Market Exposure and Moral Universalism: Evidence from Online and Laboratory Experiments

In the aboratory experiment, participants are randomly assigned to a Market prime or a Non-market prime. The Market prime asks participants to read a short vignette describing a highly marketized society and then write 2–4 sentences imagining themselves in that context. The Non-market prime asks participants to read a vignette describing a non-market society and write 2–4 sentences imagining themselves in that context. This random assignment allows us to identify the causal effect of market exposure on moral decision-making.

The online survey component is not an intervention. It relies on naturally occurring variation in respondents’ market exposure, measured by the marketization index of their province of residence. These data are used to examine correlational relationships rather than causal effects.

2025-09-10

2025-12-31

1. Moral universalism in honesty.

2. Personal morality and norm evaluation. We examine several dimensions of individual morality and the evaluation of social norms:
(1) Trade-offs between personal interests and moral (or norm-based) considerations
(2) Willingness to comply with social norms

In our mind-cheating allocation game, each respondent completes 20 decision rounds（12 in the online survey, to reduce participants’ attention burden). In each round, they are paired with one of two randomly assigned partner types: an in-group member or a out-group member( a completely random stranger or a co-citizen from the same city in the online survey). Each partner type appears 10 times (6 times in the online survey), with the order of appearances randomized across rounds.

Moral universalism in honesty will be measured as the average difference between the total amount allocated to out-group members and the total amount allocated to in-group members across ten mind-cheating allocation tasks (six tasks in the online survey). Therefore, this method is better suited for identifying differences in moral universalism at the group level, but it has certain limitations in capturing individual-level variation.

Trade-offs between personal interests and moral (or norm-based) considerations will be measured in the following questions:
(1) Suppose there are two job opportunities. One job offers a lower income than the other but allows you to help others more effectively. What is the maximum percentage of income you would be willing to give up to take the job that enables you to help others more?
(2) Suppose there are two job opportunities. One job offers a lower income than the other but carries more prestige (i.e., greater social recognition). What is the maximum percentage of income you would be willing to give up to take the more prestigious job?
(3) If you had to choose between following social rules and obtaining monetary gain in a certain situation, how would you decide? Please use the slider to indicate which side you are more inclined toward.
(4) Agreement with statement: "I believe that personal interests are much more important than morality."

Willingness to comply with social norms will be measured with the following questions:
(1) Imagine a person who normally does not drink is invited to a dinner. Others at the table insist he should drink according to local custom, but he is very reluctant. To what extent do you support him firmly refusing to drink?
(2) Agreement with statement: "To improve efficiency, it is acceptable not to strictly follow the rules."
(3) Agreement with statement: "The education I received from my parents emphasized that achieving goals is more important than following rules."
(4) Agreement with statement: "In my daily life, I am a person who strictly abides by rules."

1.Measures of moral universalism. We also elicit three other measures moral universalism adapted from prior literature that are non-incentivized.
(1) Universalism in altruism
(2) Universalism in trust
(3) Moral universalism assessed through qualitative psychological questionnaires

The measures of universalism in altruism and trust will follow the approach of Enke (2022), with the modification that we consider only three distinct social groups—(i) a member of one’s extended family, (ii) someone from one’s hometown, and (iii) a friend of one’s relative. This restriction is intended to ensure that the questions remain broadly applicable across populations while keeping the design simple and easy to implement.

In the experiment, in-group and out-group membership is predetermined based on participants’ initial preference scores for paintings, with participants split into two equal-sized groups using the sample median. Subsequently, participants are randomly assigned to either a market or non-market condition. After being primed with the corresponding scenario, they complete incentivized mind-cheating allocation tasks and additional behavioral tasks.

The online survey uses naturally occurring variation in respondents’ market exposure, measured by their province’s marketization index. Survey design details are available at: https://osf.io/236qc.

The in-group and out-group membership is predetermined based on participants’ initial preference scores for paintings. Participants are first shown pairs of abstract paintings created by two artists (from Chen and Li, 2009) and asked to indicate their preference on a scale from -5 to 5 (five rounds in total). Based on their cumulative scores, participants are split into two equally sized groups using the sample median as the cutoff. After grouping, participants are told which paintings belong to which artist and are then asked to guess the artist of two new paintings. During this stage, they have five minutes to discuss with in-group members via a group-specific chat channel (communication is voluntary and unrestricted, but generally centers on the paintings). Each correct guess yields a monetary reward. This procedure is designed to strengthen group identity before the main tasks.

The core experimental design adapts the “mind cheating game” (Jiang, 2013). In this task, respondents allocate money between themselves and a randomly matched participant. Each respondent first privately chooses a letter (A–F), then reports it after observing the payoffs associated with each letter. Because only the respondent knows their true choice, they have the opportunity to misreport in order to secure higher payoffs. Matched partners are either a randomly assigned in-group member or a randomly assigned out-group member. The game consists of 20 repeated rounds (12 in the online survey), with the payoff schedule associated with each letter randomized in every round. Half of the rounds involve a partner from the in-group, while the other half of the rounds involve a randomly assigned out-group member. The order in which the two partner types appear is randomized across the rounds.

Prior research (e.g., Enke, 2022) has operationalized moral universalism as the difference in altruistic behavior toward in-group members versus strangers, typically measured through allocations of money or trust points. Our new approach has three advantages compared to previous methods: (1) real monetary incentives create tension between self-interest and moral norms, (2) it directly measures the moral decision toward in-group versus out-group members, and (3) it avoids confounding with distribution preferences (compared to the measure of universalism in altruism).

Participants are randomly assigned to either the market or non-market priming condition using a computer-based randomization procedure implemented in the laboratory.

Randomization is conducted at the individual level. Each participant is independently assigned to either the market or non-market priming condition.

No

Not applicable, as randomization is conducted at the individual level.

Laboratory experiment: 200 participants Online survey: 1,200 participants

Laboratory experiment: 100 participants of Market group, and100 participants of Non-market group.
Online survey: Planned 40 participants per province across 30 provinces (total 1,200), with a conservative minimum target of 20 participants per province.

40 individuals per province

We aim to detect a minimum effect size of Cohen’s d = 0.4, corresponding to a small-to-medium effect, for the primary outcome, moral universalism in honesty. Based on a two-sided test with α = 0.05 and power of 0.80, we require approximately 98 participants per group (196 in total) in the laboratory experiment. This ensures adequate power to detect an effect of the expected magnitude.