National parochialism is ubiquitous across 42 nations around the world

Cooperation within and across borders is of paramount importance for the provision of public goods. Parochialism – the tendency to cooperate more with ingroup than outgroup members – limits contributions to global public goods. National parochialism (i.e., greater cooperation among members of the same nation) could vary across nations and has been hypothesized to be associated with rule of law, exposure to world religions, relational mobility and pathogen stress. We conduct an experiment in participants from 42 nations (N = 18,411), and observe cooperation in a prisoner’s dilemma with ingroup, outgroup, and unidentified partners. We observe that national parochialism is a ubiquitous phenomenon: it is present to a similar degree across the nations studied here, is independent of cultural distance, and occurs both when decisions are private or public. These findings inform existing theories of parochialism and suggest it may be an obstacle to the provision of global public goods.


Supplementary Notes 1.Model results
In this section, we provide the full report of the results of the models. First, we run the model with only the Contrast 1 and public vs private variables (Supplementary Table 1), then we add models with controls and attention checks (Supplementary Table 2 Table 4 and 5). We also include models of the nations where we used incentives (Supplementary Table 6), models with expectations as an outcome variable (Supplementary Table 7), models that tested cross-cultural predictions on national parochialism and cooperation (Supplementary Table 8 Table 10), and a model that tests whether we observe outgroup derogation (Supplementary Table 11).

Full main model
In this section, we report the model with the main treatments and their interaction predicting cooperation: Contrast 1 (Ingroup = 1, Outgroup and Stranger = 0), Observability (public choice = 1, private choice = 0). Results show that people are more cooperative with ingroup compared to outgroup and stranger (national parochialism), and that people cooperate more when their choice is public, compared to when their choice is private. The interaction between the two treatments is not significant. Nsubjects 18,411 Note. Contrast1 = ingroup vs outgroup + stranger treatments; Observability = public vs private treatments; × = interaction term. *total observations are slightly less than the expected number of decisions (12) multiplied by the number of subjects (18,411) due to some missing cases in Japan (in Japan, due to a randomization error participants were randomly allocated to fewer decisions). The table shows estimates without adjustments for multiple comparisons. All tests were twosided. Source data are provided as a Source Data file.

Controls and robustness checks
In this section, we report two models. In Supplementary Table 2  Nsubjects 18,189 Note. Contrast1 = ingroup vs outgroup + stranger treatments; Observability = public vs private treatments; Quick Understanding = people who provided the correct answer in their first attempt + others); × = interaction term. The table shows estimates without adjustments for multiple comparisons. All tests were two-sided. Source data are provided as a Source Data file.
Results show that controlling for quick understanding, age, gender and education, people are still more cooperative with ingroup compared to outgroup members and stranger (national parochialism), and that people cooperate more when their choice is public, compared to when their choice is private. The interaction between the two treatments also remain not significant.
We also find that men, compared to women, cooperate more with strangers, that age has a positive relation with cooperation and that more educated people cooperate more compared to less educated people.
In Supplementary Table 3 Nsubjects 27,527 Note. Contrast1 = ingroup vs outgroup + stranger treatments; Observability = public vs private treatments; Attention check = people who passed the attention check + people who did not pass the attention check); × = interaction term. The table shows estimates without adjustments for multiple comparisons. All tests were two-sided. Source data are provided as a Source Data file.

Gender and education
In this section, we report two models. In Supplementary Table 4 Results show that controlling for quick understanding, age, and gender, people are still more cooperative with ingroup compared to outgroup members and stranger (national parochialism), and that people cooperate more when their choice is public, compared to when their choice is private. We also find that men, compared to women, cooperate more with strangers, and that age has a positive relation with cooperation. Importantly, we find an interaction between Gender and Contrast 1, suggesting that national parochialism is stronger among men, compared to women. Nsubjects 18,189 Note. Contrast1 = ingroup vs outgroup + stranger treatments; Observability = public vs private treatments; Quick Understanding = people who provided the correct answer in their first attempt + others); × = interaction term. The table shows estimates without adjustments for multiple comparisons. All tests were two-sided. Source data are provided as a Source Data file.

Supplementary
In Supplementary Table 5, we report a model with the main treatments and controls: Results show that controlling for quick understanding, age, and education, people are still more cooperative with ingroup compared to outgroup members and stranger (national parochialism), and that people cooperate more when their choice is public, compared to when their choice is private. We also find that higher educated people are more cooperative than lower educated people. Importantly, we find an interaction between Education and Contrast 1, suggesting that national parochialism is stronger among lower educated people, compared to higher educated people. Nsubjects 18,189 Note. Contrast1 = ingroup vs outgroup + stranger treatments; Observability = public vs private treatments; Quick Understanding = people who provided the correct answer in their first attempt + others); × = interaction term. The table shows estimates without adjustments for multiple comparisons. All tests were two-sided. Source data are provided as a Source Data file.

Payment model
In this section, we report the model with the main treatments and their interaction with There is also no significant main effect of Payment on cooperation, although as expected, if any, people in the incentive treatment cooperate less compared to people in the hypothetical scenario.
As incentives are hypothesized to decrease experimental demand effects 1 , these findings provide support for the idea that our study design which relies on hypothetical decisions was not contaminated with demand characteristics. This result, together with past research that found no differences across between-subjects and within-subjects designs in the extent of national parochialism and observability 2,3,4 , makes it very unlikely, in our view, that demand characteristics might constitute a potential confound. Nsubjects 2,443 Note. Contrast1 = ingroup vs outgroup + stranger treatments; Observability = public vs private treatments; Payment = real stakes + hypothetical choices); × = interaction term. The table shows estimates without adjustments for multiple comparisons. All tests were two-sided. Source data are provided as a Source Data file.

Expectations model
In this section, we report the model with the main treatments, their interaction, and = graduate school or higher). Nsubjects 18,194 Note. Contrast1 = ingroup vs outgroup + stranger treatments; Observability = public vs private treatments; Quick Understanding = people who responded correctly in their first attempt + people who did not respond correctly in their first attempt); × = interaction term. The table shows estimates without adjustments for multiple comparisons. All tests were two-sided. Source data are provided as a Source Data file.

Supplementary
Results show that people expect more cooperation when their partner is an ingroup member compared to when the partner is an outgroup member or a stranger, and that people expect more cooperation more when the choice is public, compared to when their choice is private. The interaction between the two treatments is not significant. Men expect more cooperation from their partners, compared to women. Older participants expected more cooperation from others. People that provided the correct answer in their first attempt in the comprehension check expected less cooperation from their partners, and there is no difference in expectations among people with different education levels.

Cross-national models
In this section, we report the results of independent mixed effects models (with nations as random intercepts) where each cross-national indicator predicts national parochialism (Supplementary Table 8) or cooperation (Supplementary Table 9). The outcome variable of Supplementary  Table 9 is the output of a principal component analysis that assumes that the data across the 12 decisions load on one factor. This can be interpreted as the cooperativeness of one person independent of the treatments (general level of cooperation). Regarding the results, the interpretation of the sign of the estimate is intuitive for most of the indicators. We list here the ones for which interpretation is counter-intuitive. Higher scores in confidence to institutions means lower confidence in institutions. Higher scores in Indulgence vs Restraint mean higher indulgent nations, while higher scores in Individualism vs Collectivism mean higher individualistic nations. Higher scores in self-expression (petition) mean lower endorsement of petitions.

Model with cultural clusters
In this section, we report independent mixed models split by cultural clusters with the main treatments predicting cooperation: Contrast 1 (Ingroup = 1, Outgroup and Stranger = 0), Observability (public choice = 1, private choice = 0). We clustered nations based on previous seminal work by Inglehart

Ingroup favoritism vs outgroup derogation
In this section, we report the model with Contrast 2 (Outgroup vs Stranger) and its interaction with Observability predicting cooperation: Contrast 2 (Stranger = 1, Outgroup = 0), Observability (Public choice = 1, Private choice = 0). Results show that people are more cooperative with outgroup members compared to strangers. These findings (together with the findings of Supplementary Moreover, we also tested whether identification with the own nationality was positively associated with national parochialism across nations. To do so, we run a mixed effect model of national identification predicting national parochialism score. National identification consists of a 1-item measure in which participants were asked to what extent they agreed or disagreed to the following statement: "I identify with my nationality" (1= totally disagree, 7 = totally agree).
Supporting the idea that a motivation of ingroup favoritism drives national parochialism, we found that national identification was positively associated with the national parochialism score (b = 0.10, p < .001).

Meta-analytic approach on national parochialism across nations
Supplementary Figure 1. Forest plot of national parochialism across nations. Forest plot displaying effect size of national parochialism predicting cooperation. For each nation, we report estimated effect size (Cohen's d) in each nation represented by a black square and 95% confidence interval. The size of the black square represents the sample size in each nation. The overall estimated population effect sizes are represented by the size of the black diamonds, which correspond to the 95% confidence intervals. Source data are provided as a Source Data file.

Cross-cultural indicators
We considered several cross-national indicators. For all of them we retrieved the most recent available data from datasets, such as the World Bank, world values survey, Hofstede etc. Rule of law Freedom house 2018

Cultural distance: analytic approach
We analyzed whether a measure of cultural distance between two nations predicted differences in national parochialism and cooperation. Cultural distance is a measure of the overall cultural differences between nations. To build this index, we retrieved bilateral cultural distance data from http://culturaldistance.muth.io/ 13

R-codes
Code of the main models.

Instructions
In this section, we include the instructions of the experiment. Regarding the decisions, we include three example of decisions (1 = partner is ingroup, public choice, 2 = partner is outgroup, public choice, 3 = partner is stranger, private choice). All the other decisions are combination of information provided in these 3 examples.

Information Sheet
Introduction. The study is being conducted by Professor ### and Professor ### We aim at testing some theories about decision making. For this reason, we kindly ask you to answer the survey seriously.
Procedures. The purpose of this research is to examine decision making in different situations. You will interact with some other participants in some decision making tasks. Then, you will be asked to answer some questions about the decision making tasks. We estimate it will take no more than 25 minutes to complete the study.

Risks/Discomforts.
There are no anticipated risks for participating in this study.

Benefits.
A potential benefit of participating is that you might learn something about decisionmaking that you might not have been aware of before. You may also be assigned to make a decision involving or being affected by someone from another nation. Anonymity. All of your answers will be anonymous. Any information you provide will be stored indefinitely on the encrypted and password protected site, and on password-protected computers only. When presenting the results of this research, we will in no way focus on individual participants' responses and will instead present the findings in summary form. You will not be asked for information that would enable to identify you personally. If you understand the information above and agree to participate in this research project, please click "I Agree" to start with this study. If you do not wish to participate right now, please close your web browser. Thank you for considering participating.

Compensation.
Welcome to the study. This is a study about decision making. The study involves participants from many countries around the world.
You will be asked to make decisions in several decision making tasks. You will be paired with a different person in each decision making task.
14. Before reading the instructions, please create a nickname for yourself (any combination of two letters and two numbers, such as x2f4)

(page break)
In these tasks, you and other people from around the world will make several decisions.
Each decision will be made with a different partner.

(page break)
To make sure you have understood the instructions, please answer the following questions: YOU: NICKNAME HERE PERSON B: is from Italy PLEASE NOTE: Your and your partner's decision in this round will be public. All participants will be given a link to see the results of your individual contribution under your nickname in a widely distributed blog: http://www.what-did-people-do.com The number of MU you expect PERSON B will send to you (before they get doubled) Set the button to a number

(page break)
__ PLEASE NOTE: in the next page, you will make a decision with a different partner __ (page break)

Example outgroup and public treatment decision
In this round you will make a decision in the following situation YOU: NICKNAME HERE PERSON B: is from one of 8 countries excluding Italy (Australia, Colombia, Germany, India, Nigeria, Serbia, Singapore, United States) PLEASE NOTE: Your and your partner's decision in this round will be public. All participants will be given a link to see the results of your individual contribution under your nickname in a widely distributed blog: http://www.what-did-people-do.com Please make a decision in the following situation YOU: NICKNAME HERE PERSON B: is from one of 8 countries excluding Italy (Australia, Colombia, Germany, India, Nigeria, Serbia, Singapore, United States) PLEASE NOTE: Your and your partner's decision in this round will be public. All participants will be given a link to see the results of your individual contribution under your nickname in a widely distributed blog: http://www.what-did-people-do.com

Example stranger and private treatment decision
In this round you will make a decision in the following situation