Broad student outreach methods seem heavy-tailed regarding converted fellows and time-effectiveness due to differences in scale.
If you have the opportunity, use a university newsletter to reach many students.
Caveat: This is just n=22 from one student group’s intro fellowship (and my first forum post). Consider EA Groups Resources or the Global Challenges Project if you need advice for student outreach.
Background
This semester EA Munich facilitated an in-person intro fellowship for the first time as part of the nationwide intro fellowship, which Evander Hammer and Moritz von Knebel organized. Twenty-two fellows participated in Munich, making the cohort the largest in Germany. I think Munich offers an untapped opportunity for community-building (our local group is searching for a full-time community builder – just DM me!) since there are two top 50 universities[1] with >100,000 students.
Outreach methods
Student outreach represents the upper part of the funnel of involvement with EA, meaning the first step in a newcomer’s journey from the audience to high impact. We used multiple outreach channels to reach students: a university mailing list to 8,500 recipients with a link to the application form[2], an intro talk with ~20 guests, ~8 in-person classroom presentations in front of ~30 students each, ~25 uploads of slides on Moodle pages of lectures, ~30 posters inside university buildings and free flatscreen advertisements at university canteens. This is roughly in line with what has been discussed for student outreach methods in Germany.
I felt that the hours of presentations and personal contacts didn’t achieve much after the intro talk, but I was pretty surprised when I heard of the number of fellows due to the uni newsletter. This feeling of surprise motivated me to write this post.
Heavy-Tailed Result
12 of 22 (=55 %)[3] fellows stated in the application form that they applied to the intro fellowship after reading one of two outreach emails (data here). This distribution looks roughly heavy-tailed. This pattern also shows up when you look at the roughly guessed time-effectiveness[4] of the different outreach methods. I argue that time-effectiveness is a good proxy for the overall costs of our student outreach since the opportunity costs are the main cost faced and our student outreach was quite labor-intensive. Furthermore, only the intro talk incurred financial costs, not altering the cost-effectiveness of the mailing list.
Interpretation
I believe the main reason the uni newsletter was so compelling is the sheer mass of its recipients and the application form as a direct call to action. ~200 people looked at the application form after the first mail.[5] This contrasts with other forms of broad outreach[6] which didn’t engage that many people. In the canteens, the flatscreen advertisement was barely present, hampering the potential of this even more general outreach method. Whether the share of recipients becoming fellows varies between the mass-outreach tools[7] can’t be stated with this data, suggesting that the percentage is higher for people who spend time at the intro talk. Furthermore, the mass-outreach approach might lead to a higher dropout rate before the first session, but dropouts are excluded from the sample. Anecdotal evidence hints that the selection effect is relevant enough even with the uni newsletter so that only people somewhat interested in EA come to the first fellowship session.
Implications of Heavy-Tailedness
In this section, I will assume that the results are valid and would replicate with more fellows. Heavy-Tailedness regarding impact has been claimed several times, e.g., by my former co-organizer and skiing mate Mathieu Putz concerning earning to give or by Toby Ord in global health. It implies that most value comes from a few people, interventions or, in this case, outreach methods. Here, at least from the utilized mass methods, the most scalable approach reaches the most potential fellows. This might not be generalizable to more personal forms of outreach, if performed properly. Community builders should evaluate their outreach methods and share, which I hope is already being done. Be aware of the varying effectiveness of broad outreach methods and that you can do a proper amount of outreach within a limited amount of time. Therefore an acceptable minimum viable product of outreach for groups with low organizing capacity and a large student body seems feasible.
What we will improve
I guess that more personal forms of outreach with higher fidelity were neglected in Munich. We didn’t reach out successfully to more value-aligned people or groups, partly because we started from scratch with only three active students at the beginning of the semester. For first outreach at universities, fidelity might not be so essential in contrast to media outreach since most students are open to hearing from student groups. Probably even more consequential, we didn’t do any tabling at all (the university rejected our applications for tabling thrice :-( ). Although there probably are even more ingenious ways of getting students into the funnel, the limitations of the methods utilized remain noteworthy, too. For most presentations at lectures, we used no slides and an invitation to the intro talk was the only call to action. The slides we uploaded on ~25 Moodle pages were quite rudimentary.
Last Words and Call to Action
Maybe the heavy-tailedness in student outreach was already clear to many, e.g., getting on the uni newsletter was strongly recommended on a CB retreat, but I didn’t internalize this much. Hopefully, some readers who organize groups can provide better and more indications of the effectiveness of different outreach methods and whether the heavy-tailedness also holds with more personal forms of outreach (e.g., tabling). Don’t take the results or the overinterpretation of the too-small sample of fellows too seriously; probably, the principal value of this post is from the exchange in the comments. Get on uni newsletters or other mass-scalable student outreach methods!
I am thankful for my peer group in Munich, Julius and Max, with whom I reached out to the students. Thanks to Max, Julius and Evander for their valuable comments. All mistakes remain my own.
Sadly, we only had access to the mailing list after the intro talk, but this luckily enables me now to estimate the effects of the methods separately-ish. The mailing list consists of 8,503 students at LMU who haven’t opted out of receiving mails from different bodies of the university and registered student groups.
At first, my System 1 counted 18 of 22 fellows as reached by the uni newsletter. LMU is the name of the main uni, so it is unclear how they found the fellowship, e.g. whether by posters or the newsletter.
The time estimates are rough and foremost based on my impressions. I think 1.5 h is a fair estimate for writing two emails, although I neglect the fixed time-cost of registering the student group at university (which took the bureaucracy five f*** months and me a dozen follow-up emails!). I think 2 h reflects the time cost of creating, giving and organizing the Intro Talk roughly. I assume 3 h for personal contact, since I and others talked to multiple people about the fellowship, in real life and over WhatsApp.
This data is for the first mail by bit.ly. For the second mail Tiny URL reported 8,600 clicks (100% rate), but I conclude with a Bayesian prior that TinyURL didn’t work.
I guess that the uploaded slides have been opened by few (probably low click rate), the posters (unclear), the screen advertisement and canteens (unclear) and the presentations (>100 hearers) have reached fewer people, while being mass-outreach tools.
I guess that the uploaded slides have been opened by few (probably low click rate), the posters (unclear), the screen advertisement and canteens (unclear) and the presentations (>100 hearers) have reached fewer people than the uni newsletter, while being mass-outreach tools. Since the share of the mailed students turned into fellows is very low (0.1 %), even a method which got us no fellows might be more effective in this ratio.
A Case Study: The heavy-tailedness of broad student outreach methods
Key takeaways
Broad student outreach methods seem heavy-tailed regarding converted fellows and time-effectiveness due to differences in scale.
If you have the opportunity, use a university newsletter to reach many students.
Caveat: This is just n=22 from one student group’s intro fellowship (and my first forum post). Consider EA Groups Resources or the Global Challenges Project if you need advice for student outreach.
Background
This semester EA Munich facilitated an in-person intro fellowship for the first time as part of the nationwide intro fellowship, which Evander Hammer and Moritz von Knebel organized. Twenty-two fellows participated in Munich, making the cohort the largest in Germany. I think Munich offers an untapped opportunity for community-building (our local group is searching for a full-time community builder – just DM me!) since there are two top 50 universities[1] with >100,000 students.
Outreach methods
Student outreach represents the upper part of the funnel of involvement with EA, meaning the first step in a newcomer’s journey from the audience to high impact. We used multiple outreach channels to reach students: a university mailing list to 8,500 recipients with a link to the application form[2], an intro talk with ~20 guests, ~8 in-person classroom presentations in front of ~30 students each, ~25 uploads of slides on Moodle pages of lectures, ~30 posters inside university buildings and free flatscreen advertisements at university canteens. This is roughly in line with what has been discussed for student outreach methods in Germany.
I felt that the hours of presentations and personal contacts didn’t achieve much after the intro talk, but I was pretty surprised when I heard of the number of fellows due to the uni newsletter. This feeling of surprise motivated me to write this post.
Heavy-Tailed Result
12 of 22 (=55 %)[3] fellows stated in the application form that they applied to the intro fellowship after reading one of two outreach emails (data here). This distribution looks roughly heavy-tailed. This pattern also shows up when you look at the roughly guessed time-effectiveness[4] of the different outreach methods. I argue that time-effectiveness is a good proxy for the overall costs of our student outreach since the opportunity costs are the main cost faced and our student outreach was quite labor-intensive. Furthermore, only the intro talk incurred financial costs, not altering the cost-effectiveness of the mailing list.
Interpretation
I believe the main reason the uni newsletter was so compelling is the sheer mass of its recipients and the application form as a direct call to action. ~200 people looked at the application form after the first mail.[5] This contrasts with other forms of broad outreach[6] which didn’t engage that many people. In the canteens, the flatscreen advertisement was barely present, hampering the potential of this even more general outreach method. Whether the share of recipients becoming fellows varies between the mass-outreach tools[7] can’t be stated with this data, suggesting that the percentage is higher for people who spend time at the intro talk. Furthermore, the mass-outreach approach might lead to a higher dropout rate before the first session, but dropouts are excluded from the sample. Anecdotal evidence hints that the selection effect is relevant enough even with the uni newsletter so that only people somewhat interested in EA come to the first fellowship session.
Implications of Heavy-Tailedness
In this section, I will assume that the results are valid and would replicate with more fellows. Heavy-Tailedness regarding impact has been claimed several times, e.g., by my former co-organizer and skiing mate Mathieu Putz concerning earning to give or by Toby Ord in global health. It implies that most value comes from a few people, interventions or, in this case, outreach methods. Here, at least from the utilized mass methods, the most scalable approach reaches the most potential fellows. This might not be generalizable to more personal forms of outreach, if performed properly. Community builders should evaluate their outreach methods and share, which I hope is already being done. Be aware of the varying effectiveness of broad outreach methods and that you can do a proper amount of outreach within a limited amount of time. Therefore an acceptable minimum viable product of outreach for groups with low organizing capacity and a large student body seems feasible.
What we will improve
I guess that more personal forms of outreach with higher fidelity were neglected in Munich. We didn’t reach out successfully to more value-aligned people or groups, partly because we started from scratch with only three active students at the beginning of the semester. For first outreach at universities, fidelity might not be so essential in contrast to media outreach since most students are open to hearing from student groups. Probably even more consequential, we didn’t do any tabling at all (the university rejected our applications for tabling thrice :-( ). Although there probably are even more ingenious ways of getting students into the funnel, the limitations of the methods utilized remain noteworthy, too. For most presentations at lectures, we used no slides and an invitation to the intro talk was the only call to action. The slides we uploaded on ~25 Moodle pages were quite rudimentary.
Last Words and Call to Action
Maybe the heavy-tailedness in student outreach was already clear to many, e.g., getting on the uni newsletter was strongly recommended on a CB retreat, but I didn’t internalize this much. Hopefully, some readers who organize groups can provide better and more indications of the effectiveness of different outreach methods and whether the heavy-tailedness also holds with more personal forms of outreach (e.g., tabling). Don’t take the results or the overinterpretation of the too-small sample of fellows too seriously; probably, the principal value of this post is from the exchange in the comments. Get on uni newsletters or other mass-scalable student outreach methods!
I am thankful for my peer group in Munich, Julius and Max, with whom I reached out to the students. Thanks to Max, Julius and Evander for their valuable comments. All mistakes remain my own.
LMU is 32rd and TUM is 41st at the Times Higher Education Ranking 2021, topping the ranking nationwide.
Sadly, we only had access to the mailing list after the intro talk, but this luckily enables me now to estimate the effects of the methods separately-ish. The mailing list consists of 8,503 students at LMU who haven’t opted out of receiving mails from different bodies of the university and registered student groups.
At first, my System 1 counted 18 of 22 fellows as reached by the uni newsletter. LMU is the name of the main uni, so it is unclear how they found the fellowship, e.g. whether by posters or the newsletter.
The time estimates are rough and foremost based on my impressions. I think 1.5 h is a fair estimate for writing two emails, although I neglect the fixed time-cost of registering the student group at university (which took the bureaucracy five f*** months and me a dozen follow-up emails!). I think 2 h reflects the time cost of creating, giving and organizing the Intro Talk roughly. I assume 3 h for personal contact, since I and others talked to multiple people about the fellowship, in real life and over WhatsApp.
This data is for the first mail by bit.ly. For the second mail Tiny URL reported 8,600 clicks (100% rate), but I conclude with a Bayesian prior that TinyURL didn’t work.
I guess that the uploaded slides have been opened by few (probably low click rate), the posters (unclear), the screen advertisement and canteens (unclear) and the presentations (>100 hearers) have reached fewer people, while being mass-outreach tools.
I guess that the uploaded slides have been opened by few (probably low click rate), the posters (unclear), the screen advertisement and canteens (unclear) and the presentations (>100 hearers) have reached fewer people than the uni newsletter, while being mass-outreach tools. Since the share of the mailed students turned into fellows is very low (0.1 %), even a method which got us no fellows might be more effective in this ratio.