Disciplinary frames to working with data
One of the goals of this blog is to help you become better positioned in understanding the networks of research data services happening across the Harvard community. There is a lot of data research happening at Harvard. There are also lot of places to get help. In our GIS support services, we encounter a wide range of disciplinary backgrounds when we are helping researchers with their data. The software, methods, and conceptual approaches to working with data can vary greatly, depending on which school you are coming from.
In this post I’d like to surface some experiences of my colleagues in Harvard’s Digital Scholarship Group , which is an interdisciplinary team of faculty and staff advancing facility with digital research methods. We talk with Gabe Pizzorno , who is the Senior Preceptor on History and is the Faculty Chair of the Digital Scholarship Group. We also talk with data visualization specialist Jessica Cohen-Tanugi , who is coming from a background in physical sciences, and has worked across the board in terms of subject specialties. Jess is well-positioned to dicuss how different approaches to working with data can impact research outcomes.
JESSICA COHEN-TANUGI: I came from the physical sciences, and one thing I have drilled in me from my training that I don’t always see in the humanities is a healthy layer of skepticism for anything I create. For example, if I make a chart or graph, I assume I did something wrong, and it’s part of my regular practice to poke around and search for inevitable mistakes. Think about doing physics homework – you naturally make a lot of mistakes. We get that training in the sciences working with complex calculations and equations. One time, I was working on a problem where I forgot to take the absolute value of something, and I found an amazing trend my adviser was so excited about and even shared with colleagues. Of course, once I fixed the mistake, the amazing trend disappeared. This critical eye helps me a lot as I support students using black box tools like Tableau, where decisions are taken out of your hands. I’m suspicious of this, and am constantly trying to understand the defaults the program might be making for you without you realizing it. So this training has stayed with me.
On the other hand, humanists do have real advantages when it comes to doing data work. I look forward to seeing what Gabe has to say about this. One thing I’ve observed is a tendency in humanisists to be able to put data questions in context. In the sciences, I would get laser focused on a particular problem and why something wasn’t working. The humanists I’ve worked with are connecting the work with deep background. I was working with one researcher who was struggling with questions about how to break 16 categories of race data down to smaller buckets. He had to make more general groupings in order to be able to see any trends, and was grappling with these choices, because of the complexities of racial data. When I was working in the sciences, we often thought, “these things are similar, let’s group them together”. I think most scientists could benefit from some data humanities training.
GABE PIZZORNO: The way we tend to understand data naturally in the context of the humanities is sort of multifaceted, similarly to how we might think about text, meaning that data are susceptible to the same kinds of human interpretation. I think that is the biggest difference from other disciplinary approaches. That’s something we can bring to the table.
JESSICA COHEN-TANUGI: In Gabe’s sessions in the Digital Scholarship Group workshops , he reminds us that data are not objective. When I first heard that, I did not like it. Gabe could see I was making a face when he said it, and he started laughing. I think I was struggling for the same reasons I am a good scientist. In science, we’re trying to take as accurate measurements of the world as possible. You want as perfect an observation of whatever you’re studying as is humanly possible. “What do you mean data is not objective? I made sure that measurement is objective. I used a great thermometer. I checked it was working. I put it in the ground myself!” Then you start thinking about it and you can see how your own research motivations do impact the data you collect. I did have to decide that three decimal places was enough. I did have to choose the brand of thermometer. The moment you take an observation of the world, even just a photo, you’re limiting the data that comes in. We can’t record everything, we have to make these decisions. It’s not a bad thing. We think of “bias” as being bad, but it’s just a fact. Rather than this making me feel like my efforts to minimize bias are futile, I now take those extra steps to communicate what my motivations were for each project, and how I think my choices may have impacted the data.