@eale-office.bsky.social @eeanews.bsky.social @aeainformation.bsky.social

10/10. CAVEAT 2: The data I used are R-packages (ChineseNames and lexicon: freq_last_names); this simple analysis hinges on the reliability of these sources. Similarly, I have been extensively helped by ChatGPT to quickly code in R (not my software), so apologies if something is off.

9. CAVEAT 1: Of course, this inter-country issue adds to the intra-country fairness (I, myself have skin in the game having a “S” surname).

8. A much simpler solution (also saving space in the main text) is to systematically impose randomization, as proposed by Ray r Robinson, 2018. Isn’t it time we adopt this as our default option?

7. Recently, journals have started listing all three authors in the main text for publications with three co-authors. This is a welcome change, but it is likely to become ineffective as the number of co-authors continues to rise (Card and DellaVigna, 2013).

6. Could this also partly explain why Chinese authors seem to be less cited on average, even when they are first authors (Young and Zhou, 2025)?

5. Indeed, Li and Yi, 2021 find that Chinese economists (but not physicists or statisticians) are more likely to go back to China, where alphabetical ordering is not the norm.

4. As a consequence, the probability that a Chinese author is first author co-authoring with an American one is only around 30%. With two American co-authors is only 19.2%.

3. Even more strikingly, in the US, very few people have surnames that start with X, Y, or Z, while these letters account for around 28.5% of the Chinese population!

2. For instance, in the 1990 US Census, around 43% of the American population had a surname starting with a letter between A and H. The same number for Chinese surnames is 23%.

Very interesting, althouth I wonder how much people weighted in the main criticism about the violation of the exclusion restriction. Personally, that is enough to take the main AJR result as at best suggestive, even if the data were all correct.

Are you sure the data for the graph on the left is correct? The turnout in Florence was 45.6%. This is indeed the case for question 5. All questions had a very similar turnout rate. Maybe some issues with the averages?

I think excluding non-voters in this case is highly misleading, as not voting was a specific recommendation from some parties to make the "no" win.
Would your argument work even in that case? Certainly, the South would not come out as a stronger supporter than the North.

@cairosofie.bsky.social @hckongsted.bsky.social

Does the fact that Musk significantly helped Trump's win not count? Are you saying that all money spent in electoral campaigns in the US does not influence it?
If not, your definition is, in my opinion, way too narrow.