tag:blogger.com,1999:blog-26951738.post6693047965830745559..comments2024-03-28T19:14:33.619-07:00Comments on The Splintered Mind: Contest Winner! A Philosophical Argument That Effectively Convinces Research Participants to Donate to CharityEric Schwitzgebelhttp://www.blogger.com/profile/11541402189204286449noreply@blogger.comBlogger24125tag:blogger.com,1999:blog-26951738.post-29005934104713021152022-05-23T13:52:54.020-07:002022-05-23T13:52:54.020-07:00Thanks Daaronr:
Yes, we're following up. We&...Thanks Daaronr:<br /><br />Yes, we're following up. We're almost set to run a big study using 80 of the submitted arguments to see what features the effective arguments share in common. Stay tuned!Eric Schwitzgebelhttps://www.blogger.com/profile/16274774112862434865noreply@blogger.comtag:blogger.com,1999:blog-26951738.post-9763424556428590842022-05-23T13:42:55.794-07:002022-05-23T13:42:55.794-07:00Has this been followed up on? It's very intere...Has this been followed up on? It's very interesting, and I'd be eager to see<br /><br />- the data and instruments shared to allow further analysis<br />- it written up in a way that could be read more easily<br />- and maybe peer reviewed. <br /><br /> daaronrhttps://www.blogger.com/profile/00008568796254623426noreply@blogger.comtag:blogger.com,1999:blog-26951738.post-65200133911444999452020-07-08T07:51:09.100-07:002020-07-08T07:51:09.100-07:00I've also used Prolific, but not for any of th...I've also used Prolific, but not for any of the charitable giving studies. Hopefully the results would generalize across different ways of recruiting participants!Eric Schwitzgebelhttps://www.blogger.com/profile/16274774112862434865noreply@blogger.comtag:blogger.com,1999:blog-26951738.post-76271741366883949942020-07-08T02:07:53.071-07:002020-07-08T02:07:53.071-07:00Hi, Eric. Apart from mturk, do you try other platf...Hi, Eric. Apart from mturk, do you try other platforms like Prolific Accademic?Callannoreply@blogger.comtag:blogger.com,1999:blog-26951738.post-80213874843198469112020-07-07T16:16:37.368-07:002020-07-07T16:16:37.368-07:00Very cool, dominicq!
Celso: Yes, it made a differ...Very cool, dominicq!<br /><br />Celso: Yes, it made a difference. In the Singer & Lindauer condition, 24% of those who donated anything donated to Helen Keller International, compared to 6-12% in the other conditions (chi-square [6x2], p < .001).Eric Schwitzgebelhttps://www.blogger.com/profile/16274774112862434865noreply@blogger.comtag:blogger.com,1999:blog-26951738.post-64692171286843833212020-07-07T07:59:37.109-07:002020-07-07T07:59:37.109-07:00This is really interesting (and useful)!
Would yo...This is really interesting (and useful)!<br /><br />Would you mind telling me if the people who were compelled by Singer's argument donated (coherently) to the Hellen Keller Foundation or if they donated to other charities as well? I am working on coherence between beliefs and actions and this piece of information would be REALLY useful. Thanks in advance!celsohttps://www.blogger.com/profile/07179965136061112854noreply@blogger.comtag:blogger.com,1999:blog-26951738.post-8861484210826050042020-07-06T13:09:39.327-07:002020-07-06T13:09:39.327-07:00I was inspired by this post and the winning argume...I was inspired by this post and the winning argument, so I decided to put it into video form: https://www.youtube.com/watch?v=9vehER01Fdg<br /><br />Maybe it's useful to someone with busy friends who might want to watch a 2.5 min video.dominicqnoreply@blogger.comtag:blogger.com,1999:blog-26951738.post-57977947630540971972020-06-30T15:01:37.826-07:002020-06-30T15:01:37.826-07:00Thanks for the continuing comments, folks!
Gerry:...Thanks for the continuing comments, folks!<br /><br />Gerry: Yes, interesting point. Since all five arguments are within statistical chance of each other, I wouldn't make the inference strongly, but it does seems plausibly to be one of the argument's strengths.<br /><br />William: Maybe so! I am actually running another study, with Josh May, that looks at exactly this issue with a set of arguments that differ primarily in the dimension you mention while keeping other factors constant. We had been planning to orally present some of our findings for the first time at conference a few weeks ago, but unfortunately the conference was cancelled. :-(Eric Schwitzgebelhttps://www.blogger.com/profile/16274774112862434865noreply@blogger.comtag:blogger.com,1999:blog-26951738.post-5352681209221272612020-06-30T14:33:25.116-07:002020-06-30T14:33:25.116-07:00(For my entry, click my name on this comment.)
On...(For my entry, click my name on this comment.)<br /><br />One thing I noticed about all five of the Top 5 arguments (though not my entry) is that they all can be interpreted as guilting the reader into donating. That is, there is an unstated implication the reader could draw that the reader would be a bad person if they chose not to donate:<br /><br />Argument #9: After reading this winning argument, the reader might think: "Now if I don't donate the $10 I'd be admitting that I don't value the suffering of children in poor countries even one-thousandth as much as my own child (or someone I know's child). What a terrible person I'd be. I don't want to feel like a bad person so I'll donate."<br /><br />Argument #3: Someone might think: "Practically everyone agrees that giving to charity is good, so if I don't donate the $10 that would make me bad. I don't want to feel like a bad person so I'll donate."<br /><br />Argument #5: "If I take the $10 rather than donate it, I'd be putting my own interest in receiving $10 above the interests of four children who don't want malaria, which would make me a bad person. I don't want to feel like a bad person so I'll donate."<br /><br />Argument $12: "I just read that I should feel good about whether I decide to 'take' or 'give' the $10. And also that I should prioritize helping a large number of people over the value of $10 for myself. So now I'm not sure that I could feel good about 'taking' the money for myself. I don't want to feel guilty over $10 so I'll donate."<br /><br />Argument #14: "'Every single day you have the opportunity to spare a small amount of money to provide a fellow human with the same basic access to food or drinking water – how often have you done this?' Clearly I'd be a bad person if I decided to take $10 that is offered to me rather than give the $10 to provide a fellow human with basic access to food or drinking water. I don't want to feel like a bad person, so I'll donate.<br /><br />My entry didn't do this, and may be that (part of) why it was bad. (Click my name to see my entry.)William Kielyhttps://forum.effectivealtruism.org/posts/TSoNFMaLWxNwEjoAa/study-results-the-most-convincing-argument-for-effective?commentId=dwT4wAixnEHF78GQHnoreply@blogger.comtag:blogger.com,1999:blog-26951738.post-78458450097346791832020-06-29T18:23:33.820-07:002020-06-29T18:23:33.820-07:00Looking at the arguments, the winner (#9) was very...Looking at the arguments, the winner (#9) was very clear and precise in explaining how curing trachoma is fantastic value for your charity dollar. One of the others (#5) hinted at this, but seemed to beat around the bush with thought experiments - the rest had a bunch of ethical arguments about why you should give. (I really didn't examine them all in detail, but in point of fact it's first impressions that matter here anyway.)<br /><br />Seems like your result makes a case for pointing out that howeffective the giver's altruism is. (Of course emotional appeals centering on an individual do something in the same category - but in a different way.)Gerry Quinnhttps://www.blogger.com/profile/04078394659680797175noreply@blogger.comtag:blogger.com,1999:blog-26951738.post-19126858522592578182020-06-29T12:33:43.152-07:002020-06-29T12:33:43.152-07:00Thanks for the continuing comments, folks!
ATab: ...Thanks for the continuing comments, folks!<br /><br />ATab: Just wait til you see a study on arguments for vegetarianism, on which I collaborated with Brad Cokelet and Peter Singer.<br /><br />Victoria: The factors you identify seem plausible. Fiery and I hope to get deeper into exploring such factors in follow-up research.<br /><br />Jim: There was no control condition in Phase 1, unfortunately, but we will hopefully be able to release the full text of all tested arguments at some point.Eric Schwitzgebelhttps://www.blogger.com/profile/16274774112862434865noreply@blogger.comtag:blogger.com,1999:blog-26951738.post-69132552060028104522020-06-29T04:39:27.949-07:002020-06-29T04:39:27.949-07:00It would be interesting to see the text of some of...It would be interesting to see the text of some of the arguments that preformed worse than the control, as well.Jimhttps://www.blogger.com/profile/13757770746129793555noreply@blogger.comtag:blogger.com,1999:blog-26951738.post-17852248425859366192020-06-28T14:14:11.210-07:002020-06-28T14:14:11.210-07:00I think there are two questions being explored thr...I think there are two questions being explored through this contest. First the structure of exchanges between groups for shared social objectives and second, how to engage the morals/ethics of individuals. The winners just understand the dynamics of the structure the best: <br /><br />https://wp.me/p8PxIA-dcVictoria Wilsonhttps://www.blogger.com/profile/04003176403006388842noreply@blogger.comtag:blogger.com,1999:blog-26951738.post-91717451535589968692020-06-27T17:42:03.017-07:002020-06-27T17:42:03.017-07:00You were motivated to do this study by being unabl...You were motivated to do this study by being unable to find philosophical arguments to change ethical behavior. Also your previous work with Rust found that ethicists are no more ethical than other professors. So in both cases, Singer is really screwing with your theories! :) ATabarrokhttps://www.blogger.com/profile/10941701423727165720noreply@blogger.comtag:blogger.com,1999:blog-26951738.post-1053346618888489052020-06-26T12:58:42.410-07:002020-06-26T12:58:42.410-07:00Thanks, Simon! Yes, we plan to make the data publ...Thanks, Simon! Yes, we plan to make the data public, though we're hoping to do some follow-up research and publish the lot in a journal, so we need to be careful about embargo and prepublication issues. Thanks for the link to Liddell and Kruschke. One impression I have after a quick skim is that they're having to seriously hop up the extremes at 1 and 5 to get the models to fail. This fits with my impression that while t-test and ANOVA are pretty robust to violations of normality, extreme violations of normality create problems, and maybe the highly non-normal distributions in their examples are part of the problem. (Please forgive me if a closer read would suggest I'm misinterpreting.) On these data in particular, I would think that dollar amounts *are* metric (even if a 5 point Likert scale might not be). Our data aren't normal, with clustering at $0, $5, and $10 -- but the violation of normality here seems less extreme than in the Liddell cases and I'd think is within the realm of violation of normality where t-tests and ANOVAs aren't too bad, especially given that it's not a close call on the p values. Does this make sense?Eric Schwitzgebelhttps://www.blogger.com/profile/16274774112862434865noreply@blogger.comtag:blogger.com,1999:blog-26951738.post-12570154486166156872020-06-26T02:54:38.838-07:002020-06-26T02:54:38.838-07:00Interesting work, Eric. However, I too have some s...Interesting work, Eric. However, I too have some statistical concerns with treating ordinal data as metric. While means in this case could be “ok” for a “quick and dirty estimation”, they have been shown to conflate BOTH type I and II errors when the data are ordinal. See for instance:<br /><br />Liddell, T. M., & Kruschke, J. K. (2018). Analyzing ordinal data with metric models: What could possibly go wrong?. Journal of Experimental Social Psychology, 79, 328-348.<br /><br />Are you planning to make the data publicly available? Given the topic of your research, I would be happy to give some input if interested. Another thing, that you may also find interesting, could be to model whether some arguments consistently can yield some donations (whereas other arguments potentially more rarely could give higher payoffs).<br /><br />Best of luck with your research.<br /><br />Simon H. Del Pin, PhD Student in cognitive psychology.Simonnoreply@blogger.comtag:blogger.com,1999:blog-26951738.post-14926225575648458392020-06-25T08:41:17.252-07:002020-06-25T08:41:17.252-07:00Thanks for the comments, folks!
Anon Jun 23: I di...Thanks for the comments, folks!<br /><br />Anon Jun 23: I disagree that marginal significance is always inappropriate. You will note that we draw no conclusions based on the statement of marginal significance, saying that it is "not clear" whether there was a difference. Of course to say that there was NO difference (of such-and-such a size) would also be a statistical claim that would require a power analysis or similar. We are quite clear about effect sizes, since we give the means and differences. I think Cohen's d is overrated, if that's what you were hoping to see, given its dependence on SD where SD might be of much less interest than the difference in means. I report the p values of < .001 not as an estimate of effect size, so I'm wondering why you attribute that to me.<br /><br />Anon Jun 24: I think that's quite possible. It's "dry" (or seems to be) relative to a vivid description of what it is like to endure a terrible disease or relative to a gripping narrative, but I doubt that it is completely emotionally neutral. To what extent emotional arousal might be driving these effects is something we hope to explore in follow-up work.<br /><br />D: Several submissions paraphrased or referred to Peter Singer, so it is interesting that he won! My guess is that he has been at this so long that he has a sense of what sorts of arguments are more or less effective on his audience. He and Lindauer also done some empirical testing of their own of the effectiveness of various arguments.Eric Schwitzgebelhttps://www.blogger.com/profile/16274774112862434865noreply@blogger.comtag:blogger.com,1999:blog-26951738.post-40528676257495168372020-06-24T15:09:22.993-07:002020-06-24T15:09:22.993-07:00It's funny-- on reading the introduction to th...It's funny-- on reading the introduction to this, I thought about what I would have written, and figured I would have paraphrased Peter Singer, as I had found one of his arguments about charity very convincing. So it's nice to see he won.Dhttps://www.blogger.com/profile/15141040566131538098noreply@blogger.comtag:blogger.com,1999:blog-26951738.post-2895655114359351362020-06-24T09:20:05.699-07:002020-06-24T09:20:05.699-07:00I wonder how "dry" the winning argument ...I wonder how "dry" the winning argument really is. While it doesn't lay out a full narrative in writing, the reference to one's own child serves to evoke emotionally charged personal attachments. While the egalitarian (or at least more-than-nothing-tarian) argument is doing work to generalize care from the near-and-dear to the charity-receiving stranger, I can't help but think that it is leveraging the power of very un-dry content.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-26951738.post-61433970958902727612020-06-24T08:58:57.745-07:002020-06-24T08:58:57.745-07:00Science Is Testing, as in relativity quantity qual...Science Is Testing, as in relativity quantity quality...Arnoldhttps://www.blogger.com/profile/02580641063222662041noreply@blogger.comtag:blogger.com,1999:blog-26951738.post-19152049260763995952020-06-23T19:46:53.077-07:002020-06-23T19:46:53.077-07:00I would like to caution against some of the statis...I would like to caution against some of the statistical language used in this article. Marginal significance is outdated and inappropriate; likewise, statistical significance is not a measure of degree (i.e., it is not highly significant or moderately significant, etc. - it just is or is not). You seem to be conflating significance with the size of the effect, which should be reported with effect size estimates instead. Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-26951738.post-76677801152906698882020-06-23T18:22:27.070-07:002020-06-23T18:22:27.070-07:00Tim: Right, we’re looking at trying to pull those ...Tim: Right, we’re looking at trying to pull those sorts of things apart.<br /><br />Anon: If I recall, there are peaks at 0 5 and 10, a declining spread over 1–4, and not much from 6-9. So not exactly a normal distribution, but t and ANOVA are fairly robust.Eric Schwitzgebelhttps://www.blogger.com/profile/16274774112862434865noreply@blogger.comtag:blogger.com,1999:blog-26951738.post-75788112038269770352020-06-23T10:26:50.290-07:002020-06-23T10:26:50.290-07:00It would be interesting to know the distribution o...It would be interesting to know the distribution of donations. Did a lot of people give around $3.50, or did 35% give $10 and the rest nothing?Anonhttps://www.blogger.com/profile/03169828144395505195noreply@blogger.comtag:blogger.com,1999:blog-26951738.post-73105352828997585152020-06-23T09:46:30.410-07:002020-06-23T09:46:30.410-07:00Really interesting!
So now can you do a study th...Really interesting! <br /><br />So now can you do a study that compares 'dry argument only' both to 'moving anecdote only' and to 'argument that includes moving anecdote'? Maybe you could find a way to have the 'dry argument' and 'moving anecdote' both be related to one another closely enough so that you could just more or less combine the text of the two to create your 'argument that includes moving anecdote.'Tim O'Keefehttp://tokeefe.gsucreate.orgnoreply@blogger.com