Search for: All records

Visualization grammars, often based on the Grammar of Graphics (GoG), have much potential for augmenting data analysis in a programming environment. However, we do not know how analysts conceptualize grammar abstractions, or how a visualization grammar works with data analysis in practice. Therefore, we qualitatively analyzed how experienced analysts (N = 6) from TidyTuesday, a social data project, wrangled and visualized data using GoG-based ggplot2 without given tasks in R Markdown. Though participants’ analysis and customization needs could mismatch with GoG component design, their analysis processes aligned with the goal of GoG to expedite visualization iteration. We also found a feedback loop and tight coupling between visualization and data transformation code, explaining both participants’ productivity and their errors. From these results, we discuss how future visualization grammars can become more practical for analysts and how visualization grammar and analysis tools can better integrate within a programming (i.e., computational notebook) environment. 

We propose a new approach to uncertainty communication: we keep the uncertainty representation fixed, but adjust the distribution displayed to compensate for biases in people’s subjective probability in decision-making. To do so, we adopt a linear-in-probit model of subjective probability and derive two corrections to a Normal distribution based on the model’s intercept and slope: one correcting all right-tailed probabilities, and the other preserving the mode and one focal probability. We then conduct two experiments on U.S. demographically-representative samples. We show participants hypothetical U.S. Senate election forecasts as text or a histogram and elicit their subjective probabilities using a betting task. The first experiment estimates the linear-in-probit intercepts and slopes, and confirms the biases in participants’ subjective probabilities. The second, preregistered follow-up shows participants the bias-corrected forecast distributions. We find the corrections substantially improve participants’ decision quality by reducing the integrated absolute error of their subjective probabilities compared to the true probabilities. These corrections can be generalized to any univariate probability or confidence distribution, giving them broad applicability. Our preprint, code, data, and preregistration are available at https://doi.org/10.17605/osf.io/kcwxm 

We conducted a longitudinal study during the 2022 U.S. midterm elections, investigating the real-world impacts of uncertainty visualizations. Using our forecast model of the governor elections in 33 states, we created a website and deployed four uncertainty visualizations for the election forecasts: single quantile dotplot (1-Dotplot), dual quantile dotplots (2-Dotplot), dual histogram intervals (2-Interval), and Plinko quantile dotplot (Plinko), an animated design with a physical and probabilistic analogy. Our online experiment ran from Oct. 18, 2022, to Nov. 23, 2022, involving 1,327 participants from 15 states. We use Bayesian multilevel modeling and post-stratification to produce demographically-representative estimates of people's emotions, trust in forecasts, and political participation intention. We find that election forecast visualizations can heighten emotions, increase trust, and slightly affect people's intentions to participate in elections. 2-Interval shows the strongest effects across all measures; 1-Dotplot increases trust the most after elections. Both visualizations create emotional and trust gaps between different partisan identities, especially when a Republican candidate is predicted to win. Our qualitative analysis uncovers the complex political and social contexts of election forecast visualizations, showcasing that visualizations may provoke polarization. This intriguing interplay between visualization types, partisanship, and trust exemplifies the fundamental challenge of disentangling visualization from its context, underscoring a need for deeper investigation into the real-world impacts of visualizations. Our preprint and supplements are available at https://doi.org/osf.io/ajq8f . 

The grammar of graphics is ubiquitous, providing the foundation for a variety of popular visualization tools and toolkits. Yet support for uncertainty visualization in the grammar graphics—beyond simple variations of error bars, uncertainty bands, and density plots—remains rudimentary. Research in uncertainty visualization has developed a rich variety of improved uncertainty visualizations, most of which are difficult to create in existing grammar of graphics implementations. ggdist , an extension to the popular ggplot2 grammar of graphics toolkit, is an attempt to rectify this situation. ggdist unifies a variety of uncertainty visualization types through the lens of distributional visualization, allowing functions of distributions to be mapped to directly to visual channels (aesthetics), making it straightforward to express a variety of (sometimes weird!) uncertainty visualization types. This distributional lens also offers a way to unify Bayesian and frequentist uncertainty visualization by formalizing the latter with the help of confidence distributions. In this paper, I offer a description of this uncertainty visualization paradigm and lessons learned from its development and adoption: ggdist has existed in some form for about six years (originally as part of the tidybayes R package for post-processing Bayesian models), and it has evolved substantially over that time, with several rewrites and API re-organizations as it changed in response to user feedback and expanded to cover increasing varieties of uncertainty visualization types. Ultimately, given the huge expressive power of the grammar of graphics and the popularity of tools built on it, I hope a catalog of my experience with ggdist will provide a catalyst for further improvements to formalizations and implementations of uncertainty visualization in grammar of graphics ecosystems. A free copy of this paper is available at https://osf.io/2gsz6 . All supplemental materials are available at https://github.com/mjskay/ggdist-paper and are archived on Zenodo at doi:10.5281/zenodo.7770984 . 

Visualizations depicting probabilities and uncertainty are used everywhere from medical risk communication to machine learning, yet these probabilistic visualizations are difficult to specify, prone to error, and their designs are cumbersome to explore. We propose a Probabilistic Grammar of Graphics (PGoG), an extension to Wilkinson's original framework. Inspired by the success of probabilistic programming languages, PGoG makes probability expressions, such as P(A|B), a first-class citizen in the language. PGoG abstractions also reflect the distinction between probability and frequency framing, a concept from the uncertainty communication literature. It is expressive, encompassing product plots, density plots, icon arrays, and dotplots, among other visualizations. Its coherent syntax ensures correctness (that the proportions of visual elements and their spatial placement reflect the underlying probability distribution) and reduces edit distance between probabilistic visualization specifications, potentially supporting more design exploration. We provide a proof-of-concept implementation of PGoG in R.