Discourses of Research: Toward Transdisciplinary Research on the Tweets of Teachers

• In Event: 68.049 - Data, Big and Small

Abstract

This paper outlines methodological challenges and some findings from a transdisciplinary study conducted by information scientists, educational theorists, and computer scientists that explores how the tweets of teachers build professional community and respond to issues related to education reform.

The transdisciplinary approach enabled us to use the methods and knowledge from a variety of disciplines; an approach that yielded new methods, questions, and findings that would not be possible using any one methodology or knowledge base alone (Wickson, Carew & Russell, 2006; Gray, 2008). We analyzed 550,508 tweets from 507 accounts of self-identified teachers using machine learning, data visualization, and theoretical inquiry.

“Big data” is used by researchers to answer questions or solve problems; it is important to note that big data is generated by the aggregation of “small data” produced by everyday online practices such as social media use and online search (Wallach, 2014). While members of the research team from traditionally quantitative disciplines approached the dataset with methodological and analytical tools that count and classify, the qualitative researchers approached the data with ethical questions about our research practices and the data themselves. Part of our work became communicating across disciplines toward an understanding of what each other conventionally wanted from data to help us work together.

Each member of the research team brought a primary or “home” discourse to the work (Heath, 1983; Gee, 1987), and each had to learn at least one new “secondary” discourse in order to communicate successfully in the research space. This did not mean that we relinquished our home discourses, but we had to ask questions beyond our home discourses, leading the project to a transdisciplinary space that enabled new ways of thinking.

This transdisciplinary project toggled between data big and small. The educational theorists drew on concepts of democracy to explain teachers’ use of Twitter; however, there was no systematic analysis of the tweets to support that claim empirically. The theoretical claim was based on significant experience with small data such as reading individual accounts and using Twitter on a daily basis. Using machine learning, the computer scientists clustered the over .5M tweets based on the similarity of the kinds of language present in the 140 characters. The information scientists created a network visualization of the clusters. Suddenly, big data met small data: interactions between individual accounts were visible and clusters represented communities based on language and engagement. As a result of this transdisciplinary analysis, the educational theorists were able to refine and operationalize their concepts of community.

Our presentation will provide examples of the contested arenas of our research, such as approaches to IRB, interpretation of clusters, and applying a philosophical lens to terms colloquially used in social media studies. By engaging with these issues across disciplines, transdisciplinary research helped us utilize the bigness and smallness of our dataset, reflect on our home disciplines, and value each other’s methods.

Authors

• Jessica Hochman, Pratt Institute

• Doris A. Santoro, Bowdoin College

• Stephen Houser, Bowdoin College

• Clare Bates Congdon, Bowdoin College

• Bridget Gavlin, Pratt Institute