Tuesday, January 10, 2006

Underestimating time requirements for ALM research

I totally underestimated the time it would take to get the main portion of my research done. That is, the actual searches in LexisNexis, and entering the data into my Excel spreadsheet. (Most of my non-primary source research -- reading books, journal articles, and other sources on Chinese media, NCNA, Chinese relations with Vietnam, foreign relations theory, content analysis techniques, etc. -- was completed in the autumn.)

I remember telling the group at the ALM thesis writers' meeting back in November that it would probably take two or three days (meaning 16-24 hours) but it's taken a lot longer than that.

Here's a rough summary of the stages involved, and the time its taken each one:

1) Learning the advanced features of LexisNexis: About four hours in late November

2) Testing LexisNexis and creating my own search terms/variables: About 30 hours over a three-week period from late November to late December

3) Creating and testing a spreadsheet to hold the data, and run calculations upon the data entered: About 10 hours to create a 17-worksheet spreadsheet representing each of the 17 years between 1977 and 1993, each with a column for each month plus the total for the year (13 columns in all) with approx. 125 rows (!) to contain search results, derived results, percentages, and ratios. That's at least 27,625 cells containing data, although I will only enter data for about between 23 and 30 rows per year (variable because of the 1000-hit limit in LexisNexis, which forces me to break up certain variables into multiple searches in a given month)

4) Creating a census of NCNA stories: About 10 hours

5) Running the searches in LexisNexis for each of my 22 variables and combinations of variables (e.g., V, V+K, V+L, V+S, V+K+S, etc.): I am not done yet, but I estimate it will take 16 hours to conduct a total of 4,488 searches in LexisNexis.

That's 70 hours total, when I can find the time (more on that later). So far, I have completed more than half of my variables/combination searches in LexisNexis over the entire time span -- I am doing it variable by variable rather than year by year, and it takes approx 45 minutes to do a search for each month from January 1977 to December 1993.

Next installment, I'll discuss some of the challenges creating my search strings.

No comments: