Big Data Insight Needs Sort

October 30th, 2014 | EVALUATION

To better help museum visitors make sense of large data sets, also called “big data”, this study investigated if there were generalizable ways in which visitors engage with and then make meaning of such data sets. This front-end study was designed to explore if there were different, distinct, and repeatable patterns intuited by individuals as they work with large data sets. This was a descriptive, process method using a complex card sort with an interview. Each card had the name of one food item written on it. Food items were diverse, including eggs, crackers, lasagna, apples, tofu and almonds. The choice of words was intentional with a goal of having foods that would disrupt simple categorization. Participants were given the cards randomly shuffled. As the individual sorted the cards, the evaluator made notes about the approach. Individuals were observed in how they conducted the sort, and then were prompted to discuss both the approach and the final categories. There were two main methods the participants used to sort the cards: Method 1. Sorting cards into emergent categories as the participants flipped through the stack of cards. Method 2. Examining all cards, deciding on categories, and sorting cards based on these pre- determined categories. This study found that in general, people are not likely to change their approach once initiated. This lack of casuistic thinking is not unique to how people make sense of large sets of data, but it is amplified and might suggest a reason that individuals have a hard time making sense of complex visualizations. Most people tend to start with the “local” and begin to force items into the established framework. In this case, local would refer to the typical content analysis framework approach of beginning analysis with the first piece of data rather than looking at a block of data before making categorical assignments. Unlike the research approach to content analysis, no individual in the study went back and recoded all cards based on changes in the categories, although a few did pull cards from existing categories to help support a “new” category, even though the card could belong in both categories. People seem to be uncomfortable with individual outliers. When single cards did not fit into their categories, people pulled other cards that fit in another category, but could also support the individual card. Given that most people initiate their categories in the sequence by which the cards are presented, it would suggest that providing dissonance initially in the data would create more critical organizing. Includes observation protocol.



Team Members

Indiana University, Contributor
Mary Ann Wojton, Author, Lifelong Learning Group


Funding Source: NSF
Funding Program: ISE/AISL

Related URLs

Pathways: Sense-Making of Big Data


Audience: Evaluators | General Public | Museum | ISE Professionals
Discipline: Computing and information science | Education and learning science | Technology
Resource Type: Evaluation Reports | Formative | Interview Protocol | Research and Evaluation Instruments
Environment Type: Exhibitions | Museum and Science Center Exhibits