Smoke and Mirrors Research Methodology

Choosing a Topic

In choosing a topic, I wanted to research something that aligned with my academic interests in Science Technology Studies (STS). The production of seemingly “objective” scientific knowledge at a renowned research institution that was (and, to a certain extent, still is) backed by industry with interests in manipulating public opinion was an extremely intriguing topic to me.

“Big Tobacco” is perhaps the industry most notorious for manipulating the public’s trust of science to sell products. Duke researchers of the 1950s and 1960s were well aware of these tactics, and sought to reap the benefits of research funding without any of the stigma of aiding a potentially-harmful industry, as evidenced by Duke researcher and administrator Marcus Hobbs’s insistence that “ through a previous understanding with the Damon Runyon Foundation, there was to be no publicity regarding grants from Damon-Runyon or work carried out under the project at Duke University” (M.E.H. – Note on Telephone Conversation 5-17-55, Marcus Hobbs Papers, Damon Runyon Correspondence 1955 Folder, 1955.05.17).

Collecting the Data

It is this obscuring of the links between the tobacco industry and the research university that inspired my research. Where was I going to find this information, if it was so obscured? Luckily, due to a 1994 lawsuit by the state of Minnesota against major tobacco companies was the impetus for the creation of an online database of thousands of tobacco-related documents. I also had access to the Duke University Archives, and found a lot of the more Duke-specific documents cited in my project in the Marcus Hobbs papers. I kept track of the dates, sources, and important quotes from the documents in different Evernote documents, and used a different “note” page for every notable person.

Representing the Data

My next question was “How do I represent what I find in a meaningful way?” I didn’t want the form of representation I chose to be linear- the connections between Duke and the tobacco industry are complex and don’t proceed in any logical or temporal way. Eventually, after asking Tim Stallmann, who specializes in countermapping and “unconventional” maps, I was led to Gephi, an open-source social network mapping software that can create complex social networks using statistics.

Gephi generates network maps using excel spreadsheets, so I set to work compiling pages of notes into lists of nodes (actors, like people or organizations) and edges (connections between actors). However, I soon found that gephi needed a little more coding and statistics experience than I had, and after consulting with Brian Norberg, I also couldn’t easily export my Gephi maps online to provide visitors of the website with the interactive experience I wanted them to have. Brian suggested I use Kumu, a proprietary social network mapping software that is web-based. I easily converted my Gephi spreadsheets to spreadsheets that Kumu could use and imported my data.

Once the data was fed into the program, I had to do something I didn’t expect- some basic HTML coding! When setting the different layouts of my maps, which included a map that was sorted by statistical communities and a map which allows you to display the data broken down by timeline, I had to fine-tune the layouts using HTML code. This wasn’t easy- I had no familiarity with HTML coming into this project, and Youtube tutorials became my best friend for a few days. The end result was something that I believe is as user-friendly as possible.

Reflections

Upon reflection, I think choosing to represent the information I collected in social network maps carries some major advantages over other forms of representation. From the researcher perspective, network maps allowed me to show complex relationships in ways more narrative-heavy forms of representation couldn’t- for example, showing directed connections with arrows or two-way connections with two different arrows. Once I had finished entering my data, the more connected points were apparent, and it provided me with direction for further avenues of research. The most important advantage, however, was the ability of the same data set to tell many different stories. I tell one such story, the story connecting Duke researchers Marcus Hobbs and Jed Rose, in our team’s presentation at the Story+ Research Symposium (linked here). I could easily have focused on a completely different relationship using the same map. Additionally, the different maps I created tell different stories by sorting or highlighting different aspects of relationships. The community analysis map gives the viewer a different takeaway than the timeline map, for example. Mapping also has the advantage to the viewer of self-guided exploration- the viewer can spend as much or as little time viewing specific connections or elements.