It’s mostly the non-EAers who give $0, it seems. Removing those datapoints would diminish the difference between groups and make EAers look worse than they are—we care about how much total was given, not a random dichotomization like $0 vs non-$0.
I don’t know what we care about. You provided, basically, an exploratory data analysis, but I don’t see a specific, well-defined question that you’re trying to answer.
And, of course, $0 vs non-$0 is not entirely random :-) I’m not arguing for “removing” datapoints, I’m arguing for something like a hierarchical model which won’t give you as single number as “the answer” but will provide better understanding.
I don’t see a specific, well-defined question that you’re trying to answer.
su3su2u1 has accused EAers of hypocrisy in not donating despite a moral philosophy centering around donating; hypocrisy is about actions inconsistent with one’s own claimed beliefs, and on EA’s own aggregative utilitarian premises, total dollars donated are what matter, not anything about the distribution of dollars over people.
Hence, in investigating whether EAers are hypocrites, I must be interested in totals and not internal details of how many are zeros.
(The totals aren’t going to change regardless of whether you model it using mixture or hierarchical or zero-inflated distributions; and as the distribution-free tests say, the EAers do report higher median donations.)
on EA’s own aggregative utilitarian premises, total dollars donated are what matter, not anything about the distribution of dollars over people.
This is a pretty narrow conception of EA. You can be an EA without earning to give. For example, you could carefully choose a career where you directly do good, you could work in advocacy, or you could be a student gaining career capital for later usage.
I’m with Lumifer on this: even though what EAists are supposed to care about is basically total money given (assuming it’s given as effectively as possible), that doesn’t mean that hypocrisy of EAists is a function of total money given.
Having said that, just eyeballing the data seems sufficient to demonstrate that EAists are (at least according to self-report) more generous than non-EAists in this population, and I bet any halfway reasonable analysis will yield the same conclusion.
total dollars donated are what matter, not anything about the distribution of dollars over people.
Really? Under this approach a hundred EAers where 99 donate zero and one donates $1m suffer from less hypocrisy than a hundred EAers each of which donates $100.
And if only total dollars matter, then there is no need for all that statistical machinery, just sum the dollars up. “Median donations”, in particular, mean nothing.
I think you’re mistaken about total dollars. If I were curious about the hypocrisy of EA people, I would look at the percentage which donates nothing, and for the rest I would look at the percentage of income they donate (possibly estimated conditional on various relevant factors like age) and see how it’s different from a comparable non-EA group.
Really? Under this approach a hundred EAers where 99 donate zero and one donates $1m suffer from less hypocrisy than a hundred EAers each of which donates $100.
Yes. If EA manages to get $1m in donations rather than $10k, then MISSION ACCOMPLISHED.
(Of course, in practice we would prefer that $1m to be distributed over all EAers so we could have more confidence that it wasn’t a fluke—perhaps that millionaire will get bored of EA next year. I have more confidence in the OP results because the increased donations are broadly distributed over EAers, and the increase is robust to outliers and not driven by a single donator, as the u-test of medians indicates. But ultimately, it’s the total which is the intrinsic terminal goal, and other stuff is more about instrumental properties like confidence.)
And if only total dollars matter, then there is no need for all that statistical machinery, just sum the dollars up.
Total over all time is what matters, but unfortunately, observations of the future are not yet available or I would simply include those too… The point of the regressions is to get an idea of future totals by looking at the estimates for age and income. Which indicates that EAs both donate total more now, and will also donate more in the future as well, consistent with Scott’s claims. (If, for example, I’d found that EAers were disproportionately old and the estimated coefficient for EA was ~0 when Age was included as a variable, then that would strongly suggest that the final total dollars would not be greater for EAs and so Yvain’s defense would be wrong.)
There are two entirely separate questions. (1) Are EAs mostly hypocrites? (2) Does EA lead to more (and more effective) charitable donations? It’s perfectly reasonable to care more about #2, but it seems fairly clear that the original accusation was about #1.
Glancing at the data, it looks like the median EA at several ages gives 0 as well as the median non-EA. You might want to separate the 0 set from everything else and then answer two questions:
what percentage of EAs/non-EAs donate any money
when they give, how much do EAs give, how much do non-EAs give.
I think this makes more sense then what is happening now- the lines don’t seem to fit the data very well.
It’s mostly the non-EAers who give $0, it seems. Removing those datapoints would diminish the difference between groups and make EAers look worse than they are—we care about how much total was given, not a random dichotomization like $0 vs non-$0.
I don’t know what we care about. You provided, basically, an exploratory data analysis, but I don’t see a specific, well-defined question that you’re trying to answer.
And, of course, $0 vs non-$0 is not entirely random :-) I’m not arguing for “removing” datapoints, I’m arguing for something like a hierarchical model which won’t give you as single number as “the answer” but will provide better understanding.
su3su2u1 has accused EAers of hypocrisy in not donating despite a moral philosophy centering around donating; hypocrisy is about actions inconsistent with one’s own claimed beliefs, and on EA’s own aggregative utilitarian premises, total dollars donated are what matter, not anything about the distribution of dollars over people.
Hence, in investigating whether EAers are hypocrites, I must be interested in totals and not internal details of how many are zeros.
(The totals aren’t going to change regardless of whether you model it using mixture or hierarchical or zero-inflated distributions; and as the distribution-free tests say, the EAers do report higher median donations.)
This is a pretty narrow conception of EA. You can be an EA without earning to give. For example, you could carefully choose a career where you directly do good, you could work in advocacy, or you could be a student gaining career capital for later usage.
I’m with Lumifer on this: even though what EAists are supposed to care about is basically total money given (assuming it’s given as effectively as possible), that doesn’t mean that hypocrisy of EAists is a function of total money given.
Having said that, just eyeballing the data seems sufficient to demonstrate that EAists are (at least according to self-report) more generous than non-EAists in this population, and I bet any halfway reasonable analysis will yield the same conclusion.
Really? Under this approach a hundred EAers where 99 donate zero and one donates $1m suffer from less hypocrisy than a hundred EAers each of which donates $100.
And if only total dollars matter, then there is no need for all that statistical machinery, just sum the dollars up. “Median donations”, in particular, mean nothing.
I think you’re mistaken about total dollars. If I were curious about the hypocrisy of EA people, I would look at the percentage which donates nothing, and for the rest I would look at the percentage of income they donate (possibly estimated conditional on various relevant factors like age) and see how it’s different from a comparable non-EA group.
Yes. If EA manages to get $1m in donations rather than $10k, then MISSION ACCOMPLISHED.
(Of course, in practice we would prefer that $1m to be distributed over all EAers so we could have more confidence that it wasn’t a fluke—perhaps that millionaire will get bored of EA next year. I have more confidence in the OP results because the increased donations are broadly distributed over EAers, and the increase is robust to outliers and not driven by a single donator, as the u-test of medians indicates. But ultimately, it’s the total which is the intrinsic terminal goal, and other stuff is more about instrumental properties like confidence.)
Total over all time is what matters, but unfortunately, observations of the future are not yet available or I would simply include those too… The point of the regressions is to get an idea of future totals by looking at the estimates for age and income. Which indicates that EAs both donate total more now, and will also donate more in the future as well, consistent with Scott’s claims. (If, for example, I’d found that EAers were disproportionately old and the estimated coefficient for EA was ~0 when Age was included as a variable, then that would strongly suggest that the final total dollars would not be greater for EAs and so Yvain’s defense would be wrong.)
There are two entirely separate questions. (1) Are EAs mostly hypocrites? (2) Does EA lead to more (and more effective) charitable donations? It’s perfectly reasonable to care more about #2, but it seems fairly clear that the original accusation was about #1.
Glancing at the data, it looks like the median EA at several ages gives 0 as well as the median non-EA. You might want to separate the 0 set from everything else and then answer two questions:
what percentage of EAs/non-EAs donate any money when they give, how much do EAs give, how much do non-EAs give.
I think this makes more sense then what is happening now- the lines don’t seem to fit the data very well.