First You Get the Money, Then You Get the Power: The Effect of Cheating on Altruism

Round 1

Reviewer 1 Report (New Reviewer)

This paper uses a laboratory experiment to examine the relationship between cheating and giving in a dictator game. In the first stage, participants compete in a real-effort task. In the Control treatment, their performance in this task is automatically scored, while in the Cheat treatment, it is self-reported, allowing the possibility to overstate their true performance. Participants with higher (reported) scores in the task are assigned the role of the dictator in a dictator game with another participant with a lower (reported) score. The results show that dictators in the Cheat treatment who did not report scores in the top quartile of the score distribution were somewhat more generous than dictators in the Control treatment. However, dictators in the Cheat treatment who did report scores in the top quartile were less generous than dictators in the control treatment.

Overall, I think this is an interesting experiment. However, there are issues that need improvement, particularly with the interpretation of the results and some other issues of clarity.

Comments:

1. It's not clear to me that the non-top-quartile dictators in the Cheat treatment were more generous because they wanted to make up for cheating. Instead, perhaps these are simply honest winners, and their honestly correlates with generosity. Participants who don't cheat given the opportunity would be more honest on average than dictators in the Control treatment, since this group would include both honest and dishonest individuals. Can a more convincing case be made that the current interpretation is correct?

2. The information at the first half of the first paragraph of Section 5 (Experimental Results) seems to fit better in Section 3, as it concerns the specifics the the experimental design and procedures.

3. It's unclear how the correlation in Hypothesis 2 between reported cheating and actual cheating is testable here, since actual cheating cannot be observed at the individual level, but only estimated in aggregate. It appears that the actual test relating to this hypothesis (bottom of page 9) actually concerns the correlation between reported cheating and scores, which is of course testable but somewhat different than the correlation hypothesized.

4. Just before Result 2, the authors discuss dictators who report "reasonable" scores (quotation marks included), but do not clearly define what they mean by this. Presumably they mean not the top quartile (scores of 11 or 12), but it would clearer to say this more precisely.

5. Immediately after Result 2, the authors mention Hypothesis 2 and reject the null. It would be helpful here (and elsewhere, right after Result 4) to remind the reader what the Hypotheses are, perhaps with brief descriptive labels.

6. In the middle of page 10, what is meant by "consistent" quiz scores? Does this mean they claimed to have not lied at all? This part could be written more clearly.

7. Just before subsection 5.1, some subjects are described as "moral cleansers." This terms does not appear to be defined elsewhere in the paper, and needs clarification.

8. At the bottom of page 15 and top of 16, it is claimed that the results show evidence that dishonesty is partly driven by expectations of dishonesty in others. This claim needs additional support. Is there a correlation between self-reported dishonesty and reported belief about others' dishonesty? Moreover, if such a correlation does exist, the causality might be the opposite (i.e. dishonesty causes belief that others are dishonest) due to the false consensus effect (see e.g. Ross et al. 1977).

9. The opening sentence of the Conclusion is too abbreviated to provide a useful summary of the results. It would be more helpful to concisely describe the manipulation and main results.

10. At the bottom of page 6 there is some text that doesn't appear to be part of any sentence, "pre-violation justifications and ii)".

11. The graphics in Figure 1 are missing.

12. There are is some informal/clichéd language that would be better re-written in more precise terms (page 7 "triumph of experience over hope", footnote 20 "take this with a grain of salt").

 

References

Ross, L., Greene, D., & House, P. (1977). The “false consensus effect”: An egocentric bias in social perception and attribution processes. Journal of experimental social psychology, 13(3), 279-301 

Author Response

Thank you for your constructive feedback. We have gone through your comments and believe they have improved the manuscript.

1) While we can never definitively rule out other characteristics that are correlated with generosity (e.g., honesty), we do have suggestive evidence. First, the HIGHSCORE and non-HIGHSCORE dictators in CHEAT both indicate that they lied (albeit HIGHSCORE dictators indicated they lied more). Second, when asked "You either lie or you don’t, there are no degrees of lying" there was no significant difference in how the two types answered the question. What is most telling however is that non-HIGHSCORE dictators were significantly more willing to tell a lie to help someone according to their survey answers. This suggests they have a preference for helping others or are giving to make up for their bad behavior and then answering survey questions in a way to justify their decisions. See the last paragraph in Section 6.1.

2) We agree. The paragraph has been moved.

3) Fair point. At the end of the experiment, we asked subjects to self-report their actual scores. We then compared these self-reported scores to their experiment scores. Overall we find that HIGHSCORE dictators have larger differences (i.e., the difference in their post-experiment survey quiz score and their experiment score is larger than the non-HIGHSCORE dictators). This is of course probably a conservative measure that under-reports the scale of lying by the biggest liars. We have expanded this discussion which can be found under Result 2.

4) Fair point. Reasonable has been defined.

5) We have cleared up the results to make the connections with the hypotheses more clear. We thank the ref. for this comment because it greatly improved the readability of the paper.

6) "Consistent" means the post-experiment score matches their experiment score. This is now defined in the text.

7) That was from an old draft. We thought we deleted all those references but one managed to slip through. As we re-read the manuscript, much of the older material did not seem to belong and we deleted it. Thank you for calling this to our attention.

8) We more formally test this using the score differentials and indeed people who cheat the most thought others would too. We also mention the potential for reverse causality.

9) The opening paragraph in the conclusion has been expanded.

10) Deleted.

11) We have made sure this is added to the zip file.

12) Removed.

Reviewer 2 Report (New Reviewer)

The paper misses some figures. That's quite weird, as it indicates the authors did not look through their final version. Please submit a properly proofread paper. Also please define cheating versus lying. Also please explain how you can measure cheating (or lying) at the individual level, or whether that's all based on self reports. In standard lying experiments it is crucially the case that individual lying canNOT be measured.

Author Response

1) Sorry for missing the figures. They were included in the PDF but the PNG files must have been left out of the zip file by mistake. We have fixed this.

2) Cheating is measured statistically (i.e., the difference between scores in the CHEAT treatment and the CONTROL treatment) and, at the individual level by the difference between reported scores in the experiment and self-reported scores at the end of the experiment. This likely means that the measure of individual level of lying is conservative, as even those who admit inaccurate reporting may not be willing to reveal the full extent of their lying.

3) We define cheating as misreporting the test score in the CHEAT sessions.

We hope this clears up any confusion and apologize for the submission mistake. 

Reviewer 3 Report (New Reviewer)

The present study examined how tournament winners treat the losers in the dictator game after tournaments in which cheating is possible. The authors conducted a 2-stage laboratory experiment in Canada and the United Arab Emirates. Subjects first completed a real effort task to earn the role of the dictator and subsequently participated in a dictator game. In treatment sessions, subjects self-reported their performance on the task. In control sessions, a computer automatically assesses subjects’ performance. Those performing in the top half in a session were assigned to be the dictator while those in the bottom half were assigned to be the receiver. The findings show that cheating significantly increases altruism in some tournament winners, but the more that subjects cheat, the less they give. Besides, cheating in Abu Dhabi sessions was significant and not significant in Calgary sessions. Differences in intelligence and beliefs were at least partially leading to cheating behavior across the two populations.

This topic is relevant and important. The manuscript is clearly written. But I still have several concerns for the current study.

1. What is the contribution of the present study? The author should further elaborate the contribution of this paper with reference to the psychological cost of cheating, moral cleanliness and the entitlement effect in the dictator game (e.g., Thielmann & Hilbig, 2019; Cochard, Gallo, Georgantzis & Tisserand, 2021).

2. Please supplement a table of descriptive statistics of giving in the dictator game, IQ test scores, and beliefs across different treatment and different populations.

3. The process of hypothesis is too simple and the theoretical foundation is not solid enough, especially for hypothesis 2, 3 and 4. If you don't have good theoretical knowledge regarding directions of effects, you have to adjust for multiple hypotheses testing.

4. The authors should discuss more about the relationship between IQ test scores, lying degree and giving in dictator game, and use a unified theory to explain the findings.

5. Figure 1 is not shown.

Author Response

Thank you for the opportunity to revise our manuscript. Your comments have significantly improved the quality of our work.

1) The primary contribution to the present work is to document the existence of pro-social cheaters as well as further report on the existence of selfish cheaters. We also note differences in decisions across cultures.

2) Done. This can be found in the appendix

3 & 4) Much of the under-pinning was discussed in a prior version of the manuscript. We have "resurrected" much of this discussion and theory, but have not included everything that was in the older version in the interest of space.

5) Thanks for calling this to our attention. Figure 1 was actually in the pdf included in the zip file but the PNG file used when compiling the text in Latex was missing. We apologize for this omission.

Round 2

Reviewer 1 Report (New Reviewer)

I am satisfied with the revised version and have no further comments.

Author Response

Thank you for giving us the opportunity to revise the paper. Your comments were very helpful and improved the work.

Reviewer 2 Report (New Reviewer)

Dear authors,

this version has improved substantially. Please make a few formal improvements:

> Please use the same rounding norm throughout. I would recommend two decimals, three max.

> Please clarify what kind of ethical approval you had.

> Please clarify where data and code for analysis are available.

> Please state whether this work was pre-registered, and if not why not.

 

Author Response

Thank you for the suggestions. They elevated the paper. 

1) Done. 2 decimal places

2) IRB approval was granted at University of Calgary NYU - Abu Dhabi

3) Data and code is on an OSF repository. 

4) The work was unfortunately not pre-registered. That seemed to become a bit more the norm after the experiments were completed. 

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

games-2098089-peer-review-v1
First you get the Money, Then you get the Power: The Effect of Cheating on Altruism.

In this paper, author shares a study on the effect of cheating on altruism in game theory models. The topic seems to be interesting. However, the quality of proposed work is not good according to the following comments:

1.      In introduction section authors do not explain the significance of proposed method.

2.      Novelty of new concept is not enough for publication.

3.      There is a major need to revise the article, because there are several grammatical mistakes in native English writing.

4.      Overall presentation of paper is not attractive.

5.      The advantages and disadvantages section is not included.

6.      I feel that although this may be a useful exercise, the paper does not come up to the international standards for publication. There is hardly any original contribution in this paper and it adds very little to the published literature.

7.      Abstract should have one sentence per each: context and background, motivation, hypothesis, methods, results, conclusions. What problem did you study and why is it important? What methods did you use? What were your main results? And what conclusions can you draw from your results? Please make your abstract with more specific and quantitative results while it suits broader audiences

Reviewer 2 Report

Comments for author File: Comments.pdf