Older people may place less moral value on the far future


Abstract

In a study ini­ti­ated by SoGive, we sought to un­der­stand to what ex­tent study par­ti­ci­pants care about (or place moral value on) peo­ple in the far fu­ture. The study sought to un­der­stand both stated opinions on the topic at an ab­stract level, stated opinions tak­ing into ac­count a con­crete (but hy­po­thet­i­cal) ex­am­ple, and (in an­other at­tempt to make this more con­crete) con­sid­er­ing the choice to donate to global poverty ver­sus cli­mate change char­i­ties. The study was in­tended to sup­port SoGive’s re­search on char­i­ties.

This pa­per also in­tro­duces a new seg­men­ta­tion of present­day­ist/​longter­mist be­liefs.

Trade-off ques­tions (which asked about choos­ing to save lives in the far fu­ture or to­day) sug­gested the fol­low­ing split of prefer­ences:

Ex­plic­itly ask­ing about val­ues sug­gested that study par­ti­ci­pants were some­what more present­day­ist than longter­mist.

Older peo­ple, es­pe­cially those in the 55 to 64 years old age cat­e­gory, tended to be more present­day­ist than younger peo­ple.

We ex­plored prefer­ences around donat­ing to cli­mate change char­i­ties ver­sus global poverty char­i­ties. We did not find a very strong prefer­ence ei­ther way.

Some of the ques­tions were then re­peated in an­other study which was na­tion­ally rep­re­sen­ta­tive. Some high­lights of this are also given be­low. They did not ma­te­ri­ally dis­agree with the find­ings of the ini­tial non-na­trep study.

Fur­ther re­search is en­couraged, es­pe­cially along the lines set out in the “Ideas for fu­ture re­search” sec­tion.

Background

SoGive com­mis­sioned Re­think Pri­ori­ties to perform this study, with SoGive stat­ing the aims, speci­fi­ca­tions, and bud­get for the study, and Re­think Pri­ori­ties sup­ported with the sur­vey de­sign and the ex­e­cu­tion of the study. This pa­per was au­thored by SoGive. The rest of this sec­tion ex­pands on the mo­ti­va­tion for the study.

Value of the far fu­ture and per­son-af­fect­ing views

A hotly de­bated topic within pop­u­la­tion ethics is the value we place on hu­mans who do not yet ex­ist. A to­tal util­i­tar­ian view may give equal weight to a fu­ture hu­man as to a per­son liv­ing to­day, while a “per­son-af­fect­ing view” typ­i­cally ar­gues that for some­thing to be good or bad, it has to be good or bad for some­one, and hence that bring­ing fu­ture happy hu­mans into ex­is­tence is no bet­ter than those peo­ple never ex­ist­ing.

One aim of the study was to un­der­stand more about the pop­u­la­tion’s at­ti­tudes to these ques­tions.

What re­search has already been done on this

As far as we know, rel­a­tively lit­tle re­search has been con­ducted on the ex­tent to which the pop­u­la­tion has per­son-af­fect­ing views, and on the ex­tent to which they pre­fer global poverty ver­sus cli­mate change as a char­i­ta­ble cause area.

How this re­search can be used for strength­en­ing char­ity research

Ri­gor­ous char­ity anal­y­sis in­volves com­par­ing differ­ent char­ity out­comes with each other. For ex­am­ple, a choice to donate to work which will help peo­ple in the fu­ture, in­clud­ing peo­ple who don’t ex­ist (such as, ar­guably, work on cli­mate change, or other work more di­rectly linked to ex­is­ten­tial risk) means money that is not spent on helping peo­ple to­day (e.g. sup­port­ing dis­ad­van­taged peo­ple in the de­vel­op­ing world).

Sup­port­ing work which has a greater propen­sity to help those who are not yet born would not be as at­trac­tive if we be­lieved that fu­ture peo­ple don’t have moral value, or have less moral value.

In or­der to make those judge­ments, hav­ing ac­cess to re­search on what the pop­u­la­tion as a whole be­lieves is use­ful.

Fur­ther­more, the study aimed speci­fi­cally to con­sider whether peo­ple (in­clud­ing donors) pre­fer to sup­port cli­mate change or global poverty. This is both in­ter­est­ing at an ob­ject level and use­ful for com­par­i­son with more ab­stract val­ues-ori­ented ques­tions.

Method

We sur­veyed 502 par­ti­ci­pants re­cruited from the Pro­lific.co plat­form. The par­ti­ci­pants were based in the UK and were not cho­sen to be na­tion­ally rep­re­sen­ta­tive.

The full sur­vey is pro­vided as an ap­pendix.

Some of the mo­ti­va­tions/​in­ten­tions be­hind the choice of ques­tions is dis­cussed here

The sur­vey ex­plores par­ti­ci­pants’ at­ti­tudes to peo­ple not yet born. It in­cludes ques­tions which ex­plic­itly ask about this at an ab­stract level, in­clud­ing the state­ments: “Fu­ture gen­er­a­tions of peo­ple, who have not been born yet, are equal in moral im­por­tance to peo­ple who are already al­ive” and “We should morally pri­ori­tise helping peo­ple who are in need now, rel­a­tive to those who have yet to be born”.

The sur­vey also con­sid­ered some slightly more con­crete ques­tions, and in­cluded trade-off ques­tions about com­par­ing a life saved now with differ­ent num­bers of lives saved in the far fu­ture (defined here as 500 years from now) -- the trade-off ques­tions.

Ques­tions about propen­sity to donate to cli­mate change and global poverty help to gauge par­ti­ci­pants’ in­ter­est in those ar­eas (which was of in­trin­sic in­ter­est for the study) and also links the stated value of un­born peo­ple to ac­tual be­havi­ours. We spec­u­lated that valu­ing the fu­ture more would be as­so­ci­ated with greater sup­port for cli­mate change.

The statis­ti­cal meth­ods used here in­clude some hy­poth­e­sis tests. In some cases ad­just­ments should be made for the fact that we are look­ing at mul­ti­ple com­par­i­sons (e.g. Bon­fer­roni or Si­dak ad­just­ments). Th­ese have not been made for rea­sons of time con­straints, al­though we sus­pect that the re­sults would not be changed if these ad­just­ments were ap­plied.

Results

1)

Mean self-rated like­li­ness (on a 1-7 scale) to donate to a Global Poverty char­ity is 4.01, a Cli­mate Change char­ity 3.60, and any other char­ity 3.52.

2)

How likely are par­ti­ci­pants to donate to Global Poverty char­ity (1) as op­posed to a Cli­mate Change char­ity (7) di­rect com­par­i­son:

Re­sults are close to the mid­point (mean 4.17). How­ever, 24.55% of peo­ple lean to­wards Global Poverty and 37.33% lean Cli­mate Change.

3)

How much good would par­ti­ci­pants ex­pect a dona­tion to a Global Poverty vs Cli­mate Change char­ity to do (on a 1-7 scale)

Here Global Poverty is ahead: mean 4.27 vs 3.94, which was statis­ti­cally sig­nifi­cant. A two-sided paired t-test shows that the data is statis­ti­cally sig­nifi­cant, given the null of no differ­ence, with a 5% long-run er­ror rate (Pr(|T| > |t|) =0.00001). The anal­y­sis is pow­ered enough to re­li­ably de­tect the effect size (0.28) over 80% of the time if a real effect ex­ists, though by Co­hen’s bench­mark 0.2 is a small effect.

4-7) The trade­off questions

Is it bet­ter to save (A) 1 per­son now or (B) 1/​2/​1,000/​1,000,000 peo­ple 500 years from now?

To un­der­stand the re­sults of this, we in­tro­duced a seg­men­ta­tion, which is defined in greater de­tail in a later sec­tion. The chart above shows how many re­spon­dents fell into each seg­ment. As de­tailed later, this seg­men­ta­tion re­quired some in­ter­pre­ta­tion of the re­sults, which prefer­ably should be val­i­dated with fur­ther re­search.

  • The left hand side of the chart in­cludes a bar for the “present­day­ist” seg­ment.

    • This seg­ment is defined as “prefers to save one life to­day ver­sus 1,000 lives in the far fu­ture, and other re­sponses are con­sis­tent” (where con­sis­tent means prefers to save one life to­day ver­sus one life in the far fu­ture or two lives in the far fu­ture; any an­swer for the “save one life to­day ver­sus 1 mil­lion lives in the far fu­ture” ques­tion is con­sid­ered con­sis­tent)

    • This cat­e­gory con­tains 91 respondents

  • The three cat­e­gories on the right (“Roughly longter­mist”, “Pure longter­mist”, “Fu­tur­ist: prefers fu­ture lives always”).

    • Th­ese seg­ments are roughly as­so­ci­ated with the prop­erty: “pre­fer to save 1,000 lives in the far fu­ture ver­sus one life to­day, and other re­sponses are con­sis­tent” (where con­sis­tent means speci­fi­cally that they also pre­fer to save one mil­lion lives in the far fu­ture)

    • Th­ese seg­ments con­tain 218 re­spon­dents.

Over­all, these re­sults sug­gest that re­spon­dents are more longter­mist than present­day­ist (p<0.001), un­der the defi­ni­tion of “longter­mist” given above.

Other defi­ni­tions of “longter­mist” may be more ap­pro­pri­ate. The sec­tion on the seg­men­ta­tion ex­plores this fur­ther.

Note that the sam­ple skewed to­wards younger re­spon­dents, and the age anal­y­sis in­di­cates that older re­spon­dents were more likely to be present­day­ist. There is a sec­tion later on which looks at a fol­low-up which was con­ducted on a na­tion­ally rep­re­sen­ta­tive (na­trep) sam­ple.

8) The ex­plicit-val­ues questions

How far do you dis­agree or agree (1-7) that:

“Fu­ture gen­er­a­tions of peo­ple, who have not been born yet, are equal in moral im­por­tance to peo­ple who are already al­ive”

“We should morally pri­ori­tise helping peo­ple who are in need now, rel­a­tive to those who have yet to be born”

Re­spon­dents largely agree with both state­ments (mean 4.63 and mean 5.02 re­spec­tively). This, de­spite the fact that the state­ments seem to op­pose each other.

The find­ings can be vi­su­al­ised with this table:

How to read this table:

The num­bers in this table show the num­ber of par­ti­ci­pants ac­cord­ing to each pos­si­ble com­bi­na­tion of re­sponses. For ex­am­ple:

  • How many par­ti­ci­pants gave the an­swer “2” out of a 7 point scale to in­di­cate their level of agree­ment with the state­ment: “We should morally pri­ori­tise helping peo­ple who are in need now, rel­a­tive to those who have yet to be born”? Ans: 26

  • How many par­ti­ci­pants gave the an­swer “6” out of a 7 point scale to in­di­cate their level of agree­ment with the state­ment: ““Fu­ture gen­er­a­tions of peo­ple, who have not been born yet, are equal in moral im­por­tance to peo­ple who are already al­ive”? Ans: 129

  • How many par­ti­ci­pants gave both of those an­swers? Ans: 15

Note that care should be taken about draw­ing in­fer­ences from the shape of this chart with­out con­duct­ing statis­ti­cal anal­y­sis (such as a sig­nifi­cance test).


Interpretation

This table in­di­cates vi­su­ally just how strongly weighted the an­swers are to­wards agree­ing with both, and just how rare it is for study par­ti­ci­pants to dis­agree with both state­ments.

This is sug­ges­tive that this sam­ple has been in­fluenced by a known effect in sur­veys—par­ti­ci­pants tend to agree to rea­son­able-sound­ing state­ments. It is un­clear whether par­ti­ci­pants care­fully con­sid­ered the ten­sions be­tween the two state­ments.

This sug­gests that care should be taken in draw­ing in­fer­ences from an­swers to the ex­plicit val­ues ques­tions.

The chart is also sug­ges­tive that the top right quad­rant (which is the part of the table as­so­ci­ated with present­day­ism) is bet­ter pop­u­lated than the bot­tom left cor­ner (which is as­so­ci­ated with longter­mism). This is un­sur­pris­ing given that the mean for the “morally pri­ori­tise peo­ple to­day” ques­tion was higher than the “fu­ture gen­er­a­tions have equal value” ques­tion.

A novel seg­men­ta­tion of present­day­ism/​longtermism

In or­der to bet­ter un­der­stand the re­sponses to the four trade­off ques­tions, we man­u­ally re­viewed the re­sponses. This re­view sug­gested a 15-cat­e­gory sys­tem of un­der­stand­ing re­spon­dents’ at­ti­tudes to the far fu­ture (+ a 16th cat­e­gory for re­sponses which we were un­able to make sense of).

The be­low table sets out those 15 (or rather 16) cat­e­gories, to­gether with the re­sponses that defined those cat­e­gories, and the num­ber of re­spon­dents in each cat­e­gory.

There­after we in­clude a table which sets out a sum­marised ver­sion of this seg­men­ta­tion. (Note that this sum­marised ver­sion is the same data shown ear­lier in a chart in the re­sults sec­tion).

Mak­ing sense of these has in­volved in­fer­ences be­ing drawn by the re­searcher about how to in­ter­pret the an­swers. Fur­ther re­search, es­pe­cially qual­i­ta­tive re­search, could help to un­der­stand this bet­ter.

Full 16-cat­e­gory segmentation


  • Note: the word­ing used is an at­tempt to in­ter­pret the re­sponses. The word­ing used may or may not re­flect the ac­tual thoughts of sur­vey re­spon­dents. For ex­am­ple there is ar­guably a differ­ence be­tween the lan­guage “can’t de­cide” (used in this table) and the lan­guage used in the sur­vey word­ing (which was “Equally good”). In other words, the de­scrip­tions are con­ve­nient la­bels, and not nec­es­sar­ily perfectly ac­cu­rate de­scrip­tions of re­spon­dents.

  • Note: For those in the un­clas­sified cat­e­gory, a re­searcher has at­tempted to un­der­stand the an­swers given, but was not able to dis­cern a logic be­hind the an­swers.


Sum­marised 8-cat­e­gory segmentation


And to sum­marise this even fur­ther we have:


An al­ter­na­tive sum­mary, per­haps less prone to the risks that arise when a re­searcher is in­ter­pret­ing the an­swers, is given below


Ap­ply­ing the longter­mism spec­trum to the ex­plicit val­ues questions

As men­tioned above, the present­day­ism/​longter­mism spec­trum is de­rived from the trade­off ques­tions (i.e. the ques­tions which com­pare sav­ing one life to­day with X lives in 500 years time).

To val­i­date this spec­trum, we tried ap­ply­ing this to the ex­plicit-val­ues ques­tions.


As can be seen from this chart, the left-hand side and right-hand side of this chart seems to sug­gest that the ex­plicit-val­ues ap­pear, at a high level, to val­i­date the present­day­ist/​longter­mist spec­trum de­rived from the trade­off ques­tions. (Note that in cases where the red line is above the blue line or vice versa, no statis­ti­cal tests were ap­plied to check whether one is re­ally higher than the other—this is in­tended as a quick vi­sual check)

There are two ex­cep­tions (high­lighted with green cir­cles):

  • The “to­tally even” group seems to side with fu­ture gen­er­a­tions some­what more than might be expected

  • The roughly longter­mist group seems to pri­ori­tis­ing peo­ple who are in need now slightly more than fu­ture gen­er­a­tions, which is the wrong way round.

Th­ese ap­par­ent in­con­sis­ten­cies could be be­cause the weakly and roughly longter­mist groups are re­ally not very longter­mist (i.e. our choice of la­bel­ling may be mis­lead­ing). Bear in mind that these groups are defined as some­thing similar to “val­ues 1000 lives in the far fu­ture over 1 life to­day”. Alter­na­tively, when dis­cussing the ex­plicit val­ues ques­tions we es­tab­lished ear­lier that re­spon­dents tend to agree to rea­son­able-sound­ing state­ments, so there’s a risk that this might be mud­dy­ing the re­sults.

In or­der to un­der­stand this bet­ter we dug into the sub­cat­e­gories of the “Roughly longter­mist” group and checked which of those are the ex­pected and the un­ex­pected way round. This is set out in Ap­pendix 3.

Age analysis

We con­ducted an anal­y­sis of both the ex­plicit-val­ues ques­tions and the trade-off ques­tions by age.

Age anal­y­sis: Ex­plicit val­ues ques­tions:

One age group is an out­lier—the 55 to 64 year old age group. The table for this age group shows that it is clearly bi­ased to­wards the “pri­ori­tise the pre­sent” (i.e. the top right) quad­rant.

The ta­bles for other age groups do not show such a marked bias, not even the 65-74 age group.

Note that cau­tion should be used in draw­ing in­fer­ences from small sub­cat­e­gories of the sam­ple.

If it re­ally is the case that the 55 to 64 year old age group is an out­lier as the more pre­sent-day-cen­tric group, it sug­gests that a sim­ple “ra­tio­nal” ex­pla­na­tion (“why care about the fu­ture when I’ll be dead soon any­way”) might not be the best ex­pla­na­tion. Other so­cio-cul­tural fac­tors may be at play.

Use­ful fur­ther re­search could ex­plore why at­ti­tudes to the far fu­ture are dis­tributed in the way they are.

Other ta­bles are shown in an ap­pendix 3.

Age anal­y­sis: Trade­off questions

To fur­ther val­i­date this find­ing, we then performed an age anal­y­sis us­ing the ear­lier trade­off ques­tions (i.e. the “would you rather save one life to­day or x lives in the fu­ture” ques­tions). This was es­pe­cially use­ful since we sus­pect that the trade­off ques­tions likely bet­ter re­flect the ac­tual opinions of par­ti­ci­pants. Th­ese seemed to val­i­date the find­ings from the di­rectly-ask­ing-about-val­ues ques­tions.

First we seg­mented the an­swers into differ­ent cat­e­gories, and then we analysed them by age:


Broadly speak­ing, this does seem to val­i­date the find­ings from the ex­plicit-val­ues ques­tions with re­gard to age skew.

Some ob­ser­va­tions based on this table:

  • Propen­sity to be present­day­ist in­creases as you go from the youngest age cat­e­gory up to the “55 to 64” cat­e­gory, and then starts to drop down again.

  • The p-value for the 35% figure high­lighted is 0.1% (null hy­poth­e­sis: propen­sity of 55 to 64 year olds to be present­day­ist is the same as the over­all group (i.e. 18%); bino­mial model, no (e.g. nor­mal) ap­prox­i­ma­tion used)

  • All age cat­e­gories from 35 years old up­wards have a higher than av­er­age propen­sity to be present­day­ist. Note that this av­er­age re­flects the age dis­tri­bu­tion of the sam­ple. This sam­ple was not na­tion­ally rep­re­sen­ta­tive, and a more rep­re­sen­ta­tive sam­ple may sug­gest that the over­all pop­u­la­tion is more present­day­ist than this sam­ple.

Na­tion­ally Rep­re­sen­ta­tive Sample

Re­think Pri­ori­ties also ex­am­ined re­sponses to the ex­plicit at­ti­tude items and char­ity prefer­ence items in a na­tion­ally rep­re­sen­ta­tive (cen­sus-matched for age, gen­der and eth­nic­ity) UK sam­ple drawn from Pro­lific.co (n=500).

As in the pre­vi­ous sur­vey, re­spon­dents were asked to ex­press their agree­ment or dis­agree­ment with (1-7, Strongly Disagree—Strongly Agree) with both the fol­low­ing state­ments (or­der ran­domised).

“Fu­ture gen­er­a­tions of peo­ple, who have not been born yet, are equal in moral im­por­tance to peo­ple who are already al­ive” [Fu­ture Equal]

“We should morally pri­ori­tise helping peo­ple who are in need now, rel­a­tive to those who have yet to be born” [Pri­ori­tise Pre­sent]

Th­ese re­sults like­wise show sig­nifi­cantly stronger agree­ment with the ‘Pri­ori­tise Pre­sent’ state­ment than the ‘Fu­ture Equal’ state­ment.



Also in line with the pre­vi­ous study, we found marginally higher self-re­ported like­li­hood to donate to Global Poverty (mean 3.93, SD=1.87) char­i­ties than Cli­mate Change char­i­ties (mean 3.85, SD=1.88), yet sub­stan­tially higher self-re­ported like­li­hood to donate to a char­ity work­ing on Any Other Char­i­ta­ble Cause (mean 4.87, SD=1.69).





We ex­am­ined cor­re­la­tions be­tween the ex­plicit at­ti­tude state­ments and the char­ity re­sponses. The Fu­ture Equal state­ment was sig­nifi­cantly pos­i­tively cor­re­lated with self-re­ported like­li­hood to donate to char­i­ties work­ing in all of the speci­fied cause ar­eas: Global Poverty, rs=0.214 (p<0.01), Cli­mate Change, rs=0.234 (p<0.01), Any Other Char­i­ta­ble Cause, rs=0.097 (p=0.031).

Conclusions

Un­der a fairly weak sense of “longter­mist” (i.e. val­ues 1000 lives in the far fu­ture over 1 life to­day) the trade-off ques­tions seem to find that the pop­u­la­tion ap­pears to be some­what more longter­mist than present­day­ist. How­ever un­der a stronger sense (i.e. lives in the far fu­ture are al­most as valuable as one life to­day) this find­ing is re­versed. This re­ver­sal would be con­sis­tent with the find­ings in­di­cated by the ex­plicit val­ues ques­tions.

Fur­ther­more, older peo­ple, es­pe­cially those in the co­hort roughly al­ign­ing with the “boomer” gen­er­a­tion (i.e. 55 to 64 year olds) were most prone to be present­day­ist. This was the re­sult which came out con­sis­tently from both the ex­plicit val­ues ques­tions and the trade­off ques­tions.

The ini­tial study didn’t in­di­cate that peo­ple had a very strong prefer­ence ei­ther way with re­gard to global poverty ver­sus cli­mate change, with the di­rect com­par­i­son show­ing a mild prefer­ence for cli­mate change, but an­other ques­tion ap­par­ently in­di­cat­ing a slight prefer­ence for global poverty. The fol­low-up na­tion­ally rep­re­sen­ta­tive study in­di­cated a slight prefer­ence for global poverty. This ex­plo­ra­tory study did not in­clude ques­tions to en­able us to un­der­stand why.

Ideas for Fu­ture Research

Un­der­stand­ing the age skew. It would be in­ter­est­ing to know whether study par­ti­ci­pants felt that it was ir­ra­tional to place moral value on events that hap­pen af­ter your death, or whether other so­cio-cul­tural fac­tors are at play. Fur­ther qual­i­ta­tive and quan­ti­ta­tive re­search could help un­der­stand this bet­ter.

Un­der­stand­ing the prefer­ences around choice of char­ity cause area. Choice of cause area is typ­i­cally in­fluenced by lots of fac­tors, and it’s un­clear to what ex­tent peo­ple’s pop­u­la­tion ethics prefer­ences were in­fluenc­ing their choice of cause area. For ex­am­ple, we don’t know to what ex­tent donors chose cause ar­eas be­cause of ex­ist­ing re­la­tion­ships with char­i­ties that they already donate to, or whether the emo­tions as­so­ci­ated with those causes varied, or to what ex­tent re­spon­dents doubted whether an­thro­pogenic cli­mate change was real, or to what ex­tent re­spon­dents as­so­ci­ated work on global poverty with money go­ing to cor­rupt dic­ta­tors.

Use of bonus in­cen­tives to un­der­stand ac­tual be­havi­ours. In de­sign­ing this study, we con­sid­ered as­sign­ing bonus in­cen­tives to par­ti­ci­pants and giv­ing them op­tions about where to donate the in­cen­tive. This would give an in­sight into *ac­tual* be­havi­ours.

Cor­re­la­tions with oth­ers in the wider moral cir­cle. We con­sid­ered also in­ves­ti­gat­ing whether propen­sity to place moral value on the far fu­ture is cor­re­lated with hav­ing a broader moral cir­cle (e.g. giv­ing moral value to an­i­mals). We thought that prop­erly un­der­stand­ing this would likely ma­te­ri­ally in­crease the length of the study.

Ex­plor­ing the present­day­ism/​longter­mism spec­trum de­rived in this pa­per. This spec­trum may be a use­ful tool for fu­ture re­search, how­ever fur­ther val­i­dat­ing and im­prov­ing the cat­e­gories may be worth­while. In par­tic­u­lar, the “roughly longter­mist” cat­e­gory seemed to have coun­ter­in­tu­itive re­sponses to the ex­plicit val­ues ques­tions; it would be use­ful to un­der­stand what is driv­ing this.

Ap­pendix 1: Survey

Char­i­ties work in a va­ri­ety of differ­ent cause ar­eas. In the fol­low­ing sur­vey we will ask you a se­ries of ques­tions about char­i­ties work­ing on differ­ent causes. There are no right or wrong an­swers, please sim­ply give your hon­est opinions.

[Page 1]

If you were to donate to a char­ity in the fu­ture, how likely do you think you think you would be to donate to a char­ity work­ing on each of the fol­low­ing causes:

Global Poverty

Cli­mate Change

A cause other than Global Poverty or Cli­mate Change

(1) Not at all likely - (7) Very likely

---

Please in­di­cate how likely you would be to donate to a char­ity work­ing on Global Poverty in com­par­i­son to Cli­mate Change:

(1) Much more likely to donate to a char­ity work­ing on Global Poverty - (4) Equally likely to donate to ei­ther - (7) Much more likely to donate to a char­ity work­ing on Cli­mate Change

---

How much good would you ex­pect a dona­tion to a typ­i­cal char­ity work­ing in each of these causes to do?

Global Poverty

Cli­mate Change

  1. None at all - (7) A great deal

[Page 2]

Is it morally bet­ter to save the life of (A) 1 per­son to­day or (B) 1 per­son who will ex­ist 500 years from now?

A is bet­ter than B

A and B are equally good

B is bet­ter than A

Is it morally bet­ter to save the life of (A) 1 per­son to­day or (B) 2 peo­ple who will ex­ist 500 years from now?

A is bet­ter than B

A and B are equally good

B is bet­ter than A

Is it morally bet­ter to save the life of (A) 1 per­son to­day or (B) 1,000 peo­ple who will ex­ist 500 years from now?

A is bet­ter than B

A and B are equally good

B is bet­ter than A

Is it morally bet­ter to save the life of (A) 1 per­son to­day or (B) 1,000,000 peo­ple who will ex­ist 500 years from now?

A is bet­ter than B

A and B are equally good

B is bet­ter than A

[Page 3]

Please in­di­cate the ex­tent to which you agree or dis­agree with the fol­low­ing state­ments:

“Fu­ture gen­er­a­tions of peo­ple, who have not been born yet, are equal in moral im­por­tance to peo­ple who are already al­ive”

“We should morally pri­ori­tise helping peo­ple who are in need now, rel­a­tive to those who have yet to be born”

  1. Strongly dis­agree - (7) Strongly agree

[page 4]

Stan­dard de­mo­graph­ics and exit questions

Gender

Age

Income

Education

Have you donated to char­ity in the last 12 months

Ap­pendix 2: Trade­off ques­tions: high-level /​ su­perfi­cial analysis

This table shows the num­ber of sur­vey re­spon­dents who gave each an­swer.

When re­view­ing this table, I looked at in­di­vi­d­ual re­sponses and con­sid­ered whether some of them looked odd/​in­con­sis­tent. Some did. E.g. Some re­sponses (8 of them) said they preferred to save 1 life in the fu­ture com­pared to 1 life to­day. How­ever when the num­ber of lives in the fu­ture in­creased, they didn’t stick to prefer­ring the fu­ture lives. (And some other in­con­sis­ten­cies) I tried analysing this table hav­ing re­moved the ap­par­ent in­con­sis­ten­cies, how­ever it did not change the over­all story much, so I pre­sent here the full pic­ture, which helps to avoid the in­fluence of re­searcher bias.

It seems from this that al­most half of the sam­ple (i.e. 240 out of 502) had a prefer­ence for pre­sent-day hu­mans over fu­ture hu­mans.

How­ever, a closer anal­y­sis based on the present­day­ist/​longter­mist seg­men­ta­tion shows that of those 240 re­spon­dents:

  • 151 of them were classed as (strongly or weakly) presentdayist

  • 87 of them were classed as (weakly or roughly) longter­mist, be­cause al­though they val­ued one per­son to­day over 1 per­son in the far fu­ture, they still placed more value on 1,000 peo­ple in the far fu­ture (for 81 of those 87) or at least placed more value on 1,000,000 peo­ple in the far fu­ture (for 6 of the 87)

  • 2 of them were un­clas­sified, mean­ing we couldn’t work out the thought pro­cesses be­hind the answers

This high­lights that an anal­y­sis as sim­ple as this doesn’t cap­ture the re­al­ity as well as the present­day­ist/​longter­mist seg­men­ta­tion.

It may ap­pear in­ter­est­ing that a small num­ber of re­spon­dents (18) ac­tively seemed to place more value on lives in the far fu­ture. How­ever this small num­ber ap­pears con­sis­tent with the Lizard­man Con­stant, so it seems rea­son­able not to make too much of this find­ing.

Ap­pendix 3: The “Roughly longter­mists” whose ex­plicit val­ues are not so longtermist

The Age anal­y­sis sec­tion men­tions that when we analyse the “Roughly longter­mist” seg­ment by the ex­plicit val­ues ques­tions, they seem to be not so longter­mist. This sec­tion ex­plores this fur­ther, as­sum­ing that the ex­plicit val­ues ques­tions were in­ter­preted in the way in­tended.

How­ever in or­der to un­der­stand this bet­ter we dug into the sub­cat­e­gories of the “Roughly longter­mist” group and checked which of those are the ex­pected and the un­ex­pected way round, which are set out be­low.

To un­der­stand how this might make a differ­ence, we con­sid­ered what would hap­pen to the bal­ance of present­day­ist/​longter­mist if we re­al­lo­cated these 81 in­di­vi­d­u­als to the “present­day­ist” cat­e­gory.

Here’s the raw (i.e. un­ad­justed) data:

If this rein­ter­pre­ta­tion of the data is cor­rect, it sug­gests that the bal­ance be­tween present­day­ists and longter­mists might be more even.

In the ab­sence of other ev­i­dence, it is likely bet­ter to treat the re­anal­y­sis (mov­ing the 81 par­ti­ci­pants) as a spec­u­la­tive al­ter­na­tive sce­nario, es­pe­cially since draw­ing in­fer­ences from ever smaller sub­cat­e­gories of re­spon­dents be­comes in­creas­ingly difficult to jus­tify statis­ti­cally.

Ap­pendix 4: Ex­plicit-val­ues ques­tions split by age

The main body of this pa­per in­cluded a table in­di­cat­ing an­swers to the ex­plicit-val­ues ques­tions for both the over­all sam­ple and the 55 to 64 age group. There now fol­lows the full ver­sion of this, i.e. the table for the over­all sam­ple and then for each age bracket.