Contributing Editor Vlada Limic writes the last column in her series about workshops and workshops, and reports on the results of her “learnering” survey, which she announced in the April 2015 issue:

I wonder how would you react to the following announcement: “The survey on learnering ran for more than four months. The data collected strongly suggests that (a) more than 70% of our peers are willing or likely willing to participate in a learnering, (b) fewer than 1% are unlikely willing or not willing to participate, and (c) all the unlikely willing or unwilling peers would still encourage a junior colleague to participate in a learning. The learnering is therefore overwhelmingly supported within our community.” This note would probably be received with skepticism, and for good reason. Indeed, to reach the above conclusions, the survey data would have to be used in a rather manipulative way, as explained below.

The first claim made above is correct—the survey was opened in mid-April and was closed on September 1. In the rest of this text, I will refer to the mathematicians (predominantly, but not exclusively, probabilists) and statisticians, whose email addresses are on my list of email contacts, as my neighbors (or neighborhood). I was also included in this neighborhood. Another peer’s neighborhood could be defined in complete analogy. The underlying assumption is that my neighborhood, with respect to responding to the learnering survey, is not much different from yours, or from that of any other peer (mathematician or statistician as defined for example in the Wikipedia article by Jean-Pierre Bourguignon1 entitled “Mathematicians in France and the World”2).

My neighborhood did expand by a little in the course of the survey, due to peers from outside contacting me specifically about the survey or workshop. The expansion rate of 1% might seem surprisingly small, but then again it might not be in view of the total response rate.

Each member of my neighborhood was sent a unique electronic invitation to the survey. A bit more than one-fifth of invitees responded by filling out the questionnaire, about 7% responded by actively opting out (either by sending me a short email message, or by clicking on a button in the SurveyMonkey widget), and the rest did neither.

The conclusions announced in the opening paragraph are misleading, even though derived from the actual survey data. For example, to arrive at the 70%+ figure, I took into consideration only the neighbors who filled out the questionnaire (in analogy to any democratic vote), while the 1% figure was obtained by dividing the number of unlikely willing or unwilling answers with the total size of my neighborhood.

In case you are wondering, the percentage who claimed that they would not encourage a junior colleague to participate in a learnering is again smaller than 1%. However, as already indicated, all of these peers would be (at least possibly if not likely) willing to participate in a learnering themselves.

There is no doubt that the gathered data is moderately if not strongly biased by the fact that the more willing a peer is to participate in a learnering (or a workshop in general), the more likely they were to fill out the questionnaire. I am not sure what this bias is called, or if there is any way to account for it. Let us then take all the opt-outs as unwilling to participate in any kind of workshop, which I sense would be somewhat unfair, yet it is a way to get objective lower bounds on the estimates.

Applying this simplification, one arrives at the following conclusion: at least one-fifth of my neighbors claimed to be (at least possibly) willing to participate in a learnering in their respective research area (this 20% is made up of 15% likely willing or willing answers, and of 5% possibly willing answers). The analogous figures corresponding to potential intra- or inter-disciplinary participation are smaller (due to limited space I do not quantify this decrease), but still non-negligible.

According to Bourguignon, there are about 80,000 active mathematicians (including statisticians) in the world, of which many work in industry3. Even by halving the 80,000, we have a small town full of mathematicians working in academic research institutions. Using the uniformity over neighborhoods assumption, we can reach the conclusion that at least a fifth (meaning multiple thousands) of active mathematicians, working in academic institutions anywhere on the planet, are enthusiastic about workshop format in general (and learnering in particular).

This seems important enough. Even though it is not clear at the moment where and how to look for a large scale workshop funding, it seems worthwhile for each member of the fifth to look for the other enthusiasts in his/her own neighborhood, in order to start thinking about co-organizing workshops with readily available resources. If you wish to know more about the survey and/or my own efforts on this path, do contact me: vlada.limic@math.u-psud.fr

—
Footnotes:
1 https://en.wikipedia.org/wiki/Jean-Pierre_Bourguignon
2 http://smf4.emath.fr/en/Publications/ExplosionDesMathematiques/pdf_en/smf-smai_explo-maths_92-97_en.pdf
3 The estimate given for France is that one mathematician in three works in industry.