EA Survey 2018 Series: How welcoming is EA?

Summary

  • A ma­jor­ity of EAs feel that the EA move­ment is wel­com­ing.

  • New EAs rank the move­ment as more wel­com­ing than long-time EAs.

  • No statis­ti­cally sig­nifi­cant trends in wel­come­ness by race, ed­u­ca­tion, or re­li­gion could be iden­ti­fied.

  • Selec­tion bias, where those who find the EA com­mu­nity par­tic­u­larly un­wel­com­ing are sys­tem­at­i­cally less likely to fill out the EA sur­vey, makes it harder to prop­erly in­ter­pret these results

In this post, we ex­plore which EAs find the EA move­ment wel­com­ing. Keep­ing the effec­tive al­tru­ist move­ment wel­com­ing is a guid­ing prin­ci­ple for many EAs. The EA Sur­vey al­lows us to see how EAs re­sponded to the ques­tion of how wel­com­ing is EA on a 5-point scale from “Very Un­wel­com­ing” to “Very Wel­com­ing”.

A ma­jor­ity of re­spon­dents in the sur­vey find the EA move­ment wel­com­ing or very wel­com­ing. This may not be very sur­pris­ing since this is from a sam­ple of peo­ple who iden­tify as EAs and are en­gaged enough to take the sur­vey. On the mar­gins are a small per­centage (6%) of EAs who be­lieve EA is un­wel­com­ing, and slightly more than a quar­ter of EAs who find the move­ment very wel­com­ing.

N.B. images can be viewed in full size if opened in a new tab.

Time in EA

Note the Y-axis: Wel­come­ness scale from 1-5(very un­wel­com­ing to very wel­com­ing) has been bounded for illus­tra­tive pur­poses.

Rat­ings of how wel­com­ing EA is are higher among EAs that joined more re­cently. As there were far fewer EAs in the sur­vey who had heard of EA be­fore 2013, group­ing these co­horts more evenly shows a lin­ear trend. Those EAs that joined many years ago and are still part of the move­ment are less likely to say it is very wel­com­ing than those who have only re­cently joined. A Kruskal-Wal­lis H test showed that there was a statis­ti­cally sig­nifi­cant differ­ence in wel­come­ness be­tween the year groups, χ2(2) = 29.295, p = 0.0001), sug­gest­ing a differ­ence be­tween at least two of the groups, but we can­not know from this test which groups.

Separat­ing EAs into just two roughly even groups by year of join­ing, we see a sig­nifi­cant differ­ence in the mean wel­come­ness score be­tween those who joined in 2015 or be­fore, and those who joined from 2016-2018. Of course, from this we can­not in­fer that the move­ment was any more or less wel­com­ing in the past when these EAs joined.

Age

The youngest group of EAs (18-22) ap­pear to have a higher mean wel­come­ness score, com­pared to older EAs. The differ­ences be­tween age groups stem from the youngest co­hort be­ing more likely to rate EA as “very wel­com­ing” than other groups, but older groups be­ing more likely to rate EA as only “wel­com­ing” rather than more of them rat­ing it as “un­wel­com­ing”. The 35+ co­hort is more likely than the oth­ers to be neu­tral on the is­sue. A Kruskal-Wal­lis H test showed that there was a statis­ti­cally sig­nifi­cant differ­ence in wel­come­ness be­tween the age groups, χ2(2) = 18.002, p = 0.0012, sug­gest­ing a differ­ence be­tween at least two of the groups.

As we noted el­se­where, the chang­ing age de­mo­graph­ics of the move­ment are likely driven by sur­vey re­spon­dents who are already in the effec­tive al­tru­ism move­ment ag­ing, rather than older peo­ple join­ing effec­tive al­tru­ism. So based on the above time-in-EA effect, we would ex­pect older EAs to view the move­ment as less wel­com­ing if they are also vet­eran EAs. Among re­cent EAs there are no sig­nifi­cant differ­ences in the mean wel­come­ness score be­tween the age groups ( Kruskal-Wal­lis H test χ2(2) = 4.744, p = 0.3145). How­ever, among EAs from ear­lier years there ap­pears to be greater vari­a­tion, but no clear trends. This may sug­gest that be­ing very new to EA has a strong pos­i­tive effect that pre­cludes the knowl­edge/​ex­pe­riences that are nec­es­sary for differ­ent age effects to take place.

Cause Selections

Those EAs with a top pri­or­ity cause of Meta Char­i­ties, AI Risk, and other Long-Term Fu­ture causes find EA more wel­com­ing than those who pri­ori­tise other cause ar­eas. A huge ma­jor­ity (92.67%) of the few re­spon­dents who pri­ori­tise Nu­clear Se­cu­rity (n=17) find EA wel­com­ing, but also are the least likely to be neu­tral. Those who pri­ori­tise Men­tal Health (n=28) were the least likely to find EA very wel­com­ing and the most likely to find EA un­wel­com­ing. A Kruskal-Wal­lis H test showed that there was a statis­ti­cally sig­nifi­cant differ­ence in wel­come­ness be­tween the Top Pri­or­ity Cause groups, χ2(2) = 27.824, p =0.0019, sug­gest­ing a differ­ence be­tween at least two of the groups.

This at first glance seems to run counter to the time-in-EA effect as those pri­ori­tis­ing Meta Char­i­ties and AI Risk tend to be long-time EAs. How­ever, those pri­ori­tis­ing An­i­mal Welfare or Global Poverty also tend to be long-time EAs but have lower wel­com­ness scores, and those pri­ori­tis­ing Nu­clear Se­cu­rity are among the newest EAs and also see the move­ment as very wel­com­ing.

First Heard of EA

One might imag­ine the source from which one learns about EA to af­fect first im­pres­sions of wel­come­ness. More in-per­son types such as per­sonal con­tact, lo­cal groups, or EA Global might be as­so­ci­ated with higher wel­come­ness than more dis­tant im­per­sonal on­line sources like books, the EA Face­book page or the EA Fo­rum. How­ever, there is no ap­par­ent per­sonal/​on­line di­vi­sion. EAs that first heard of EA via REG/​EAF/​FRI/​the Swiss group have the high­est av­er­age wel­come­ness rat­ing, fol­lowed closely by The Life You Can Save or­gani­sa­tion. Those com­ing via An­i­mal Char­ity Eval­u­a­tors feel EA is the least wel­com­ing. How­ever, a Kruskal-Wal­lis H test showed that there was not a statis­ti­cally sig­nifi­cant differ­ence in wel­come­ness be­tween the First Heard groups, χ2(2) = 14.250, p = 0.7126. Th­ese differ­ences are nei­ther sig­nifi­cant nor very sub­stan­tial, with the ex­cep­tion of those hear­ing of EA via ACE which may re­flect the lower pri­or­ity of An­i­mal Welfare/​Rights in EA, the or­gani­sa­tion’s in­volve­ment in both EA and tra­di­tional an­i­mal ac­tivist com­mu­ni­ties, and that this only com­prises a small num­ber of sur­vey re­spon­dents (n=17).

Coun­try/​City

Swiss EAs have a high mean wel­come­ness score, in con­trast to the low mean of neigh­bour­ing Ger­man EAs who are among the most likely to find the move­ment un­wel­com­ing. EAs from the UK and USA, who make up a plu­ral­ity of the move­ment, are very similar in how they view the wel­come­ness of EA. A Kruskal-Wal­lis H test showed that there was a statis­ti­cally sig­nifi­cant differ­ence in wel­come­ness be­tween the coun­try groups, χ2(2) = 28.209, p =0.0004, sug­gest­ing a differ­ence be­tween at least two of the groups. There is no ap­par­ent trend be­tween coun­tries, av­er­age time-in-EA, and wel­come­ness.

22% of Ber­lin EAs see EA as un­wel­com­ing or very un­wel­com­ing com­pared to only 4% of Melbourne EAs. 32% of Melbourne EAs view the move­ment as very wel­com­ing. EAs in Lon­don and the San Fran­cisco Bay Area, which are the most com­monly re­ported lo­ca­tions, have rel­a­tively lower mean wel­come­ness rat­ings. A Kruskal-Wal­lis H test showed that there was a statis­ti­cally sig­nifi­cant differ­ence in wel­come­ness be­tween the city groups, χ2(2) = 24.185, p =0.0011, sug­gest­ing a differ­ence be­tween at least two of the groups.

Politics

We have shown el­se­where that most effec­tive al­tru­ists are on the left/​cen­ter-left, and there is in­ter­est in how Effec­tive Altru­ists can be wel­com­ing to con­ser­va­tives. There are too few EAs on the “Right” (n=9) to say any­thing mean­ingful about how wel­com­ing they rated EA, how­ever, their ab­sence in raw num­bers from the sur­vey is per­haps more tel­ling. Al­most no Liber­tar­ian EAs (0.65%) rate the move­ment as very un­wel­com­ing, while 6.67% of those on the Cen­ter-Right and 9.06% of Left EAs rate the move­ment as un­wel­com­ing or very un­wel­com­ing. How­ever, a Kruskal-Wal­lis H test showed that there was not a statis­ti­cally sig­nifi­cant differ­ence in wel­come­ness be­tween at least two of the poli­tics groups, χ2(2) = 5.958, p =0.3103.

Group Membership

Mem­bers of lo­cal EA groups are sig­nifi­cantly more likely to rate EA as wel­com­ing than non-mem­bers, as are mem­bers of the EA Face­book page. There is not a sig­nifi­cant differ­ence be­tween Face­book and non-Face­book mem­bers among Lo­cal Group mem­bers, though there is among non-lo­cal group mem­bers. Mak­ing EA groups more wel­com­ing may then seem a use­ful way to keep the EA move­ment open.

The mem­ber­ship in­volve­ment clusters we iden­ti­fied in a pre­vi­ous post show that the two groups (clusters 4 and 5) with 100% lo­cal group mem­ber­ship see EA as the most wel­com­ing. This makes sense given the differ­ences iden­ti­fied above. The un­in­volved group (cluster 1) are un­sur­pris­ingly the least likely to view EA as wel­com­ing. A Kruskal-Wal­lis H test showed that there was a statis­ti­cally sig­nifi­cant differ­ence in wel­come­ness be­tween the in­volve­ment groups, χ2(2) = 61.070, p =0.0001, sug­gest­ing a differ­ence be­tween at least two of the groups.

Note the Y-axis: Wel­come­ness scale from 1-5 has been bounded for illus­tra­tive pur­poses.

Gender

There ap­pears to be ev­i­dence of a gen­der differ­ence, with women rank­ing EA in 2018 as less wel­com­ing on av­er­age than men do. This differ­ence is statis­ti­cally sig­nifi­cant, how­ever, it seems to stem mainly from more men rat­ing EA as very wel­com­ing, and more women be­ing neu­tral, rather than sub­stan­tial differ­ences in the be­lief that EA is un­wel­com­ing. The ap­par­ent differ­ence may also partly stem from women be­ing less likely to be in­volved across a wide range of di­men­sions in EA, lead­ing to more neu­tral rat­ings. Re­gard­ing other de­mo­graphic mea­sures, there were no statis­ti­cally sig­nifi­cant trends in wel­come­ness by race, ed­u­ca­tion or re­li­gion. This doesn’t mean that these trends don’t ex­ist, just that we couldn’t iden­tify them with the data that we have.

Career

Those re­spon­dents that in­di­cated they ex­pect to shift their ca­reer based on EA are more likely to view EA in 2018 as wel­com­ing than those who don’t, how­ever, there is no differ­ence be­tween those who have already changed ca­reer based on EA and those who haven’t.

8.40% of those in Direct Char­ity or Non-Profit jobs view the move­ment as un­wel­com­ing or very un­wel­com­ing, com­pared to ~5% for other ca­reer groups. Those in Re­search ca­reers are the most likely to rate the move­ment as wel­com­ing. A Kruskal-Wal­lis H test showed that there was a statis­ti­cally sig­nifi­cant differ­ence in wel­come­ness be­tween the ca­reer types, χ2(2) = 10.463, p = 0.0150, sug­gest­ing a differ­ence be­tween at least two of the groups.

Regression

Us­ing an or­dered re­gres­sion model, these pre­dic­tors on the EA wel­come­ness scale only ex­plained 2% to 3% of the vari­a­tion. This means al­most all of the differ­ences in peo­ple’s per­cep­tions of the wel­come­ness of EA come from fac­tors that aren’t rep­re­sented in the model or data we cap­tured in the sur­vey. A sep­a­rate model with Top Pri­or­ity cause was made due to the rel­a­tively large n of re­spon­dents who offered no or mul­ti­ple top causes and were ex­cluded. Over­all, the effect sizes ap­pear to be quite small and many of the sig­nifi­cant group differ­ences sug­gested above are not sig­nifi­cant so ap­pear to be con­founded rather than true as­so­ci­a­tions. It sug­gests statis­ti­cally sig­nifi­cant in­creases in wel­come­ness among those who have joined EA more re­cently, are men, are Lo­cal EA Group mem­bers, or ex­pect to shift their ca­reers be­cause of EA (though con­founded by top pri­or­ity cause it ap­pears). Notably, age group is not sig­nifi­cant. For rea­sons of space and sim­plic­ity, Coun­try and Top Pri­or­ity Cause are pre­sented in the table be­low as the most pop­u­lar re­sponse cat­e­gory (USA and Global Poverty) in com­par­i­son to all oth­ers. To view a re­gres­sion table with the cat­e­gories de­mar­cated, click here. In Models 3 and 4, there are no sig­nifi­cant differ­ences be­tween any sin­gle top pri­or­ity cause when each is com­pared to all the oth­ers com­bined. Similarly, while the USA and Ger­many ap­pear sig­nifi­cantly differ­ent when each are com­pared to the com­bined other coun­tries, this is not sig­nifi­cant once top pri­or­ity cause is con­trol­led for. How­ever, in Model 6 there is a sig­nifi­cant (p<0.05) pos­i­tive differ­ence in wel­come­ness be­tween AI Risk and Global Poverty, and be­tween Meta Char­i­ties and Global Poverty.

Selec­tion Bias

One is­sue when in­ter­pret­ing these re­sults is that there is a strong po­ten­tial for our find­ings to be un­der­mined by se­lec­tion bias. We can only col­lect data from those peo­ple we can see—those who find the EA com­mu­nity so un­wel­com­ing that they do not fill out the EA Sur­vey at all will not be in­cluded in our re­sults and we thus can­not learn from them. This could mean that we miss im­por­tant trends that would be no­ticed if we some­how had a sam­ple that was perfectly rep­re­sen­ta­tive of all cur­rent and prior EAs. Mea­sur­ing the EA-re­lated sen­ti­ment among po­ten­tial EAs and/​or peo­ple who left EA was un­for­tu­nately im­pos­si­ble with the main sur­vey and would re­quire ac­tively reach­ing out to these highly dis­persed groups. There­fore, we can­not here make claims about how wel­com­ing the move­ment is to those out­side it, and what poli­cies might be best en­acted to im­prove this. The rel­a­tive low num­ber of some groups or cat­e­gories may be sug­ges­tive of this, but not defini­tive. In­stead, we can offer in­sights into which groups are rep­re­sented in EA, even if they haven’t found it very wel­com­ing, and which groups feel es­pe­cially wel­come.

Mul­ti­ple comparisons

Given the num­ber of hy­poth­e­sis tests we do in this piece, we may run into prob­lems of mul­ti­ple com­par­i­sons, so it is also in­for­ma­tive to do Bon­fer­roni cor­rec­tion—that is count the num­ber of hy­poth­e­sis tests (both Kruskal-Wal­lis and T-tests) in the anal­y­sis and note whether any p-val­ues sig­nifi­cant at the p<0.05 level re­main sig­nifi­cant at the p<0.05/​(N of tested hy­pothe­ses) level. We ran 17 such tests (in­clud­ing tests done but not in­cluded in this post), mak­ing our ad­justed thresh­old p<.0029. Of all the re­sults we found here, only the find­ing of wel­come­ness by ca­reer type be­comes statis­ti­cally in­signifi­cant un­der this more stringent crite­rion.

Conclusion

The vast ma­jor­ity of EAs feel that the move­ment is wel­com­ing and the con­tinued growth of the move­ment is a pos­i­tive sign. Wel­come­ness-re­lated differ­ences be­tween var­i­ous groups tend to be minor, so no ma­jor cluster ap­pears to be sub­stan­tially ex­cluded in com­par­i­son with the baseline rat­ing. How­ever, we do not have con­fi­dent es­ti­mates of at­tri­tion rates from the data available here to dis­cuss any se­lec­tion or sur­vivor­ship bias. Among those still in the move­ment, we can dis­cern some differ­ences in terms of wel­come­ness rat­ings.

EAs in lo­cal groups, in par­tic­u­lar, view the move­ment as more wel­com­ing than those not in lo­cal groups, though there are also po­ten­tial se­lec­tion effects at work here. That EAs who have been around the longest tend to view the move­ment as less wel­com­ing is in­ter­est­ing, as it may sug­gest the ini­tial pos­i­tive bump in wel­come­ness rat­ings among new EAs could sub­side once they learn or ex­pe­rience more and give way to the other differ­ences high­lighted here.

Up­dates and Corrections

Upon helpful feed­back from Greg Lewis, ini­tially pre­sented chi-squared tests have been re­placed with Kruskal-Wal­lis Test re­sults. We have also added a dis­cus­sion of Selec­tion Bias and ad­just­ment for mul­ti­ple com­par­i­sons.

Text was added to offer more dis­cus­sion of se­lec­tion bias, re­gres­sion de­sign, and is­sues of mul­ti­ple com­par­i­sons.

We had pre­vi­ously con­cluded that “EAs pri­ori­tis­ing Meta Char­i­ties, and Long-Term Fu­ture causes are more likely to find EA wel­com­ing.” Upon fur­ther re­flec­tion on feed­back from Greg, we have de­cided to no longer em­pha­size this con­clu­sion. In fact, we can only con­clude that EAs pri­ori­tis­ing Meta Char­i­ties, and Re­duc­ing Risks from AI are more likely to find EA wel­com­ing rel­a­tive to EAs pri­ori­tis­ing Global Poverty. This still seems im­por­tant as a plu­ral­ity of EAs rank Global Poverty as their top cause, but the pre­vi­ous phras­ing was not ac­cu­rate.

Credits

This post was writ­ten and with anal­y­sis by Neil Dul­laghan, with con­tri­bu­tions from David Moss.

Thanks to Peter Hur­ford, Luisa Ro­driguez, and Marcin Kowrygo for com­ment and Kim Cud­ding­ton for data vi­su­al­i­sa­tion as­sis­tance.

The an­nual EA Sur­vey is a pro­ject of Re­think Char­ity with anal­y­sis and com­men­tary from re­searchers at Re­think Pri­ori­ties.

Sup­port­ing Documents

Other ar­ti­cles in the 2018 EA Sur­vey Series:

I—Com­mu­nity De­mo­graph­ics & Characteristics

II—Distri­bu­tion & Anal­y­sis Methodology

III—How do peo­ple get in­volved in EA?

IV—Sub­scribers and Identifiers

V—Dona­tion Data

VI—Cause Selection

VII- EA Group Membership

VIII- Where Peo­ple First Hear About EA and Higher Levels of Involvement

IX- Geo­graphic Differ­ences in EA

XI- How Long Do EAs Stay in EA?

XII- Do EA Sur­vey Tak­ers Keep Their GWWC Pledge?

Prior EA Sur­veys con­ducted by Re­think Char­ity:

The 2017 Sur­vey of Effec­tive Altruists

The 2015 Sur­vey of Effec­tive Altru­ists: Re­sults and Analysis

The 2014 Sur­vey of Effec­tive Altru­ists: Re­sults and Analysis