What questions reveal about novices’ attempts to make sense of data visualizations: Patterns and misconceptions

• Published in: Computers & graphics 2021-02, Vol.94, p.32-42
• Main Authors: Moraes Bueno Rodrigues, Ariane, Diniz Junqueira Barbosa, Gabriel, Côrtes Vieira Lopes, Hélio, Diniz Junqueira Barbosa, Simone
• Format: Article
• Language: English
• Subjects: Classification; Data-related questions; Empirical study; Literacy; Questions; Recommender systems; Scientific visualization; Visualization; Visualization literacy; Visualization sense-making
• ISSN: 0097-8493; 1873-7684
• DOI: 10.1016/j.cag.2020.09.015

•22 participants wrote 1058 questions about data represented in 20 visualizations.•We derived 249 unique question templates and 20 different types of problems.•Resources for improving the design of question-answering systems.•Resources for improving the design of visualization recommender systems.•Resources for education on how to better explore data in visualizations. [Display omitted] Data visualization literacy has attracted widespread interest due to the urgent need to analyze unprecedented volumes of data we have nowadays. Much work on visualization literacy focuses on asking people to answer specific questions about the data depicted in a visual representation, in an attempt to try to understand how people make sense of the underlying data. In this work, we investigate, through a user survey, the initial questions people pose when first encountering a visualization. We analyzed a set of 1058 questions that 22 participants created about 20 different visualizations, deriving templates for the recurring types of questions that emerged as information-seeking patterns, and classifying the various kinds of errors they introduced in the questions. By understanding the common mistakes they made when asking data-related questions, we now feel better equipped to inform further research on producing and consuming data visualization concepts. The results of the study reported in this paper can be used in teaching data visualization, as they uncover and classify frequent errors people make when trying to make sense of data represented visually. The study may also contribute to the design of visualization recommender systems, as the question patterns revealed what people expect to answer with each visualization.