Designing equitable and inclusive visualizations: An underexplored facet of best practices for research and publishing

Abstract

Equitable and inclusive publishing practices for engineering education research have received increased attention in recent years. JEE editorials and guest editorials have raised awareness about multiple challenges, including the problematic Whiteness and maleness of much research and the need to make diversity the default condition (Pawley, 2017), racially biased citation patterns (Holly, 2020), and other general aspects of publishing ethics (Loui, 2016). There are also ongoing discussions in an engineering education journal editors' group about how to increase the inclusivity of our collective publishing practices. For instance, topics such as inclusive pronouns, positionality statements, and how to better involve scholars of color without overburdening them have been discussed. However, inclusive visualization practices have not yet received the same critical attention. Importantly, visualizations can play a number of key roles in manuscripts, such as synthesizing frameworks or literature (Eppler, 2006), showing relationships between core variables (Tufte, 1997), providing illustrative examples of focal phenomena (e.g., see Schimpf et al., 2020), or enabling comparisons of intervention outcomes (Gleicher et al., 2011). Thus, their influence has a wide reach. Just as other aspects of publishing can serve as mechanisms for either exclusion or inclusion, so too can our choices when designing visualizations. In this guest editorial, we highlight the heretofore unexamined topic of visualization to add to those ongoing efforts to increase the inclusivity of engineering education research publishing practices. The three inclusivity dimensions we discuss are (1) communicating to an interdisciplinary audience, (2) representation equity within visualizations, and (3) readers' physical dis/abilities and differences. In discussing these dimensions and how their associated design decisions can affect the inclusivity of engineering education research, we aim to raise awareness, provide reflective prompts for designing and reviewing visualizations, and ultimately decrease the unintentional use of exclusionary practices. These dimensions are not a definitive list but are intended to encourage a wider discussion within the community about inclusive visualization practices. Our first dimension of inclusivity involves communicating to an interdisciplinary audience. Engineering education is an interdisciplinary field that brings together scholars from engineering disciplines, education disciplines, and social science fields among others. While some types of complex visualizations (e.g., multivariate box plots or threedimensional bar graphs) may be standard or common in some of these fields, there are others that very rarely use any visualizations at all. Therefore, not all of the interdisciplinary contributors to engineering education research are equally familiar with all visualization approaches. As such, we need to ensure that visualizations are discernable to the full community so that they do not become inadvertent gatekeepers. For example, to read the boxplot in Figure 1, a reader would need to understand the meaning behind the length of the box, the horizontal line and the glyph within the box, the lines extending below and above the box, the dots beyond the lines, and so forth. If a reader is not familiar with these conventions, he or she is likely to be confused by the figure. Authors can use several strategies to increase the likelihood that their visualizations will be understood by readers from any discipline. First, visualizations in manuscripts should be accompanied by thorough in-text explanations and descriptive captions of the graphic. These explanations or captions should describe central variables, concepts, or categories depicted and provide guidance on how to read the visualization. Authors should likewise clearly and thoroughly label key components of the graphic (Tufte, 2001). Second, while it may be tempting to display more data by incorporating additional variables, categories, or dimensions into a single graphic to allow visualization-savvy readers to dig deeper, authors should not design overly complex visualizations that incorporate information peripheral to the point(s) DOI: 10.1002/jee.20388