[gnso-wpm-dt] WPM: ANNEX-Draft #1
WPM Team Members:
Attached, as promised, is a first draft of the ANNEX in which Liz' feedback
is incorporated. As you will note, the steps and instructions are a bit
shorter less technically complex than I originally envisioned, which I
attribute directly to our model simplification - a welcome outcome!
I made a slight alteration (see below) to Jaime's original methodology, but
I believe that I kept it in the spirit of what he recommended.
Jaime:
a) if Range<=2 there will be no discussion.
b) if Range>2, extreme value raters (and only them) will have a set
amount of time (2 to 3 minutes each) to justify their rating and a new
iteration occurs.
c) After the second voting session If Range<=3 take the median and
stop, otherwise extreme raters defend their rating and a new iteration
occurs.
d) After 3 voting sessions, if Range<=4 take the median and stop.
Ken:
a) After individual Councilor ratings are aggregated, if Range <=2,
compute Median; DONE
If Range > 2, discussion ensues. [Note: see my comments below about
"extreme" positions]
b) After 1st Round, if Range <=2, compute Median; DONE
If Range > 2, proceed to Round #2.
c) After 2nd Round, if Range <=3, compute Median; DONE
If Range > 3, proceed to Round #3.
d) After 3rd Round of discussion and polling, compute Median and STOP
[Note: there is no need for another Range calculation unless we recommend
continuing to a 4th Round].
As noted above, Jamie has recommended that ONLY the extreme ratings should
be "defended" in the group discussion. While I concur that time is a
challenging constraint, I am not sure this approach will lead to consensus
more effectively than an airing of all views. As we saw in our own
testing, an extreme position can be persuaded by a "middle of the road"
explanation; but, if that argument is never heard, the opportunity is
missed. If only the extremes justify their positions, it is almost like the
majority is being convinced to move further away from consensus. Let me
posit an example to illustrate what I mean. Suppose, for a particular
project, the following arrangement of ratings occurred after the individual
Councilor ratings were processed:
Councilors Rating
2 2 (10%)
4 5 (20%)
13 6 (65%)
1 7 ( 5%)
The Range starts out at 5 (7-2), so there isn't consensus by our definition.
In the first discussion round, should the 7s and 2s (extremes) be asked to
persuade the 5s and 6s (85%!) to move their ratings up/down or have the
majority try to convince the outliers that their values are too high/low?
I don't believe either approach will be ideal in all cases. From my
vantage facilitating the team's test discussions, it seemed most helpful
when those with deep knowledge and experience provided their rationale,
which led to greater group understanding and, ultimately, consensus. It
may take longer to do it this way, but, since our goal is group consensus,
the more productive discussion that occurs in Round 1 may prevent the need
for Rounds 2 and 3.
I look forward to our session on Monday. Based upon early Doodle results
<http://www.doodle.com/xfzdc8w9d2ht27nr> , it looks like time will be either
1800 or 1900 UTC. Please check the WPM email list tomorrow (after 1200
UTC) at which time Gisella will close the poll and let us know the time slot
for Monday's call. I will upload the latest versions of both documents to
the Adobe Connect room <http://icann.na3.acrobat.com/gnsowprioritization/> .
Ken Bour
Attachment:
Annex-Work Prioritization (KBv1-LGv1).doc
|