For the past few years, I’ve been fretting about how to begin to analyze the large body of relatively unstructured data collected from our research in the Bakken oil patch. This includes thousands of photographs, hours of video, interviews, and various notes. Most of our preliminary analysis has drawn upon our field notes and selective and impressionistic readings of the data that we collected. This is not to suggest that our analysis is wrong, but it lacked a certain amount of nuance in part because we were overwhelmed by the quantity of data that our methods produced.
The issue is partly because we decided early on to collect data at the regional level largely because we we were not entirely sure what our sites would look like and how to best document them. After a few trips, however, we had identified over 50 workforce housing sites across the region that clearly housed workers associated with the oil boom, had a diversity of units (generally RVs) and approaches to life in “the patch,” and showed signs of change through time. We then used photography and video to document these sites over the course of numerous short field seasons of only a few days. The resulting archive captured the dynamism of the Bakken oil patch through time and a remarkable level of detail about individual workforce housing sites and units.
Over the summer, I had a few fascinating conversations with a Colorado Ph.D. student, Erin Baxter, whose dissertation research used Atlas.ti to organize and analyze photographs that formed the only historical record of a century-old excavation by Earl Morris in the American southwest. She explained to me how she used the software to track various features of the excavation through multiple photographs. Atlas.ti also made it easier for her to organize and analyze the photographs including certain features or chronological indicators that would allow her to reconstruct the history of the excavation. (I’m sure it much more complex than what I described, but that was my take away!)
This prompted me to write a little grant and get a copy of Atlas.ti (which isn’t cheap!) and to begin to use it to code my photographs from the Bakken. This week, I ran a pretty basic trial of 70 photos taken in October 2014. These photos produce the following list of codes which correspond either to features or conditions visible in the photographs:
The code list is still in a bit of flux and will undoubtedly be expanded, but after even just 70 photos, it is a pretty good summary of objects and conditions associated with workforce housing in the Bakken.
The photographs that produced this code list are group according to date and camp number and when possible by unit in a camp. This will allow me to consider changes through time and across different camps while also controlling for our tendency to take more photographs of particularly interesting units or units with substantial number of associated features and objects. While we are not coding images to produce explicitly quantifiable data, it looks like we can use the grouping function in Atlas.ti to allow us to document the distribution of features proportionately across our study sites.
Finally, Atlas.ti will also allow us to code video and text which we can also group according to site. With any luck this allows us to connect more explicitly our evidence from interviews and systematic video with our photographic documentation.
So, stay tuned as I explore how Atlas.ti can create a more nuanced image of workforce housing during the Bakken boom.