Dear Diane,
you wrote:
>I am at the stage of reading up on method to decide how to handle 40 text
>documents looking for ideology within the text - ie how do the texts
>represent the institute from which they came to the public?
>
>I would appreciate comments on this, I think I can do one of two things -
>1. count the words which are ideological indicators such as economic,
>values, or community type words and see how the counts compare or
>2. code the documents and then count these indicators
I would suggest a combined quantitative plus qualitative approach. You may
do 'traditional' content analysis for a start, counting words that appear
most frequently in your dataset. In this way, you get an idea of important
concepts, values, or symbols that are used to represent the specific
institute to the public. And you may also compare the documents among
themselves and create some sort of taxonomy. Besides, you may thus map the
'discursive universe' in which these documents subscribe (i.e. the
various values, symbols etc. used for the purpose of representing an
institute to the public)
As a second step, I would suggest that you isolate short segments of text
(e.g. a few lines, a small paragraph) which concentrate on the
representation of the institute to the public and analyse them
qualitatively. The type of analysis I would propose is either frame
analysis or discourse analysis in general (if you are not familiar with
these, I can give you some references). THus, you may decipher the
particular rhetoric/discursive strategies used in each text, the specific
way in which a given term or value is presented, the elements it is related
to etc.
In my view, this double strategy can avoid the shortcomings of traditional
content analysis where a word or wordstring is isolated from its context.
At the same time it also prevents you from falling into common traps of
qualitative analysis such as overseeing important elements in the text
and/or your findings being too much researcher dependent (problems of
reliability).
I believe that both Atlas.ti and Nudist would be of help in doing both the
above tasks because they both allow for quantitative operations (counting
of words or word strings, counting of co-occurence etc.) and they can also
facilitate the retrieval of segments and the organisation of your
qualitative analysis.
Best,
Anna Triandafyllidou
---------------------------------------------------
| Dr. Anna Triandafyllidou |
| Marie Curie Post-Doctoral Fellow |
| Institute of Psychology |
| Italian National Research Council (CNR) |
| viale Marx 15, Rome 00137, Italy |
| tel. +39 06 86090220 |
| fax +39 06 824737 |
| email: [log in to unmask] |
----------------------------------------------------
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|