At 11:43 AM 4/18/2013, Brian E Carpenter wrote: >Indeed. Ideally, though, we need a statistician to look at the >historical ratios (e.g. M/F ratios) in the attendee lists vs the >I* membership, to see whether there is a statistically significant >bias in the selection process over the years. > > Brian
Yup. We have: 1) The difference in composition between the world as a whole and the IETF. 2) The difference in composition between the industrialized nations as a whole and the IETF. 3) The difference in composition between the set of small medium and large networking companies as a whole and the IETF. 4) The difference in composition between the IETF (for some measure of that) and the composition of the set of IETF working group chairs. 5) The difference in composition between the set of IETF working group chairs and the set of members of the IAB and IESG. 6) The difference in composition between the IETF and the composition of the IAOC. All of these are somewhat interesting statistical analysis problems. My guess (without looking at the numbers) is the (1) is pretty large. (2) is still large, but substantially less than (1), (3) is quite a bit smaller than (1) or (2) - all of these for pretty much any axis you want to evaluate. For your question, I differentiated between the IAOC and the IAB/IESG for one specific reason - an injected bias for the second group that mostly says that paying your dues as a working group chair is a pre and sometimes co requisite for becoming a member of the IAB or IESG. I'm pretty sure the injected bias for the IAOC is "business experience" of some sort. I haven't seen anything on the list that suggests there's selection bias for working group chairs - it's possible there is, but I can't actually recall any complaints going back say 10 years where the selection of a working group chair was intimated to be biased based on gender or any other criteria. Given that the positions are about 90% self-selection and 10% area director approval, I don't know how you could do bias analysis there and get meaningful results. That's question (4). We can provide a statistical number for the difference, but the "why" is going to be elusive. So going forward, what I think we should be more interested in is not so much the composition of the IETF, but the composition of the set of WG chairs and how that population compares to the set of IESG and IAB members. That's question (5). That question doesn't need to be asked on the registration forms for Berlin. I'm not sure we have enough history yet for question (6). I will note that besides the "must have been a WG chair" bias, we also have a "sell by date" bias, especially for IESG members. By that I mean that a perfectly qualified candidate for the IESG who has been serving in that position for say 6 years is less desirable than one who has similar qualifications but hasn't yet served and may end up being considered "disqualified" even if fully qualified. That bias is written into the 3777 considerations as is the WG chair bias. It may be useful to take the set of current (or last 5 year) chairs, and subtract out any chair who has served at least 2 or 4 years as IAB or IESG and do the comparison with the IAB/IESG again. Mike
