Summary 

Objective: I am interested in working with others to develop and host a web-survey, or series of web-surveys, for the purpose of revealing why might people decide to change the cause-areas they currently work on, or are interested in. Additionally, what factors contribute to an individual changing their cause-area prioritization?  

The aim of this survey would be to determine how both objective metrics (e.g. the cause-area's scale, tractability, and neglect as measured by an EA-affiliated organization) for a cause-area's importance and personal values (subjective interest in a cause-area) influence cause-area switching.  

I suspect that in the process of obtaining answers to the above questions, the following questions will also have to be addressed; these can be thought of as secondary objectives of the survey. 

• Which cause-areas do people prioritize, coming into the survey.
• Which cause-areas do people work on or donate to, if any, coming into the survey.
• What people estimate it would take for them to alter this prioritization.
• How the prioritization changes based on how the cause-areas are presented.

My goal with this post is to collect feedback and ideas on the survey, especially in terms of this survey's potential effectiveness, or on whether / how it should be conducted. 

I am open to feedback on any dimension of this proposal, and welcome any criticisms. I am looking for thoughts concerning the framing of the objectives, the structure and content of the survey questions, and considerations regarding the scope, potential impact, and implementation of the survey. Feedback of the form, "It might be better to phrase this question this way.", "This has basically been done before, see: [x]", "[x] is not clear enough; try rephrasing like this: [y]", etc... would be great as well. 

Search 

It is first worth seeing whether any work pertaining to this topic has been performed before, in order to avert wasting effort and resources. Research on cause-area prioritization, on people's motivations for making donations, and on what people define as cause-areas are topics which I believe are directly related to, or run tangential to, this one, and are therefore important to consider in an initial search for existing work on cause-area switching. 

This search is not comprehensive, but I believe it is a decent start. It covers a large swath of territory, and didn't require too much effort on my end. 

Cause-Area Prioritization 

I searched "survey cause" into the search bar on this forum and found the four existing EA cause-area surveys, including the one published recently (July 29). 

• EA Survey 2017 Series: Cause Area Preferences
• EA Survey 2018 Series: Cause Selection
• EA Survey 2019 Series: Cause Prioritization
• EA Survey 2020: Cause Prioritization

Motivations for Donating

I searched "What do people donate to" on Google Scholar and found these two articles on the topic. Everything considered, I have not read up too much on the psychological underpinnings of altruism in humans. 

• Predictors of young people's charitable intentions to donate money: An extended theory of planned behavior perspective
• The development and validation of the motives to donate scale

Defining Cause-Areas 

Below are the immediate resources I thought to look to for work on defining cause-areas. I spent the least time searching for resources on this topic; I may synthesize information from more sources in a future edit. 

• Oxford Prioritization Project Review
• 80000 Hours: Cause Prioritization Research
• Counterproductive Altruism: The Other Heavy Tail
• 80000 Hours: Problem profiles
• OpenPhilanthropy: Focus areas

Evaluations 

For any of the previous topics, feel free to refer me to more resources. 

My evaluation is that a survey on cause-area switching would be novel, and would provide useful information, despite there being some overlap with existing cause-priorization surveys. 

I believe that it would be most effective to employ scales and listings similar or identical to those employed by existing cause-priorization surveys; this keeps things consistent and should permit for cross-analyses. 

Small Selection of Existing Work on Cause-Area Shifting

Here I paraphrase sections of some of the resources I have included above. 

EA 2020 Survey - describes a real change in people's cause areas, but does not consider factors that drive this shift :

Comparing ratings across separate EA surveys, we observe a decrease in support for global poverty over time, and an increase in support for animal welfare and AI risk.

EA 2o20 Survey - describes a factor - COVID - that influenced one shift in people's cause-area prioritization:

One result that may be of particular interest, however, is support for Biosecurity (and pandemic preparedness), given the pandemic. This increased from 4% in 2018 and 2019 to 6% in 2020.

EA 2020 Survey - describes cause-rating / engagement difference between gender; differences in cause-area switching might also correspond to differences in gender:

...suggesting that the relationship between engagement and cause ratings varies by gender, though the overall pattern is quite similar.

EA Survey 2020 - describes cause-area preference over time, but does not consider factors that drive this shift:

[see section: Cause preferences over time]

Measurements and Objectives

I think it would be useful to

• Utilize the same questions pertaining to demographics and EA involvement that the EA cause-prioritization surveys (see previous links) use.
• Utilize nearly the same questions pertaining to cause-area prioritization that the EA cause-prioritization surveys (see previous links) use.
• Utilize nearly the same questions pertaining to donation behavior that the EA cause-prioritization surveys (see previous links) use.

My hope is that some responses from this survey can be grouped with responses from other EA cause-priorization surveys, and that the new data - the characterizations of cause-area switching - can be used to inform trends, such as cause-area preference shifts over time, found in the previous surveys. 

Here is a list of general objectives for this survey (not ordered by importance, but by their appearance in the survey):

1. Determine the demographics (including past EA involvement) of the survey participants.
2. Determine what cause-areas people (believe are worth donating to) / (prioritize).
3. Determine what people believe it would take for them to change their existing prioritization of cause-areas.
4. Determine (how people might) / (whether people will) change (append or delete) the EA-pruned cause-area selection.
5. Determine how cause-area presentation affects cause-area prioritization in people unfamiliar with (EA) / (EA-pruned cause areas).
6. Determine what variables influence the direction of cause-area switching (e.g. why does someone switch their top cause-area from Climate Change to Mental Health instead of from Climate Change to AI Alignment), the frequency of cause-area switching, and the motivations underlying cause-area switching.

Considerations for the Objectives

I am open to more considerations of this nature.

For (2): Should these be EA pruned cause-areas (cause areas promoted by EA affiliated organizations), general cause areas (something people might list in a short answer response question), or a mixture of both?

For (4): This might not be necessary. I included this because I can imagine a scenario where people unaffiliated with EA believe the selection of cause-areas EA has generated is way off-base, and because of this, become suspicious about the survey and the movement as a whole. If people had the opportunity to append cause-areas or remove cause-areas, this might reduce this problem, and might make a small group of people more inclined to answer honestly and complete the survey. 

For (6): These should be made more precise. 

Survey Questions

Potential Sequences for Survey Questions 

Imagine that the word "Person" is the first word in each piece of the sequence. Here, a sequence denotes a portion of the survey that could consist of just a single question or of several related questions. The order of the sequence follows the order of the question sets in the survey.

• lists demographics → ranks EA-pruned cause-areas → lists which cause-areas they've historically cared about → estimates what it would take for them to change their current cause-area ranking → 
• lists demographics → ranks EA-pruned cause-areas → takes an exercise (exposure to new information) that manipulates their cause-area ranking → estimates what it would take for them to change their ranking → 
• lists demographics → ranks EA-pruned cause-areas → imagines if they were in a different stage of life, and then re-ranks cause-areas based on that → estimates what it would take for them to change their ranking → 
• lists demographics → ranks EA-pruned cause-areas → predicts how their cause-areas might change over time → estimates what it would take for them to change their current cause-area ranking → 
• lists demographics → ranks EA-pruned cause-areas → explains the reasons behind their cause-area ranking → estimates what it would take for them to change their current cause-area ranking → adds or removes EA-affliated organizations' cause-areas to capture what they feel humanity should focus on → 

Considerations for the Survey Questions

Here are some considerations (in the form of questions) for the potential survey questions I list soon. Ideally, these questions should be asked of each survey question. Please provide suggestions for these meta-questions, should any come to mind. 

• What data does this question seek to collect?
• What measures should be taken to minimize the potential misinterpretations of this question?
• How can the scope of this question be broadened without decreasing the quality of the answer / signal?
• Is there any reason someone would be motivated to answer dishonestly? If so, how can this be mitigated?

Question Sets 

The sets of questions are aggregated by objective. Again, there is a loose order to the objectives here, but this is subject to change, depending on the suggestions. Within each objective, the questions are unordered. There is some redundancy in the questions; I do this to increase the size of the space of potential questions - there might be slight changes in response based on the phrasing of a particular question. 

• Short Response: [SR]
• Fill in Blank: [FB] (very short response)
• Scale: [lowest value - highest value]
• Multiple Choice: [<values>]

Demographics

• EA Survey 2020: Cause Priorization - use same questions
• Are you involved with Effective Altruism (EA)? [0 - 5]
• What is your age? [FB]
• What is your gender? [<genders>]
• What is your highest level of education? [<levels of education>]
• What regions are you located? [<regions>]
• What is your race? [FB]

Cause-areas prioritization

[<cause-areas>] : insert all (esp. from EA Survey) cause-areas pruned by EA affiliated orgs. 

• EA Survey 2020: Cause Priorization - use same questions
• What do you believe is worth donating to? [SR]
• If you had to choose only one of these cause-areas, which would you choose? [<cause-areas>](choose one). Why? [SR]
• How would you rank these cause-areas? [<cause-areas>]
• Select the top three cause-areas you find most pressing. [<cause-areas>]
• How do you think other people would rank these cause-areas? [<cause-areas>]
• Have you ever made a donation to an issue or cause you found important? If so, what did you donate to? [SR]
• Would you remove any cause-areas on this list? [<cause-areas>] Why? [SR]
• Are there any cause-areas that you think should be added to this list? [SR]
• What factors do you consider when prioritizating a cause-area? [SR]
• What resources do you use to help prioritize cause-areas? [<resources>]+[SR]
• If you were to give advice on cause-area priorization to someone less experienced than yourself, what would you say? [SR]

Estimations for changing prioritization of cause-areas

• What do you believe it would take for you to shift your #1 priority cause-area? [SR]
• How do you think your cause-area priorization will change over your lifetime? [SR]
• What would have to occur for [random cause-area, varies between surveys] to be your highest priority cause-area? [SR]
• Do you believe your cause-area priorization is more based in personal interest or quantitative metrics?

Cause-area presentation 

• [present quantitative estimates of cause-area urgency for the cause-areas the person didn't priorizate] Have these descriptions changed your priorization of cause-areas? [SR]
• [present EA affliated organizations' cause-area priorizations] Is there anything about this cause-area prioritization that you would change? [SR]
• Which factors, unrelated to the cause-areas, do you feel influence your cause-area priorization? [<potential factors>]+[SR]
• Which factors, unrelated to the cause-areas, do you feel influence other people's cause-area priorizations? [<potential factors>]+[SR]

Direction and frequency of cause-area switching

• If you have switched cause-areas, how long did it take for this to occur? [FB]
• If you have switched cause-areas, which cause-area did you switch from, and which did you switch to? [FB]
• If you have switched cause-areas multiple times, how often did these switches occur? [FB]
• If you have switched cause-areas, what was the primary factor influence this/these switch/switches?  [FB]
• If you have switched cause-areas, how rational were you in switching? [SR]
• If you have switched cause-areas, do you feel there were factors out of your control that engendered the switch? [SR]

Implementation

Here I describe my initial plans and ideas for executing the survey, and for maximizing its reach and impact.

Colleges

Myself: Given my circumstances, I am 75% confident that I will be able to submit a survey on this topic to at least two of the following departments [Economics, Neuroscience, Computer Science, Mathematics, Sociology, Anthropology, Philosophy] in my college. Additionally, I am somewhat confident that I will be able to convince EA groups at two other colleges in the area to host this survey in at least one department. I am not sure how well surveys fair at my college, but my estimate is that a double-digit amount of people (out of the ~3000 people attending my college) will take the survey. The survey would likely be presented as "An Altruism Survey". 

My friends:  My non-EA affiliated friends and family will likely take the survey. A subset of my friends will likely propose this survey to at least one of the departments in their college, but I don't have an estimate for how this will pan out.

Online Forums

The same forums and sites that the EA Survey used to disseminate their surveys can be used for this survey. 

At this current moment in time, I would not know how to scale the survey or maximize / diversify participation. 

Other Survey Considerations

• Total number of questions
• Complexity of questions (e.g. written responses are more tiring to do)
• (Whether to) / (how much to) pay people to participate in the survey
• Accessibility of survey
• What service is used to create the survey
• Whether the survey has interactive exercises

Final Remarks

I believe a survey of this nature will confer useful information for global priorities research and for building effective altruism. Understanding the factors that influence people's cause-area switching could be important for promoting interest, engagement, and action in cause-areas that are quantitatively evaluated to be pressing, or in need of further research, but which the general public or EA community doesn't find pressing. 

I am motivated to conduct this survey with others out of  (1) curiosity and (2) a desire to build effective altruism. 

Acknowledgements

Thank you for reading this everyone; I hope something interesting, and potentially impactful arises from this post. This idea was spurred by the post: Research into people's willingness to change cause areas? Thank you again, Davis_Kingsley.