EA Survey 2017 Series: Demographics II

By: Katie Gertsch and Tee Barnett


The annual EA Survey is a volunteer-led project of Rethink Charity that has become a benchmark for better understanding the EA community. This is the fifth article in our multi-part EA Survey 2017 Series. You can find supporting documents at the bottom of this post, including our previous piece on community demographics, prior EA surveys, and an up-to-date list of articles in the EA Survey 2017 Series. Get notified of the latest posts in this series by signing up here.


This article brings EA demographics back by popular demand. As in, demand for the metrics not covered in the previous post. We hope you enjoy this second look.


The survey respondents identified as white by a wide majority. Among the 1,069 who self-identified regarding race, 88.9% identified as white, 0.7% identified as black, 3.3% identified as hispanic, 7.0% identified as asian, and 621 respondents preferred not to answer the question. It was possible to identify with as many races as one wanted, but only 3.59% answered ‘Yes’ to self-identify as more than one race, and only one person (0.09%) identified with three races.


While diversity comes in many forms, especially in a definitional sense, EA is unlikely to be characterized as racially diverse according to this survey. There may be considerable margin for error in these findings, not the least because such a large proportion of respondents did not answer. But the trope of EA being a predominantly white (89%) and male (70.1%) community, however, is not likely to fade anytime soon without directed effort.


A longitudinal analysis of the community’s racial composition cannot be conducted because no data on race was gathered in the 2015 survey.


Want to contribute more to this discussion? We recommend joining the Diversity & Inclusion in EA group on Facebook.

Race and Geographic Location

A crosstab of declared racial identity according to location revealed a vast white majority across the top five EA hubs around the world. New York City emerged as the most racially diverse EA hub in the community. This was statistically significant with p = 0.02, but it’s not clear how much we can read into this.


Left-leaning EAs composed 64.8% of respondents, while ‘Centre’ (8.1%), ‘Centre Right’ and ‘Right’ (3.3%) accounted for a considerably smaller portion of the sample. Libertarian EAs constitute a sizeable proportion of the sample (8.7%)  a small group (6%) explicitly chose not to answer, and 9% refused to identify with any of the political spectrum. These percentages do not include the 785 people who took the survey but did not answer this question.


Data on political preference was collected but not published in the 2015 EA Survey report, allowing us in 2017 to present longitudinal data on community-wide shifts in political orientation.

From 2015 to 2017, the survey indicates a slight shift away from the political left in the EA community. The tables above show 27.27% of the 2015 ‘Left’ moved to the ‘Centre Left’, and 5.88% of the ‘Centre Left’ went “Centre”. There was also some polarization, as 46.15% of the “Centre” moved “Centre Left”.


Want to contribute more to this discussion? We recommend joining the Effective Altruists Discuss Politics group on Facebook.

Politics and cause area preference

When looking at the relationship between politics and other areas, we broke down political orientation into whether someone identified with the “Left” (i.e. they said they were “Left” or “Centre Left”) or did not identify with the left (i.e., they picked a different option like, “Centre”, “Centre Right”, “Right”, “Libertarian”). “Other” and “Prefer not to answer” were dropped from this variable. We found 682 respondents who were associated with a left-leaning position (left), 212 respondents who were not associated with a left-leaning position (non-left), and 943 people with no position.


A crosstab of political orientation and cause area preference revealed that individuals on the left are more likely to be interested in politics (28% of people on the left rate politics as a top or "near top" cause, compared to 22% of people not on the left), poverty (78% of people on the left rate poverty as a top or "near top" cause, compared to 72% of people not on the left), animal welfare (41% of people on the left say animal welfare is top or near top compared to only 28% of the non-left), and environmentalism (42% of people on the left say environmentalism is top or "near top". compared to 21% of non-left).


Conversely, people on the left are less likely to care about AI (42% of people on the left rate AI as top or "near top" compared to 47% of people not on the left).

Politics and geographic location

Despite the San Francisco Bay Area being anecdotally associated with libertarians, it had the highest amount of people identifying with the left, with 82.9% of Bay Area respondents. Of the other five largest EA cities, London was 80.85% left, Oxford  was 76.92% left and Boston was 73.53% left, and New York City was 63.64% left. However, despite these percentages of left appearing quite different, there was no statistically significant trend in left vs. non-left that we could pick up in our data.


Politics and dietary habits

Results show a significant difference according to political affiliation, where 48.9% on the left identified as vegetarian or vegan, while only 29% on the non-left did.


This makes sense in the light of the above, looking at politics and cause area preference, where we see a significantly greater proportion (41%) of people on the left putting a high priority on animal welfare, compared to a smaller proportion sharing that level of priority from those on the non-left (28%).

Age and cause area preference

Using the median age of 27 as a dividing point, those below the median  grouped as ‘younger’ and those above the median as ‘older’, we compared cause area preference in these two groups. The group younger than the median age showed a preference for AI (53.1% compared to 37.9% of older people) and less of a preference for poverty (72% vs. 78% of older people).

Employment status

Employment status responses were lead by for-profit work (43.7%) and non-profit organizations (17.0%). There were similar numbers for self-employed (9.5%) and academics work (9.6%). Unemployed respondents made up 7.7%, while 6.8% reported working for a government entity, and 1.2% were homemakers. Those who are financially independent, through savings, passive income or a providing partner accounted for 4.6%.

Field of study

Respondents were allowed to select more than one field of study. Most popular fields among EA’s, by a significant margin, proved to be computer science (18.9%) and maths (16.1%). Following that, philosophy (9.9%), other sciences (9.2%), social sciences (8.6%) and economics (8.4%). Less often chosen were the fields of humanities (7.1%), engineering (6.9%), physics (6.7%) and finally medicine (2.8%).

Year joined EA

Pardoning 2017 for being the current year, the last few years appear to have been strong for EA recruitment, though there may also be a survivorship bias with EAs who joined in previous years no longer identifying with EA or take the EA survey. Post-2013, we see double-digit percentage growth in the number self-identified EAs joining the community.


Some additional metrics on  EA movement growth from Peter Hurford and Joey Savoie is available in “Is EA Growing? Some EA Growth Metrics for 2017”.


Post written by Katie Gertsch and Tee Barnett, with edits and analysis from Peter Hurford.


A special thanks to Ellen McGeoch, Peter Hurford, and Tom Ash for leading and coordinating the 2017 EA Survey. Additional acknowledgements include: Michael Sadowsky and Gina Stuessy for their contribution to the construction and distribution of the survey, Peter Hurford and Michael Sadowsky for conducting the data analysis, and our volunteers who assisted with beta testing and reporting: Heather Adams, Mario Beraha, Jackie Burhans, and Nick Yeretsian.


Thanks once again to Ellen McGeoch for her presentation of the 2017 EA Survey results at EA Global San Francisco.


We would also like to express our appreciation to the Centre for Effective Altruism, Scott Alexander via SlateStarCodex, 80,000 Hours, EA London, and Animal Charity Evaluators for their assistance in distributing the survey. Thanks also to everyone who took and shared the survey.

Supporting Documents

EA Survey 2017 Series Articles

I - Distribution and Analysis Methodology

II - Community Demographics & Beliefs

III - Cause Area Preferences

IV - Donation Data

V - Demographics II

VI - Qualitative Comments Summary

VII - Have EA Priorities Changed Over Time?

Please note: this section will be continually updated as new posts are published. All 2017 EA Survey posts will be compiled into a single report at the end of this publishing cycle

Prior EA Surveys conducted by Rethink Charity (formerly .impact)

The 2015 Survey of Effective Altruists: Results and Analysis

The 2014 Survey of Effective Altruists: Results and Analysis

Comments (14)

Comment author: Vincent-Soderberg 19 September 2017 11:57:52AM 4 points [-]

I'm curious about disabilities demographics in EA (both mental health and physical health). As far as i can remember, the EA survey never asked that, but it seems like something relevant to ask.

It might not be as relevant as i think though.

Comment author: Kelly_Witwicki 03 October 2017 03:35:19AM 3 points [-]

Seeing how e.g. depression correlates with cause area preferences would be interesting.

Comment author: concerned_ 18 September 2017 09:14:52PM 4 points [-]

I'd be curious to see how "year joined" correlates with cause area preference.

Comment author: Peter_Hurford  (EA Profile) 18 September 2017 11:01:47PM 5 points [-]

We actually have a post on that coming up soon, looking at how cause area preferences change over time!

Comment author: Kelly_Witwicki 03 October 2017 03:29:49AM 2 points [-]

Mentioned this to Tee -- I would love to see both (1) turnover data, and (2) previous (pre-EA) work with or donations to non-profits/other previous altruistic pursuits.

For the latter, the survey could maybe ask how much volunteer time, career hours, and money respondents previously put into charity, maybe in the last year and/or five and/or lifetime before finding EA. It could also offer categories for the charities' cause areas.

Relatedly, whether the respondent was vegetarian or vegan before finding EA would be interesting, and/or some scale about how important they thought issues faced by nonhumans were. Maybe for each of several cause areas (not just the big three) a scale of how important (not necessarily relatively, just "not at all important" to "very important"?) they thought the issue was before coming to EA.

Comment author: Kelly_Witwicki 03 October 2017 03:36:34AM 3 points [-]

I would also be interested to see a question for what value people assign to life on Earth at present.

I imagine for instance this is much higher, and may even be the difference between highly positive and highly negative, for EAs who are most concerned with x-risk as compared to those more interested in animal farming. (And more obviously, s-risk, but it would still be interesting to quantify the difference, if just in terms of e.g. "highly negative" to "highly positive".)

Comment author: Rick 19 September 2017 06:17:31PM 1 point [-]

Are there any theories about what is driving the really high non response rate for race? Or any cross tabs about what groups or locations are more likely to have a non response for race? Racial demographics in EA is an important topic, and it's a shame that we can't get better data on it.

Comment author: Peter_Hurford  (EA Profile) 19 September 2017 11:21:34PM *  4 points [-]

I can see how the non-response rate looks alarming and I definitely owe some context for that.

One thing we tried this year was a separate donations only survey, where people only reported their donations and a few other questions. Race was not on this slimmer survey. 554 did not answer this race question because they were never asked it.

Another source of apparent non-response is the fact that we asked people Yes or No for four different races (White, Black, Asian, and Hispanic). It looks like some people checked "Yes" for one race but did not explicitly check "No" for the others. This accounts for another 120 people.

Combining these first two reasons, there are only 67 people genuinely ignoring the race question. You then have to account for survey fatigue, where people answer some of the questions at the beginning of the survey, but then get bored, busy, or distracted and quit the survey without answering the rest of the questions. Given that race was at the bottom of the seventh page of the survey, this could be acute. I couldn't find anyone who neglected to answer the race question but did answer a question after the race question, so it looks like these three factors may fully account for all non-response.

Comment author: Bernadette_Young 20 September 2017 01:11:29PM 4 points [-]

That's still a very important point that doesn't seem to have been made in the analysis here: the demographic questions were not included in the questions put to all respondents. Since there are good reasons to think that people taking the "full" and "donations only" survey will differ systematically (e.g. more likely to have been involved with EA for longer). If the non responses are not random that's an important caveat on all these findings and very much limits any comparisons that can be done over time. I can't seem to see it discussed in the post?

Comment author: Peter_Hurford  (EA Profile) 20 September 2017 07:50:10PM *  0 points [-]

Yeah. I personally think that offering the donations only survey was a bad idea for the reason that you said and a few other reasons.

Even if everyone took the full survey, the non-response would still be pretty non-random -- you still have to have the tenacity to persist to page seven, which I imagine correlates with being more involved in EA and you also have to have taken the survey in the first place, which we also know is not random. It would have been nice to not make this worse, though.

Comment author: Rick 19 September 2017 06:56:23PM 0 points [-]

Sorry to fixate on this, but I've just never seen non response rates this high before - 10% is high in most cases of surveys, 40% is absurd. Like, yes you always have groups who feel like the answers don't accurately capture their reality, but given that you did allow for multiracial answers (and given the homogeneity of EA from a race stand point), this usually would be only a very small fraction of respondents. There's also the population that, for lack of a better term, "don't believe in race" and never answer this question, but given how small this population is in general, unless an absurdly high number of them are EAs then this should also only be a very small fraction.

I really, really hope this isn't the explanation, but I could see at least some of these answers coming from the perspective of "I don't think race is a problem in EA, and people should stop asking about it, so I'll just not answer at all as a protest or something." As someone who sees data collection as sacred, I would be appalled by this - so please, someone, for the sake of my sanity explain what could possibly drive a 40% non response rate that is not this.

Comment author: ELW 19 September 2017 10:09:44PM 1 point [-]

I wondered if the oddly high portion of refusal to answer was ideological, too. I hope this isn't the case and inclined to think it's unlikely; though there seem to be some EAs who are wary of questions regarding this kind of diversity, as they reject it is something to be tackled within the movement, I would not have thought that proportion (or rather the proportion who hold the view to the point of refusing the data altogether) would be as large as this.

It feels rather optimistic to suggest this is an issue of categorisation e.g. though the answers on race were not exclusive, which would be obviously problematic, most race sections on this type of survey that I'm used to have a more fine-grained response options. However, it seems even if this were an issue it ought not cause problems for such a large proportion of respondents.

It's maybe worth noting the comparative proportion of respondents who did not answer political leanings (42.7%). If nothing else I think the number of people who refused response on race would be more bizarre/worrying if there was a very high response rate on everything else. My first thought was that refusal to answer on political leanings is likely idealogical (wariness of EA being especially associated with any specific political position) but on second thoughts I wonder if this is more likely to be a category problem? (It may be that people are reluctant to select "other" because their political stance is subsumed within left/right/centre etc, but they feel it is not well described by these options...? However, I'm not confident how likely this is and do not have a background in the intricacies of data gathering - unlike, I assume, those and ReThink who put this together.)

Comment author: Peter_Hurford  (EA Profile) 19 September 2017 11:22:23PM 1 point [-]

The answer looks to be pretty simple and unimportant, as I explain in this comment.

Comment author: Rick 20 September 2017 02:31:42AM 0 points [-]

Oh thank goodness, I am glad it is nothing worrying. Thank you for clarifying!