Security for Seniors
I worked with a team to conduct semi-structured interviews with 46 older adults (65+ years). In coding and analyzing these transcripts, we identified a complex range of privacy and security needs specific to this population, along with common threat models, misconceptions, and mitigation strategies. Supported by a research grant from the Center for Long-Term Cybersecurity, the results were published and presented at the 2019 Symposium on Usable Privacy and Security (SOUPS). The paper was also highlighted during the USENIX keynote by Alex Stamos, chief information security officer at Facebook.
Qualitative Coding & Analysis
Older adults (aged 65+) are emergent users of smart systems, especially in health care. However, these technologies are often not designed for older users; moreover, they can pose serious privacy and security concerns due to their novelty, complexity, and propensity to collect vast amounts of sensitive information. Due to infrequent use, limited technical knowledge, and age-related declines in ability, older adults can be particularly vulnerable to privacy and security risks.
To study older adults’ privacy and security attitudes, we conducted 1–1.5 hour semi-structured interviews. In these conversations, we discussed: (1) current technology use, (2) privacy- and security-related concerns and (3) user management strategies. Participants were recruited via nursing homes, senior centers, and organizations for retired people in the San Francisco Bay Area.
We screened potential participants using surveys in several formats — online, phone, paper, and in person — but excluded individuals with serious cognitive impairments and limited English comprehension. With approval from the Institutional Review Board, we conducted interviews from May to June 2018; interviews were conducted with 46 participants in person, at their residences or public senior centers (their choice). Sessions concluded with exit surveys about participants’ individual characteristics, either self-administered or by the researcher (also their choice).
The interviews were audio recorded, professionally transcribed, then coded. Three coders iteratively developed a codebook by independently coding subsets of transcripts and jointly resolving conflicting codes. To maximize the value of thematic analysis, 4 researchers used a holistic coding approach, in which at least 2 coders per interview applied open coding to entire interviews, independently selecting excerpts to annotate. All 4 coders then resolved disagreements at the interview level (so that at least 3 out of 4 agreed).
︎A participant examines some graphics used as probes to understand potential technologies and their uses.
︎Participants were recruited through tear-off flyers and in-person announcements at senior centers. Paper screeners included self-sealed envelopes for return to the researchers.
︎One participant, an avid Fitbit and iPhone user, fills out the paper exit survey.
We identified a range of complex privacy and security behaviors specific to this population – from use of public and secondhand devices to caregiving responsibilities both for themselves and others. Our work adds depth to current models of how older adults’ limited technical knowledge and age-related ability declines amplify vulnerability to certain risks; we found that health, living situation, and finances play a notable role as well.
We also identified usability issues with current protections, learning and troubleshooting approaches, and misconceptions regarding security and privacy. We found that older adults often struggle to mitigate known and unknown risks — and that managing them frequently consists of limiting or avoiding technology use. Building on the preferences of older adults, we conclude by offering privacy- and security-enhancing recommendations for product developers and for educational programs.
For more information, see the resulting SOUPS paper and presentation.
︎One blind participant shares her digital audiobook player as a technology she frequently uses.
︎Another participant shares her Jitterbug phone. She described frequently forgetting to carry it with her – in this instance, leaving it in the trunk of her car.
︎ The paper was highlighted during the 2019 USENIX keynote by Alex Stamos, chief information security officer at Facebook.
The project was a valuable exercise in recruiting participants of a somewhat difficult-to-reach population. It also highlighted the need to use a variety of approaches — i.e. online, phone, paper, and in person — to administer screeners and surveys for the population of study. Lastly, given that the interviews required the discussion of sensitive topics, it helped to develop skills in establishing rapport and gaining trust.
During a 12-week summer internship, I conducted design research for the buyer experience team at Atlassian. This included 25 stakeholder interviews as well as 3 iterative approaches to communicate customer perspectives in novel, actionable ways.
Synthesis & Ideation
Atlassian is an enterprise software company with a unique self-service sales model. This means pricing is transparent, and customers are able to make purchases directly on the website, without having to haggle with a sales person. This approach helps to lower product prices, as the company does not bear the cost of employing a traditional sales force.
Despite these advantages, some customers could benefit from the guidance of a contact person while making the decisions required to purchase Atlassian’s software. The purchasing process is complex, dependent on company size (or number of licenses), geographic market, and whether the customer is new or returning – not to mention the various product bundles and deployment options.
Given Atlassian’s growth and success, there had been a longstanding sense of “tribal knowledge” that the self-service model was sufficiently easy to use. Recent work to better understand buyer journeys had revealed major pain points, however, and was informing strategic decisions for the upcoming fiscal year. I was asked to deliver an artifact for internal stakeholders that would help keep the buyer’s perspective top of mind while making everyday decisions. Thus, the primary research questions were:
- How might we generate customer empathy while considering stakeholder needs?
How might we communicate the buyer experience in a clear and compelling way?
How might we generate customer empathy while considering stakeholder needs?
Reviewing secondary research
To gain a better grasp of organizational context, I began by reviewing prior visualizations that had been well socialized, as well as existing buyer journey work done by internal product teams. Though all teams had started with the same template, the resulting artifacts took divergent forms, spanning three slide decks, nine Trello boards, and too many intranet pages to count.
These existing journeys were difficult to digest for those not already “deep in the weeds.” Furthermore, they were based on various types of data sources – including surveys, experiments, customer interviews, and internal workshops – and did not consider the overall Atlassian purchasing process. To make sense of all this qualitative data, I created a few Trello boards of my own, looking for overarching – and when possible, product-agnostic – themes.
︎Synthesizing common pain points in the purchasing process. Each colored bar indicates a product label; cards with multiple colors thus indicate a common theme across products.
Cross-functional stakeholder interviews
As I was diving into these buyer journeys, I also conducted a listening tour with 25 stakeholders across design, research, development, product management, marketing, and customer advocate roles. During this process, I came to see that that while people are motivated to improve the buyer experience, their priorities are focused on their own workflows or products rather than the overall customer experience.
These conversations helped to identify four major requirements for any type of customer-centric messaging or artifact: brevity, flexibility, actionability, and credibility. Furthermore, the interviews led to an epiphany: while Atlassian’s mission is to “unleash the potential of every team,” many stakeholders didn’t realize that even before product use, purchasing is also a team effort, with input from legal, security, and procurement departments.
︎Requirements identified from interviews with 25 cross-functional stakeholders.
︎A key takeaway from conducting interviews.
How might we communicate the buyer experience in a clear and compelling way?
Pain points as comics
In an attention-scarce environment, brevity seemed both a novel and necessary contrast to the documentation that already existed. One approach I pursued was to create comics, inspired by webcomicname’s three-panel comics with a blob as the protagonist, always ending with “Oh no.” These succinct visual reminders of customer pain would ideally be propagated in relevant everyday communication channels such as intranet pages and chat rooms.
Furthermore, using humor to communicate pain points would help to diffuse any blame or shame that people might feel when pain points are constantly emphasized, promoting empathy over negativity. While not specifically actionable (nor well illustrated, for that matter), the format offers a simple, visual reminder of customer perspectives that can be easily shared – and ultimately aims to shift culture at a grassroots level.
︎Illustrating how purchasers often face internal resistance to adopt new software.
︎Customers often struggle to evaluate Atlassian’s complex products, within the short timeframe of the allotted trial period.
︎At the time of this project, Atlassian only accepted a limited number of currencies, creating friction for its international customers.
Sports field visualization kit
Another more flexible approach I took was inspired by board games and the World Cup, both of which I observed employees engaging with at the office. To highlight the complexity and players of the purchasing process, I wanted to leverage Atlassian’s existing emphasis on team work to reframe customer-centric messaging in a way that would be relatable to internal stakeholders.
I imagined movable components on a sports field background to show the buyer experience in a quick and engaging way, during meetings or onboarding sessions. A digital version could be stitched together into a GIF, then – like the comics – saved and shared in common communication channels.
︎An early outline of the journey on a sports field, with buyers crossing from left to right, across the purchasing phases: Identify Need, Research, Trial, and Convert.
︎An example GIF of a visualization kit prototype in use on a whiteboard.
Top tasks benchmarking survey
Revisiting the identified requirements, however, I realized that both of these approaches were brief or flexible, but not necessarily actionable or credible. Additionally, I had reservations about whether either format would live on after the duration of my summer internship, once I had returned to my graduate program.
At this point, I luckily became aware of the Research & Insights team’s new benchmarking initiative, using Jerry McGovern’s top tasks method. This approach would be credible, given that it is customer led, and it would dentify actionable priorities. It would also ensure longevity beyond my internship as the Research and Insights team intended to continue the work over the next few years.
I thus redirected my efforts to develop a top tasks survey, revisiting existing documentation to generate a master list of tasks; I also returned to internal stakeholders I had previously interviewed to gather feedback in adjusting language and granularity of tasks.
I was unfortunately not able to see the top tasks survey through to completion before my internship ended. However, my work in the final few weeks did generate a master list of tasks that the Research and Insights team would use to administer the top tasks survey; this would ultimately meet the goal of identifying a brief, actionable, and credible list of customer priorities. After my departure, I was also informed that my comics were discovered by the Head of the Design Studio, who found them resonant and had them printed on posters for every office.
︎Master list of top tasks generated, rendered on Mural.
︎Printed poster of comics in the San Francisco office.
This project was valuable in learning not to be afraid to question a prompt, as well as to exercise rigor in validating any existing data. Given that the project was largely self-directed, I also gained experience in managing expectations as well as seeking assistance when necessary. Furthermore, despite the desire to demonstrate the value of my work as an individual, I came to understand the importance of folding efforts into larger-scale initiatives to ensure longevity of impact.
Auditory Data Design
Informed by an interdisciplinary literature review, we developed an analytical framework of data sonification practices as well as a voice user interface (VUI) representing data from the U.S. Census and the American Community Survey. We then conducted usability test sessions with this VUI to evaluate the potential of data exploration via conversational interfaces, as well as to present recommendations for future work.
UX Writing (Voice Design)
According to National Public Media, one in six Americans own a smart speaker as of January 2018, and this population has more than doubled in the course of one year. With the rise of both virtual assistants and software-embedded devices, audio-first interactions are becoming more prevalent in daily life. However, there are not yet industry standards for sharing data via sound experiences – particularly through emergent smart speaker interfaces.
This project was motivated by open data initiatives, particularly those run by government agencies. In anticipation of the 2020 U.S. Census, we began evaluating existing web-based data exploration tools, conducting cognitive walkthroughs of the U.S. Census’s 2010 website, interactive maps, and “Profile America” audio stories. While this informed a general understanding of data exploration, we imagined a future in which one would be able to make sense of data exclusively by ear. With support from the Berkeley Center for New Media, we focused on the following research question: how might audio enable us to understand complex datasets?
︎Experience map of existing census data exploration tools.
How might audio enable us to understand complex datasets?
We first conducted an in-depth literature review of prior work related to “data sonification,” or using non-speech audio to convey information. Our analysis was initially based on “auditory graphs,” shaped by visual analogs (i.e., histograms, scatter plots, pie charts, etc). Spanning across various disciplines – including human-computer interaction, accessibility, music, and art – the range of papers demonstrated great variety: some demonstrated more creative intent with memorable, musical designs, while others focused on accurate, scientific representations. Papers also demonstrated differences in whether the researchers mapped sounds to real or simulated data.
Observed differences in the data sonification literature manifested in two primary areas:
- the rigor – or absence – of experimental procedures, and
- the quality of stimuli used – whether researchers used abstract MIDI sounds or audio mapped to semantic meaning (e.g. using the sound of rain to communicate precipitation).
For our framework, we thus plotted selected papers along two dimensions: Objective vs. Subjective Sonification Approach on the vertical axis, and Abstract vs. Functional Data on the horizontal axis. Key papers were grouped by the following representational characteristics: Trends, Clusters, Spatial Relationships, and Distributions.
︎Conceptual framework for existing auditory data representations.
While reviewing the literature, however, we discovered that prior research often lacked associated audio files, since much of it is decades old. We therefore reproduced three different sonification methods, using the softwares Ableton Live and Audacity:
- an audio choropleth map proposed by Zhao et al, using pitch to represent population data by state and region from the 2010 Census,
- an audio line graph inspired by Brown and Brewster, using timbre to represent employment data by age group from the 2015 American Community Survey (ACS), and
- an audio pie chart designed by Franklin and Roberts, using rhythm to represent employment data from the ACS again, but by education level.
Using the prototyping software Invocable (formerly Storyline), we used these auditory data designs to develop our own voice user interface, an Alexa skill called “Tally Ho.”
︎Summary of corresponding visual representations.
︎Overview of the conversational prototype.
With this VUI, we conducted in-person usability testing to evaluate the potential of auditory data exploration via a contemporary, conversational interface. We sought and recruited a five-person sample that would demonstrate varying degrees of familiarity with: smart speakers, musical knowledge, and census data. During each moderated session, the participant was presented each of the three different audio representations in a randomized order, then asked follow-up questions about initial impressions, perceived difficulty, and user expectations.
︎ Conducting usability testing on the Alexa-enabled Amazon Tap.
Our prototype used pitch, timbre, and rhythm to represent data points, category differences, and overall trends: users were able to hear these distinctions and interpret them mostly correctly after hearing a scripted explanation from the VUI. (More documentation, including the full report and audio files, can be found on the Berkeley School of Information project page. )
Our results suggest that users generally enjoyed the experience of hearing data – finding it “cool,” “fun,” and even “powerful” – but also had difficulty remembering insights as passive listeners. Even with repetition, participants lacked precise recall; 4 out of the 5 participants noted that it was “confusing” to remember what they had just heard.
Given that the average human primarily relies on capabilities of sight first and sound second, it is unsurprising that VUIs tend to require a greater cognitive load and more training than traditional visual interfaces. Among the participants who already owned smart speakers, the devices were mainly used to perform simple tasks like playing music and setting alarms. Understanding sonified data was an entirely new type of experience. One way to overcome this novelty factor would be to conduct a longitudinal study to assess changes in both performance and enjoyment of the experience over repeated interactions.
With the growth of conversational interfaces – in tandem with the rise of ubiquitous computing – we remain optimistic about the way forward in designing auditory data representations. As Hermann and Ritter suggest, humans are “capable [of] detect[ing] very subtle patterns in acoustic sounds, which is exemplified to an impressive degree in the field of music, or in medicine, where the stethoscope still provides very valuable guidance to the physician.” Continuing to develop new interaction patterns for auditory data exploration will benefit not only those who are visually impaired or limited in numeracy skills, but also anyone who is curious about making sense of data through alternative means.
Life in a Minute
“Life in a Minute” is a critical making installation that facilitates playful, tangible interactions to encourage reflection on personal time allocation. With support from the Berkeley School of Information and the Berkeley Center for New Media, the project was exhibited in the gallery of the 2018 Ethnographic Praxis In Industry Conference (EPIC).
In an era when both time and attention are increasingly scarce, there is a need for design that encourages active, thoughtful reflection. Concerned about the social implications of technology, our aim with “Life in a Minute” was to make time tangible, probing people to think critically about the value of time and to reveal underlying attitudes.
The metaphor of financial spending is an act that directly represents people’s priorities: we hypothesized that feelings of time scarcity are a particularly fertile ground for arriving at cultural knowledge, given that time is inherently finite. This project thus addresses a novel intersection of time as a theme, reflective design, and tangible user interfaces.
The participant’s lifetime is embodied in pennies, translating the average worldwide 71.5-year lifespan into 715 pennies. Over the course of 60 seconds, participants “spend” these pennies that represent their lifetime by allocating them into five different life target jars – Career, Community, Education, Family, and Play: these include servo-powered lids that open and close at random, adding a carnivalesque quality to the experience.
When time is up, players receive a receipt as evidence that embodies the “lifetime” spent and enables reflection after the experience has ended. The receipt is printed by an Arduino-powered thermal printer, with coin allocation weights mapped to years in a lifespan.
︎ Overview of the installation setup, including both the participant and an attendant.
︎An attendant deposits coins onto a conveyor belt for the participant.
︎The participant allocates coins picked up into jars of “life target” areas, whose servo-powered lids open and close at random.
︎When a minute has elapsed, each “life target” jar is weighed and recorded.
︎ Each participant receives a receipt of how they “spent” their time, with coin allocation weights mapped to years in a lifespan.
Rich feedback from 25 participants suggest the value of “Life in a Minute” as a cultural probe to uncover attitudes toward time scarcity and decision-making processes. For instance, some participants “spent” their coins as soon as they were dispensed; others waited at the end of the conveyor belt to collect as many coins as possible, delaying their time allocation. Another key contrast was in relation to the life target jars, opening and closing at random: some people waited for certain jars to open, while others allocated opportunistically to any jars available. Asking the participant about such behaviors afterward effectively sparked self-reflection and candid dialogue about their motivations, conscious or unconscious.
The game-like design of the interaction also layers in a sense of fun, occasionally inducing a competitive mindset. One participant remarked, “People with bigger hands have an advantage.” Another participant was observed peeking at another’s receipt, asking, “How did you do?” But the most unexpected method of engagement with the system involved pairs playing together, adding in a new layer of observable interaction. Among one duo, we overheard the following question, “What’s our strategy?” To which her partner replied, “To waste as little time as possible.” When asked why she chose not to play solo, another participant remarked, “Life is easier with a partner.”
︎ Exhibiting in the gallery of 2018 Ethnographic Praxis In Industry Conference.
︎ One visitor’s response to the project at EPIC.
Given the quality of such data, we propose that this work may be repurposed as a projective interviewing method. A key improvement would be to allow participants to self-identify categories of personally valuable “life targets,” as self-designated categories would downplay any prescriptive intention and permit greater interpretability for the participant. Additionally, the receipt or idealized “life summary” artifact could be compared to actual behavioral evidence, similar to Anderson et al.’s “ethno-mining” approach: for example, participants could be asked to map their real online behavioral data with broader “life targets.”
We offer “Life in a Minute” as a novel approach to understanding subjective perspectives about the value of time. By introducing an interactive installation, we open a space for thinking more broadly about new methods of collecting data which may more accurately mirror real sentiment and lived experiences.