What is Effective Thesis?

Effective Thesis is a project that directs students’ research towards EA causes by offering them EA-related topics for their final dissertations and theses. The aim is to deliver three valuable outcomes:

• Changing student trajectories at a particularly crucial juncture in their lives and directing some of their research careers to more impactful paths.
• Generating additional research in EA cause areas, with little cost to EA funding sources.
• Making students more knowledgeable in a specific EA cause and overall more involved in EA.

The hope is that this could be potentially high-leverage, even more so than usual career coaching, as we are focusing on a particularly crucial juncture. Other potential beneficial effects of the project are diversifying the scope of people involved in EA by bringing some from non-mainstream backgrounds into the community (e.g. mechanical engineers, archaeologists, history students, etc.); making current academics (students’ supervisors and committees) more familiar with the EA perspective through students’ work; and potentially creating a gateway for academic-minded students interested in improving the world but not knowing about or not willing to identify with EA directly.

More generally, in the goal of influencing research, this project focuses more on the junior side of research career trajectories, trying to create new junior researchers who are not necessarily dependent on the EA funding. For other approaches like funded fellowships or engagement with senior academics, there are likely better-suited organisations like Global Priorities Institute or Forethought Foundation.

I founded the project in June 2017 supported by the Czech EA Association and received an EA Grant in August 2018 to test the new project design - thesis coaching. Thesis coaching is meant to provide more tailored advice to students via connecting them with experienced coaches (usually, researchers from the EA community). Coaches participate on a voluntary basis and usually have only one call with the student (devoting about 1 hour per student). The hypothesis was this would lead to a higher number of interested students and a higher quality of students’ project choice decisions than the original design, where there was a list of specific topics on the website and students were expected to match themselves with a topic. Read here to get more details on the background of the project and why we have decided to change the design.

After the student applied, was matched and connected with a coach and received the coaching, I followed up with them to have an impact interview and explore in which ways Effective Thesis influenced them. Here are the questions I used. I plan to do another follow-up after they finish their theses (approx. 6-12 months after they received the coaching) to get a better sense of the long term impact and to test some of the hypotheses mentioned above.

Results

Thesis Coaching

The advice we were able to give students seems higher quality. Students report getting reasonably high value from the project: about 7.3/10 of usefulness on average and about 50 % of counterfactual impact in their decision-making regarding their study/research/career plans. However, sources of this value are mixed: 1/3 of students appreciated general advice on research direction, 1/3 appreciated academic career-related advice and 1/3 appreciated guidance in the topic they came up with themselves.

The main source of counterfactual impact seems to be in getting specific topic ideas; getting sources (readings, etc..) to start with; getting the general direction to focus on within their study discipline (e.g. biosecurity for maths students); and being introduced to the main idea that they can do EA-related research in their thesis. On the other hand, the main hindrance for counterfactual impact seems to be not getting a specific-enough topic suggestion from the coach.

Promotion of the project

In contrast with the first year when the project was promoted mainly via online ads and directed outwards from the EA community, this year I focused on promoting it in the EA community via various channels: presenting it at the EAG London and EAGxNordics, informing student group leaders about the project in cooperation with LEAN; and 80,000 Hours mentioning Effective Thesis in one of their posts.

This strategy, together with the new thesis coaching design, attracted more students. There were 111 applications in the first 5 months, out of which I wasn’t able to help (or had very low-quality applications from) about 1/3 of people, another 1/3 stopped communicating during the process and about 1/3 used the coaching advice and had an impact interview with me in the end.

I have collected data on 5 descriptive variables from the people who applied for coaching: Whether someone considers research career as their top career option, how much involved in EA they are, which country they are from, what degree level and their university ranking. Under this outreach strategy, the project attracts mostly students whose career plan is to become researchers - roughly 2/3 of people who applied for coaching stated they want to take careers with a large research component. About 1/3 of people who applied were very involved in EA, 1/3 people we moderately involved, 1/3 people were not familiar with EA. Dropout was mainly among those who were not involved in EA and those who didn't plan to become researchers.

People from 22 countries applied for coaching with most people applying from the UK (16 %) and the rest of Europe (41 %), US (15 %) and Australia (7 %).
Most of the students who applied studied masters degree (36 %), then undergraduate degree (24 %) and then PhD degree (19 %), with 17 % not reporting their degree levels.
28 % of students who applied were from top 100 university rated by this website. Dropout was close to even on these variables and didn't show any significant trend.

Impact

When considering the final impact of the project, I estimated the impact of the project on each student’s decisions based on all the information I had. Impact assessed this way had two components: 1) how good is the change that the student has done; and, 2) how much of that change was counterfactually caused by Effective Thesis. When I tried to plot all student cases, I came to the conclusion that the distribution is likely log-normal, with few best cases accounting for most of the project’s impact. These best cases were students who planned to become researchers in the long-term and have substantially changed their focuses mainly thanks to Effective Thesis. I expect this trend to continue and expect that the main value that Effective Thesis will bring will be in influencing few people who will become top researchers and shift their research focus towards more important problems.

Forthcoming plans

Since the last update, I have also made some changes to the content - I’ve stopped promoting specific topics and moved towards promoting general high-impact research paths for each study discipline (see e.g. biology). Also, I have created a new page Agendas showing all research agendas in the EA community, and the page Finished showing finished as well as in-progress theses of students we advised.

As discussed in the Results section, prescribing students a specific topic doesn’t work very well in the current setting. On the other hand, most students benefited from being introduced into a general direction for their discipline and getting general academic career advice. This suggests either that coaching is not yet designed well-enough to deliver specific topics; or that the coaching is not the optimal strategy for delivering specific topics; or that offering specific enough topics is just too difficult and intractable. However, based on my impression from impact interviews, it seems that offering specific topics would still be the most important factor in raising the counterfactual impact of Effective Thesis. Therefore, it might be worth it to afford more thinking to the question of how to deliver students specific enough topics and continue experimenting with coaching design in order to learn how to do it.

My other plans are to focus on redesigning the website to make it appear more credible, trustworthy, official and prestigious; and communicate more clearly that what we are doing is important. I also plan to focus on improving the coaching process to generate even more counterfactual value, providing students with recommendations on future courses and supervisors and summarizing the advice on how to choose a thesis topic in a couple of blog posts. Recently, I have also launched an online community of students to enable them to connect with each other, solve some of their problems without my assistance and to enable me to communicate relevant job/grant offers, calls for papers and updates on thinking regarding how to start with high-impact research more readily.

In the long-term, I’d like to cooperate more closely with EA research orgs and get some further feedback on the project, since visiting Oxford was very helpful for setting direction on future development.

Regarding the promotion of the project, I would like to keep focusing on EA community members since they seem to benefit most from the coaching. Still, I’d like to devote a small portion of resources to experiment with promoting the project to non-EAs, since this has large expected value if done right.

If you have any feedback on this project, I’ll be happy to read it in the comments or at david@effectivethesis.com. If you have a good way to promote this project or have ideas on how to, please reach out as well.

Acknowledgements

I would like to thank all coaches involved who allowed this project to happen, CZEA for constant support, especially Jan Kulveit who inspired the project and came up with the idea to change design to coaching and Dan Hnyk who helps with programming, CEA for their financial support and together with EAGxNordics organisers, LEAN and 80,000 Hours for promoting the project and many community members who provided valuable feedback and ideas for future development of the project.