In terms of recreating my project under the lens of a small–n qualitative methodology, one of the main things to consider is how I would change my dependent variable so that it is now qualitative, instead of quantitative, and how I would use my data source to operationalize it. With respect to this, there are a couple of ways that I can complete this. I can either classify my indicators as “whether or not a system is operational 24 hours or not”, or I could create “low, medium, and high” categories that would correspond to the number of hours that each system is open. For example, in the first classification, I would use my data to place Berlin as a “yes”–indicating it is open 24 hours a day–and I would put Paris as a “no”–signifying that it is not open all day.[1] In the second system, I would put Brussels in the “low” category, having fewer than 19 operational hours; Paris in the “medium” category for having in between 19 and 21.5 operational hours; and Berlin in the “high” category for having more than 21.5 operational hours.[2]
Within this system however, there are some difficulties with deciding where different systems belong. This comes from the fact that some systems have a different number of hours for different lines, or the fact that they operate for different times on different days. One solution would be to indicate that a case could be in two categories with a “sometimes” for the first system (yes/no) or a “low/medium”, “medium/high”, or “low/high” for the second system (low/medium/high). Another would be to simply prioritize one schedule over the other, although that is not ideal because that intentionally excludes some of the data that I am finding.
I am still having a little difficulty with case selection, because I am attempting to use Mills Method of Difference to figure out how systems end up with the same dependent variable result while being drastically different in my independent/intervening variables. Despite this, I am looking towards using two systems that are “yes” or “high” and two systems that are “no” or “low”. I can also expand this to include two cases of “medium”. I would also try to make sure that each pairing has different values of the independent variable, so that I can look at how the intervening variables have an effect on the operational hours of metro systems.
Bibliography:
“Mapa Metro,” Mapa Metro, last modified 2010, accessed October 24, 2017, http://mapa-metro.com/en/.
[1] “Mapa Metro,” Mapa Metro, last modified 2010, accessed October 24, 2017, http://mapa-metro.com/en/.
[2] Ibid.
You are off to a good start here, Nathan, but you might think about shifting your DV even more to leverage the strengths of this methodology. Instead of just converting your numerical scale of operating hours into a word scale, think more in terms of the specific case outcomes that you want to explain (successes or failures, counterintuitive results in a specific case, etc.). In other words, shifting to a small-n mentality requires you to rethink the *purpose* of the research as well as the form of the DV.
With that in mind, you might begin by asking which cases are particularly puzzling? Remember that in small-n research you need to *know* the outcomes that you are explaining (just as Howard knew, from the outset, that she was explaining the difference between some successful and some failed PKOs, or how Saunders knew that she was explaining as particular shift in US strategy). Then, you think about the particular ways in which you’d capture the DV in the case or cases that you propose to analyze.