Computer Aided Content Analysis
Content analysis software is used to analyze texts in two related, but different
kinds of sociological studies. Most typically, sociologists use content analysis
for qualitative projects that use a broad research method called discourse
analysis. Typically, they choose texts of interest produced by a community
of actors whose interpretive practices are the subject of study. A goal common
to these studies is identifying basic trends or main categories. Typically,
such approaches are supplemented by other qualitative research methods used
to explain why different subpopulations in a community choose particular narrative
devices.
Some sociologists are interested in using the results of content analysis in
quantitative multivariate data analysis, to model outcomes using statistical
software. Typically, regression is used when the outcome variable is continuous.
A number of statistical models have been developed that allow the application
of "regression" like modeling in situations where the outcome is
either nominal or ordinal. The two that have been most popular in the sociological
literature are the "logit" or "logistic regression" and
the "probit" models. Logit and probit can be fairly easily applied
to cases where the dependent variable is either nominal or ordinal, and has
two or more levels; and the independent variables are any mix of qualitative
and quantitative predictors.
One of my main goals in this project was to choose a relevant software package
that could be used in either of these kinds of computer aided content analysis
(following established conventions, I refer to the first as qualitative, and
the second as quantitative, although clearly these terms may not be entirely
accurate).
The Murdock grant was used to acquire, and begin to develop a teaching protocol
for content analysis as a research method. In June 2002, I recruited two students
whom I later taught to use basic content analysis, and who were then to decide
which kind of software was best suitable for qualitative as well as quantitative
applications. We tried out a software program available for the Macintosh platform
called Nudist. In the future we hope to look at software for the Windows platform
called Ethnograph and N*vivo, two packages that are recommended by most advocates
of qualitative sociological methods.
Our experience with Nudist, as well as our surveys of reviews of equivalent
software programs, led us to conclude that in many cases, computer-aided content
analysis required a great deal of investment (in time and effort) for a relatively
small pay-off. At minimum, a user must invest at least a month in getting a
feel for content analysis, then a further two months in adapting texts to be
coded by Nudist, and I could tell that my two students, faced with a three
month period of preparation for an uncertain reward, would probably never conduct
their own content analysis.
Over the course of the summer during which the first phase of my Murdock project
was conducted, my students struggled to use Nudist to analyze six interviews
with informants. However, they made no progress toward carrying out their own
independent projects (both students went on to do excellent, rigorous quantitative
theses).
For the near future, I recommend that most students use the Excel spreadsheet
software, rather than Nudist, for basic content analysis. My experience is
that Excel is much easier to learn to use, and that most student needs are
better served by this fairly quick method of recording data analysis.
Nudist may be a good option for those students who wish to do a simple “search
and tally” analysis. However, if they wish to analyze a text that requires
hand-coding, they must be warned that developing a sampling method, appropriate
units of analysis, and a good coding scheme will be extremely time consuming.
Many kinds of qualitative data analysis will require hand-coding, and can probably
be more easily conducted using Excel spreadsheets alone, or in conjunction
with Stata statistical analysis software.
One final comment: Nudist was developed for Mac OS9 and earlier, and its developers
have indicated that they do not expect to develop future versions for Mac OSX.
Back to List of Projects
Return to Murdock Grant home
page