Table of contents:
​
1. What is a prior view?
2. Why is it helpful?
3. Why it is not our only endline perspective
4. How much weight do we give prior views
5. Different types of prior views
5.1. Cause area prior informedness
5.2. Organizational priors
5.3. Research lead priors
6. Deeper reading

1. What is a prior view?

A prior view is a view that is held before a research process is started. Every researcher and organization comes in with background assumptions and expectations even before the research begins. In most cases, our goal with this prior view is to replace it with a far more informed view formed by a detailed and lengthy research process.

2. Why is it helpful?

Priors can range widely both in usefulness and in level of informedness. When looking at the cause areas we are covering in 2020, our priors will be much stronger in some areas than others. This is largely based on the information that our organization and individual researchers are coming into the area with. We try to lay out our priors’ transparency before conducting the deepest level of research so that it is clear how large our update is compared to our baseline expectations. This also makes us more sensitive to confirmation bias and ensures that statistically, we conduct updates based on new research. We also feel it is important for people to know our organizational priors to understand how our process works.  ​

3. Why it is not our only endline perspective

Priors carry many of the same weaknesses as IC but generally are more weakly informed.

4. How much weight we give prior views

Priors are stated at the outset and can range from being given zero weight to holding one-fifth weight. We do not expect priors to be a deciding factor in any intervention, but we do expect them to influence our research process to some degree. For example, if we have strong priors in a given area, less broad research will need to be done in the IC phase and thus more time will be able to be put into crucial consideration work.  

5. Different types of prior views

5.1. Cause area prior informedness

The following describes our informedness regarding the four cause areas:
  • Animals
    • Organizational background: We have focused our 2019 study primarily on animals. A similar research process was used that we expect to be fairly applicable to our research process this year. Since its founding in 2013, our organization has been interested in and connected to the animal community.  
    • Staff background: Of our director team, one-third came from an animal background before joining CE. Of our research team, three-fifths would describe animals as their primary cause before joining CE, with one-fifth having a strong interest in animals for an extended period of time. The lead researcher on animals describes her primary cause as animals.
    • Level of pre-2020 research year informedness: High
  • Mental health
    • Organizational background: Our organization has incubated one mental health-focused charity in 2019 but has not engaged in research in prior years with a mental health focus.
    • Staff background: Two-thirds of directors studied psychology as their major in university. One-fifth of our research team has a background in the economics of subjective well-being. These staff members will lead our mental health research.
    • Level of pre-2020 research year informedness: Moderate 
  • Contraceptives
    • Organizational background: We conducted a shallow review into contraceptives in 2019 and determined it sufficiently promising to be worked on more extensively this year. 
    • Staff background: No staff member has a significant background in contraceptives.
    • Level of pre-2020 research year informedness: Low
  • Health policy 
    • Organizational background: Our organization has conducted some health policy research focused on NCDs, specifically tobacco taxation, but this was in 2016. 
    • Staff background: No staff member has a significant background in health policy.
    • Level of pre-2020 research year informedness: Low

5.2. Organizational priors

Below is a nonextensive list of our priors going into a given charity idea or research area. Every organization has certain background assumptions or worldviews that affect its process and results. We feel that stating this transparently has significant benefits for understanding. 

Skeptical: We come from the position of skeptical priors. In the absence of strong evidence of impact, we assume a lack of effectiveness. When researching an area we generally start from the view that the area is likely to be less impactful upon further investigation and that gaps in data will more often reveal weakness than strength. As a result, we are much quicker to determine that an area is not worth further research than we are to determine that an idea is worth recommending. It also means that for areas with major exposed uncertainties, we are more likely to be negative than would be someone who is less skeptical. These priors have been formed through conducting charity research and speaking to others in the field. Our perception is that models such as CEAs almost always get less promising as more information becomes available, and this is true across a wide range of cause areas and organizations.  

Scientific: In general, we think that science is the top methodology for determining truth in the world. It seems to have the strongest track record historically. Not all problems are conducive to scientific studies or methodology, but even in cases where the data are not directly research-informed, we still use heuristics common in the scientific community such as “extraordinary claims require extraordinary evidence” and “falsifiability.”   

Utilitarianism: Our team primarily holds utilitarian and consequentialist views; although we have a wide range of these views. In accounts of high moral uncertainty, we use the parliamentary model. Additionally, when an idea could be highly impactful due to other common ethical views, we try to state these clearly. 

All lives have ethical value: We ethically value humans regardless of what country they are from, which often leads our researchers to recommend charities focused on lower-income countries. We also value non-human animals, including both those commonly ethically valued (such as dogs) and less commonly valued, such as fish. We generally have similar intuitions to the ones described in this report. We ethically value far future humans and animals but have historically been skeptical of the ability to affect them. 

Cluster vs sequence: We are generally quite skeptical of the sequence approach and charity ideas that are based on a single argument, such as Pascal’s wager. We are far more positive toward something that appears positive from many angles and explains weight convergence when considering charity ideas. 

Convergence: One principle we see as underused in traditional weightings of evidence is the concept of convergence or triangulation. The basic idea behind this is that multiple pieces of evidence should converge on the same answer, bringing additional strength of evidence. Convergent data is seen as a good indicator by most experts and is a considerable part of why replication studies are performed. However, we find too often that one strong RCT comes out and people update or change their minds too rapidly even if the convergent data (e.g., previous weaker studies or macro-level data) are pointing the other way. We tend to use a highly convergence-focused approach when evaluating evidence. 

5.3. Research lead priors ​

Our research lead priors are stated in a small section of the report and are generated before writing the eighty-hour report but after the two-hour-and-twenty-minute prioritizations. Because our “priors” are generated after some initial research, they will likely seem more predictive than our true priors. Instead, they reflect our initial conclusion based on the small amount of information gathered in the first round. This means that every researcher will likely have variable priors based on a wide but nonexhaustive range of sources. 

6. Deeper reading