Differentially private scatterplots enable the plotting of two attributes while guaranteeing a specified level of privacy. What a user sees from the scatterplot can be affected by which privacy algorithm is used and how it adds noise to the data. However, there is no existing work that quantifies this effect. We present the results of a pilot data study that compares the visual utility of algorithms that create differentially private scatterplots. We compare five popular algorithms across a range of parameters. The results indicate that DAWA and Geometric Truncated are the best algorithms for visual utility. Future research could focus on optimizing the different parameters to maximize utility of the visual representations.
Data visualization is pervasive in the lives of children as they en-counter graphs and charts in early education and online media.In spite of this prevalence, our guidelines and understanding ofhow children perceive graphs stem primarily from studies con-ducted with adults. Previous psychology and education researchindicates that children’s cognitive abilities are different from adults.Therefore, we conducted a classic graphical perception study on apopulation of children aged 8–12 enrolled in the Ivy After SchoolProgram in Boston, MA and adult computer science students en-rolled in Northeastern University to determine how accuratelyparticipants judge differences in particular graphical encodings. We record the accuracy of participants’ answers for five encodingsmost commonly used with quantitative data. The results of ourcontrolled experiment show that children have remarkably similargraphical perception to adults, but are consistently less accurateat interpreting the visual encodings. We found similar effective-ness rankings, relative differences in error between the differentencodings, and patterns of bias across encoding types. Based on ourfindings, we provide design guidelines and recommendations forcreating visualizations for children. This paper and all supplementalmaterials are available at https://osf.io/ygrdv.
Data visualization is a young and growing field with many unresolved, debated, and untested theories that need to be tested, validated, and strengthened. Replication studies are a common way to achieve this validation by investigating the credibility, rigor, and generalizability of previously published research. The execution of replication studies is crucial to strengthen the field of data visualization and to ensure that the field’s foundations are solid with accurate methods and theories. Although visualization researchers acknowledge the epistemological significance of replications and its necessity to establish trust and reliability, the field has made little progress to support the publication of such studies and importantly provide methods to the community to encourage replications. In this paper, we contribute Vis Repligogy, a novel framework to incorporate replication studies within visualization course curricula that not only teaches students replication and evaluation methodologies but also results in executed replication studies to validate prior work. To validate the feasibility of the framework, we present the results of four replication studies that were conducted in a graduate data visualization course that implemented the Vis Repligogy framework. We also provide a meta-analysis of the four replication studies that replicated and validated, a seminal work in data visualization investigating the effects of visual embellishments on chart interpretability and memorability. Finally, we reflect on our experience of conducting the replication studies through the Vis Repligogy framework and reflect on the experience to provide useful recommendations for future use of the framework. We envision that this framework will encourage instructors to conduct replications in their own courses and will help facilitate more replication studies in the visualization community to support a culture shift where replications are commonplace.
Over the last decade, remote experiments have become a widely used and integral method for many human-computer interaction domains. Nonetheless, extended reality (XR) researchers have been slow to adopt remote research methods. This can largely be attributed to standard remote experimentation techniques being ill-suited for the unique XR domain constraints. Existing research, albeit limited, has aimed to overcome these constraints and demonstrate the viability of traditional remote research methods for XR studies, yet most XR experiments have remained in-lab. This gap in XR methodology has never been more evident or detrimental than during the ongoing global COVID-19 pandemic. During the pandemic, safe and ethical co-present in-lab experimentation has become increasingly difficult, if not impossible. Many researchers struggled to transition to remote research methods resulting in delayed, canceled, or unsatisfactory experiments. Beyond this current crisis, remote research methods present several advantages, such as obtaining a larger sample and accessing specific user populations that have not been leveraged in XR research — leading to missed opportunities and potentially less rigorous results.Our previous research demonstrated the efficacy of using existing social virtual reality (VR) platforms to implement and conduct remote VR experiments. Social VR platforms provide an experienced and VR-equipped user base to recruit from and customizable distributed synchronous virtual environments to implement experiments, which makes them a natural fit for VR experiments. They allow researchers to be co-present in the same virtual environment as participants to proctor experiments, similar to how they would during a co-present in-lab study. However, existing social VR platforms were not built with this use-case in mind, resulting in several limitations, such as the inability to easily save data externally. These limitations prevent existing social VR platforms from being a viable long-term XR research method. Our previous work identified two potential paths towards establishing long-term social VR remote research methods. The first potential path is to partner with existing social VR platforms to create official channels for remote studies. The second potential path is to build a bespoke social VR platform specifically for conducting XR remote experiments. We believe both of these paths have their respective strengths and weakness and are viable long-term solutions for remote XR studies. In this position paper, we aim to contribute a detailed discussion of both of these paths, their benefits, limitations, and potential architecture. In so doing, we hope to provide the XR community our insights into how social VR research methods can be expanded and inspiration for the potential future of remote XR research.
Increasingly, visualization practitioners are working with, using, and studying private and sensitive data. There can be many stakeholders interested in the resulting analyses—but widespread sharing of the data can cause harm to individuals, companies, and organizations. Practitioners are increasingly turning to differential privacy to enable public sharing of data with a guaranteed amount of privacy. Differential privacy algorithms do this by aggregating data statistics with noise, and this now-private data can be released visually with differentially private scatterplots. While the private visual output is affected by the algorithm choice, privacy level, bin number, data distribution, and user task, there is little guidance on how to choose and balance the effect of these parameters. To address this gap, we had experts examine 1,200 differentially private scatterplots created with a variety of parameter choices and tested their ability to see aggregate patterns in the private output (i.e. the visual utility of the chart). We synthesized these results to provide easy-to-use guidance for visualization practitioners releasing private data through scatterplots. Our findings also provide a ground truth for visual utility, which we use to benchmark automated utility metrics from a variety of fields. We demonstrate how multi-scale structural similarity (MS-SSIM), the metric most strongly correlated with our study’s utility results, can be used to optimize parameter selection. A free copy of this paper along with all supplemental materials is available at https://osf.io/wej4s/.
Increasingly, visualization practitioners are working with, using, and studying private and sensitive data. There can be many stakeholders interested in the resulting analyses-but widespread sharing of the data can cause harm to individuals, companies, and organizations. Practitioners are increasingly turning to differential privacy to enable public data sharing with a guaranteed amount of privacy. Differential privacy algorithms do this by aggregating data statistics with noise, and this now-private data can be released visually with differentially private scatterplots. While the private visual output is affected by the algorithm choice, privacy level, bin number, data distribution, and user task, there is little guidance on how to choose and balance the effect of these parameters. To address this gap, we had experts examine 1,200 differentially private scatterplots created with a variety of parameter choices and tested their ability to see aggregate patterns in the private output (i.e. the visual utility of the chart). We synthesized these results to provide easy-to-use guidance for visualization practitioners releasing private data through scatterplots. Our findings also provide a ground truth for visual utility, which we use to benchmark automated utility metrics from various fields. We demonstrate how multi-scale structural similarity (MS-SSIM), the metric most strongly correlated with our study's utility results, can be used to optimize parameter selection. A free copy of this paper along with all supplemental materials is available at https://osf.io/wej4s/.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.