Another "uh-oh" moment for my thesis

I spent the better part of the weekend reading The Content Analysis Guidebook, by Kimberly Neuendorf. It was originally assigned by Doug and Joe Bond in my ALM proseminar back in 2003, but I've revisisted it a few times since then. (Check out the book's official website, it's an excellent resource)

Anyway, rereading the entire book forced me to consider how I am going to treat variables in my research. In the field of computer-assisted content analysis (a subset of which is computer assisted text analysis, or CATA) this is a very rigorous process. Unfortunately, my simplisitic view of foreign policy issues as currently formulated cannot easily fit into a logical system of measurable dependent and independent source variables and corresponding message variables. I had been considering issues like "overlapping territorial claims in the South China Sea", "Vietnam's treatment of ethnic Chinese," "Vietnam's relations with Kampuchea", "Vietnam's relations with the Soviet Union" to be equal variables in terms of measuring them in NCNA coverage. But they cannot be. "Ethnic Chinese" is a single variable, but something like "relations with the Soviet Union" encompasses a mass of sub-issues, including military cooperation, economic aid, etc.

This led me to a fork -- should I break up "country" variables into smaller, more easily compared pieces? That would be problematic, I concluded. There are just too many of them over the 15-year period under study. They are not only are hard to catalog, but some may also not lend themselves to database searches and measurement.

Another option I am considering is simplifying my methodology to only measure countries (Vietnam, Kampuchea, and the two superpowers) and including Laos as a basis for comparison with Kampuchea in my research. As many NCNA articles' formats are corrupted in Lexis Nexis, and don't have segmented lead paragraphs, I would use the next best thing -- article headlines -- to identify the focus of articles about Vietnam, and then full-text searches of the other country-associated terms to create a matrix that reflect the importance to NCNA (and by extension, what Lampton calls China's "leading nucleus") of certain country-variables, and correlations between multiple country-variables.

I'll post an update later this week about how this plays out. I would like to get started on the actual number crunching before the next thesis-writers' meeting on the 20th.