Data Clothing and BigBarChart: designing physical data reports on indoor pollutants for individuals and communities

Abstract

In response to participant preferences and new ethics guidelines, researchers are increasingly sharing data with health study participants, including data on their own household chemical exposures. Data physicalization may be a useful tool for these communications, because it is thought to be accessible to a general audience and emotionally engaged. However, there are limited studies of data physicalization in the wild with diverse communities. Our application of this method in the Green Housing Study is an early example of using data physicalization in environmental health report-back. We gathered feedback through community meetings, prototype testing, and semistructured interviews, leading to the development of data t-shirts and other garments and person-sized bar charts. We found that participants were enthusiastic about data physicalizations, connected them to their previous experience, and had varying desires to share their data. Our findings suggest that researchers can enhance environmental communications by further developing the human experience of physicalizations and engaging diverse communities.

OVER THE past decade, personal data reporting to individuals participating in studies has become more prevalent due to research findings on ethics and the community experience of data reporting, community advocacy, and new institutional practices. At the same time, there has been increased interest in the field of data physicalization, which was defined in 2015 as “a physical artifact whose geometry or material properties encode data” [1]. Since then, authors have described a continuum between visualization and physicalization “which includes media such as paper that have both virtual and physical qualities, and hybrid representations that combine solid objects with video-projected overlays” [2]. These approaches use a wide range of materials and scales and include tangible and embedded data displays [1], [3],[2]. Possible advantages include their ability to engage with the emotional aspects of data [4], their ability to bring data to a broader audience through artistic or educational approaches [5], and the hands on nature of many of these projects [6]. Data physicalizations have also been used to provide an opportunity for self reflection on one’s own data, including projects on time management (e.g. Michael Hunger’s “On LEGO Powered Time-Tracking”) and personal finances (e.g. Hampus Edström’s “Dynamic Bar Chart to Visualize One’s Finances”) [7] and may better engage the embodied nature of data. Yet the role and impact of these approaches are not fully understood as research in these areas is recent.

This study combines progress in data physicalization and data reporting to explore the potential of data physicalizations for sharing data with participants in an environmental health study. We describe three data physicalizations based on research on indoor pollutants in Boston, Massachusetts; Cincinnati, Ohio; and Northern California. This includes individual pieces—Dressed in Data and Data Shirts—that are designed to share personalized data with each study participant, and a collective physicalization—BigBarChart—intended to create an interactive experience that allows the community to explore the overall study data. We discuss the design and fabrication of these data physicalizations and our findings from using them in the real world.

DATA REPORTING IN THE GREEN HOUSING STUDY

The Green Housing Study (GHS) was designed to measure the impact of energy-efficient “green” renovation on indoor environmental quality and the health of children with asthma living in public housing. The study included green-renovated and control public housing units in Cincinnati (n = 28) and metro-Boston (n = 44) and measured indoor environmental chemicals and allergens through air, dust, urine, and blood samples, as well as collecting data on children’s asthma. Air and dust samples were analyzed by gas chromatography mass spectrometry at Southwest Research Institute and urine samples were analyzed via high performance liquid chromatography tandem mass spectrometry at the CDC Environmental Health Laboratory [8]. Sampling occurred from 2011 to 2013 and was led by researchers from the Harvard T.H. Chan School of Public Health, the University of Cincinnati, and Silent Spring Institute. Over one hundred environmental chemicals were measured including pesticides, flame retardants, fragrances, polycyclic aromatic hydrocarbons, polychlorinated biphenyls, parabens, and phthalates. Some of these chemicals impact asthma symptoms or outcomes while others have been linked to a range of potential health impacts from hormone disruption to cardiac events to cancer [8], [9].

GHS study results were shared with participants through community meetings, personalized paper reports, and prototype data physicalizations. The paper reports were designed by the GHS and Personal Report-back Ethics study teams based on past research [10], [11]. Study participants received their personal reports at community meetings, during home visits, or by mail. Interviews with participants after they received their reports showed that they appreciated getting their results, learned from them, and were motivated to change exposure-related behaviors [8].

Data physicalization reporting prototypes were designed to compliment and build from the information presented in the community meetings and personal exposure reports. This provided an opportunity to test prototypes and better understand how data physicalization could be used with a broad group of people, including public housing residents, environmental health researchers, and artists. Interviews and ethnographic observations with these communities (Table 1) explored the three data physicalizations,—BigBarChart, Data Shirts, and Dressed in Data—and centered on understanding how data physicalizations may be used in data reporting and how they are received by a general audience. This work suggests that data physicalization can be an important part of reporting exposure data to study participants and that data physicalization researchers can learn from working with these studies where researchers are already collecting data and sharing it back to diverse communities.

Table 1.

Data collected for three data physicalizations for reporting of environmental exposure data.

Project	Data collected
	Event	Participant population	Data
BigBarCliart	GHS community meeting, Cincinnati	~20 community members ~5 environmental researchers	Ethnographic observation
	2014 International Society of Exposure Science conference (Figure 3a)	~25 environmental health researchers, exposure scientists, policy makers	Ethnographic observation
	Prototype testing	2 graduate students	Semi-structured interviews
Dressed in Data	2013 MIT Fashion Show (Figure 3c)	~15 university students, faculty, staff, artists, and industry employees	Ethnographic observation
	2013 LINKS conference at MIT	~10 researchers in computer science, data visualization, and industry	Ethnographic observation
	Prototype testing	2 graduate students	Semi-structured interviews
Data Shirts	GHS community meeting, Cincinnati	~20 community members ~5 environmental researchers	Ethnographic observation
	2014 International Society of Exposure Science conference (Figure 3a)	~25 environmental health researchers, exposure scientists, policy makers	Ethnographic observation
	Prototype testing	3 families living in public housing in Boston	Semi-structured interviews
	GHS community meeting. Old Colony, South Boston	~12 GHS study participants and community members ~5 environmental health researchers	Ethnographic observation
	GHS community meeting, Castle Sq, Boston	~20 GHS study participants and community members ~8 environmental health researchers	Ethnographic observation
	Post-study interviews	10 GHS Boston area study participants who received a data shirt	Semi-structured interviews

DESIGN CONSIDERATIONS

The aim of these data physicalizations was to create an engaging entry point to environmental exposure data. We focused on designing pieces that were aesthetically interesting (e.g. exploring color, interaction) and approachable (e.g. soft objects with organic forms) while also leveraging novelty of data physicalizations to engage audiences.

BigBarChart aimed to add a hands-on collective way to explore data during community meetings and Data Shirts sought to supplement the individual paper data reports by creating shirts that each individual participant could take home representing their own results. BigBarChart and Data Shirts physicalized data from a class of chemicals relevant to asthma, a central concern of participants in the study, while Dressed in Data physicalized an entire suite of chemical exposures for one individual and is an artistically focused precursor to Data Shirts. All physicalizations sought to address the central questions that environmental health study participants identified in earlier report-back research [10]:

1. What did you find?

2. How much?

3. Is that high?

4. Is it safe?

5. Which of the many chemicals and exposure sources should I focus on?

6. Where did the chemical come from?

7. What can/should I do?

PHYSICALIZATIONS & FABRICATION

BigBarChart: interactive data physicalization for community exploration

BigBarChart (Figure 1) is a human-sized interactive bar chart that was designed to facilitate collective discussion and exploration of environmental data, for example, at community meetings. It is intended to be an adaptable physicalization system that can be used with any small dataset that can be represented through bar charts.

Figure 1.

Design sketch for BigBarChart. Human-sized bars fill a room to create an immersive space for people to experience and discuss data together. (From [8], with permission.) (img: Laura Perovich, Pip Mothersill).

BigBarChart builds from work exploring 3D printed bar charts [12], dynamic physical bar charts [13], and co-created physical bar charts (e.g. Lucy Kimball’s “Inverted Participatory Bar Charts”) [7], among others. Documented benefits of physical bar charts include an increase in the speed with which people could read information off a graph relative to a virtual plot and support for physical ways of thinking [12]. Research includes handheld and tabletop sized bar charts, as well as VR explorations on the impacts of size and scale in data visualization [14].

A number of interaction modes and mechanisms for the bars were tested during the design process, including origami paper tubes and pop-up laundry hampers. Initial prototyping underlined the potential of form factors that were soft, a bit amorphous, and human sized, leading to further development of the pop-up laundry hamper prototypes. Human sized bars offered the affordance of comparing the height of the bars against one’s own height and could impact people’s sight lines as embodied forms of measurement. A number of interaction modes were tested, including direct manipulation of bars through pushing, squeezing of bars, and physical variable selection buttons (Figure 2). Selection buttons were used for the initial prototype as their affordances best matched the dataset and simplified the control system.

Figure 2.

Interaction modes prototyped for BigBarChart included (a) soft buttons where variables could be selected using bean bags (b) handprints on the bars that could allow data selection through squeezing (c) squishing bars to understand relationships within the dataset. (img: Laura Perovich)

Each bar in BigBarChart included a wood base with a geared motor system that wound strings to compress the pop-up hampers to the appropriate height to represent quantitative variables. Qualitative variables were represented through color changing LEDs within the pop-up hampers, and the system was controlled by networked Arduino Unos and a python script. Pop-up hampers were custom manufactured in collaboration with Yiwu Kenong Daily Commodity Factory and were 5 feet 5 inches tall and just over two feet wide, with a translucent white nylon fabric exterior and telescoping poles inside the bars.

Dressed in Data: aesthetic data overviews to create intuitions through fashion

Dressed in Data (Figure 4) is a set of artistic clothing pieces representing the indoor air exposure data from one participant in the Northern California Household Exposure Study [15]. It aims to provide a visually intriguing overview of an individual’s exposures across chemical classes.

Figure 4.

The front and back of four Dressed in Data pieces that represent chemical levels in the indoor air of one individual in the California Household Exposure Study. (img: Laura Perovich).

Dressed in Data contributes to fashion-based work expressing data about the body on the body. This includes Xuedi Chen’s and Pedro Oliveira’s x.pose that physicalizes individual privacy data through changing the transparency of panels in a 3D printed top [7] and a tshirt that represents real-time exposure to air pollution through LEDs [16], among others.

Dressed in Data uses data from the Northern California Household Exposure Study, as the GHS data was still being gathered at that time. It was our first effort in physicalizing environmental exposure data, and we focused on the “how many chemicals” and “how much” questions frequently posed by study participants [10]. A number of ways of displaying the data were tested and ultimately we grouped the 104 indoor air chemicals and 6 chemical classes into four outfits (~5 chemicals were excluded due to analytic issues or redundancies). Both rectangles (length) and squares (area) were tested for representing amounts of each chemical and squares were chosen for aesthetic reasons, though data perception research indicates that area may be more difficult to interpret [17].

We prototyped designs using both measured and relative concentrations to determine the size of the squares. This reflects challenges faced by researchers in interpreting data on emerging contaminants when little is known about the potential health impacts, and it is unclear at what level exposures may cause harm. We decided to scale the squares relative to the GHS median detected value for each chemical, in order to provide some normalization across chemical classes to respond to the “how much?” and “is that high?” questions. This design decision emphasizes outlier chemical measurements within the study population but does not account for the fact that this community may be universally higher than another community for a particular chemical. Ideally, all chemicals would be scaled relative to a human health standard or a national median but that data does not currently exist for most of these chemicals.

Data lace patterns were computationally designed using the R statistical programming language, laser cut with supporting pieces, glued onto a laser cut translucent fabric grid, and finalized through cutting away the supports (Figure 5). A rough response to the “where did the chemical come from?” report-back question was represented through the materials; for example, garments representing pesticide data used green and purple fabric rastered with a petal design in a nod to flowers where pesticides may be used. Data lace materials representing chemical exposures were chosen to create dissonance in the piece as a gesture to the negative impacts of these chemicals.

Figure 5.

Lace patterns on the Dressed in Data garments indicate the number of chemicals detected and the relative amount detected for (a) industrial chemicals (b) pesticides (c) phthalates (d) metals. More squares indicate that more chemicals were detected and larger squares indicate a higher amount of that chemical relative to other homes in the study. (img: Laura Perovich).

Data Shirts: personalized clothing for individual exposure data

Data Shirts (Figure 6) is a set of customized shirts showing individuals their relative exposure to nine phthalates, chemicals related to asthma. Phthalates are common in some plastics and fragranced cleaning and personal care products, among other sources. Forty-four Boston area GHS participants received a personalized shirt by mail or at two community meetings. Data Shirts builds from feedback from Dressed in Data and textile data physicalizations including Kristen Cooper Nutbrown’s temperature scarves and embroidered personal data [12], [18].

Figure 6.

The final Data Shirt design for the Boston GHS report-back included (a) a lollipop graph that represents the relative amount of 9 phthalates found in the dust of the individual’s home (b) a duck shaped QR code to access more detailed information on a website (c) a tag that describes the graph and the chemicals. (From [8], with permission.) (img: Laura Perovich).

Data Shirts was designed to address the previously identified questions asked by exposure study participants [10] as part of the GHS data reporting. Information on the number and relative amount of chemicals found in dust was placed on the front of the shirt (Figure 6a), information about the health impacts and sources of the chemicals was included in an inside tag on the shirt (Figure 6c), and a duck-shaped QR code on the sleeve allowed participants to access more detailed information through a website (Figure 6b).

This design was intended to provide an entry point to the larger data report-back and to be more easily interpretable than Dressed in Data, while also recognizing the aesthetic requirements of clothing.

We tested a number of fabrication processes (Figure 7) and found screen printing using vinyl cut stickers produced the best outcomes; other processes led to messy prints and unpleasant textures, were time consuming, or could not be sufficiently customized. We tested prototypes using different personal items (e.g. scarves, phone cases, ties, socks) and amounts of data, and saw a preference for simple designs on shirts during community testing (Table 1). Our team sketched over 20 designs based on sample data and created full-size shirts of three designs (Figure 8) spanning a range of approaches:

Figure 7.

Many Data Shirt fabrication processes were tested including (a-b) laser rastering fabrics (c) sun dyed fabrics (d) thermochromic pigment (e) spray painted paper stencils (f) hand painted sticker stencils. (img: Phoebe Cai, Amber Guo).

Figure 8.

We tested three prototype Data Shirts designs with Boston families including (a) a skull, snakes, and roses design (b) a Chernof faces design (c) a lollipop plot design. (img: Laura Perovich).

Lollipop chart:

each line represents a chemical; line height represents the amount (literal; abstract pattern)

Modified Chernoff face plot:

each face represents one chemical; expression and color represent the amount (literal-metaphorical; recognizable form)

Punk rock-like skull design:

snake length represents the amounts of detected chemicals; flowers represent chemicals that are not detected (metaphorical; recognizable form).

We selected a design based on feedback from three families living in Boston public housing. All families found the skull design to be “creepy” and culturally undesirable while they were positive about the other two designs. We selected the lollipop chart as there was a slight preference for the interpretability of this design, and one family perceived it to be more gender neutral. Families also expressed a preference for black long sleeve shirts. As in Dressed in Data, we represented the data for each participant relative to the median value for that chemical to contextualize the information. Additional information on the sources and health effects of the chemicals was included on the inside of the shirt, so participants could choose whether to disclose the meaning of the shirt, as we saw that some people expressed ambivalence about sharing personal data while others wanted to use the garment to start conversations about asthma.

DATA AND DATA ANALYSIS

BigBarChart, Dressed in Data, and Data Shirts were evaluated qualitatively through ethnographic observation and semi-structured interviews during and after the design process. We obtained feedback from community members, artists, environmental scientists, and data visualization researchers, detailed in Table 1.

The GHS included families living in public housing, many of whom have had limited access to economic resources or formal education, or experience other stressors. Participants in Cincinnati were 98% African-American and participants in Boston were about 40% Chinese-American and 40% African-American or Hispanic which was roughly representative of community meeting attendees. All participants in the semi-structured Data Shirt interviews identified as female; 50% were Hispanic . and 10% Chinese-American. Interviews on Data Shirt prototypes were conducted with Chinese-American families and BigBarChart and Dressed in Data interview participants were 50% Hispanic.

Interview transcripts were originally analyzed by three members of the GHS research team to evaluate experiences of exposure study participants receiving their reports [8]. We developed codes based on interview questions and conceptual themes, grouping and summarizing them to identify patterns across participants. New codes were added as themes emerged from the transcripts. Transcript excerpts were identified, collected, and linked to codes using Excel and systematic application of the codes was verified by comparing across researchers. This iterative process surfaced roughly 15 themes. Six themes focused on participants’ experiences of receiving their results and how they used the results to try to make changes for themselves and their communities, as were previously reported [8]. The original codes and transcripts were then re-reviewed by the first author to reassess the relevance to data physicalization themes. New themes and variations on themes not included in the first publication were added, for example the ways in which participants used data physicalizations to think about privacy and results sharing with others. Data physicalization codes and themes in the transcripts were reviewed by the authors and complemented by community meeting notes, team emails, and design documents previously collected from the GHS research team and within the data physicalization team

FINDINGS AND OBSERVATIONS

We identified four main themes across BigBarChart, Data Shirts, and Dressed in Data that can inform future work in data physicalization in the wild and in data reporting.

Theme 1: Researchers and community members from diverse backgrounds were curious and excited about the data physicalizations

Data physicalizations often caught people’s attention. During the Cincinnati community meeting, many people commented on BigBarChart very shortly after entering the meeting room. Dressed in Data also attracted considerable speculation at LINKS 2013 and during the 2013 MIT fashion show.

Community members who provided feedback on the early data shirt prototypes were very interested in the concept and eager to give their opinions and shape the outcome. GHS participants who received the shirts found them to be memorable. When asked about her report back results, one participant quickly mentioned, "I got a T-shirt" (G16). Another reported that the shirt immediately attracted her child’s curiosity, “[she] was in love with her sweater…The minute I opened it, she put it on and she was asking me about the writing and what it meant” (G19).

Overall we saw that the data physicalizations caught people’s interest in a positive manner. Positive emotional response is important, because many environmental health researchers have been reluctant to share personal exposure results with study participants because they feared generating alarm, and these expectations have been a barrier to adopting ethics recommendations for report-back [19]. Spontaneous engagement is also important in the context of data reporting in public housing where many participants have had limited access to formal education or experience language barriers. Data physicalization’s ability to provide an exciting way to begin engaging with environmental data supports the efforts by the research team to create reports that are easy to understand, respond to participants’ needs, build ownership, and create health-protective action.

Theme 2: Data physicalization is novel and people attempted to understand it by connecting it to their prior experience

Many people were not familiar with the idea of data physicalization and tried to connect it to their prior experience to make sense of it. At first, people didn’t know they were seeing “data.” During the Cincinnati community meeting we observed a number of people comparing BigBarChart to other things they had seen, with one individual relating it to the creatures in the subway scene in The Wiz. A number of data professionals attending the LINKS conference were also unfamiliar with data physicalization techniques and preferred to focus on the lace part of Dressed in Data and to interpret it as a grid with data points as opposed to considering the outfit more holistically.

Similarly, participants perceived the Data Shirts to be novel and puzzled over them. One participant reported that her children thought the Data Shirt was both interesting and odd:

[My son] told me it was cool. [His brother] said it was weird…like he never had a tshirt like that. He thought it was strange. Cool.

(G17)

Another participant described her interpretation of the design on the shirt:

It look like a robot on the arm…or a…it look like a eye or something…And this is like some poles or something going up, with face…with little holes…

(G15)

Data physicalization approaches are novel to a wide range of communities and led to a variety of interpretations as people attempted to connect them to their prior experiences.

Theme 3: Participants were enthusiastic about customized designs that were particular to themselves or the study

A number of Boston GHS participants brought up that their Data Shirts were customized for them, for example, saying “I think iťs cool. It really is…iťs custom made for him” (G10) and that “[my daughter] was like impressed. She’s like, ‘It was made just for me.’ I’m like, ‘Yeah, it was made exactly for you.’” (G19) Receiving an object unique to them was unexpected and interpreted as a sign of effort and respect by the research team.

On the flip side, our team also noticed the challenges of designing a generic physical interface to express a particular data set while prototyping BigBarChart. It was created to be an adaptable physicalization system that, similar to most visualization systems, is decoupled from the world so as to be “compatible with a range of concrete datasets” and “used in a range of contexts” [2]. Yet we observed that it was difficult for people to relate to the physicalization partially because there was not a visual or metaphorical connection to GHS; the form factor did not have any connection to people, homes, environmental pollutants, or asthma. These challenges may be particularly impactful for physicalizations as people have little familiarity with these ways of expressing data, and suggests that custom designs may be more appropriate. As researchers, we also found it difficult to design a physical interface that was easily adaptable to the particularities of all datasets—e.g. scale and spread of quantitative variables, number of categories in qualitative variables.

Theme 4: People’s desire to share their own data varied

The Data Shirts interviews surfaced the varying visibility levels that participants wanted for their data and the communities with which they may be interested in sharing it. As personal data is embedded in the physicalizations, considerations of privacy and desire for sharing is important in determining form factors and interpretability of physicalizations. One participant expressed their eagerness to share their data with their colleagues, saying "I will wear it [the] first day back to work" and describing an interest in sharing study information with her social circles (G14). Another participant mentioned that they only wanted to share their information in particular contexts, "we don’t usually share our t-shirt with other people, but that day at the meeting, we did talk about it with other participants briefly" (G12).

Other participants limited when or where they wore their Data Shirts. Two participants (G17 and G19) said that they or their children had worn the shirt inside the home but not in public. Another participant discussed her concern that wearing the shirt might cause problems for her son:

I donť know, maybe it would be like targeting him, if he wears something like that because people going to be like looking at him like, "What is that?"…For people that maybe have done the study, they would notice that it was from the study but you know kids. They can be brutal.

(G16)

Participants used the Data Shirts to have conversations and share information with the communities they are comfortable with. For some participants this is confined to the home while others wanted to engage more broadly.

DISCUSSION

This study shows that data physicalizations can be a useful tool in reporting data to study participants and, at the same time, the data physicalization community can learn from engaging in these contexts where researchers are already collecting and sharing data with communities. These partnerships enable data physicalization researchers to offer their expertise for community benefit while also advancing the field by gaining feedback in a real-world context.

We found people were curious and enthusiastic about exploring the data physicalizations and giving feedback on the prototypes. Prior work has shown that participants value data report-back in environmental health studies [10], [11] and data physicalizations may be a way to expand on this enthusiasm to promote further engagement. Collective data physicalizations, such as BigBarChart, provide an opportunity to engage people adjacent to the study who are not receiving their own reports, for example extended family. Previous work on data reporting found that family members who were skeptical about the role of environmental chemicals in health could create barriers for the study participants who wanted to adopt changes [8], so engaging relatives and community members through data physicalizations might be beneficial in creating broader action.

The process of creating data physicalizations in an environmental health study provides an important opportunity to get broader feedback on data physicalizations. Working with multicultural communities (described in “Data and Data Analysis”) provided important insights on how people connect physicalizations to cultural anchors that may or may not be desirable in the design, for example the comparison of BigBarChart to The Wiz and the negative connotations of skulls in the Data Shirt prototypes. Working with communities that are not familiar with the concept of data physicalization or prototyping technologies provided an important reminder of how uncommon these approaches can be and underlined the importance of creating a framing for these experiences that leverage excitement while also creating a path towards interpretation. While novelty can create considerable enthusiasm it may also lead to confusion or resistance. Data physicalization researchers could engage more intentionally with experience designers in order to craft clear paths towards engaging with physicalizations. Some research does speak to the experience of physicalizations, such as the data mural processes [20], community environmental installations [21], and constructive visualizations [22], though much of the field emphasizes the translation of data into physical form and data perception.

Creating data physicalizations as part of a data reporting study builds a space to exchange knowledge with other research communities working on data communication. Feedback on BigBarChart prototypes in Cincinnati led to the suspension of its design for GHS, as other approaches were more logistically adaptable (easier transportation, fewer requirements around space) and seemed easier for participants to engage with, as they did not have the unreliability of early interactive prototypes. These meetings also guided the simplification of Data Shirts. In addition, partnering with a research study that used community engaged techniques allowed us to access community members who gave important feedback on the three finalist shirt designs and helped us select an appropriate option. Future data physicalization work in report-back contexts could take these approaches further through codesign processes, though different populations may have varying abilities to participate due to time or resource constraints.

Finally, our work points to the importance of considering privacy and sharing when creating data physicalizations for study participants. This is relevant to any physicalization based on a person’s data, but may be particularly important in data report-back where some individuals are eager to share their study experience broadly and others would like to receive their results themselves or share them only with their families. Participants were able to use the Data Shirts to meet all these goals by varying the contexts in which they wore the garments. Our early prototypes also leveraged the affordances of various garments for varying privacy—for example, socks are much less visible than shirts. Future work could allow participants to select a garment for displaying their data that best fits their sharing preferences.

LIMITATIONS

A strength of this study is the community context in public housing in two different cities, a population that needs improved data access to address environmental health disparities. However, the interviews were limited to the Boston area and may not be representative of other public housing communities. Many preferences around data physicalization may be community specific and further research using codesign processes for data physicalization in report-back would surface additional insights. GHS data physicalizations grew from previous research on data communication for exposure report-back in communities and therefore did not explore perception; for example, how standing near a human sized bar chart versus walking through it may impact how participants understand data from the physicalization. These topics could be explored in complementary lab based studies, however, community-based case studies using data of interest to diverse populations offer unique perspectives that are important to this developing field.

CONCLUSION

This study describes three data physicalizations used to share environmental health data with communities and study participants. It builds from ongoing work in data reporting and research on physical ways to share data with diverse communities in the wild. We found that participants were curious and excited about the data physicalizations, supporting hypotheses on the value of these approaches for engaging people. At the same time, data physicalizations are still unfamiliar to many groups and require additional design work to build an experience around the physicalization that leverages this enthusiasm into engagement and problem-solving. Creating physicalizations for data report-back is one way to work with diverse communities and learn from other data communication researchers. It also surfaces possible participant concerns and design opportunities in data physicalization, including data privacy and sharing, and the challenges of scaling up physicalizations relative to digital or paper reporing.

Figure 3.

Data physicalization prototypes were tested at (a) the 2014 International Society of Exposure Science conference (BigBarChart, environmental researchers) (b) GHS community meeting in Cincinnati (Data Shirts, community members and environmental researchers) (c) the 2013 MIT Fashion Show (Dressed in Data, artists). (img: Silent Spring Institute, Andy Ryan).

Biography

LAURA J. PEROVICH is currently an Assistant Professor in Art+Design with Northeastern University, Boston, MA, USA. She was previously a Research Assistant with Silent Spring Institute. She received a Ph.D. degree in Media Arts and Sciences from the Massachusetts Institute of Technology, Cambridge, MA, USA. She is the corresponding author of this article. Contact her at perovich@media.mit.edu.

PHOEBE CAI is currently working toward a Ph.D. degree in Economics with Harvard University, Cambridge, MA, USA. She was previously an Undergraduate Researcher with the Massachusetts Institute of Technology. Contact her at pcai@g.harvard.edu.

AMBER GUO is currently a Program Manager with Microsoft, Seattle, WA, USA. She was previously an Undergraduate Researcher with the Massachusetts Institute of Technology. Contact her at atguo@alum.mit.edu.

KRISTIN ZIMMERMAN is currently a Shop Artist with the Weber Group, Sellersburg, IN, USA. She was previously an Undergraduate Researcher with the Massachusetts Institute of Technology. Contact her at ninjissimo@alum.mit.edu.

KATHERINE PASEMAN is currently the Head Of Customer Experience at Fix The Mask, Davis, CA, USA. She was previously an Undergraduate Researcher with the Massachusetts Institute of Technology. Contact her at katherine.paseman@gmail.com.

DAYANNA ESPINOZA SILVA is currently a Software Engineer with Oracle, Santa Clara, CA, USA. She was previously an Undergraduate Researcher with the Massachusetts Institute of Technology. Contact her at despinoza@alum.mit.edu.

JULIA G. BRODY is currently the Executive Director and a Senior Scientist with Silent Spring Institute, Newton, MA, USA. Contact her at brody@silentspring.org.