Dan raises some interesting points:
> There is something that concerns me about the discussion of name
recognition
> in software, and how acceptable (or unacceptable) reported research may be
> based on what software was used to analyze it. It seems to me that while
> the software used certainly will effect the analysis process, the
> *methodology* used is much more salient to the question of whether or not
> conclusions drawn can be related in a meaningful way to the data. As the
> use of a type of software becomes ubiquitous, it slowly fades from view.
As
> Jens noted, no one asks you which statistical analysis package you used,
or
> which word processor. This should be relevant only when it effects the
> methodology.
I would also add that we may have forgotten the technological developments
in qualitative data collection techniques. Some of us remember the days
before the tape recorder became a prerequisite (OK nearly) of qualitative
data collection. We were taught how to quickly excuse ourselves so we could
slip into the ladies (or gents) to write up our observations. Uwe Flick
reminds us of this in her Introduction to Qualitative Research, Sage,1998.
She argues that while the introduction of the tape recorder demoted the
value of observer's notes (for example), it made other types of analysis
(for example, conversational analysis) possible.
Regarding Dan's further point -
>
> This, of course, raises the question of what methodological biases
different
> software packages have, and even if those biases exist, whether or not
they
> have a major impact on the research conducted using that package.
Flick is rather pessimistic. She raises the possibility that the spread of
computer programmes for qualitative analysis may marginalise certain
methodological approaches because they are incompatible with computers.
Although she does recognise the value of these programmes in making
transparent the analytical process.
I am more optimistic than Flick about the spread of software packages for
qualitative analysis. First of all, it is important to remember that they
are computerised versions of the manual tools which were used before - index
cards, card sorting machines, cut and paste, colour coded highlighting,
writing codes in the margins of transcripts, tables to show patterns in the
data, maps of relations between themes etc. Some of these tools were easier
to computerise than others - so coding and cut and paste became computerised
first. But over time more and more of these manual tools are being
computerised plus new tools are being introduced, such as text search and
hyper-links which were not possible manually.
These tools which I have listed above are tools common to a number of
approaches used in analysing qualitative data. Renata Tesch described them
as falling within what she calls interpretational qualitative analysis and
theory-building qualitative analysis. It is just you would use these tools
differently depending on the approach you are taking. (I recommend having a
look at Tesch's book Qualitative Research: Analysis Types and Software
Tools, The Falmer Press, 1990- although the software part is completely out
of date, the first half of the book on the different approaches to
qualitative analysis is very interesting - I wouldn't take it as gospel but
that was not her intention.) Now all of the software packages usually
discussed on this list: Atlas.ti, the Ethnograph, NUD*IST, NVIVO, WinMax (in
alphabetical order!!) use a similar set of tools which are aimed at this
area of analysis. And a lot of people who subscribe to this list are using
approaches to analysis which fall into these broad categories. Hence, the
number of people who said they had not heard of a different set of packages
someone recently posted on the list. I suspect (but I don't know!) the
reason why those packages are not so well known on this list is because they
offer a different set of tools for a completely different approach used in
analysing qualitative data. However, I do feel that within the
Atlas-to-WinMax range of software there are subtle differences and that some
of these packages are better suited to certain approaches to analysis. I
can only comment on the packages I know very well - the ones for which I run
training workshops - which are Atlas, NUDIST, NVIVO and Decision Explorer -
(Decision Explorer is a clear example of a different approach to analysing
(and collecting) data called cognitive mapping - I won't discuss it further
here.) I am familiar to a certain extent with the Ethnograph and WinMax but
I would like to have time to play with them more before making definite
statements about them.
Some approaches to analysis, such as Grounded Theory, you can use with any
of these packages. But because the tools are slightly different in each
package, you would do it in different ways. (Just like in my first
grounded theory study back in 1972, I used a filing cabinet, typed several
carbon copies of interview notes and observations, cut and paste them into
different folders, piled up bits of paper into different piles, all the time
writing in my journal my analytical thoughts how it all hanged together - in
a study I did later in the 70s I dispensed with the filing cabinet and
carbon copies and used colour coded index cards instead, which I could
shuffle in different ways, and large analytical tables to help me see
patterns in the data - and I still wrote up my analytical thoughts in a
journal) The point is I used different manual tools for the same approach to
analysis. The important thing is to be clear about the methodological
approach you are taking.
But I did say there are differences between them. Atlas's great strength is
its network tool. Because you can specify the types of relations between
elements in the network, I think that it is particularly well-suited to some
forms of ethnographic research -for example, domain and taxonomic analysis
as described by James Spradley. NUD*IST's great strength has been its Index
Search tool but that has now been eclipsed by its cousin, NVIVO's Search
tool which is even more powerful. This tool is used when you want to cut
your data in different ways to discern patterns. NUD*IST is still a good
workhorse in that its simpler structure makes it easier to handle large
amounts of qualitative data. Some discourse analysts have told me that they
are excited by the possibilities of NVIVO as it takes rich text so they can
use their preferred style of using formatting and colour to mark up their
text. I know that some people are using WinMax as they feel that its clear
four window display makes it easy to use with undergraduates who are
learning how to do qualitative analysis.
I could say more but I think that this e-mail has grown long enough.
(Apologies if this is a bit of a ramble but it is late at night.) I'll just
make one more point. If you are interested in any methodological biases in
these software, read what the developers themselves have written. They all
have web-pages you can access - see the webpage for this list to find them.
I was running a NUD*IST workshop recently and one of the participants
reminded me that Lyn Richard's article 'From Filing Cabinet to Computer' was
a great introduction to understanding the logic behind NUD*IST (sorry I
don't have the complete reference but I am writing this away from my home
base - but I am sure the article is on the QSR website).
Dr. Silvana di Gregorio
SdG Associates
Research and Training Consultants
Tel/Fax:+44-(0)181-806-1001
E-mail: [log in to unmask]
Http://www.sdgassociates.demon.co.uk
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|