At 09.37 -0700 97-10-16, Francoise Herrmann wrote:
>Hi all, I thought yesterday that I would be most interested in seeing your
>coding schemes
>What are you coding for?
Oh, when it comes to countables I have so far adhered to the fairly
uncontestable basics. I have gone through the archives and pulled out
Date:
=46rom:
Subject:
and
List:
from the mail headers. Hard trivial work. I keep thinking of Latour &
Woolgar's "Laboratory life" about the importance of labelling test tubes
with the right stickers etc. Even so there are coding decisions to make:
like what to do with the same message sent to several lists simultaneously
(keep multiple entries). And discerning what is a user mis-send of doubles
(keep) and doubles that must be a freak of the machinery or slip of the
archiving (delete). I'm sort of done, but there are gaps, notably a BIG
data loss from the procmail server of June 96 through to... what was it...
beginning of October. There I had to rely on my own archives. And I only
saved what interested me... (Anybody got records of the stuff that wasn't
interesting?)
>How are you training your raters?
The cadre of raters consists of me... It has been an interesting in-job
training, however, mediated by a couple of very responsive artifacts: Word
5, to which I am already very entrained -- I have tamed it and it me. And
=46ilemaker, where *it* was certainly the boss to start with. It's getting
more well-behaved now. Time, struggles and feedback.
>what is the scale that you are using? what kind of inter-rater reliability
>coefficients did you come up with?
Umm... ummm...
>In contrast to expository writing
>or perhaps not-so dialogic writing as email communication, I am wondering
>the kinds of adaptations that you have done.
Not done yet. Not quite thinking of it that way either. When I get into the
content, which I surely will, for some selected theme, I will probably not
go for coding to a scheme, but stay at a more exploratory stage of
research. As you have read my dissertation you will know that I have been
trained in the phenomenographic approach, which is basically about spending
your wits on finding categories that emerge from the data. Today I am not
so sure I would deploy my resources in this fashion forever, always
constructiong new categories for ever new phenomena. What I'm saying is
just that this training has made me wary of adopting and adapting coding
schemes. And presumably, since I have spent so much time doing this kind of
analysis, I have developed skills in attending to the material that still
are useful -- it would be a pity to waste what this "human tool" has been
calibrated to.
Will keep you posted when something happens.
Eva