Information Technology Reference
In-Depth Information
student will temporarily bestow upon him/her a form of
situated expertise
(Zhou,
Zemel, & Stahl, 2008) such that he/she is expected to provide an extended sequence
of
expository
postings (Mercer & Wegerif, 1999).
The discussion during a data session can be quite unorderly. Different people
see different possible understandings of the log and propose alternative analyses.
Generally, discussion of a particular posting continues until a consensus is tenta-
tively established or someone agrees to look into the matter further and come back
next week with an analysis. Notes are often taken on the data session's findings,
but the productive result of the discussion most often occurs when one researcher is
inspired to write about it in a conference paper or dissertation section. When ideas
are taken up this way, the author will usually bring the more developed analysis
back for a subsequent data session and circulate the paper.
In coding analysis, it is conventional to train two people to code some of the same
log units and to compare their results to produce an inter-rater reliability measure
(Strijbos & Stahl, 2007). In our chat interaction analysis, we do not pretend that the
log can be unproblematically partitioned into distinct units, which can be uniquely
assigned to a small number of unambiguous codes. Rather, most interesting group
discourse segments have a complex network of interwoven references. The analysis
of such log segments requires a sophisticated human understanding of semantics,
interpersonal dynamics, mathematics, argumentation and so on. Much is ultimately
ambiguous and can be comprehended in multiple ways—sometimes the chat par-
ticipants were intentionally ambiguous. At the same time, it is quite possible for
analysts to make mistakes and to propose analyses that can be shown to be in error.
To attain a reasonable level of reliability of our analyses, we make heavy use of
data sessions. This ensures that a number of experienced researchers agree on the
analyses that emerge from the data sessions. In addition, we try to provide logs—or
even the entire session data with the Replayer—in our papers so that readers of our
analyses can judge for themselves the interpretations that are necessarily part of chat
analysis.
Describing Group Practices (Generalizability)
The research question that drives the VMT Project is: What are the distinctive mech-
anisms or processes that take place at the small-group level of description when
the group is engaged in problem-solving or knowledge-building tasks? Therefore,
we are interested in describing the inter-personal practices of the groups that inter-
act in the VMT environment. There are, of course, many models and theories in
the learning sciences describing the psychological practices of
individuals
involved
in learning. At the opposite extreme, Lave and Wenger's (1991) theory of sit-
uated learning describes social practices of
communities
of practice, whereby a
community renews itself by moving newcomers into increasingly central forms of
legitimate peripheral participation. However, there are few descriptions specifically
of how
small groups
engage in learning practices.
