We present findings from design research on disclosure at the intersection of data visualization, digital storytelling, and feminism. While there is an increased awareness of power structures in data science, computing, and design, there is little design research to confront these. This work explores the potential of disclosing context information of data stories, i.e., digital storytelling formats utilizing data visualizations, to enable critical-feminist readings of and reflections on these stories. Drawing from a growing body of feminist scholarship in human-computer interaction, data science, and beyond, we identify key aspects and forms of disclosure for embedding them into visual data story interfaces. We devise and validate these aspects and forms within a case study: a Web-based scrollytelling article explaining the feminist concept of intersectionality using a combination of animated illustration, data visualization, and text. With this work, we demonstrate and discuss the potentials and pitfalls of disclosure practices in data storytelling.Contents
1. Introduction
2. Background
3. Towards disclosure as a critical practice
4. Case study: A data story explaining intersectionality
5. Discussion
6. Conclusion
We are witnessing a growing number of works in computing and design raising questions about social issues of power, bias, and exclusion in data science (e.g., boyd and Crawford, 2012; D’Ignazio and Klein, 2020b; Buolamwini and Gebru, 2018; Kitchin and Lauriault, 2018). Visual data stories, for instance, are increasingly used to discuss real-world problems, report on current topics, and advocate for social transformation. Data stories and narrative visualizations are characterized by their approach to storytelling rooted in data and combining visual representations and story elements to enable readers to comprehend complex issues. Distinct narrative structure tactics are used for communication, such as the author-driven approach (linear, less interactive, explanatory), the reader-driven approach (non-linear, interactive, exploratory), or a combination of both (Segel and Heer, 2010; Weber, et al., 2018; Stolper, et al., 2016). Despite an increased interest in data stories in academia and practice (Riche, et al., 2018), there are also skeptical voices challenging promises of empowerment (Parvin, 2018).
Recent work on critical, ethical, and feminist data visualization emphasizes the need for considering power, context, and emotion in data visualization research and design, and proposes various approaches and principles toward a critical practice of data visualization (D’Ignazio and Klein, 2016; Dörk, et al., 2013; Correll, 2019). One central and recurring aspiration in such work is the need for disclosure — providing and revealing background and context information of a visualization project and its creation, i.e., how it was done, why it was done, and who did it. Disclosure emerges as a central principle towards critical and feminist data visualization as it contributes to a fair acknowledgment of otherwise hidden or undervalued labor (D’Ignazio and Klein, 2020b). It may also help to challenge the idea that data and their visual representations are given, neutral, and universally true, but instead are authored, partial, and situated (Drucker, 2011; Drucker, 2017; Loukissas, 2019). While prior work on this matter has been largely theoretical, there has been surprisingly little applied research on disclosure as a critical practice that strives to empower viewers to critically read and reflect on visualizations and their contexts. There is limited understanding of how to effectively achieve disclosure in visual interfaces and representations of data. What are the key aspects of disclosure and how can they be usefully integrated into a data visualization?
To address this question, this research makes two main contributions: a framework and a case study. The former provides a theoretical foundation and the latter is the practical context, in which the framework is validated and refined. Focusing on the format of visual data stories — a common Web-based article format incorporating interactive storytelling and visual data representations — we contribute a framework for the critical practice of disclosure in data visualization. With the framework, we identify key aspects of disclosure, devise representation forms for disclosure information in visual data story interfaces, and invite scholars and designers to critically consider disclosure during the design and study of visual data stories. We apply this framework in the case study of a Web-based scrollytelling article explaining the feminist concept of intersectionality using a combination of animated illustration, data visualization, and textual descriptions, while also disclosing certain making-of information about the article and its creation. We present the article’s disclosure-specific interface design as well as findings from a log analysis of user activity and an online questionnaire. We close with a critical discussion of implications and limitations of our research.
Our research is informed by feminist epistemologies and a growing body of research and design elevating ethical, critical, and feminist principles in human-computer interaction and data visualization. It also relates to transparency in data journalism, data provenance in visual analytics, and reflexivity in qualitative research.
2.1. Feminist standpoint theory and intersectional feminism
Feminist philosopher Haraway (1988) and her concept of “situated knowledge” emphasize that all knowledge in the world is situated and partial, produced by individuals with specific perspectives within certain social, political, economic, and cultural contexts. These perspectives determine what we know about an object of interest, what questions we ask, and what meanings we make. The perspectives that are particularly visible and claim to be neutral are those of middle-class white men since they are the most privileged in capitalist, patriarchal, and racist structures. Haraway describes this as the “god trick”: Objectivity and “the view from above, from nowhere” obscures that it merely embodies a particular view (male, white) and makes this the seemingly neutral, universal, and default view. Harding introduces the term “strong objectivity” to acknowledge the particular position of the scholar. She argues to center the perspectives and lived experiences — standpoints — of women and other marginalized people who have not been traditionally included in knowledge production (Harding, 1993, 1991).
Intersectional feminism acknowledges that people are shaped by multiple, simultaneously interacting, and interwoven dimensions of social power relations — gender, gender identity, sexual orientation, race, religion, class, disability, age, etc. — that determine one’s lived experiences in the world. Within prevailing social systems — capitalism, white supremacy, and patriarchy — a white, abled, cis-woman has different experiences of discrimination, marginalization, and oppression than, for instance, a Black, disabled, trans woman (Hill Collins, 2009; Crenshaw, 1989; hooks, 2014; Combahee River Collective, 1977).
Prior research in the field of human-computer interaction transfers feminist epistemology into technology development and the design of interactive systems (Suchman, 2002; Bardzell and Bardzell, 2011). More recent work focuses, for instance, on gender as non-binary in technology infrastructures (Spiel, 2021), drag, queer, and trans technology design tools (Baeza, et al., 2021), and accessibility of makerspaces for intersectionally marginalized women (Hedditch and Vyas, 2021). Meanwhile, there is a kindred effort in data visualization by researchers and practitioners formulating a feminist critique against a kind of data determinism, advocating for feminist data studies, and presenting novel strategies concerned with social power in data science and practice (D’Ignazio and Klein 2020a, 2020b; Hill, et al., 2016; Hill, 2020).
2.2. Critical data studies and practices
Scholars of critical data studies demonstrate how data practices tend to maintain and reinforce structures of power and domination (boyd and Crawford, 2012; Dalton, et al., 2016; Iliadis and Russo, 2016; Kitchin and Lauriault, 2018). Data are always subject to particular circumstances, contexts, and complex processes with decisions involving people with particular positionalities and perspectives (D’Ignazio and Klein, 2020b). Bowker (2005) and Gitelman (2013) emphasize this by suggesting a specific wording: Data are never raw, not a natural phenomenon, instead they are always cooked. Working with data requires attention, care, and many decisions all shaping the insights and outcomes derived from data. To highlight that data are never simply given — as the Latin meaning of data suggests — but require the deliberate selection and recording of observations, Drucker (2011) introduces the term capta: “Data are capta, taken not given, constructed as an interpretation of the phenomenal world, not inherent in it.”
However, many visual representations of data portray data as given and universally true. Kennedy, et al. (2016) describe four conventions that make visualizations appear objective, transparent, and factual: two-dimensional viewpoints, clean layouts, geometric shapes, and the inclusion of data sources. Yet, data visualizations are subject to particular circumstances, processes, as well as decisions and assumptions of their designers. They have the potential to influence viewers and to depict a certain message more persuasively (Pandey, et al., 2014). Visualizations can manipulate or empower depending on the designer’s intention and the contexts in which visualizations are viewed (Dörk, et al., 2013), and can even be deceptive and misleading (Pandey, et al., 2015; Cairo, 2015). Hullman and Diakopoulos (2011) demonstrate that narrative data visualizations hold a rhetorical force and involve a range of editorial choices such as omission, emphasis, or ambiguity operating at different layers of data visualization to frame an issue or advance an argument.
In the context of data journalism, Bounegru and Gray (2021) formulate 12 challenges for a critical data practice, some of which constitute forms of disclosure. Disclosure practices are particularly exemplified in the open and transparent communication of data sources and analysis methods employed in the production of data stories. In narrative data visualizations, citing and linking data sources and methodological choices signals transparency and trustworthiness and builds credibility (Hullman and Diakopoulos, 2011). Weber, et al. (2018) define the meta-story — i.e., being transparent about data sources, methodologies, and processes — as one of seven key features of journalist data stories. They conducted a qualitative interview study on uses of data visualization in news organizations in Europe demonstrating that the “how-we-did-it” element is conventionally positioned at the end of a story in a fact box or as an external article linked in the data story. Further work argues for the transparent communication of “how tos” of data journalistic work and how conclusions are drawn: Providing links to data sources, making the code available, and explaining the methods employed can increase the quality of data journalism and deepen the data literacy of journalists and the broader audience (Mazotte, 2021; Radcliffe and Lewis, 2021; Rinsdorf and Boers, 2016).
In the fields of visual analytics and data science, provenance refers to the history of data transformations as part of analytical processes; several provenance tools have been developed to support the recording, tracking, and communicating of reasoning processes and computational workflows (Ragan, et al., 2016; Davidson and Freire, 2008; Freire, et al., 2008; Gotz and Zhou, 2009; North, et al., 2011; Callahan, et al., 2006). Provenance information may refer to the data, visualization, interaction, insights, and intentions for a range of purposes including recall, replication, recovery, collaboration, presentation, and reflection (Ragan, et al., 2016).
2.3. Reflexivity and transparency in research
The practice of reflexivity is an important part of qualitative research. Based on the perspective that knowledge production is situated and subjectively constructed, reflexivity practice means the conscious self-awareness, identification, and analysis of the researcher’s own role, biases, assumptions, contexts, and motivations through self-reflection and how these, for instance, affect the research process and what knowledge researchers create (Tracy, 2010; Finlay, 2002b; Meyer and Dykes, 2019). Reflexivity is accounting for the subjectivity of researchers and therefore essential for a disclosure practice in visual data storytelling. Feminist approaches of reflexivity are particularly concerned with asymmetric power relations, for instance, between researchers and those being researched, and argue for self-awareness of researchers’ own positions and situated nature within the research (Finlay, 2002a). Another related practice is that of transparency — the honest and critical description of how the research was conducted including contexts, decisions, activities, methods, actors as well as challenges, failures, and unexpected changes. The acknowledgment of participants, funding, colleagues, etc. is also part of it (Tracy, 2010). Such transparency in research resembles our understanding of disclosure for data stories. Sturdee, et al. (2020) propose to make social and material contexts and environments, in which “in the wild” research takes place, visible and transparent using “pictorial” publications.
Meyer and Dykes (2019) examine visualization design studies as applied visualization research and propose criteria that contribute to rigor in and constructed knowledge through design studies. Building on reflexive practices devised in social sciences, Meyer and Dykes argue to apply reflexive considerations to visualization design studies and propose guiding questions for researchers regarding readiness and biases as well as two explicit tools to encourage and support reflexivity. Additionally, transparency considerations throughout a design study are highlighted such as the justification of design decisions, disclosure of reflexive notes and processes, provision of description of analysis process, reflective documents, and report on failures.
2.4. Disclosure in critical, ethical, and feminist data visualization
Akin to feminist epistemology and critical data studies, prior work on critical, ethical, and feminist data visualization build on the premise that data and visualizations are always partial and situated. Dörk, et al. (2013) propose four principles for a critical approach to information visualization: disclosure, plurality, contingency, and empowerment. Here, disclosure is described as the revealing of human decisions, assumptions, and intentions about data, representation/design, and interaction in order to enable viewers to (critically) engage with the data visualization and the addressed issue. As part of the presented ethical challenges of visualization work, Correll (2019) argues for making the provenance of data and the various decisions of analysts visible to enable transparency and criticism. He also advocates for communicating and visualizing the often hidden labor that a visualization project rests upon as well as the design and user research work as a contribution to a fair valuation of labor and as an added value for reproducibility and the field’s progress in general. D’Ignazio and Klein (2020b, 2016) advance the strategy of showing the work behind data products and thus contributing to the fair acknowledgment of otherwise often hidden, undervalued labor as part of their data feminism principle “Make labor visible”. Furthermore, another data feminism principle named “Embrace pluralism” involves the practice of disclosing the project’s method, decisions, and one’s own positions and identities. Akin to reflexivity and transparency in qualitative research, D’Ignazio and Klein advocate for acknowledging one’s own partial and situated positionalities and thus the value of multiple perspectives. Building on the convention in data journalism to provide access to the used data and methods they see value in extending this practice to reveal human decisions, intentions, and involved identities. Burns, et al. (2021) propose to accompany visualizations with metadata about data sources, data cleaning and processing, design elements, creators, and audiences and discuss advantages and disadvantages of disclosing these metadata.
Prior research on positionality, criticality, provenance, and reflexivity provides a broad theoretical foundation for disclosure as a critical practice in data visualization, interface design, and digital storytelling. While we extensively draw from this work, we noticed that there has been little applied visualization research focusing particularly on the practice of disclosure. There is a surprising lack of practical approaches to identifying and integrating disclosure information into visual interfaces, such as data stories, to enable reflective and critical engagements with data and visualization.
3. Towards disclosure as a critical practice
With this research we are centering on disclosure as a critical practice that strives to enable and encourage people to read, view, and examine data visualizations in a critical-feminist way. We argue that disclosure holds the potential to help challenge the idea that data and their visual displays are neutral and universally true, but instead always situated in particular contexts implying a cascade of decisions by people with partial standpoints and perspectives. We use the term “critical practice” to refer to Agre’s term of “critical technical practice” which “require[s] a split identity — one foot planted in the craft work of design and the other foot planted in the reflexive work of critique” [1]. Focusing on the format of visual data stories we aim to unfold a framework for the critical practice of disclosure by suggesting both key aspects of disclosure as well as distinct representation forms for disclosure information in visual data story interfaces (see Figure 1).
Figure 1: Visual overview of the proposed framework including six key aspects clustering disclosure information and six representation forms for disclosure information in visual data story interfaces. Any representation form can be applied to any key aspect.
3.1. Key aspects of disclosure
Based on prior work in qualitative research, data science, data journalism, and data visualization related to transparency, reflexivity, provenance, and disclosure, we identify six key aspects of disclosure information. We consider all of these necessary to enable a critical-feminist reading of a visual data story. Each aspect is explained below and accompanied by guiding questions in Table 1.
Team positions
One part of self-reflexivity can be achieved through acknowledgment, conscious self-awareness, and critical reflection on the partiality of ones own position and knowledge. This includes researchers’ and practitioners’ social, institutional, racial, etc. positions in the world and the impact on the project. It may also include the critical self-reflection of one’s position and impact within a project, particularly in terms of the relation to those being researched, and the identification of those voices that may be missing. The team’s motivations, intentions, expectations, and assumptions about the project can also be part of this aspect (D’Ignazio and Klein, 2020b; Bardzell and Bardzell, 2011; Meyer and Dykes, 2019; Tracy, 2010).
Project process
The disclosure of the project process, i.e., the description of how the project was conducted, within which contexts it came into the world, and how it progressed over time, can be accomplished by giving insights into the research and design process and making it transparent to viewers. This could include project goals, applied methods and activities, key project decisions, challenges, unexpected changes, failures, and interim results. The reflection of considered, but eventually discarded methods, activities, studies etc. can also be part of it (Tracy, 2010; D’Ignazio and Klein, 2020b; Weber, et al., 2018; Correll, 2019).
Hidden labor
Tracing back invisible labor, which the project rests upon, and making it visible contributes to a fair acknowledgment and valuation of the various efforts that make a data story possible. In data visualization projects, this especially includes labor of people contributing to the data collection, preparation, and analysis process which are often hidden in the end products (Correll, 2019; D’Ignazio and Klein, 2020b). This disclosure aspect also includes the crediting of people who were temporarily involved in the project such as participants of evaluations, interviewed experts, consulted colleagues, etc. (Tracy, 2010).
Data settings
Providing access to the underlying data sources and data sets of a data visualization project can be considered a fundamental form of disclosure, in that it allows, in principle, to use other visualization techniques for confirmation or comparison. Additionally, the disclosure of data settings (Loukissas, 2019), i.e., the communication of the data sets’ contexts and the transformation steps, methods, and human decisions involved in aggregating, analyzing, and processing data, may demonstrate the situated, specific, and biased nature of data (D’Ignazio and Klein, 2020b; Correll, 2019; Burns, et al., 2021; Weber, et al., 2018).
Design decisions
The intentions, decisions, and reasons for specific visual representations of data, the visual design, and interaction are often not visible in the final results. Disclosing these decisions may contribute to visualizations being perceived less as given, neutral, or universally true and help recognize that they hold a rhetorical force and are subject to particular contexts, preferences, and intentions of their creators (Dörk, et al., 2013; Burns, et al., 2021).
Critical reflections
This final key aspect focuses on the critical reflection on the project outcomes. Limitations, known problems, open questions, data gaps, etc. are to be defined within the scope of the project subject and topics presented in data visualization. Also relevant is the critical examination of potential reproductions of biases, discrimination, structures of power, and oppression.
Table 1: Disclosure information can be distinguished into six key aspects each raising a different set of guiding questions. Key aspect Guiding questions Team positions Whose positions are represented in the project team? Whose perspectives are excluded but would be important for the project? How do the social, professional, political, etc. positions of the team (members) affect the project? Why are team members involved in this project? What relationships do researchers/authors have with those being represented? Project process What are the initial goals of the project? What methods and activities are carried out? What key decisions are made? What are the challenges and unplanned changes in the process? Could it be helpful for viewers to see the process? Hidden labor Who collects, curates, archives, digitizes, cleans, maintains, analyzes, etc. the data? Which people are involved in the project process and who makes it possible in the first place? Could it be helpful for viewers to see people’s efforts that made the project possible? Data settings How and why was the data collected? What are known limitations and biases of the data? How is the data used? What are potential impacts of the data? Can the data sources and data sets be published? What are transformation steps, methods, and human decisions involved in analyzing, aggregating, and processing data? How could these steps be communicated? Design decisions What decisions are made concerning the visualization and interaction design? Why are particular design choices made? What are discarded alternatives? Why have they been discarded? What are possible limitations of the selected visual data representation? Critical reflections What are known limitations and gaps of the project outcomes? What necessary next steps does the project need? How could the project outcome lead to exclusions and/or discrimination? At which points are biases, stereotypes, structures of social power and oppression potentially reproduced and reinforced?
The key aspects do not constitute a complete or exhaustive list of disclosure information. The details may shift considerably for any given visualization or storytelling project. While the key aspects refer to what kinds of disclosure information a visualization project should address, in the following we outline how these aspects could take form in a data story.
3.2. Representation forms of disclosure
Complementing the key aspects of disclosure, we propose six representation forms that enable an integration of disclosure information into visual data stories. Conceptually, any representation form can be applied to any key aspect. The forms can be used individually or in different combinations with each other. The representation forms differ particularly in how disclosure information is connected with the primary information layer of the data story, whether it is backgrounded, foregrounded, or interwoven in the viewing experience, what types of disclosure information they are suitable for, and in the design effort required. The representation forms of disclosure vary along four indicators: function, placement, prominence of disclosure information, and effort to design and implement it (see Table 2). We distinguish between low, medium, and high levels of prominence and effort.
Table 2: Overview of the representation forms' characteristics along four indicators: functional role, placement with regard to the primary information layer, prominence, and the effort to design and implement them. Form Function Placement Prominence Effort Preamble Introductory disclosure information Before High Low Complement Further disclosure information previously introduced using other disclosure forms Along Medium Medium Thread Disclosure information that directly refers to an element of primary information layer Interwoven High High Disruption Disclosure information that directly refers to an element of primary information layer and is of particular importance for the critical reading Interruptive High High Colophon Further disclosure information previously introduced using other disclosure forms After Medium Low Appendix Supplementary material linked and introduced within other disclosure forms External Low Low
Preamble
When the disclosure information is placed before the primary information layer, it is foregrounded in the viewing experience and has a high prominence since it is the first information displayed. The leading position of the preamble could affect the reader’s perception and interpretation of the content (Hullman and Diakopoulos, 2011). The preamble is used preferably for introductory information that also contributes to the understanding of the story. Compared to other forms, only little effort is required as the primary information layer does not need to be adjusted.
Complement
Disclosure information can also accompany the primary information layer as a secondary, simultaneous layer that is accessible from the primary information layer but is not integrated with primary information. The complement is not necessarily foregrounded in the viewing or reading experience, but accessible on demand. This form is preferably applied for disclosure information that has already been introduced using other forms that have a higher prominence. Compared to other disclosure forms, medium effort may be needed here, as complement requires only a weak connection with the primary information layer. However, the effort needed ultimately depends on the concept and design of the individual project and how disclosure information is to be presented.
Thread
Disclosure information can also be interwoven into the primary information layer of a visual data story interface, with disclosure elements embedded directly into an article. The thread is highly prominent since it is foregrounded in the viewing experience and placed in the immediate context of primary information. It is thus preferred to be used for disclosure information that directly refers to specific primary information, e.g., information about data contexts and design decisions. A high design effort is required for the thread compared to other disclosure forms presented since the disclosure information is directly interwoven into the primary information layer and thus strongly influences the concept and design of the overall article.
Disruption
A more intermittent variant of the thread is disruption. The perusal of the primary information layer is interrupted by the automatic display of disclosure information without viewers intentionally requesting it. This representation form has a high prominence since viewers are disrupted in their viewing experience and obliged to react to disclosure information. Akin to pop-up dialogs requesting Web page visitors to sign up to newsletters or turn off ad blockers, there is a risk of irritating and frustrating readers. It is recommended to carefully consider the inclusion of a disruption in order to avoid persistent annoyance for viewers. The disruption is again preferably applied for disclosure information that directly refers to particular primary information and is of specific importance and interest for the critical reading of the primary information layer. In comparison to the other forms, a high design effort is needed.
Colophon
Disclosure information can also be placed after the primary information layer. As such the colophon has only a medium prominence since viewers may already be saturated with new information at this point. While there might be a higher probability that viewers may overlook disclosure information or consider it less relevant, some viewers may also expect disclosure information at this point in an article. It is advisable to use this form to further disclose information previously introduced in other forms. In comparison to other disclosure forms, a colophon requires only little effort. Typical information provided in the colophon is the methodology used to source and process data for visualizations used in a story. Often, the team behind the project is mentioned in the colophon.
Appendix
This representation form describes disclosure information that is placed separately from the visual data story and is accessible via a link to external resources, e.g., an extra article, often used in digital journalism practice. It has a low prominence since it is separate and not directly part of the viewing experience. This appendix is optimally used for supplementary materials, e.g., data sets, source code, workshop and evaluation materials, which should be linked and introduced within another disclosure form. Relatively considered, the appendix also requires little effort.
The characterization of representation forms for disclosure information is meant as a support structure when authoring, designing, and implementing a data story. These forms of disclosure emerged during the case study of a visual data story about intersectionality, which we introduce in the following section.
4. Case study: A data story explaining intersectionality
To assess the viability of the disclosure framework we present a case study that sits at the overlap between advocacy, data visualization, interface design, and digital storytelling. One major outcome of this is the scrollytelling article Inter...what? Intersectionality! A visual Introduction explaining the concept of intersectionality using a combination of animated illustration, data visualization, and textual description (see Figure 2). After briefly introducing the article’s creation context and structure, we explain its disclosure-specific interface in detail and share findings from its evaluation.
Figure 2: The scrollytelling article Inter...what? Intersectionality! A visual Introduction explains the concept of intersectionality, highlights the inverse correlation between discrimination and privilege, illustrates three examples from Germany, and supports further engagements with these topics: uclab.fh-potsdam.de/intervis/.
4.1. Context and structure
The research and design process of the case study was iterative and characterized by a constant fruitful exchange between the involved participants — with backgrounds in interface design, information visualization, Web development, and feminist studies — from devising initial visualization ideas, refining design concepts, developing the narrative to implementing a Web-based scrollytelling article.
Experts in intersectionality were invited to critically review the visual encodings. Feedback sessions of 45 minutes each were carried out with three experts in the form of video calls, during which they expressed helpful feedback, which was followed by design revisions. The data story comprises the narrative layer and disclosure layer (see Figure 3). The narrative layer consists of an introduction and five chapters through which viewers can scroll and two voluntary overlays in chapters 1 and 2 with further information about the presented subject. The article introduces the concept of intersectionality, explains the inverse correlation between discrimination and privilege, illustrates these with three examples of intersectional discrimination in a German context, and supports further engagement with these topics. For the visualization of intersectionality, discrimination, and privilege, the visual metaphor of interwoven strokes was developed, which is used in all illustrations and data visualizations throughout the data story. Different orientations of strokes present different social categories, which are interwoven and thus create a sort of net, which is visible throughout the entire article. Color is used to highlight discrimination (orange) and privilege (gray-violet).
The disclosure layer reveals making-of information about the scrollytelling article and its creation, i.e., how it was done, why it was done, and who did it. People directly or indirectly involved are introduced and certain visualization decisions are made transparent. The perspectives and positions of the authors, the project’s process, and the outcomes are shared and critically reflected. Next, we discuss specific visual interface elements designed to embed key aspects of disclosure information in various forms into the visual data story.
Figure 3: Page structure and navigation paths of the visual data story: The narrative layer (black) and the disclosure layer (orange).
4.2. Disclosure-specific interface design
When initially accessing the scrollytelling Web page, an overlay with introductory information is displayed (preamble). The overlay consists of the article’s headline and a text box that is used for general disclosure information such as the main goal of the visual data story, and who and in which context it was developed (see Figure 4). Selecting the “Go to article” button below the text box closes the overlay and reveals the beginning of the data story. The button with the label “More insights into project” provides access to further disclosure information on the virtual backside of the scrollytelling page. Viewers can decide how they would like to continue. If viewers decide to go to the article they can start scrolling through chapters, read text, and view illustrations and visualizations.
Figure 4: When accessing the data story, an overlay appears which is used for introductory disclosure information. Viewers can decide how to proceed: with entering the visual article or viewing further disclosure information.
In the first three chapters, an icon with a light bulb is placed in the upper right corner which stands out due to its orange color (see Figure 2). By clicking on it, viewers can reveal slide-ins that are inserted from the right and used for explaining design decisions and data contexts (thread) (see Figure 5). The content of the slide-ins refers to the visual elements in view at a given scrolling position. Each slide-in also includes a “More insights into project” button, which takes viewers to further disclosure information.
Figure 5: On-demand slide-ins can be opened from the side at certain positions in the data story. These slide-ins explain design decisions and provide additional context information about article elements currently in view.
The peculiarity of the slide-in in chapter 3 is that it inserts automatically when viewers reach a particular scrolling position at the beginning of the chapter and therefore interrupts the reading flow in the primary information layer (disruption). It is solely used for this particular slide-in since the context of the used data on intersectional discrimination and the decisions and intentions about its visual displays are of particular importance for the critical reading of the data story’s third chapter. At the end of the scrollytelling article, a text box provides a general outline of the limitations of the article (colophon). Below this, an orange button labeled “Insights into the project” is placed.
From various points in the article more in-depth background information about the project can be accessed via this type of button. Regardless of the position in the article, clicking on it triggers a flip animation revealing the virtual backside of a Web page containing further disclosure information (complement) (see Figure 6). Some of this disclosure information has been introduced before, e.g., in the initial overlay, the slide-ins, or the text box at the end of the article. Five subheadings structure information about the project phases and methods, design decisions and data context, people who contributed to the project, creators’ positionalities and intentions, project reflections as well as external links to supplemental material such as prior publications, source code, and data sources (appendix).
Figure 6: The backside of the scrollytelling Web page can be accessed from five positions in the data story. Several key aspects of disclosure function as menu items clustering and structuring information.
4.3. Evaluation design
To understand how readers respond to disclosure information and engage with various forms provided with the visual data story, we conducted an evaluation of the case study. We aimed to observe: 1) engagement with disclosure elements; 2) appreciation of particular positions of disclosure information; and, 3) the usefulness of disclosure information for the critical reading of the visual data story.
The evaluation was carried out by combining two methods. A logging mechanism anonymously tracked selected interactions carried out on a page, and visitors were asked to participate in an online questionnaire linked via a button displayed when reaching the last chapter of the article or after five minutes of reading. To enable a coupled analysis of logging data and questionnaire results, a session identifier based on the time of the Web page visit was included in the questionnaire form. For each session, we anonymously logged the view-port size, vertical scroll positions, and time at which the questionnaire button was clicked. The page was divided into 24 parts (see Table 3) aiming to derive individual reading paths as well as the time spent in each of the parts from the scroll positions and other parameters such as the activation of overlays and slide-ins. By distinguishing between parts containing the primary content vs. disclosure information we were able to derive time spent in both layers.
The questionnaire was structured into three main parts: attention check, background, and disclosure forms. The attention check consisted of three questions about the data story and its main elements. The next set of questions referred to participants’ demographic and professional backgrounds. This was followed by the main part featuring questions about the disclosure forms implemented in the case study. Each section consisted of three different statements about the level of engagement, appreciation, and usefulness with consistent wording for all forms via a five-point Likert scale and a free-form input field for feedback on the respective disclosure information and elements.
Participants were recruited via social media accounts of the team members and research group and via an internal communication platform of our university. Additionally, we contacted organizations that are of relevance and interest for the content and form of the visual data story. Aiming to reach more people that may be interested in the article’s topic and format we placed an advertisement in a German daily newspaper with a call to read the article and participate in the online questionnaire.
4.4. Evaluation findings
Over the course of 12 weeks (June–August 2021), we collected log data of 1,981 sessions in total with a bouncing rate of 42.2 percent (sessions in which the visited Web page is left without any further interactions). For our analysis, we include 382 sessions, during which at least one third of the previously defined page parts were visited (considered sessions). In this subset, the median visit duration is 5.1 minutes (1.8–18.2 IQR) and the ratio between the time spent on the narrative layer and disclosure layer of the data story is 7:1. In 86.4 percent of the sessions, the narrative content outweighs disclosure content. Besides the forced exposure to the initial overlay (preamble) and the automatic slide-in (disruption), interactions that go beyond scrolling occurred in 39.8 percent of the sessions.
The most read parts are the mandatory initial overlay (preamble): 100 percent, the first narrative content of the article: 98.2–95.8 percent, and the automatic slide-in in chapter 3 (disruption): 91.9 percent (see Table 3). Chapters 4 and 5 and the project information text box at the end of the article (colophon) are less frequently visited than the narrative chapters before. The least seen parts are the overlays with additional narrative information in chapters 1 and 2 (8.1 percent and 7.3 percent), the on-demand slide-ins in chapters 1 and 2 (thread) (4.2 percent and 9.9 percent), and the backside pages 2-5 (complement) (between 7.6 percent and 8.1 percent). The first backside page was visited more often (20.7 percent).
Table 3: List of page parts of the visual data story with the percentage of page visits containing the respected part in relation to all for analysis considered page visits. Page parts in italics are those that belong to the disclosure layer. Page parts Visits in percent Initial overlay (preamble) 100.00 Intro 98.2 Chapter 2 97.4 Chapter 1 96.6 Title & teaser 96.6 Case study 1 95.8 Automatic slide-in (disruption) 91.9 Chapter 3 89.5 Case study 2 83.5 Case study 3 75.1 Chapter 4 71.5 Chapter 5 69.1 Project text box at end (colophon) 61.5 References 58.6 Imprint 46.9 Backside page 1 (complement) 20.7 On-demand slide-in chapter 2 (thread) 9.9 Overlay chapter 1 8.1 Backside page 2 (complement) 8.1 Backside page 3 (complement) 7.9 Backside page 5 (complement) 7.6 Backside page 4 (complement) 7.6 Overlay chapter 2 7.3 On-demand slide-in chapter 1 (thread) 4.2
When considering the overall time spent across different disclosure elements of the data story relative to the total time spent on the disclosure layer, the following distribution is revealed: 37.8 percent initial overlay (preamble), 36.8 percent automatic slide-in (disruption), 9.3 percent backside pages (complement), 8.2 percent project text box at end (colophon), and 7.8 percent on-demand slide-ins (thread).
Eighteen people participated in the online questionnaire after reading through the data story. One questionnaire response is not included in the analysis since, according to a comment, the majority of the disclosure components did not function technically on that person’s smartphone. Two other sessions were discarded because they did not fulfill the requirement of a considered session and, in addition, did not have at least two correct answers in the attention check of the questionnaire. The remaining 15 questionnaire responses provide the basis for the subsequent analysis as their associated sessions are considered sessions coupled with at least one attention check question being answered correctly. Quotes translated from German are indicated with (†).
Fourteen of the 15 questionnaire respondents are living in Germany, one in Switzerland. Ten of them identify as female and five as male. The age distribution of participants spans from young to older adults: 18–24 (2), 25–29 (1), 30–39 (4), 40–49 (3), 50–59 (3), 60 and older (2). Eleven of the 15 participants were familiar with the subject matter presented in the scrollytelling article while all of them have some or a lot of experience with digital formats and devices. One of the participants consumed such a visual data story several times per week, two of them once a week, eight of the participants every few weeks, and four of them never. Most of the participants (11) used a computer/laptop to view the data story, while two used a tablet and two a smartphone.
About half of the questionnaire participants appreciated the initial overlay (preamble) (8) and the majority considered its information to be useful for the critical reading of the data story (10). See Figure 7 for an overview of the questionnaire responses. While five agreed with the statement that the overlay is engaging, four disagreed with that, and six were neutral towards it. Responses from the free input field suggest that the overlay consisted of too much text for the beginning, one participant stated that it does not evoke that much curiosity for the article.
Figure 7: Butterfly chart visualizing the questionnaire results along statements about engagement with, appreciation of, and usefulness of distinct disclosure elements implemented in the case study and a Likert scale.
Eight questionnaire participants found the on-demand slide-ins (thread) and their triggers to be engaging and encouraging, four of them disagreed with that. Eleven of 15 appreciated the on-demand slide-ins at their respective positions, almost all participants (14) found their information useful for the critical reading of the data story, and no one disagreed with these two assertions. One participant commented on the on-demand slide-ins: “The focus was very much on explaining the design, pulling my attention away from the content a bit” (†). A similar comment was submitted for the automatic and thus mandatory slide-in at the first case study in chapter 3 (disruption): “In parts, it distracted me” (†), whereas another participant considers this distraction rather positively: “It’s great that it’s so disruptive. Really lets you focus on it.” Overall, the automatic slide-in received the highest level of agreement on its engaging and encouraging quality (13), while a considerable number of participants appreciated the position in the article (11) and found it useful for contextual and critical reading (12). Less people found the project text box at the end of the article (colophon) engaging or encouraging (7), however, most participants responded favorably about its placement (13) and utility (11). One statement underlines this sentiment: “The display could be more engaging. It’s currently just a lot of text and doesn’t do justice to the interesting page that opens when clicking the button.” The feedback on the backside (complement) differs from the responses on the other disclosure elements: All three statements received between three and five agreements, while most of the participants were neutral towards them (6, 7, and 5) and three or four did not vote on these statements. This could be partly explained by the fact that several participants did not view the backside in detail, which they commented on in the free input field. One questionnaire participant stated “I didn’t read ‘Insights into project’ because I was busy reading the article for longer than I thought and then I wasn’t able to absorb it anymore” (†). In contrast, the backside was highly appreciated by some participants: “Very interesting content that (in this detail) is rarely seen in other projects;” or “Insights into the process I always find very powerful! Thank you” (†).
We now discuss the main insights from the case study and reflect on the proposed framework for disclosure practice demonstrating its added value as well as its limitations and implications for research and design.
Disclosure for critical reading
The questionnaire findings suggest that providing disclosure information can contribute to a critical reading of visual data stories and stimulate reflections on the context of their making. Disclosure may hold the potential to foster critical data literacy of viewers by communicating data and their visualizations as situated in certain contexts, complex processes, and decisions involving various individuals with specific and partial positionalities. However, while our case study resulted in encouraging findings about the usefulness of disclosure information for critical reading, it is reasonable to assume that the rather few questionnaire responses are not sufficient to prove this conclusively. One promising approach would be, for instance, to conduct qualitative interviews with viewers investigating the experience of disclosure in more depth.
Not visited, but appreciated
The questionnaire results indicate a generally high appreciation of disclosure information embedded into the article. Viewers value being trusted with disclosure content and considered as qualified to engage with it. This may also apply to those who have not engaged with the content much or not at all. Weber, et al. emphasize that “even readers who will not spend time on such activities may appreciate being positioned as qualified for doing so” [2]. Prior research in data journalism and narrative visualization demonstrate that providing links to data sources, data sets, etc., and showing applied methods and processes increase trustworthiness and build credibility (Hullman and Diakopoulos, 2011). It is likely that this is true even if readers do not actually interact with it knowing they could if they chose to. However, while we consider the appreciation of disclosure as essential, this alone is not sufficient to promote a critical reading of a data story, which requires actual engagement with disclosure information.
Limited attention
With 39.8 percent of the tracked user sessions we did experience more interaction with additional information than expected. Drucker, et al. (2018) cited 10–15 percent of Web page visitors who use the interactive functions of data-driven stories. Despite this, in comparison to the primary information layer, the disclosure content resulted in lower levels of engagement, which might be explained simply by the competition for attention between the two layers. Readers may be overwhelmed with the amount of content and, if time is limited, are more likely to peruse the “actual” story and not its making-of. By foregrounding disclosure information in the reading experience and thus enforcing a focus on it, one may influence the battle for attention between disclosure and primary content. This can be achieved, for example, by placing it before the primary information layer or by interrupting the reading flow. We assume that a balance is needed for foregrounding disclosure information in the viewing experience since it entails the risk of irritating viewers. Leaving disclosure information aside, to the end, or entirely separate could in turn cause viewers to pay too little attention to it, not view it at all, perceive it as irrelevant, or be already saturated with information.
Interconnected effort
Preparing and providing for disclosure also implies an increased effort in terms of data collection, content creation, documentation, concept ideation, and implementation. Due to its interconnected character with the article creation process, the practice of disclosure makes almost all steps more complex. We assume that the more the disclosure aspects of our framework are interwoven into an article, the greater the effort to harmonize primary content and disclosure information. Adding, for instance, a text box with disclosure information at the end of a data story (colophon) or placing it externally (appendix) does not require much integration compared to other disclosure forms that are more deeply embedded or even interwoven in the narrative. However, this ultimately depends on the concept of the individual project and editorial role of the disclosure content. Early on in a given data story project, it is advisable to consider the status of disclosure information and explore if and how it can be integrated into the main narrative.
Into the future
While the disclosure layer presented in our case study was relatively text-intensive, we see great potential in exploring other representation forms for disclosure information that enrich also the modality of the article. For example, dynamic linked videos coupled with interactions in the primary information layer could communicate a range of disclosure content without requiring additional reading (Nelson, 2021). Future work could explore further options to integrate disclosure more into the authoring and design of data stories, in particular the forms that refer to particular elements in the primary information layer (thread and disruption). Although in our case study we implemented such elements, the potential of interweaving has not yet been fully explored and could be taken further. Making disclosure part of the actual narrative could mitigate the battle for attention between disclosure and primary content. Advancing the critical practice of disclosure in the creation of visual data stories requires more work on the interplay of the various representation forms and their integration into the primary information layer.
Confounding factors
As with any research method, there are limitations to our research that may have affected our findings. No one outside German-speaking countries answered the questionnaire. The number of page visitors and respondents to the questionnaire might represent a very small sample to derive a conclusive picture of how disclosure is perceived and experienced. The way the Web study was set up, did not allow for the distinction between initial and follow-up visits of the page, as each visit generated a unique session id. This means we do not know how often readers and respondents have revisited the article, something which may have occurred multiple times considering its format. Another possible inaccuracy of logging data, caused by the lack of a logging event indicating the end of the reading sessions, could have an influencing effect. As a consequence, we built our quantitative analysis on the sequence of visited page parts and not on the time spent on them, which also prevents the influence of article characteristics like text length and visual appearance. A factor influencing the perception of the dynamically shown disclosure information is the usage of a mobile and a desktop version of the article which we accepted to widen the readership. By using intersectionality as the topic of the article, we might have addressed an audience potentially biased toward disclosure as a critical-feminist design practice. As always with case studies, the presented research was carried out with one specific instance of a visual data story and cannot cover the variety of forms and contents (to be) found in practice.
With this work we centered on disclosure as a critical practice that strives to empower data story readers and viewers to appreciate and reflect on data visualizations and their contexts from a critical-feminist perspective. Drawing on prior research on reflexivity, transparency, provenance, and disclosure, we proposed a principled framework for disclosure practice in visual data storytelling that is meant as a scaffolding for design and research of disclosure in data stories. Building on literature of the field, we identified six key aspects: team positions, project process, hidden labor, data settings, design decisions, and critical reflections and complementary representation forms: preamble, complement, thread, disruption, colophon, and appendix embedding disclosure information into visual data story interfaces. We validated this framework with a case study and presented its disclosure-specific design as well as findings from an evaluation. While the results are largely encouraging, our research reveals that time poses a major challenge for disclosure as a critical practice in data visualization. On the one hand, visualization creators have to plan in additional resources to produce and integrate disclosure information, and on the other hand, readers need to have sufficient attention and interest to read or view the additional information provided. Ideally, the status of disclosure in a data visualization should become a central editorial decision. To address open challenges and questions, we would like to see the proposed framework and terminology as an invitation to further develop the critical practice of disclosure.
About the authors
Hannah Schwan recently completed her Master’s thesis at the University of Applied Sciences Potsdam at the intersection of data visualization, interface design, and intersectional feminist theory.
E-mail: hannah [dot] schwan [at] fh-potsdam [dot] deJonas Arndt is a research associate at UCLAB at the University of Applied Sciences Potsdam with a focus on the visual and dynamic arrangements of data stories as part of the VIDAN research project.
E-mail: jonas [dot] arndt [at] fh-potsdam [dot] deMarian Dörk is a research professor in the Department of Design of University of Applied Sciences Potsdam and co-director of the UCLAB, a transdisciplinary research space at the intersection between computing, design, and the humanities.
E-mail: marian [dot] doerk [at] fh-potsdam [dot] de
Acknowledgments
We thank Sandra Cartes for supporting the case study project with her valuable content feedback and advice. We thank the participants of the evaluation questionnaire for their time and generosity in sharing insightful feedback. We thank our colleagues Francesca Morini, Sara Akhlaq, and Nicole Hengesbach for their thoughtful feedback and suggestions on the paper draft. We would like to acknowledge the German Federal Ministry of Education and Research for funding the research project “VIDAN: Visual and dynamic arrangements of narratives” associated with the presented research.
Notes
1. Agre, 1997, p. 155.
2. Weber, et al., 2018, p. 202.
References
Philip E. Agre, 1997. “Toward a critical technical practice: Lessons learned in trying to reform AI,” In: Geoffrey C. Bowker, Susan Leigh Star, Les Gasser, and William Turner (editors). Social science, technical systems and cooperative work: Beyond the great divide. Mahwah, N.J.: Lawrence Erlbaum Associates, pp. 131–157.
Sal Baeza Arguello, Ron Wakkary, Kristina Andersen, and Oscar Tomico, 2021. “Exploring the potential of Apple Face ID as a drag, queer and trans technology design tool,” DIS ’21: Designing Interactive Systems Conference 2021, pp. 1,654–1,667.
doi: https://doi.org/10.1145/3461778.3461999, accessed 20 October 2022.Shaowen Bardzell and Jeffrey Bardzell, 2011. “Towards a feminist HCI methodology: Social science, feminism, and HCI,” CHI ’11: Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, pp. 675–684.
doi: https://doi.org/10.1145/1978942.1979041, accessed 20 October 2022.Liliana Bounegru and Jonathan Gray, 2021. “Introduction,” In: Liliana Bounegru and Jonathan Gray (editors). Data journalism handbook: Towards a critical data practice. Amsterdam: Amsterdam University Press, pp. 11–23.
doi: https://doi.org/10.5117/9789462989511, accessed 20 October 2022.Geoffrey C. Bowker, 2005. Memory practices in the sciences. Cambridge, Mass.: MIT Press.
danah boyd and Kate Crawford, 2012. “Critical questions for big data: Provocations for a cultural, technological, and scholarly phenomenon,” Information, Communication & Society, volume 15, number 5, pp. 662–679.
doi: https://doi.org/10.1080/1369118X.2012.678878, accessed 20 October 2022.Joy Buolamwini and Timnit Gebru, 2018. “Gender shades: Intersectional accuracy disparities in commercial gender classification,” Proceedings of Machine Learning Research, Conference on Fairness, Accountability and Transparency, volume 81, pp. 77–91.
Alyxander Burns, Thai On, Christiana Lee, Rachel Shapiro, Cindy Xiong, and Narges Mahyar, 2021. “Making the invisible visible: Risks and benefits of disclosing metadata in visualization,” 2021 IEEE Workshop on Visualization for Social Good (VIS4Good), pp. 1–15.
doi: https://doi.org/10.1109/VIS4Good54225.2021.00008, accessed 20 October 2022.Alberto Cairo, 2015. “Graphics lies, misleading visuals,” In: David Bihanic (editor). New challenges for data design. London: Springer, pp. 103–116.
doi: https://doi.org/10.1007/978-1-4471-6596-5_5, accessed 20 October 2022.Steven P. Callahan, Juliana Freire, Emanuele Santos, Carlos E. Scheidegger, Cl´udio T. Silva, and Huy T. Vo, 2006. “VisTrails: Visualization meets data management,” SIGMOD ’06: Proceedings of the 2006 ACM SIGMOD International Conference on Management of Data, pp. 745–747.
doi: https://doi.org/10.1145/1142473.1142574, accessed 20 October 2022.Combahee River Collective, 1977. “The Combahee River Collective Statement,” at https://americanstudies.yale.edu/sites/default/files/files/Keyword%20Coalition_Readings.pdf, accessed 20 October 2022.
Michael Correll, 2019. “Ethical dimensions of visualization research,” CHI ’19: Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems, paper number 188, pp. 1–13.
doi: https://doi.org/10.1145/3290605.3300418, accessed 20 October 2022.Kimberlé Crenshaw, 1989. “Demarginalizing the intersection of race and sex: A Black feminist critique of antidiscrimination doctrine, feminist theory and antiracist politics,” University of Chicago Legal Forum, volume 1989, number 1, article number 8, at http://chicagounbound.uchicago.edu/uclf/vol1989/iss1/8, accessed 20 October 2022.
Craig M. Dalton, Linnet Taylor, and Jim Thatcher, 2016. “Critical data studies: A dialog on data and space,” Big Data amp; Society, volume 3, number 1 (20 June).
doi: https://doi.org/10.1177/2053951716648346, accessed 20 October 2022.Susan B. Davidson and Juliana Freire, 2008. “Provenance and scientific workflows: Challenges and opportunities,” SIGMOD ’08: Proceedings of the 2008 ACM SIGMOD International Conference on Management of Data, pp. 1,345–1,350.
doi: https://doi.org/10.1145/1376616.1376772, accessed 20 October 2022.Catherine D’Ignazio and Lauren F. Klein, 2020a. “Seven intersectional feminist principles for equitable and actionable COVID-19 data,” Big Data & Society, volume 7, number 2 (30 July).
doi: https://doi.org/10.1177/2053951720942544, accessed 20 October 2022.Catherine D’Ignazio and Lauren F. Klein, 2020b. Data feminism. Cambridge, Mass.: MIT Press.
doi: https://doi.org/10.7551/mitpress/11805.001.0001, accessed 20 October 2022.Catherine D’Ignazio and Lauren F. Klein, 2016. “Feminist data visualization,” Proceedings from the Workshop on Visualization for the Digital Humanities at IEEE VIS Conference; version at http://www.kanarinka.com/wp-content/uploads/2015/07/IEEE_Feminist_Data_Visualization.pdf, accessed 20 October 2022.
Marian Dörk, Patrick Feng, Christopher Collins, and Sheelagh Carpendale, 2013. “Critical InfoVis: Exploring the politics of visualization,” CHI EA ’13: CHI ’13 Extended Abstracts on Human Factors in Computing Systems, pp. 2,189–2,198.
doi: https://doi.org/10.1145/2468356.2468739, accessed 20 October 2022.Johanna Drucker, 2017. “Information visualization and/as enunciation,” Journal of Documentation, volume 73, number 5, pp. 903–916.
doi: https://doi.org/10.1108/JD-01-2017-0004, accessed 20 October 2022.Johanna Drucker, 2011. “Humanities approaches to graphical display,” Digital Humanities Quarterly, volume 5, number 1, at http://www.digitalhumanities.org/dhq/vol/5/1/000091/000091.html, accessed 15 June 2022.
Steven Drucker, Samuel Huron, Robert Kosara, Jonathan Schwabish, and Nicholas Diakopoulos, 2018. “Communicating data to an audience,” In: Nathalie Henry Riche, Christophe Hurter, Nicholas Diakopoulos, and Sheelagh Carpendale (editors). Data-driven storytelling. Boca Raton, Fla.: CRC Press, pp. 211–232.
doi: https://doi.org/10.1201/9781315281575, accessed 20 October 2022.Linda Finlay, 2002a. “Negotiating the swamp: The opportunity and challenge of reflexivity in research practice,” Qualitative Research, volume 2, number 2, pp. 209–230.
doi: https://doi.org/10.1177/146879410200200205, accessed 20 October 2022.Linda Finlay, 2002b. “‘Outing’ the researcher: The provenance, process, and practice of reflexivity,” Qualitative Health Research, volume 12, number 4, pp. 531–545.
doi: https://doi.org/10.1177/104973202129120052, accessed 20 October 2022.Juliana Freire, David Koop, Emanuele Santos, and Claudio Silva, 2008. “Provenance for computational tasks: A survey,” Computing in Science and Engineering, volume 10, number 3, pp. 11–21.
doi: https://doi.org/10.1109/MCSE.2008.79, accessed 20 October 2022.Lisa Gitelman, 2013. “Raw data” is an oxymoron. Cambridge, Mass.: MIT Press.
doi: https://doi.org/10.7551/mitpress/9302.001.0001, accessed 20 October 2022.David Gotz and Michelle X. Zhou, 2009. “Characterizing users’ visual analytic activity for insight provenance,” Information Visualization, volume 8, number 1, pp. 42–55.
doi: https://doi.org/10.1057/ivs.2008.31, accessed 20 October 2022.Donna Haraway, 1988. “Situated knowledges: The science question in feminism and the privilege of partial perspective,” Feminist Studies, volume 14, number 3, pp. 575–599.
doi: https://doi.org/10.2307/3178066, accessed 20 October 2022.Sandra G. Harding, 1993. “Rethinking standpoint epistemology: What is ‘strong objectivity’?” In: Linda Alcoff and Elizabeth Potter (editors). Feminist epistemologies. London: Routledge, pp. 49–82.
doi: https://doi.org/10.4324/9780203760093, accessed 20 October 2022.Sandra G. Harding, 1991. Whose science? Whose knowledge? Thinking from women’s lives>. Ithaca, N.Y.: Cornell University Press.
Sonali Hedditch and Dhaval Vyas, 2021. “A gendered perspective on making from an autoethnography in makerspaces,” DIS ’21: Designing Interactive Systems Conference 2021, pp. 1,887–1,901.
doi: https://doi.org/10.1145/3461778.3462015, accessed 20 October 2022.Rosemary Lucy Hill, 2020. “What is at stake in data visualization? A feminist critique of the rhetorical power of data visualizations in the media,” In: Martin Engebretsen and Helen Kennedy (editors). Data visualization in society. Amsterdam: Amsterdam University Press, pp. 391–405.
doi: https://doi.org/10.1515/9789048543137-027, accessed 20 October 2022.Rosemary Lucy Hill, Helen Kennedy, and Ysabel Gerrard, 2016. “Visualizing junk: Big data visualizations and the need for feminist data studies,” Journal of Communication Inquiry, volume 40, number 4, pp. 331–350.
doi: https://doi.org/10.1177/0196859916666041, accessed 20 October 2022.Patricia Hill Collins, 2009. Black feminist thought: Knowledge, consciousness, and the politics of empowerment. Second edition. London: Routledge.
bell hooks, 2014. Feminist theory: From margin to center. Third edition. New York: Routledge.
doi: https://doi.org/10.4324/9781315743172, accessed 20 October 2022.Jessica Hullman and Nick Diakopoulos, 2011. “Visualization rhetoric: Framing effects in narrative visualization,” IEEE Transactions on Visualization and Computer Graphics, volume 17, number 12, pp. 2,231–2,240.
doi: https://doi.org/10.1109/TVCG.2011.255, accessed 20 October 2022.Andrew Iliadis and Federica Russo, 2016. “Critical data studies: An introduction,” Big Data & Society, volume 3, number 2 (16 October).
doi: https://doi.org/10.1177/2053951716674238, accessed 20 October 2022.Helen Kennedy, Rosemary L. Hill, Giorgia Aiello, and William Allen, 2016. “The work that visualisation conventions do,” Information, Communication & Society, volume 19, number 6, pp. 715–735.
doi: https://doi.org/10.1080/1369118X.2016.1153126, accessed 20 October 2022.Rob Kitchin and Tracey P. Lauriault, 2018. “Towards critical data studies: Charting and unpacking data assemblages and their work,” In: Jim Thatcher, Andrew Shears, and Josef Eckert (editors). Thinking big data in geography: New regimes, new research. Lincoln: University of Nebraska Press, pp. 3–20.
Yanni Alexander Loukissas, 2019. All data are local: Thinking critically in a data-driven society. Cambridge, Mass.: MIT Press.
doi: https://doi.org/10.7551/mitpress/11543.001.0001, accessed 20 October 2022.Natalia Mazotte, 2021. ”Working openly in data journalism,” In: Liliana Bounegru and Jonathan Gray (editors). Data journalism handbook: Towards a critical data practice. Amsterdam: Amsterdam University Press, pp. 138–142.
doi: https://doi.org/10.5117/9789462989511, accessed 20 October 2022.Miriah Meyer and Jason Dykes, 2019. “Criteria for rigor in visualization design study,” IEEE Transactions on Visualization and Computer Graphics, volume 26, number 1, pp. 87–97.
doi: https://doi.org/10.1109/TVCG.2019.2934539, accessed 20 October 2022.Garrett Dash Nelson, 2021. “Roll the tape with Moviemaps,” Leventhal Map & Education Center at the Boston Public Library (18 May), at https://www.leventhalmap.org/articles/roll-the-tape-with-moviemaps/, accessed 15 June 2022.
Chris North, Remco Chang, Alex Endert, Wenwen Dou, Richard May, Bill Pike, and Glenn Fink, 2011. “Analytic provenance: Process+ interaction+ insight,” CHI EA ’11: CHI ’11 Extended Abstracts on Human Factors in Computing Systems, pp. 33–36.
doi: https://doi.org/10.1145/1979742.1979570, accessed 20 October 2022.Anshul V. Pandey, Katharina Rall, Margaret L. Satterthwaite, Oded Nov, and Enrico Bertini, 2015. “How deceptive are deceptive visualizations? An empirical analysis of common distortion techniques,” CHI ’15: Proceedings of the 33rd Annual ACM Conference on Human Factors in Computing Systems, pp. 1,469–1,478.
doi: https://doi.org/10.1145/2702123.2702608, accessed 20 October 2022.Anshul V. Pandey, Anjali Manivannan, Oded Nov, Margaret Satterthwaite, and Enrico Bertini, 2014. “The persuasive power of data visualization,” IEEE Transactions on Visualization and Computer Graphics, volume 20, number 12, pp. 2,211–2,220.
doi: https://doi.org/10.1109/TVCG.2014.2346419, accessed 20 October 2022.Nassim Parvin, 2018. &rlquo;Doing justice to stories: On ethics and politics of digital storytelling,” Engaging Science, Technology, and Society, volume 4.
doi: https://doi.org/10.17351/ests2018.248, accessed 20 October 2022.Damian Radcliffe and Seth C. Lewis, 2021. “The datafication of journalism: Strategies for data-driven storytelling and industryacademy collaboration,” In: Liliana Bounegru and Jonathan Gray (editors). Data journalism handbook: Towards a critical data practice. Amsterdam: Amsterdam University Press, pp. 314–330.
doi: https://doi.org/10.5117/9789462989511, accessed 20 October 2022.Eric D. Ragan, Alex Endert, Jibonananda Sanyal, and Jian Chen, 2016. “Characterizing provenance in visualization and data analysis: An organizational framework of provenance types and purposes,” IEEE Transactions on Visualization and Computer Graphics, volume 22, number 1, pp. 31–40.
doi: https://doi.org/10.1109/TVCG.2015.2467551, accessed 20 October 2022.Lars Rinsdorf and Raol Boers, 2016. “The need to reflect: Data journalism as an aspect of disrupted practice in digital journalism and in journalism education,” In: Joachim Engel (editor). Promoting understanding of statistics about society. Proceedings of the Roundtable Conference of the International Association of Statistics Education (IASE), Berlin, pp. 1–7, at https://iase-web.org/documents/papers/rt2016/Rinsdorf.pdf, accessed 20 October 2022.
Edward Segel and Jeffrey Heer, 2010. “Narrative visualization: Telling stories with data,” IEEE Transactions on Visualization and Computer Graphics, volume 16, number 6, pp. 1,139–1,148.
doi: https://doi.org/10.1109/TVCG.2010.179, accessed 20 October 2022.Katta Spiel, 2021. “‘Why are they all obsessed with gender?’ — (Non)binary navigations through technological infrastructures,” DIS ’21: Designing Interactive Systems Conference 2021, pp. 478–494.
doi: https://doi.org/10.1145/3461778.3462033, accessed 20 October 2022.Charles D. Stolper, Bongshin Lee, Nathalie Henry Riche, and John Stasko, 2016. “Emerging and recurring data-driven storytelling techniques: Analysis of a curated collection of recent stories,” MSR-TR-2016-14, at https://www.microsoft.com/en-us/research/publication/emerging-and-recurring-data-driven-storytelling-techniques-analysis-of-a-curated-collection-of-recent-stories/, accessed 15 June 2022.
Miriam Sturdee, Sarah Robinson, and Conor Linehan, 2020. “Research journeys: Making the invisible, visual,” DIS ’20: Proceedings of the 2020 ACM Designing Interactive Systems Conference, pp. 2,163–2,175.
doi: https://doi.org/10.1145/3357236.3395590, accessed 20 October 2022.Lucy Suchman, 2002. “Located accountabilities in technology production,” Scandinavian Journal of Information Systems, volume 14, number 2, pp. 91–105, and at http://aisel.aisnet.org/sjis/vol14/iss2/7, accessed 20 October 2022.
Sarah J. Tracy, 2010. “Qualitative quality: Eight ‘big-tent’ criteria for excellent qualitative research,” Qualitative Inquiry, volume 16, number 10, pp. 837–851.
doi: https://doi.org/10.1177/1077800410383121, accessed 20 October 2022.Wibke Weber, Martin Engebretsen, and Helen Kennedy, 2018. “Data stories: Rethinking journalistic storytelling in the context of data journalism,” Studies in Communication Sciences, volume 18, number 1, pp. 191–206.
doi: https://doi.org/10.24434/j.scoms.2018.01.013, accessed 20 October 2022.
Editorial history
Received 11 July 2022; revised 8 September 2022; accepted 20 October 2022.
This paper is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.Disclosure as a critical-feminist design practice for Web-based data stories
by Hannah Schwan, Jonas Arndt, and Marian Dörk.
First Monday, Volume 27, Number 11 - 7 November 2022
https://firstmonday.org/ojs/index.php/fm/article/download/12712/10721
doi: https://dx.doi.org/10.5210/fm.v27i11.12712
