Hello Oliver & others---
CDC's two software programs, "AnSWR" and "CDC EZ-Text," were explicitly
designed to help solve problems encountered during large, team-based, and
multi-site qualitative studies. Both programs can generate inter- and
intra-coder reliability reports (including kappa statistics for how codes
are used by multiple coders). Both also include features for interfacing
qualitative results with quantitative datasets & statistical analysis
software. Given your list of needs described below, I'd suggest you might
want to look at AnSWR. It is intended to be a software system for
coordinating and conducting large-scale, team-based analysis projects that
integrate qualitative and quantitative techniques. If you have more
specific questions, you can email the AnSWR user support at: [log in to unmask]
Both AnSWR & EZ-Text, along with their documentation, are distributed free
of charge from their websites accessible from:
http://www.cdc.gov/hiv/software.htm
Although one might run into hardware limitations (especially on less
powerful or out-of-date computers), as far as I know neither AnSWR nor
EZ-Text have serious limits on the amount of text per respondent or the
number of respondents. It mostly depends on your research needs, as well as
your available time & resources. For example at the high end, in one large
study I know about that successfully used EZ-Text, the database included
roughly 40 megabytes of text. A different study using EZ-Text included
information from 479 respondents. In citing these figures, I am not
necessarily recommending that people generate voluminous datasets; these two
software programs are used much more frequently on small or medium sized
projects. But when large datasets are needed, AnSWR and EZ-Text may help
users coordinate the efforts of numerous staff working at separate research
locations (e.g., multiple teams in various cities in the United States).
Finally, for further suggestions regarding team-based multi-site research,
you might refer to a journal article by Kate MacQueen and her colleagues
(1998) entitled "Codebook development for team-based qualitative analysis,"
published in Cultural Anthropology Methods 10(2): 31-36. Reprints are
available from the AnSWR & EZ-Text websites.
Cheers to all,
Jim
James W. Carey, PhD, MPH
Methods Research Team Leader
Behavioral Intervention Research Branch
Division of HIV/AIDS Prevention
Centers for Disease Control & Prevention
1600 Clifton Road, Mailstop E-37
Atlanta, GA 30333
USA
emails: [log in to unmask]
[log in to unmask]
> -----Original Message-----
> From: Oliver Treib [SMTP:[log in to unmask]]
> Sent: Tuesday, November 14, 2000 6:59 AM
> To: [log in to unmask]
> Subject: File size: Atlas.ti and NVivo compared
>
> Dear list members,
>
> we are a group of four researchers planning to analyze quite a large
> number
> of expert interviews (around 200 interviews, each with a length of about
> 30
> pages) with the help of a qualitative software package. Important features
> for us include
>
> * teamwork compatibility (i.e. easy traceability of codings etc. done by
> different team members)
> * editing text without invalidating previous coding
> * quantitative data analysis (that feature does not have to be too
> sophisticated, but things like "how often does this or that coding appear
> in the interviews or in certain subgroups thereof?" should be possible)
> * rich text formatting (incl. importing rtf files from word processors)
> (would be nice, is, however, not essential)
>
> From the range of available software packages, NVivo and Atlas.ti seem to
> be the most appropriate packages for the purpose of our research. As I
> understand it, Atlas seems to have a slight advantage in quantitative
> analysis, while only NVivo provides the possibility to edit text without
> invalidating previous coding and has the additional feature of rich text
> formatting. Therefore, we currently tend towards buying NVivo.
>
> What I am wondering about, however, is the question of file size. If I
> understand it correctly, in Atlas you do not import the text to be
> analyzed
> into the project file but leave it in a txt file and store it decentrally.
> This should keep the file size of the main project file quite small. In
> NVivo, however, the text has to be imported and stored in the project file
> together with all the codings, notes etc. I expect this to increase file
> size to a considerable degree, which might make data portability and
> safety
> copying problematic and could possibly also end up in program instability.
>
> Does anybody working with either of the two packages (or any other package
> providing the features mentioned above) have an idea how big a file for
> our
> 200 interviews would become? Any comments would be very helpful.
>
> Regards,
>
> Oliver Treib
>
>
> ¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬
> Oliver Treib
> Max Planck Institute for the Study of Societies, Cologne
> mailto:[log in to unmask]
> http://www.mpi-fg-koeln.mpg.de/people/ot/
> ¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬¬
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|