I went through the Replication data provided by the authors in the Acknowledgments, especially the part about China. I have a few comments to share:
1. About email usage. The authors used World Bank data on "how often business firms use email to contact their customers" to control the results and found it has no big effect. The number for China is 85%, and is pretty high among the countries which have this data (for example, only ~76% for Germany). Apparently using such data to control the results wouldn't help China's ranking. Even if this data is reliable, it is not clear how it reflects the usage of email for people involved in this study in their daily life. I suggest a survey about email usage in daily communication should be conducted at the same time in institutions similar to those the experiments were performed at.
2. According to the data, all 400 "wallets" were "turned in" in China by a single study assistant (#1, male, age 27.7) between July 7-27 2015. Each day he delivers 30-40 "wallets", each to a different institution, which I feel very intense though may not be impossible. About 55% of the recipients didn't speak English, or 45% did, which is pretty high for a country like China, though still much lower than the average among all 40 countries (~70%). According to explanation by the authors I found somewhere else, if the recipients do not speak English, the study assistant will use a cellphone app to communicate with him/her. So far I haven't found exactly what Chinese sentences the assistant used in China, which may be crucial for most the recipients to understand the situation. However, the assistant did report a pretty high percentage of recipients as understood the situation (average 5.2, with 0 the least understood and 6 the best understood). I suggest the authors explain in some details how the assistant, a German-speaking university student, communicate with the local recipients and assess if the recipient did understand what is going on.
3. I found the refuse rate is extremely low. Among all 17000+ "wallets", only about 100 were refused (0.588%). In China, the refuse rate is even 0. As someone has pointed out below, because of a scam, in which a guy will turn in a "found on street" wallet and later another guy will show up to claim it, and says he had more money in it, many people in China tend to not accept wallets turned in by a stranger. More likely they will suggest you to turn it in to police. I would like to see the authors explain why not even a single recipient refused to accept their "wallets" in China. Is it because you didn't give them any chance? If that is true, I wouldn't expect the recipients will take the "wallets" very seriously.
Last but not least, I saw the authors in several occasions denying their intention to study the country-level difference of civic honesty. But in the published paper they clearly stated "our findings also represent a unique data set for examining cross-country differences in civic honesty". And the title, the figures, and the analysis in the Supplementary Materials also indicate the opposite. If they really didn't intent to do it, and only wanted to study the material incentive effect, they should have used some way to treat their raw data first rather than simply throwing it out. For example, they could use (return_rate_with_money - return_rate_with_no_money)/return_rate_with_no_money to eliminate some country-specific factors which reduce the overall returning rate but has nothing to do with civic honesty.
Thanks.
