Abstract

Data-driven decision-making has become the gold standard in science, industry, and public policy. The trouble with data is that it frequently provides only an imperfect and partial representation of a phenomenon of interest. The gap between data and reality makes data alone insufficient to make good analysis decisions and data interpretations, and, as a result, analysts and experts utilize personal knowledge from various sources to fill in the gap between data and reality. In practice, personal knowledge is typically not incorporated in analysis tools in a structured way, which is problematic if others who lack that knowledge interpret the data. This dissertation centers around the topic of data hunches, an analyst’s knowledge about how and why data is an imperfect and partial representation of the phenomena of interest, and investigates how experts’ knowledge about data is utilized in data analysis and how interactive data visualizations can facilitate the process of recording and communicating experts’ knowledge for the analysis process. The dissertation makes three contributions to the topic: 1) an analysis of interview studies with analysts from a wide range of domains and with varied expertise and experience inquiring about the role of contextual knowledge and the process of incorporating various sources of knowledge into analyses; 2) defining, theorizing, and characterizing experts’ knowledge about data and data as data hunches, and positioning data hunches in the existing understanding of uncertainty; 3) proposing a framework and guidelines to design visualizations that support recording and communicating data hunches through visual- izations intuitively and effectively. This dissertation aims to elevate the role of experts’ knowledge in data analysis and provides guidelines and techniques to design visualiza- tions to support externalizing knowledge explicitly. Through the analysis and proposed guidelines, it is envisioned that data hunches will empower analysts to externalize their knowledge, facilitate collaboration and communication, support the ability to learn from others’ data hunches, and ultimately, lead to better data-driven decision-making.

Citation

BibTeX

@phdthesis{2023-thesis-lin,
  title = {Data Hunches: Expressing Personal Knowledge in Data Visualizations},
  author = {Haihan Lin},
  school = {University of Utah},
  month = {December},
  year = {2023}
}