Monday, May 07, 2012

Research Triangle

Over the years, various colleagues have explained to me their personal beliefs about what they consider the optimal number of PIs involved in research projects, not for any particular research project, but just in general. The proposed optimal number of collaborators is most typically 1 or 2, although some people make a case for higher numbers.

Note that these hypotheses refer to general situations, not necessarily to any particular individual or project. This is mostly a "thought exercise"; fodder for musing and discussion. In real life, of course, the nature of the project, the culture of the discipline, the amount and type of funding available, the type of institution, and the personalities and career stages of the people involved are important in determining the Optimal Number of Collaborators (ONC). Some people prefer to work alone, so for them, the ONC =  0. In my own case, the ONC is a very stable 1-2, but I am happy to work in larger groups for some projects.

I say "mostly" a thought exercise instead of entirely, because I know of some cases in which the organizational structure of a research unit (and even the design of research space) has been planned based on a hypothesis about the ONC.

The theoretical limit of certain types of grants can also affect this. For example, with most NSF grants (at least in my field), it's possible to have 2-3 collaborators, but more than that would blow the budget up beyond a reasonable (fundable) limit.

Eveb so, is there a general ONC that applies in many cases, considering mostly human factors such as how many people are likely to get along well and have good, productive discussions and overall collegial relationships? That is, can we say, without being too far from unrealistic, "The most productive and collegial collaborations involve n people"? Does anyone want to make a case, for yourself or for the wider world of researchers, for:

ONC = 1: collaboration with one other person, perhaps even the same person on many different projects over the course of a career (= research line, or dyad?);

ONC = 2: collaboration with two other people (= research triangle);

ONC = 3: collaboration with three other people (= research quadrilateral);

.. and so on, with various polygons describing higher ONCs.

And does anyone want to generalize about ONC > 3? Do these collaborations tend to be more/less productive than smaller groups owing to their larger size (the bigger the group, the more research results), or does the increased chance of personality clashes, miscommunication etc. make (some of) them more unwieldy than the lean, mean research machines of smaller teams?

I am veering back and forth between the general and the personal here because there are two different levels of questions I am posing:

(1) Do you think that in general the ONC concept is relevant to the World of Research? (and if so, what is the ONC?); and

(2) What is your personal ONC for most projects (or does this number vary a lot?), and do these collaborators tend to be the same ones for project after project, or do you play the field with collaborators and work with many different people (even if your ONC doesn't change)?


Anonymous said...

My personal ONC is pretty low (0-2), but I am involved in a larger group that has assembled to apply for a grant program that specifically funds larger groups of collaborators. Initially I was concerned that it might be distracting and stressful to work with a group size so far above my ONC, but so far the major challenge has been logistical: with a larger group, I can tell we will need to rely on more sophisticated and regimented mechanisms for communicating and sharing than the ones I use with just a single collaborator.

nordicTT said...

1- from my experience, the ONC varies very much in different research fields. Also in the same topic, I see a lot of differences weather the research is experimental or theoretical.

2- My ONC varies a lot, from 2-3 to 15-20. The people involved tend to be the same for several years, especially for projects that involve a big experimental set-up. However I find more efficient to work in smaller groups, as the amount and quality of research doesn't generally scales with the number of people.

Anonymous said...

ONC = 1. In the special case of long-time collaborators who reside at my current institution (esp. former mentees of mine, we have a real research faculty track where they can be promoted and become independent) 2-3 can be ok, because we have somewhat established roles and know how to work together.

More and more the funding agencies want large groups. I think this is bad. Usually the $ increases with group size but less than linearly. So each PI gets not that much and as a result feels little responsibility, and the management becomes very difficult.

Anonymous said...

I'm in pure math and I mostly prefer ONC=1 or 2. However, a few years ago I started a collaboration with ONC=3, all of us female mathematicians. As a group, we go much slower than if we had smaller ONC, but we reach further than any subset of us could. I'm very fond of this collaboration because we went through a lot of things while writing these papers (getting tenure-track jobs, promotions, 2-body problems, a baby, etc), and we always work with the same energy and enthusiasm as the first day.

Anonymous said...

I have had a long and productive collaboration with one particular colleague. When I was younger I worried that people would think that he was the "brains" behind our collaborative work (he is male, I am female), and I know that this did happen. However, over time I think this problem mostly went away, although every once in a while I meet someone who refers to our work as my colleague's work.

Anonymous said...

An ONC of ~3 'feels' about right for me. Small enough that contributions are obvious/fair & everyone gets a slice of the action. More than that & your distance from actual science decreases (you're the expert on X models called in to beef up the section on X models).
Less than that & the workload is very large & there's less sanity-checking.

Anonymous said...

I'm a new researcher, currently a Co-PI for the first time. In our case, there are 4 partners (myself included) and it feels too large. The partners are not all pulling their weight equally, and I feel that the problematic partner would have trouble getting away with this in a smaller group (although maybe not - as I say, I'm new to this).

I think a smaller partnership would be ideal at my current career stage. It's challenging to work collaboratively for the first time (not counting collaborative work as a grad student, which was real but felt less high-stakes). Jumping into a 4-way collaboration is not a good way to ease yourself into this part of research life.

Anonymous said...

This may be geographic. I think the expected ONC for many EU funding programs is fairly large (> 5) even for the smallest calls.

GMP said...

My ONC is 2-3. With NC=1, I admit I tend to be controlling and if the other person in the same, it gets pretty bad; I get too irritated with that one person and they probably do with me. With NC=2 or 3, the collaboration is small enough that we can still be efficient, but I know I need to let go and things will move at a somewhat slower pace than I'd like.

Anonymous said...

I like NCs of 2 or so best but it's really common in my field to have really large collaborations. I find that these work best (for me and in terms of output) if they are really collaborations of subcollaborations. For example, in one project I'm just finishing, I have a very tight coupling with one other person for data collection and subject specific publication, I have a couple of tight collaborative groups (NC~3) where each of our respective research foci in the same project overlap to say something new, then I have synthesis-type collaborations with the larger group. So, it's sort of a nested hierarchy of collaborations. It means something on that project is always moving forward and keeps me from being utterly frazzled by the chaos of the entire group.

plam said...

I've only had experience with NC=1 so far (and quite a bit); I've never tried larger numbers. I should write a couple more NC=0 papers.

gasstationwithoutpumps said...

1 or 2.

Not 0. I have trouble getting papers written if I work alone (though I often get more work done that way, it doesn't get published).

Large projects spend all their time on unproductive meetings.