This continues our series of student reflections and analysis authored by our research team.
When tPP was still in its coding stage, I enjoyed perusing through documents to find the codable variables within our dataset. I was consistently interested in where defendants fell in terms of ideology and group affiliation. I would often notice similarities between certain groups. I had questions about what may have an influence on the patterns I was seeing. When deciding on a topic for my analysis, I knew that I wanted to look into group affiliation on a deeper level.
A defendant’s affiliation generally has a big impact on a multitude of other variables like othered status, location of attack, foreign affiliation and more. The challenging part of focusing on group affiliation was deciding on what other variables to investigate in combination with it. My first attempt at an analysis of the dataset looked at the effect of othered status and foreign affiliation on group affiliation versus no affiliation. I was able to use the numbers from the dataset to produce a large amount of facts and figures, but it quickly proved to be too many variables and far too many patterns to analyze. I did not find relevant patterns in the areas I was expecting, either. I was interested in far too many variables to produce an intuitive paper that explored group affiliation in the manner it deserves.
I started to organize the entire dataset by utilizing pivot tables within Excel. Pivot tables are tables of statistics used to summarize another, larger data table such as tPP’s. With this feature, I was able to easily insert or remove any variable from the dataset in order to see what kind of findings it would produce. I stuck with using group affiliation as a base and Excel provided me with massive tables, grouping defendants into rows. It showed me precisely how many of them were involved with each group. When inserting a variable like foreign affiliation, I was able to easily see which defendants had a connection to a foreign country. The titles shown on the table were no, yes and unknown which are the three options for foreign affiliation coding. Underneath each title was a list of group affiliations that fit into the category. Each group affiliation row showed the number of defendants that applied to one of the categories. While using each of the 1,194 defendants data produced an overwhelming amount of numbers, I was able to identify a handful of clear and relevant patterns. For example, 91.6% of the defendants who are affiliated with Al Qaeda are foreign affiliated (132 people). When looking at the 348 defendants who have no group affiliation, the PivotTable clearly shows an overwhelming majority (330 of 348) have no foreign affiliation. This method of organizing the data’s numbers was very helpful in deciding what specific groups and variables would produce information worth investigating further.
The sample of tables below show a part of a pivot table that looks at group affiliation in relation to foreign affiliation as described above.
While I enjoyed looking into each of the variables, I struggled to write an analysis on group affiliation in relation to just one of the other variables. I ended up using a lot of the relevant findings in my writing, but it did not come together well; it was too broad. I eventually decided to analyze the data in a way that I had not considered before. I chose to use heat mapping to represent where individual attacks had occurred. Rather than putting all of the data into the system, I chose to sort by group affiliation. Through this method, I have begun to reveal findings through analyzing group affiliation and ideology in relation to geographic location which has already proven to show relevant patterns of attack.
– Jessica Enhelder