries spanning 6 continents, we examined how much money people wanted in their absolutely ideal life. In 86% of countries the majority of people thought they would achieve their absolutely ideal lives with US10millionorless,andinsomecountriesaslittleasUS10 million or less, and in some countries as little as US1 million or less. However, a substantial minority (8–39% across countries) wanted as much money as they could obtain, indicating unlimited wants. Limited and unlimited wealth ideals were not related to country differences in economic development, but those with unlimited wants tended to be younger, city-dwelling people who valued power, success and independence, and lived in countries with a greater collective focus and acceptance of power differences. The results suggest that transformative approaches relying on limiting wealth and growth to achieve sustainability may be more consistent with human ideals and aspirations than commonly believed. If people had truly unlimited wants as assumed in economics, pursuing sustainability would be a real challenge. This study provides evidence that most people’s wants are moderate rather than unlimited, suggesting that policies limiting wealth in pursuit of sustainability could be acceptable to many.

Full text

1
Evidence from 33 countries challenges the assumption of unlimited wants
Bain & Bongiorno (2022)
in Nature Sustainability
Supplementary Information
Table of contents
S1. Example statements about unlimited wants in economics textbooks
2
S2. Materials and Methods
3
S2.1. Ethical approvals
3
S2.2. Data and materials availability
3
S2.3. Sampling overview and exclusion criteria
3
S2.4. Sample demographics and country information
4
S2.5. Measures
9
S3. Ideal life lottery responses by country (Supplementary Text)
13
S3.1. Study 1
13
S3.2. Study 2
14
S4. Country level associations and meta-analyses (Supplementary Text)
16
S4.1. Explanation of analytical approach
16
S4.2. Country level associations
17
S4.3. Meta-analyses
18
S4.4. Analysis of text responses for main use of wealth (Study 1)
22
S5. Explaining variation between those with limited wants (Supplementary Text)
27
S6. Supplementary Information references
31

2
S1. Example statements about unlimited wants in economics textbooks
Below are quotes from introductory textbooks on Economics or Microeconomics from major
publishers describing unlimited wants as a basic human characteristic.
Source
Quote
Acemoglu, Laibson, & List. Microeconomics
(Global Edition). (Pearson Education
Limited, 2019).
Scarcity exists because people have unlimited wants in
a world of limited resources. (p. 46)
Anderton (2000) Economics (3rd Ed).
(Pearson Education Limited, 2000).
People’s needs are finite. However, no one would
choose to live at the level of basic human needs if
they could enjoy a higher standard of living. This is
because human WANTS are unlimited. It doesn’t
matter whether the person is a peasant in China, a
mystic in India, a manager in the UK or the richest
individual in the world, there is always something
which he or she wants more of…. Resources are scarce
but wants are infinite. It is this which gives rise to The
BASIC ECONOMIC PROBLEM… (p. 1)
McConnell, Brue, & Flynn. Microeconomics:
Principles, problems, and policies (21st Ed).
(McGraw Hill, 2018).
For better or worse, most people have virtually
unlimited wants (p. 7)
Parkin. Economics (Global Edition).
(Pearson Education Limited, 2018).
Wants are the unlimited desires or wishes that people
have for goods and services.… Scarcity guarantees
that many— perhaps most— of our wants will never
be satisfied. (p. 97)
Sloman, Garret, & Guest. Economics
(Updated 10th Edition). (Pearson Education
Limited, 2020).
Ask people if they would like more money, and the
vast majority would answer ‘Yes’. But they don’t want
more money for its own sake. Rather they want to be
able to buy more goods and services, either today or
in the future…As countries get richer, human wants
may change but they don’t disappear. Wants are
virtually unlimited…. So this is the fundamental
economic problem: human wants are virtually
unlimited, whereas the resources available to meet
those wants are limited. (p. 7)
Table S1.
Example statements about unlimited wants in economics textbooks.

3
S2. Materials and Methods
S2.1 Ethical approvals
Study 1: Queensland University of Technology (Approval Number 1600000223).
Study 2: University of Queensland (Approval Number 2018001124).
S2.2 Data and materials availability:
Study 1: https://osf.io/25398/ (publicly available)
Study 2: https://osf.io/k3wdp/ (publicly available)
S2.3 Sampling overview and exclusion criteria
Study 1.
Participants were sourced through a commercial survey company (Survey Sampling
International), who were contracted to provide 220 adult community participants per country
(with an aim to obtain usable samples of 200 allowing for a 10% exclusion rate). These 12
countries were selected to span regions and to include both economically developed and
developing countries (sometimes known as BRICS: Brazil, Russia, India, China, South Africa),
up to the limit of the project budget. Data were collected in March 2018.
The majority of measures in the survey were focused on people’s understanding of sustainable
development (published in 1). The ideal life lottery measure was presented near the end of the
survey after measures of sustainable development goals, values, societal change, and future
orientation.
Exclusions. Exclusions were based on responses to the longer survey. These were:
(i) Missing data on the ratings of Sustainable Development Goals (SDGs) on sustainability
domains;
(ii) Pattern responding, evidenced by no variation in responses for all 17 SDGs when rating
correspondence to sustainability domains (environmental, social, or economic); and
(iii) Non-citizens who had lived in a country less than 5 years.
The first two exclusion criteria were the same as those used in Bain et al., 2019 1. The third
criterion was added to make data from Study 1 more directly comparable with Study 2 (the
rationale for this exclusion is explained in the description for Study 2 below).
Study 2
Participants were sourced through a commercial survey company (Dynata), who were contracted
to provide 200 adult community participants representative on gender and age for each country.
Including thirty countries allows for stronger tests of cross-country differences and was the limit
of the project budget. The specific countries were chosen to provide a good distribution across
geographical regions, levels of economic development, and country-level innovativeness as
measured by the 2019 Global Innovation Index 2. Data were collected in May-June 2020.

4
The majority of measures in the survey were focused on people’s reactions to innovations. The
ideal life lottery measure was presented near the end of the survey after rating these innovations,
values, societal change beliefs, and self-reported innovativeness.
Countries in this study varied widely in citizenship laws and restrictions. We found that in
United Arab Emirates and Saudi Arabia more than 50% of the sample were not citizens.
However, we also measured how many years they had lived in the country and identified that
many non-citizens were very long-term residents, and thus were reasonably likely to be
embedded within the country and culture.
Rather than excluding the majority of participants from these countries, we decided to include
non-citizens who were residents for a minimum of 5 years (“long-term” residents). After
applying this criterion, the average years of residence for included non-citizens in United Arab
Emirates was 14.9 (SD = 8.5), and in Saudi Arabia was 16.6 (SD = 10.7). We then applied this
criterion consistently across all countries in both studies, and Tables S1 and S2 show how many
non-citizens were included in each country using this criterion.
Exclusions. Exclusions were based on responses to the longer survey.
(i) Pattern-responding. This was evidenced by no variation in ratings of the societal change
or self-reported innovativeness measures;
(ii) Attention check failure. An attention check item was included at the end of the values
measure: “To confirm you are a human being, please select not at all like me for this
item.” Those who did not select this option were excluded; and
(iii) Non-citizens who had lived in a country less than 5 years.
S2.4 Sample demographics and country information
Sample sizes and basic demographics before and after exclusions are shown in Table S2 (Study
1) and Table S3 (Study 2). These tables also show country-level data used in cross-country
comparisons.
The rate of exclusions was higher in Study 1 (20.9%) than in Study 2 (6.5%), attributable to a
higher level of pattern-responding. Where both studies obtained samples from the same country,
the average age was always higher in Study 2 suggesting that the directive for representative
sampling across age for Study 2 was successful in obtaining more older adults.
Sources of country-level indicators are provided below:
Development indicators:
HDI: Human Development Index; United Nations Development Programme (2019 values)3
GDP (PPP): GDP per capita adjusted for purchasing power parity, expressed in international
dollars (World Bank 2019 figures:
https://data.worldbank.org/indicator/NY.GDP.PCAP.PP.CD);
GINI inequality index (World Bank https://data.worldbank.org/indicator/SI.POV.GINI). We
note that GINI data was not available for some countries.

5
Lottery experience:
These measures were a later addition in response to a reviewer’s comment that cultural
experience with lotteries may influence lottery choice.
We used figures from Garrett 4 to identify:
Lottery use: The countries that operated lotteries (89 countries, of which 25 countries were in
our studies). The 8 countries not identified as using lotteries were Nicaragua, India, Indonesia,
Vietnam, Ukraine, Uganda, UAE, Saudi Arabia.
%GDP spent on lotteries: For the 25 countries where lotteries were reported, lottery sales as a
percentage of country GDP. This ranged from .004% (Russia) to 2.478% (Singapore).
Cultural indicators:
Cultural dimensions and scores were based on research by Hofstede and colleagues 5, where full
descriptions are provided. However, we also provide brief descriptions below derived from this
source. Long-term Orientation was a later addition to Hofstede’s model, with few countries
having scores on this dimension in Hofstede’s original research. Therefore, we used scores from
a more extensive investigation of this dimension using the World Values Survey 6. However, it
can be seen in Tables S1 and S2 that cultural dimension scores are still not available for all
countries in these studies.
Individualism(-Collectivism): Where individuals are focused on outcomes for themselves and
their families (Individualism), versus a focus on the outcomes for the wider community
(Collectivism). Sometimes describe as a difference in focus between “I” and “We”.
Power Distance: The acceptance of power differentials in society. In high Power Distance
countries there is acceptance that some people rightly have authority over others. In low Power
Distance countries there is a focus on equality between people.
“Masculine-Feminine”: A preference for achievement and assertiveness (Masculine) versus a
preference for cooperation, modesty, and caring for others (Feminine). We place this term in
quotation marks as objections have been raised to labeling cultures as “masculine” (or
“feminine”) based on gender stereotypes relating to achievement and power that are
increasingly outdated 7,8. However, we appreciate that this is the commonly used term for this
dimension so retained it to avoid confusion.
Uncertainty Avoidance: A focus on achieving certainty and predictability, with strongly
codified behavior, versus more flexible, context-dependent, and pragmatic expectations for
behavior.
Long-term orientation: A focus on preparing for the challenges of the future through education,
self-control, and self-reliance, versus preserving the present and the past with a focus on
tradition and service to others.
Table S2 shows that the samples showed a good spread across demographic, development, and
cultural dimensions. Table S3 shows greater regional variation, including countries in Africa and
the Middle East, which are regions that have often been excluded in cross-cultural psychology

6
research. The countries also showed a good spread across levels of human/economic
development and on cultural dimensions. We note that the ratio of females is much lower in the
UAE and Saudi Arabia, but this reflects a true difference in gender ratio in these populations.

7
Sample size (n)
Demographics
Development indicators
Cultural dimensions
Country
Before
exclusions
After
exclusions
Non-
citizens
included
Age #
M (SD)
Gender
(% female)
HDI
GDP
(PPP)
GINI
Individu
alism
Power-
Distance
“Masculine-
Feminine”
Uncertainty
avoidance
Long-term
orientation
North America
USA
220
181
0
37.9 (13.5)
50
.93
65298
41.4
91
40
62
46
29
South America
Brazil
222
146
1
36.5 (11.2)
47
.77
15300
53.9
38
69
49
76
65
Argentina
222
178
2
37.6 (12.8)
46
.85
23040
41.4
46
49
56
86
-
Asia
China
220
188
0
32.7 (8.4)
49
.76
16830
38.5
20
80
66
40
118
India
221
186
2
32.5 (10.8)
49
.65
6997
37.8
48
77
56
40
61
South Korea
220
201
0
38.5 (10.9)
49
.92
43143
31.6
18
60
39
85
75
Europe
Russia
216
177
0
39.2 (11.5)
47
.82
29181
37.5
-
-
-
-
-
Sweden
221
159
2
39.1 (14.2)
52
.95
55820
28.8
71
31
5
29
33
United Kingdom
220
168
7
42.6 (15.4)
48
.93
48698
34.8
89
35
66
35
25
France
231
172
1
39.9 (14.7)
54
.90
49435
31.6
71
68
43
86
-
Africa
South Africa
243
190
8
36.2 (12.1)
53
.71
13034
63.0
65
49
63
49
-
Oceania
Australia
215
166
7
45.7 (17.5)
49
.94
53469
34.4
90
36
61
51
31
# Some people mistakenly wrote their year of birth (e.g., 1973) rather than age, and these were converted to ages using a birthdate of January 1 of
the corresponding year. We deleted the age for one outlier case with an age well over 100 that we attributed to a typographic error.
Table S2:
Study 1 sample descriptions and country characteristics.

8
Sample size (n)
Demographics
Development
indicators
Cultural dimensions
Country
Before
exclusions
After
exclusions
Non-
citizens
included
Age
M (SD)
Gender
(% female)
HDI
GDP (PPP)
GINI
Individu
alism
Power-
Distance
Masculine-
Feminine”
Uncertainty
avoidance
Long-term
orientation^
North America
USA
202
175
11
47.8 (16.3)
50
.93
65281
41.4
91
40
62
46
39
Canada
205
188
6
45.9 (15.2)
52
.93
51342
33.8
-
-
-
-
45
Mexico
206
199
2
40.0 (15.1)
51
.78
20411
45.4
30
81
69
82
40
Nicaragua
201
195
0
34.5 (11.9)
51
.66
5631
46.2
South America
Brazil
207
204
1
39.7 (14.6)
52
.77
15259
53.9
38
69
49
76
-
Chile
207
186
3
41.7 (15.3)
50
.85
25155
44.4
23
63
28
86
47
Colombia
207
176
6
39.8 (15.0)
51
.77
15644
50.4
13
63
64
80
-
Asia
China
205
200
0
40.3 (14.5)
49
.76
16785
38.5
20
80
66
40
91
India
205
194
4
39.4 (15.6)
48
.65
7034
37.8
48
77
56
40
66
Indonesia
203
199
0
38.3 (14.1)
50
.72
12302
39.0
14
78
46
48
67
Japan
204
188
0
48.6 (16.0)
52
.92
43236
32.9
46
54
95
92
99
Philippines
204
195
4
37.4 (14.3)
50
.72
9277
44.4
32
94
64
44
42
Singapore
204
188
15
43.9 (14.7)
49
.94
101376
-
20
74
48
8
75
Vietnam
207
193
0
36.9 (12.7)
51
.70
8374
35.7
-
-
-
-
75
Europe
Russia
202
190
0
44.8 (15.8)
54
.82
29181
37.5
-
-
-
-
-
Sweden
204
195
7
47.7 (17.4)
49
.95
55815
28.8
71
31
5
29
37
United Kingdom
205
187
5
48.2 (16.7)
52
.93
48710
34.8
89
35
66
35
-
Hungary
205
196
1
46.4 (16.1)
54
.85
33979
30.6
55
46
88
82
-
Spain
208
200
7
47.5 (15.8)
52
.90
42214
34.7
51
57
42
86
46
Greece
204
202
2
43.0 (15.5)
50
.89
31399
34.4
35
60
57
112
-
Ukraine
205
194
6
32.0 (9.7)
55
.78
13341
26.1
-
-
-
-
-

9
Sample size (n)
Demographics
Development
indicators
Cultural dimensions
Country
Before
exclusions
After
exclusions
Non-
citizens
included
Age
M (SD)
Gender
(% female)
HDI
GDP (PPP)
GINI
Individu
alism
Power-
Distance
Masculine-
Feminine”
Uncertainty
avoidance
Long-term
orientation^
Africa/Middle East
South Africa
203
192
6
37.9 (15.0)
52
.71
12999
63.0
65
49
63
49
48
Kenya
205
204
0
30.1 (8.6)
50
.60
4509
40.8
27
64
41
52
-
Uganda
202
198
5
27.5 (7.2)
48
.54
2271
42.8
-
-
-
-
33
Tunisia
204
198
1
31.0 (9.0)
51
.74
11201
32.8
-
-
-
-
-
Morocco
208
196
3
37.4 (12.6)
52
.69
7826
39.5
-
-
-
-
-
UAE
203
149
62
35.5 (10.7)
26
.89
69901
32.5
38
80
52
68
-
Saudi Arabia
210
186
85
32.2 (9.7)
36
.85
48909
-
38
80
52
68
37
Oceania
Australia
202
186
11
46.8 (16.1)
50
.94
53320
34.4
90
36
61
51
-
New Zealand
207
190
15
46.3 (16.1)
52
.93
43953
-
79
22
58
49
-
^ from World Values Survey as data is available from more countries than in Hofstede et al.’s research.
Table S3.
Study 2 sample descriptions and country characteristics.

10
S2.5 Measures
Translations
All surveys used back-translation, where the English language materials were translated by one
translator, and independently translated back to English by a second translator. Discrepancies
were discussed and resolved between translators to produce the final translated version.
Translations are available from the first author upon request.
In Study 1 the following translations were used:
Spanish-Latin (Argentina)
Portuguese-Brazilian (Brazil)
Simplified Chinese (China)
French (France)
Russian (Russia)
Korean (South Korea)
Swedish (Sweden)
In India and South Africa, which have multiple common languages, English versions of the
survey were used on the recommendation of the survey company, as their panels comprised of
people competent in English and experienced in completing English-language surveys.
In Study 2 the following translations were used:
Modern Standard Arabic (Morocco, Tunisia, United Arab Emirates, Saudi Arabia)
Spanish-Latin (Chile, Colombia, Mexico, Nicaragua)
Portuguese-Brazilian (Brazil)
Simplified Chinese (China)
Greek (Greece)
Hungarian (Hungary)
Indonesian (Indonesia)
Japanese (Japan)
Swahili (Kenya)
Russian (Russia)
Spanish-Castilian (Spain)
Ukrainian (Ukraine)
Vietnamese (Vietnam)
In Canada, India, Philippines, Singapore, South Africa, and Uganda, which have multiple
common languages, English versions of the survey were used on the recommendation of the
survey company as their panels comprised of people competent in English and experienced in
completing English-language surveys.

11
“Ideal life” lottery (both studies)
This measure is fully described in the main paper (Fig. 1).
US dollar amounts were converted to local currency in each country using an indicative
conversion rate at the time of survey development, rounded to the nearest large number (see
Tables S4 & S5).
We did not adjust lottery amounts for purchasing power parity across countries for two reasons.
First, at high levels of wealth country borders largely no longer apply, e.g., in 2020 a US Green
card can be obtained by investing less than 1 million US dollars in a US business, and with
effectively no restriction on where to live in one’s ideal life, purchasing power in their current
country is not a practical restriction. Second, the differences in magnitude between options far
exceed relative purchasing power parity, and would be highly unlikely to affect the rank-order of
the options selected.
The measure was presented on a single page, and we recorded the time participants viewed that
page as a broad indicator of the extent to which they attended to the question. In Study 1 the page
included an open-ended question about the most important change they would make in their life
with the money.
Values
The measurement of values was based on the model of value content and structure developed by
Schwartz 9, which has been used extensively across countries/cultures. The original and most
widely used version of the model proposes ten value domains (short descriptions below, derived
from 9), with each domain represented with a number of more specific values (e.g., the
Conformity domain is represented by values such as obedience and self-discipline). Due to
survey length constraints we used short measures of this value model.
Conformity: Upholding social expectations or norms, including restraining impulses and
actions likely to upset others.
Tradition: Commitment to societal customs and traditions, including those derived from
religion
Benevolence: Concern for the welfare of close others
Universalism: Concern for the welfare of all people and the environment
Self-direction: Independent thought and action, and freedom to follow one’s chosen path in life.
Stimulation: Variety and new experiences, focusing on excitement and novelty
Hedonism: Pleasure and gratification for oneself
Achievement: Demonstrating personal success according to prevailing social standards
Power: Dominance and control over people and resources, along with high social status
Security: Safety and harmony within relationships, family, and society as a whole

12
Study 1
We used the 10-item SSVS - Short Schwartz’s Value Survey 10. This measure asks people to
rate the 10 major value domains in Schwartz’s model as life guiding principles, on a 9-point
scale labeled 0 “Opposed to my values”, 1 “Not important”, 4 “Important”, and 8 “Of
supreme importance”. Each value domain is described with a domain label, along with
specific values that represent that domain, e.g., Achievement (success, capability, ambition,
influence on people and events), Benevolence (helpfulness, honesty, forgiveness, loyalty,
responsibility), or Power (social power, authority, wealth).
Study 2
One issue with the value measure in Study 1 is that the Power value domain explicitly uses
wealth as example value, which may inflate correlations between Power and the lottery
measure of desired wealth. For Study 2 we used the TIVI - Ten Item Value Inventory 11 that
does not have this issue. This measure is based on the Portrait Values Questionnaire 12, a
different approach to measuring Schwartz’s value model to improve its applicability across
cultures, especially in less developed nations. Instead of asking about values directly,
participants read a description of a person that emphasizes a value, and rate how similar that
person is to them. In the TIVI one description is provided per value domain, e.g., for
Achievement – “Being very successful is important to this person. This person likes to
impress other people”, and for Power “It is important to this person to be in charge and tell
others what to do. This person wants people to do what they say”. These descriptions are
rated on a 6-point scale labelled 1 “Not at all like me”, 2 “Not like me” 3 “A little like me”, 4
“Somewhat like me”, 5 “Like me”, and 6 “Very much like me”.
Demographic measures
Study 1
Age: People wrote their age (whole years only) into a text box.
Gender: People selected from four options: Female; Male; I do not identify with either gender, I
identify as [text box for response]; I prefer not to say. From the entire sample 15 participants did
not identify as either gender, and 20 preferred not to say. As sample sizes for these latter options
were not large enough to treat as categories for analysis, we focused just on comparing females
and males.
Citizenship: Are you a citizen of [country]. Yes / No.
Years in country: How many years of your life have you lived in [country]? [text box for
response, whole years only].
Liberal-Conservative: “In political matters, people sometimes talk about "liberals" and
"conservatives." How would you place your views on this scale, generally speaking?” They
selected from seven responses: Very liberal; Liberal; Slightly liberal; Moderate/Middle of the
road; Slightly conservative; Moderately conservative; Very conservative.
Left-Right: In political matters, people sometimes talk about "left" and "right." How would you
place your views on this scale, generally speaking? They selected an option on a 7-point bipolar
scale bounded by “Left” and “Right”.

13
Relative income: Based on recent statistics, the average annual income for households in
[country] is approximately [amount in local currency from national statistics]. Considering all
the sources of income of members of your household, is your income:” They selected from 7
responses: Very much above the average; Above the average; A little above the average; About
average; A little below the average; Below the average; Very much below the average.
Rural-urban: “Do you live in a more urban/city area, or a more rural/country area?”. This was
rated on a 5-point scale from Very rural/country to Very urban/city.
Other demographic information collected including religion/religiosity, and whether they were a
university student.
Study 2
Age, Citizenship, Years in country, Liberal-Conservative, Left-Right, and Rural-Urban used
identical measures to Study 1, although we were not permitted to include Liberal-Conservative
or Left-Right items in the surveys for Tunisia and Morocco. Relative income from Study 1 was
replaced with Socio-economic status. Religion/religiosity and student status was not measured.
Below we describe the changes and new measures.
Gender: This measure was the same as Study 1 except the “I prefer not to say” option was not
presented.
Socioeconomic status: We used the “ladder” measure of subjective socio-economic status 13.
They were presented an image of a ladder with 10 rungs, accompanied by this description:
“Think of the ladder below as representing where people stand in your country. At the top of the
ladder are the people who are the best off - those who have the most money, the most education
and the most respected jobs. Those at the bottom of the ladder are the worst off - those who have
the least money, least education and the least respected jobs or no job. The higher up you are on
this ladder, the closer you are to the people at the very top; the lower you are on this ladder, the
closer you are to those at the very bottom. Where would you place yourself on this ladder?” They
then selected a number corresponding to a rung on the ladder, from 1 (the lowest rung) to 10 (the
highest rung).
Years of education: About how many years of education have you completed, whether full-time
or part-time? Please report these in full-time equivalents and include compulsory years of
schooling. Responses were made in a text box using whole numbers only.
Additionally, as data was collected in 2020 during the worldwide Covid-19 pandemic, we
included two items to assess the possible impact of the pandemic on responses.
Covid experience: Participants rated the statement: “Compared to the experiences of other people
in your country, have your experiences of the COVID-19 pandemic, or policies to restrict spread
of the virus, been:” using a 5 point scale labeled: A lot worse than average; A little worse than
average; About average; A little better than average; A lot better than average. These responses
were coded 1 to 5.
Covid wellbeing: Participants rated the statement: “How has the COVID-19 pandemic, or
policies to restrict spread of the virus, affected your wellbeing?” using a 7-point scale labeled:
Very negatively; Moderately negatively; Slightly negatively; No meaningful difference; Slightly
positively; Moderately positively; Very positively. These responses were coded 1-7.

14
S3. Ideal life lottery responses by country (Supplementary Text)
The information displayed in Fig. 1 of the article is shown in Tables S3 (Study 1) and S4 (Study
2). For each country the mode (most common response) is boldfaced, and the median (point
including the 50th percentile) is shaded.
While survey companies endeavored to achieve representative samples, sampling biases are still
inherent in internet-based surveys using company survey panels, such as requiring access to the
internet and a time/willingness to complete surveys for a small fee. So it is possible that the
proportion of Unlimiteds in these populations is underestimated. However, we note that there
was little evidence that lottery choice varied according to socio-economic status or income
within these samples (See Section S4.3 below). Moreover, identifying that many people were
Limiteds, even within somewhat restricted demographics, is evidence against the assumption that
unlimited wants is a truism about human nature.
S3.1 Study 1
Table S4 shows that the USA appears to be an outlier, with the only median/mode above 10
million and where the maximum lottery value was the most common response (indicating the
highest proportion of Unlimiteds). The most commonly chosen lottery across countries was 1
million, and the most common median was 10 million, indicating that the majority of participants
wanted 10 million or less in their absolutely ideal lives, including 46% of US participants.
Country
Currency
conversion
rate
10
thousand
100
thousand
1
million
10
million
100
million
1
billion
10
billion
100
billion
North America
USA
1
6
6
16
18
15
6
2
31.7
South America
Brazil
3
8
12
25
27
12
4
1
11.6
Argentina
20
4
20
33
19
6
2
1
14.6
Asia
China
5
8
11
30
25
15
3
2
8.1
India
50
5
14
33
17
9
2
1
19.4
South Korea
1000
7
6
24
27
8
3
1
22.4
Europe
Russia
50
10
6
36
19
9
4
1
14.7
Sweden
10
3
11
32
23
4
6
1
20.3
United Kingdom
1
3
10
26
24
13
5
1
19.0
France
1
4
12
27
23
12
8
2
13.4
Africa
South Africa
10
2
9
31
20
8
4
5
21.7
Oceania
Australia
1
2
7
22
35
14
3
1
16.9
Table S4.
Percentage of people who chose each lottery amount in the ideal life lottery measure across
countries (US Dollar approximate equivalence, currency conversion rate shown). For each
country the modal response is boldfaced and the median is shaded.

15
S3.2 Study 2
Country
Currency
conversion
10
thousand
100
thousand
1
million
10
million
100
million
1
billion
10
billion
100
billion
North America
USA
1
6
5
18
27
14
5
2
24.6
Canada
1.5
1
6
20
34
9
6
1
23.9
Mexico
25
10
8
12
26
12
7
2
24.6
Nicaragua
30
9
13
19
21
10
1
5
22.1
South America
Brazil
5
4
12
18
22
15
4
2
22.5
Chile
800
7
6
19
17
15
5
3
27.4
Colombia
4000
8
7
17
16
11
7
5
28.6
Asia
China
7
4
9
33
23
11
3
1
15.0
India
80
12
9
31
18
8
4
3
15.5
Indonesia
15000
16
12
6
12
8
6
2
39.2
Japan
100
18
6
21
18
8
3
2
25.0
Philippines
50
10
13
22
13
9
4
2
27.2
Singapore
1.5
6
3
20
22
7
4
4
32.4
Vietnam
20000
10
16
13
17
17
8
3
17.1
Europe
Russia
70
18
13
24
16
11
4
3
11.1
Sweden
10
5
10
37
16
8
5
2
16.4
United Kingdom
0.8
4
3
27
35
10
5
2
13.9
Hungary
300
5
14
17
19
18
4
1
20.9
Spain
0.9
6
7
13
25
13
8
3
24.0
Greece
0.9
2
8
22
26
9
5
2
24.3
Ukraine
30
8
15
21
18
9
4
3
21.6
Africa/Middle East
South Africa
20
4
9
30
18
6
3
5
26.3
Kenya
100
13
10
17
12
11
2
4
30.4
Uganda
4000
13
15
13
9
5
9
3
33.8
Tunisia
3
8
11
11
15
12
6
4
33.8
Morocco
10
17
12
10
12
8
6
3
33.2
UAE
4
9
11
13
21
15
7
2
20.8
Saudi Arabia
4
12
8
16
18
6
5
5
30.1
Oceania
Australia
1.5
5
5
26
33
11
3
0
16.1
New Zealand
1.5
2
4
17
42
11
4
2
18.9
Table S5.
Percentage of people who chose each lottery amount in the ideal life lottery measure across
countries (US Dollar approximate equivalence, currency conversion rate shown). For each
country the modal response is boldfaced and the median is shaded.

16
Table S5 shows that in the majority of countries in Study 2 most participants would be satisfied
in their absolutely ideal lives with 10 million or less (medians of 10 million or lower), similar to
Study 1. We note that those countries with higher medians (100 million) tended to be less
economically developed (e.g., Colombia, Indonesia, Uganda), but so were the countries with the
lowest medians of 1 million (Argentina, India, Russia), suggesting no systematic link between
economic development and lottery selection. We examine links between Unlimited/Limited
responses and country indicators more systematically below (Section S4.2).
Unlike Study 1, Table S5 shows that the most common modal response across countries was the
maximum (100 billion), observed in 15 of the 30 countries. The context of each study may have
contributed to the apparently higher incidence of Unlimited lottery selections in Study 2. In
Study 1 the broader survey context was sustainable development, which may have influenced
participants to think about living within constraints and adopting more moderate economic
ideals. In contrast, in Study 2 the broader survey context was about innovation and new products,
which may have influenced participants to think more about their capacity to consume new
consumer products and technologies which may have increased the economic ideals for some
people.

17
S4. Country-level associations and meta-analyses (Supplementary Text)
S4.1. Explanation of analytical approach
We conducted two types of analyses of the data, focusing on:
(i) examining factors between countries (development and culture) that could explain why
some countries had a higher proportion of Unlimiteds (who chose 100 billion) v Limiteds
(who chose lower amounts in the lottery); and
(ii) examining factors within countries (demographics and values) that could explain why some
individuals were more likely to be Unlimiteds or Limiteds, using meta-analyses to identify
whether these explanations were similar or different across countries.
We chose meta-analysis primarily because it is easier for people across diverse disciplines to
understand compared to alternatives (e.g., logistic multilevel modeling), especially when
examining simple relationships such as correlations (as is the case in this research). In meta-
analysis, the overall effects across countries is just the (weighted) average correlation. This
approach was used in previous research intended for a diverse scientific audience 14.
Meta-analysis allows for functionally similar tests to other techniques such as multilevel
modeling. For those versed in multilevel modeling, this includes:
-Random slopes at Level 1 (individual-level): We used “random effects” meta-analysis which
does not assume that there is a single “true” effect size being estimated, but rather that effect
sizes may differ across samples, distributed as a random variable.
-Cross-level interactions: We used meta-regression to test whether country-level variables
helped explain why correlations (effect sizes) differed across samples, often described as cross-
level interaction (or in this case country-level moderation). We used “method of moments”
meta-regression which represents a random-effects variable at level-2.
-Random intercepts at Level 2. To explain variation at Level 2 (country level), we used simple
zero-order correlations between country level information and the proportion of Unlimiteds for
each country.
Multi-level modeling is uniquely suited to complex models or when analyzing simultaneous/
independent contributions of a large number of predictors, but for our study we were interested
in simple relationships. Another advantage of multilevel modeling is the ability to easily
compare the amount of variance explained at individual and group levels, but that was not a goal
of this research.
Given our focus on unlimited wants, we aimed to understand what may account for differences
in the proportion of Unlimiteds across countries.
We focused on Study 2 because of its superior distribution of countries across regions. However,
we conducted robustness checks by adding data from Study 1, using different approaches to
incorporating this data to account for the overlap of countries between studies. These were:
Study 2+: Study 2 data plus the three unique countries from Study 1 (Argentina, France,
South Korea).
Both studies (averages): For the nine countries in both samples, we used the average
proportion of Unlimited across both studies as the best estimate of the population value for
that country.

18
Both studies (all samples): Every sample was used, leading to multiple samples from nine
countries. We believe this a less justified way to combine studies for country-level
correlations because for the repeated countries the data can only vary on one dimension
(development indicators and cultural dimensions will be identical for both samples),
influencing the degree of variation observed. However, we include it because we think it is
nonetheless likely to be requested, especially because we treat each sample separately for
meta-analyses of within-country associations, where multiple samples from the same country
do not exhibit this problem.
S4.2 Country level associations
Table S6 shows these country-level correlations. The most robust finding was for Individualism-
Collectivism, with a higher proportion of Unlimiteds observed in countries relatively more
focused on community outcomes (higher Collectivism). One interpretation is that people in more
collectivistic countries (low Individualism) may be more likely to think about the wider
community benefits they could produce with unlimited wealth, whereas in countries high in
Individualism people think more in terms of their own/family’s needs, believing these could be
satisfied with more limited wealth.
A positive but less robust relationship was found for Power Distance. This indicates that
Unlimiteds were more common in countries with greater cultural acceptance of inequality.
Other indicators did not show relationships with the proportion of Unlimiteds.

19
Development indicators
Lottery experience
Cultural dimensions
HDI
GDP
(PPP)
GINI
Lottery use
%GDP spent
on lotteries
Individualism
(lower scores=
Collectivism)
Power-
Distance
“Masculine-
Feminine”
Uncertainty
avoidance
Long-term
orientation
Study 2
r
-.34
-.15
.26
-.09
.10
-.60***
.44*
-.09
.12
-.22
n
30
30
26
30
22
22
22
22
22
12
Robustness checks
Study 2+
r
-.37*
-.17
.28
-.14
.09
-.58***
.40*
-.04
-.03
-.22
n
33
33
30
33
25
25
25
25
25
16
Both studies
r
-.32
-.10
.18
-.16
.14
-.46*
.31
-.06
-.08
-.30
(averages)
n
33
33
30
33
25
25
25
25
25
16
Both studies
r
.24
-.05
.16
-.16
.16
-.31
.20
-.03
.04
-.37
(all samples)
n
42
42
39
42
33
33
33
33
33
21
* p < .05; *** p < .005
Table S6.
Country-level correlations between proportion of Unlimiteds and cultural-level variables

20
S4.3 Meta-analyses
In the main article we focus on meta-analyses using point-biserial correlations between a
dichotomized lottery measure (1 = 100 billion [Unlimiteds ]; 0 = all other values [Limiteds]). For
the associations reported below, positive scores denote that those with higher scores on the
variable or scale were more likely to be Unlimiteds. Due to the large number of meta-analyses
and their exploratory nature, as well as the large overall sample sizes, we set a conservative
criterion for significance: p = .005.
All meta-analyses were based on random-effects models using non-iterative method of moments,
calculated using a macro designed for SPSS 15,16.
For each meta-analysis we report three statistics (plus k which indicates the number of samples):
Mean effect (and its significance): The average effect size across samples (correlation,
weighted by sample sizes) and an indicator of whether this was significantly different from
zero.
Q (and its significance): An index of variation in relationships across country, with an indicator
of whether this was significantly greater than would be expected by chance
I2: This is a descriptive index of the percentage of variation between samples attributable to real
differences rather than random variation.
Our focus was on Study 2 due to the larger and more representative sample of countries,
especially in comparison to Study 1 which had too few countries to assess cross-country
differences. However, in Tables S7 and S8 we report robustness checks (“Study 2+” and “All
samples”). We do not report the “Both studies (averages)” robustness check because both studies
contribute to the overall average effect in the “All samples” meta-analysis.
Table S7 shows associations between choosing the Unlimited/Limited lotteries with
demographics and methodological variables. Two demographic indicators were significantly and
robustly associated with lottery choice. While overall effects were objectively weak (average
correlations of less than 0.1), the strongest association of all variables was with age, with
younger people more likely to be Unlimiteds than older people. However, this varied
significantly across countries, and below we use meta-regression to identify reasons for cross-
country variation.
Table S7 also shows that those living in urban areas were more likely to be Unlimiteds than
those living in rural areas, and this did not vary significantly across countries. We surmise that
this is likely to be due to self-selection – cities tend to be where major corporations locate their
offices, which gives people the best chances of working in high wealth-creating industries.
It is important to also consider where we did not find evidence of associations. Table S7 shows
that Limited/Unlimited lottery choices were not significantly related to political orientation,
socioeconomic status, or education, nor did these effects vary significantly across samples.
Lottery choice was also not related to the time taken to answer the lottery question.
Covid did not appear to have had an overall impact on lottery choice, although the effect of
experience of Covid did vary significantly across countries. Below we use meta-regression to
examine whether this was related to country-level development and cultural indicators.

21
Age
Gender
(higher=
female)
Liberal-
Conservative^
(higher=
Conservative)
Left-
Right^
(higher=
Right)
Socio-
economic
status
Rural-
urban
(higher =
urban)
Years of
education
Covid
experience
Covid
well-
being
Lottery
page
time
Study 2
Mean effect
-.09***
.01
-.02
.01
.03
.06***
.02
.02
-.01
-.01
Q
60.3***
30.9
18.6
39.9
40.7
32.3
31.7
52.4***
24.1
47.7
I2
52
6
0
27
29
10
8
45
0
39
k
30
30
28
28
30
30
30
30
30
30
Robustness checks#
Study 2+
Mean effect
-.09***
.01
-.02
.01
as
.06***
as
as
as
as
Q
62.9***
31.2
21.9
42.4
above
36.1
above
above
above
above~
I2
49
0
0
29
11
k
33
33
31
31
33
Both studies
Mean effect
-.09***
.01
-.02
.00
as
.05***
as
as
as
as
(all samples)
Q
87.3***
36.1
30.3
54.4
above
48.9
above
above
above
above~
I2
53
0
0
28
16
k
42
42
40
40
42
*** p < .005
^Political orientation questions were not included in surveys for Tunisia or Morocco.
#Not all indicators were included in Study 1.
~Lottery page times are not comparable across studies because in Study 1 the page included an open-ended question about the most
important change they would make with the money. However, in Study 1 the association with lottery page time showed no significant
mean effect or cross-country variation, M = .07, Q = 20.2, I2 = 46.
Table S7.
Random-effects meta-analyses for point-biserial correlations between higher lottery choice and demographic/methodological
indicators.

22
Table S8 shows associations between choosing the Unlimited/Limited lotteries with values. Again, overall relationships were quite
week (overall correlations less than .08) but there were some significant and robust relationships that were relatively consistent across
countries. Unlimiteds were more likely to hold stronger Openness to change values (Self-direction, Stimulation) and Self-
Enhancement values (Hedonism, Achievement, Power), values categorized as having a personal focus 17. However, Unlimiteds and
Limiteds did not differ in the importance they placed on more socially-focused values relating to Conservation (Security, Conformity,
Tradition) or Self-Transcendence (Benevolence, Universalism).
Personal focus
Social focus
Value
Quadrant
Openness to change
Self-Enhancement
Conservation
Self-Transcendence
Value
Domain
Self-
direction
Stimulation
Hedonism
Achievement
Power
Security
Conformity
Tradition
Benevolence
Universalism
Study 2
Mean effect
.07***
.06***
.06***
.07***
.06***
.02
-.01
-.02
.01
.04
(k=30)
Q
34.1
29.1
33.9
20.4
40.7
38.2
44.0
15.2
32.8
42.4
I2
15
0
14
0
29
24
34
0
11
32
Robustness checks
Study 2+
Mean effect
.06***
.06***
.06***
.06***
.06***
.01
-.02
-.02
.01
.03
(k=33)
Q
49.0
36.4
42.9
22.4
46.5
42.8
47.1
18.3
34.2
46.6
I2
35
12
25
0
31
25
27
0
6
31
Both studies
Mean effect
.05***
.06***
.05***
.06***
.05***
.01
-.02
-.02
.01
.03
(all samples)
Q
52.9
51.1
61.1
25.7
58.3
52.9
56.2
27.9
38.5
51.7
(k=42)
I2
22
20
33
0
30
22
27
0
0
21
*** p < .005
Table S8.
Random-effects meta-analyses for point-biserial correlations between Limited-Unlimited lottery choice and values.

23
Explaining cross-country variation
Significant variation across countries/samples was identified for Age and Covid experience.
To better understand reasons for this variation, we performed meta-regressions (random-
effects) that examine associations between variation in these effects across countries and
country-level indicators. Separate meta-regressions were used to understand the three
development indicators: HDI, GDP(PPP), GINI; and the five cultural dimensions. As these
were follow-ups to explain significant variation and are conducted at a country level (and
hence a smaller dataset), we adopted a more conventional significance level of p = .05.
Age: Table S9 shows that two development indicators (HDI, GDP(PPP)) and one cultural
dimension (Power Distance) were significant and robust predictors of variation in effect
sizes of age on Unlimiteds lottery choices.
For HDI and GDP (PPP), this means in countries with higher levels of development
younger people were more likely to make the Unlimited choice (e.g., Singapore, rage = -.30
for HDI). In other words, in richer countries the young are more likely to have unlimited
wants compared to the old. In less developed countries, Unlimiteds were more evenly
distributed across the young and old (as correlations were close to 0, e.g., Uganda, rage = -
.02 for HDI).
For Power Distance, this means that in countries with greater acceptance of inequality in
society older people were more likely to be Unlimiteds than younger people (e.g.,
Philippines, rage = .09), whereas in countries with more focus on equality younger people
were more likely to be Unlimiteds (e.g., Hungary, rage = -.20).
Development indicators
Cultural dimensions
HDI
GDP
(PPP)
GINI
Individualism
(lower scores=
Collectivism)
Power-
Distance
“Masculine-
Feminine”
Uncertainty
avoidance
Long-term
orientation
Study 2
-.52***
-.52***
.20
-.25
.42*
.13
.06
-.08
Robustness checks
Study 2+
-.50***
-.52***
.21
-.18
.39*
.13
.09
-.08
Both studies
-.51***
-.51***
.30
-.22
.43**
.22
.12
-.01
(all samples)
* p < .05; ** p < .01; *** p < .005
Table S9.
Meta-regression (beta coefficients) predicting cross cultural variation in effect sizes between
age and the Unlimited lottery choice.
Covid Experience: No country-level indicators explained significant cross-country
variation for Covid experience. We note descriptively that those countries where the
experience of Covid was most negatively related to the Unlimited lottery choice was in
Japan (r = -.13), Mexico (r = -.12), and Indonesia (r = -.12), and was most positively
related in India (r = .24), Sweden (r = .21), and UAE (r = .16).

24
S4.4. Analysis of text responses for main use of wealth (Study 1)
In Study 1, immediately below the lottery question was an open-ended text-box response
question: “Briefly, what is the most important change you would make in your life with that
money?”. Here we report our analysis of the written responses to this question.
We first note our impression that people interpreted this question in multiple ways. In
particular, many people seemed to refer to their most immediate priority or the first thing they
would do, rather than how they would use the majority of the prize amount. For instance, it
was relatively common to write that they’d pay off their debts or mortgage, even for those
who selected 100 billion (we assume their debts were not this large). While our coding below
should be understood in this context, we do believe that these responses give a useful
indication of the basic priorities people had for their money.
Four percent of participants (n=91) did not answer the question directly, providing only
tangential commentary (e.g., “life is too expensive), or no interpretable text (no response or
random letters1). As most responses were short and these broad categories required only
basic coding, Google Translate was considered adequate for translating non-English
responses.
For those who provided codable responses (n=2043), we used the following four codes
(responses could be coded in more than one category). Responses were coded independently
by two coders (the authors). Inter-coder reliability was examined using Krippendorff’s α,
where α > .8 represents acceptable agreement 18. As responses could be span multiple codes,
we calculated Krippendorff’s α for each code (reported with each code below), showing very
good reliability. Disagreements were resolved through discussion.
1. Personal (79% of codable responses; α = .87). This included resources for the person
themselves and/or their families and friends (and occasionally pets). Typical examples
were buying houses, paying debts, traveling, or investing for their children’s future
(such as paying for their education or houses)
2. Society (21%; α = .90). This included resources for local, regional or national
communities, e.g., feeding the poor or housing the homeless
3. World (9%; α = .83). This included resources for the non-human world such as caring
for animals on the street, or environmental conservation such as cleaning the oceans
4. Don’t know (1%; α = .96). We coded this separately to identify people indicating no
clear idea about how they would use their ideal wealth (or chose to respond but were
unwilling to disclose their uses). This included “don’t know”, and similar responses
such as “no idea” or “not sure”.
There was also an “other” category (x%) for responses too ambiguous or general to code
(e.g., everything; safe; lots). Most coding disagreements involved whether to use the focal
codes or “other”. Our favorite “other” response was “Buy even more tickets” (from an
Unlimited participant), nicely illustrating an Unlimited mindset.
Some common responses were ambiguous, and for these we used the following coding rules:
Charity/donations. Where the recipient of charity or donations was not specified
(common) or could not be categorized (rare), we recorded codes for both Society and
World.
Education. Responses typically referred to providing for the education of one’s children or
grandchildren (coded Personal) or educating the poor (coded Society). Where the target of
education was not specified (e.g., “Education”) we coded this as Society.

25
Employment. Some responses referred to using their wealth to provide employment to
others or fight unemployment (coded Society). Where responses just referred to
employment, we coded this as Society.
Finally, we highlight below some responses that provided insights into people at the lower
level of Limited lottery choices:
“small thinking is low loss” (10 000)
“only want what is needed” (100 000)
“It’s enough” (1 million)
“would be nice but not greedy” (1 million)
“Get what I want without financial restrictions and give some to my children” (10 Million)
Comparing Unlimiteds and Limiteds.
In making comparisons between Unlimiteds and Limiteds in their intended uses, we coded
each participant’s response into a single code. Drawing on work on moral expansiveness 19,
where a person’s response was coded in multiple categories we selected the code that
reflected the greatest moral distance from the self (so the priority was World > Society >
Personal). For example, responses coded as both Personal and World were coded as World.
This means the codes were oriented towards reporting more altruistic uses of the prize.
For analyses, we initially used χ2 to examine relationships between Limiteds/Unlimiteds and
the three substantive codes (Personal, Society, World). However, the low number of
responses for World meant treating this as a separate category led to all analyses severely
failing χ2 assumptions of an expected frequency of 5 per cell. Hence, we combined the
Societal/World categories as they both represented altruistic uses.
Based on coded responses, the percentage of participants in each code in each country is
shown in Table S10, with a χ2 test to identify systematic differences in patterns of responses.
The patterns were similar and significant for all countries except China. In general,
Unlimiteds were significantly more than Limiteds likely to refer altruistic uses
(Society/World). In China, the patterns were more consistent across Limiteds and Unlimiteds,
with Personal uses much more prevalent in both groups.
Additionally, the number of “Don’t know” responses were too small for formal analysis, but
we note that Unlimiteds provided 18% of the coded responses in Table S10 but accounted for
27% of “Don’t know” responses. Although only indicative, this suggests that a higher
proportion of Unlimiteds than Limiteds did not have a clear impression of how the money
would help them achieve their ideal lives. Instead, it may be that more people in this category
treated this as a “default” (you should always want as much money as possible) without
considering its uses.

26
Limiteds
Unlimiteds
Country
n
Personal
%
Society/
World
%
Personal
%
Society/
World
%
χ2
p
North America
USA
159
48
18
19
14
4.13
.042
South America
Brazil
133
78
11
7
5
8.25^
.004
Argentina
174
64
21
5
10
16.85
<.001
Asia
China
171
85
8
6
1
0.80^
.381
India
162
44
36
6
14
5.98
.014
South Korea
177
71
7
16
6
7.08
.008
Europe
Russia
165
68
18
7
7
10.69
.001
Sweden
138
60
21
10
9
4.15
.042
United Kingdom
154
69
10
13
7
8.65
.003
France
145
71
16
6
7
11.10^
<.001
Africa
South Africa
181
52
26
6
16
21.38
<.001
Oceania
Australia
146
68
14
10
7
6.35
.012
Boldfaced numbers denote adjusted standardised residuals above +2.0 (more frequent than
expected). Underlined numbers denote adjusted standardised residuals below -2.0 (less
frequent than expected).
^ one cell had an expected value less than 5, failing to meet a typical χ2 test assumption.
Table S10.
Percentage (of total country sample with coded responses) of Limiteds and Unlimiteds who
wrote about personal or societal/world changes they would make with their chosen lottery
prize.

27
S5. Explaining variation in limited wants
While our primary focus was on the distinction between those expressing Unlimited and
Limited wants, a reviewer expressed interest in understanding differences between different
degrees of Limited wants. We agree this is an interesting supplementary question and report
these analyses below.
Our approach to the meta-analyses mirrors that used for the main meta-analyses reported in
S4.3. We used Spearman rank-order correlations as the effect size metrics for these analyses
in recognition of the ordinal nature of the lottery responses.
In summary, the relationships were similar to analyses comparing Limiteds and Unlimiteds.
That is, the same factors that predicted who would choose the Unlimited lottery prize tended
to also predict those who chose higher (but limited) lottery prizes.
Table S11 shows meta-analyses examining associations between Limiteds’ lottery choices
and demographic/methodological variables. While no overall effects were strong, choosing
larger prizes was more likely for people who were younger, reported higher socioeconomic
status, more urban and more educated. People who spend longer on the lottery question page
were also more likely to select larger prizes. Where robustness checks could be performed
(for age and rural/urban) these associations were robust. However, while effects of urban did
not vary significantly across countries, effects of age did and explanations of this variation is
reported later in this section.
Table S12 shows meta-analyses examining associations between Limiteds’ lottery choices
and values. Again, overall mean effects were not strong, but larger lottery choices were more
likely for people with self-direction, stimulation, hedonism, achievement, and power values,
reflecting the broader value orientation of Openness to Change and Self-enhancement. On
contrast, lottery prize choice was not associated with values reflecting Conservation (security,
conformity, tradition) nor Self-transcendence (benevolence, universalism). These effects
persisted in the robustness checks, and in no analysis did we find evidence for significant
cross-country variation.

28
Age
Gender
(higher=
female)
Liberal-
Conservative^
(higher=
Conservative)
Left-
Right^
(higher=
Right)
Socio-
economic
status
Rural-
urban
(higher =
urban)
Years of
education
Covid
experience
Covid
well-
being
Lottery
page
time
Study 2
Mean effect
-.08***
.02
-.03
.03
.08***
.05***
.06***
.02
-.01
.09***
Q
75.8***
25.8
21.0
44.7
34.4
31.4
22.8
41.1
39.3
60.4***
I2
62
0
0
40
16
8
0
29
26
57
k
30
30
28
28
30
30
30
30
30
27
Robustness checks#
Study 2+
Mean effect
-.08****
.03
-.03
.03
as
.05***
as
as
as
as
Q
78.4***
33.3
23.9
51.2
above
32.9
above
above
above
above
I2
59
4
0
41
3
k
33
33
31
31
33
Both studies
Mean effect
-.07***
.03
-.02
.04
as
.05***
as
as
as
as
(all samples)
Q
110.0***
37.6
35.0
60.8
above
44.5
above
above
above
above
I2
63
0
0
36
8
k
42
42
40
40
42
*** p < .005
^Political orientation questions were not included in surveys for Tunisia or Morocco.
Table S11.
Random-effects meta-analyses for Spearman rank-order correlations between Limiteds’ lottery choices (higher number = larger prize) and
demographic/methodological indicators.

29
Personal focus
Social focus
Value
Quadrant
Openness to change
Self-Enhancement
Conservation
Self-Transcendence
Value
Domain
Self-
direction
Stimulation
Hedonism
Achievement
Power
Security
Conformity
Tradition
Benevolence
Universalism
Study 2
Mean effect
.07***
.09***
.07***
.11***
.10***
.01
-.02
-.01
.02
-.01
(k=30)
Q
26.4
42.4
33.3
51.2
38.6
31.4
29.8
47.9
27.7
29.5
I2
0
32
13
43
25
8
3
39
0
2
Robustness checks
Study 2+
Mean effect
.07***
.09***
.07***
.11***
.10***
.01
-.02
-.01
.02
-.01
(k=33)
Q
30.5
45.2
36.3
52.3
41.8
32.9
31.4
50.5
28.2
30.5
I2
0
29
12
39
23
3
0
37
0
0
Both studies
Mean effect
.07***
.08***
.07***
.10***
.09***
.03
-.01
-.01
.02
.00
(all samples)
Q
35.8
49.0
41.1
61.7
49.0
57.4
39.4
61.3
32.3
41.0
(k=42)
I2
0
16
0
34
16
29
-4
33
0
0
*** p < .005
Table S12.
Random-effects meta-analyses for Spearman rank-order correlations between Limiteds’ lottery choices (higher number = larger prize) and
values.

30
To understand cultural variation in associations of age with Limiteds’ lottery choices, we
performed meta-regressions, with results in Table S13. Younger people were more likely to
choose larger prizes in more developed countries (higher HDI) and in countries with more
power-distance, and these were maintained in the robustness checks. Younger people were
more likely to choose larger prizes in more individualist countries, although this did not
withstand robustness checks. Finally, we note a possible divergence from the equivalent
Limited/Unlimited meta-regressions, as here economic productivity (GDP) was not
significantly associated with age effects whereas it was when comparing Unlimiteds to
Limiteds.
Development indicators
Cultural dimensions
HDI
GDP
(PPP)
GINI
Individualism
(lower scores=
Collectivism)
Power-
Distance
“Masculine-
Feminine”
Uncertainty
avoidance
Long-term
orientation
Study 2
-.38*
-.27
.03
-.42*
.55***
.21
.13
.13
Robustness checks
Study 2+
-.36*
-.25
.01
-.33
.55***
.19
.14
.13
Both studies
-.34*
-.25
.12
-.29
.48***
.08
.14
.08
(all samples)
* p < .05; ** p < .01; *** p < .005
Table S13.
Meta-regression (beta coefficients) predicting cross cultural variation in effect sizes between
age and the lottery choice for Limiteds.

31
S6. Supplementary Information References
1 Bain, P. G. et al. Public views of the Sustainable Development Goals across countries.
Nature Sustainability 2, 819-825, doi:10.1038/s41893-019-0365-4 (2019).
2 Cornell University, INSEAD & WIPO. The Global Innovation Index 2019: Creating
Healthy Lives—The Future of Medical Innovation. (Ithaca, Fontainebleau, and Geneva,
Ithaca, Fontainebleau, and Geneva, 2019).
3 United Nations Development Programme. Human Development Report 2020. (United
Nations Development Programme, New York, NY, 2020).
4 Garrett, T. A. An international comparison and analysis of lotteries and the distribution
of lottery expenditures. International Review of Applied Economics 15, 213-227,
doi:10.1080/02692170151137096 (2001).
5 Hofstede, G., Hofstede, G. J. & Minkov, M. Cultures and organizations: Software of
the mind. (McGraw-Hill, 2010).
6 Minkov, M. & Hofstede, G. Hofstede’s Fifth Dimension. Journal of Cross-Cultural
Psychology 43, 3-14, doi:10.1177/0022022110388567 (2010).
7 Twenge, J. M. Changes in masculine and feminine traits over time: A meta-analysis.
Sex Roles 36, 305-325 (1997).
8 Bain, P. G. & Bongiorno, R. Are Individualism and “Masculinity” related when
controlling for regional proximity? A reappraisal of Barry (2015). Journal of Cross
Cultural Psychology 46, 1226 –1231 (2015).
9 Schwartz, S. H. in Advances in Experimental Social Psychology Vol. 25 (ed Mark P.
Zanna) 1-65 (Academic Press, 1992).
10 Lindeman, M. & Verkasalo, M. Measuring values with the short Schwartz's Value
Survey. Journal of Personality Assessment 85, 170-178 (2005).
11 Sandy, C. J., Gosling, S. D., Schwartz, S. H. & Koelkebeck, T. The development and
validation of brief and ultrabrief measures of values. Journal of Personality Assessment
99, 545-555, doi:10.1080/00223891.2016.1231115 (2017).
12 Schwartz, S. H. et al. Extending the cross-cultural validity of the theory of basic human
values with a different method of measurement. Journal of Cross-Cultural Psychology
32, 519-542 (2001).
13 Adler, N. E., Epel, E. S., Castellazzo, G. & Ickovics, J. R. Relationship of subjective
and objective social status with psychological and physiological functioning:
Preliminary data in healthy white women. Health Psychology 19, 586-592 (2000).
14 Bain, P. G., Milfont, T. L., Kashima, Y. & et al. Co-benefits of addressing climate
change can motivate action around the world. Nature Climate Change 6, 154-157,
doi:10.1038/NCLIMATE2814 (2016).
15 Lipsey, M. W. & Wilson, D. B. Practical meta-analysis. (Sage Publications, 2000).
16 Wilson, D. B. Meta-analysis macros for SAS, SPSS, and Stata. Retrieved October 28,
2011 from http://mason.gmu.edu/~dwilsonb/ma.html (2005).
17 Schwartz, S. H. et al. Refining the theory of basic individual values. Journal of
Personality and Social Psychology 103, 663-688, doi:10.1037/a0029393 (2012).
18 Hayes, A. F. & Krippendorff, K. Answering the call for a standard reliability measure
for coding data. Communication Methods and Measures 1, 77-89 (2007).
19 Crimston, C. R., Bain, P. G., Hornsey, M. J. & Bastian, B. Moral expansiveness:
Examining variability in the extension of the moral world. Journal of Personality and
Social Psychology 111, 636-653, doi:10.1037/pspp0000086 (2016).