MB870

Content Analysis Paper

The objective here is to code a set of 7+50 texts (or more) on at least 5 variables and correlate the codes with aspects of the persons or entities associated with each text. Each text must be coded by at least 3 different coders. The first 7 texts are used to get the kinks out of the coding process, and are coded collaboratively. The next 50 texts are coded independently. Inter-rater reliability is calculated on both sets of texts. The codes are then related statistically to other variables based on your hypotheses. Examples:

  • Personal Ads. Evolutionary theory says women will offer (and men will seek) youth, looks, sex appeal while men will offer (and women will seek) age, status, security. Code a set of personal ads and then correlate presence of absence of these items with gender.
  • Student Essays. Get a bunch of undergraduates to write essays on some topic, such as what they did last summer. Make some predictions about the kinds of activities that male vs female students (or any other groups) will mention more often.
  • Perceptions of Political Parties. Get undergraduates of different political affiliations to discuss the economy. Examine differences in frequency of themes by party.
  • Abstracts of Articles. Analyze abstracts from two different journals in the same field to understand differences in editorial interests.
  • Web sites. Code corporate web sites and relate the presence/absence of codes to some attribute of the organizations, such as size or international focus, or social responsibility.

Notes

Keep the texts short -- otherwise you will spend too much time on the actual coding. Make sure the texts are comparable -- don't mix together web sites with brochures. Best to do this in large teams (say, 4 people) in order to reduce work load.

You might want to use text coding software such as EZ-TEXT, which is designed for this kind of text analysis, or use Atlas.ti, which was not (but it might be better to keep up your Atlas skills).

You need to specify some hypotheses that the research is intended to test. Also, it is *required* to report reliability -- it is a key part of the study.

This assignment may be done by individuals or groups. I suggest groups.

Report

The report should be very concise. A single paragraph motivating the study. A single paragraph outlining your theory. A couple of hypotheses derived directly from the theory. A somewhat longer section outlining each step of the data collection and analysis. A few paragraphs outlining the results (use tables to simplify presentation of information). A single paragraph drawing larger conclusions. The whole thing should be 5-10 pages.

Please submit the report by e-mail.