Data-Induced Groupings and How To Find Them

TL;DR

The paper tackles how data values interact with visualization design to create Data-Induced Groupings, potentially biasing interpretation in dot plots. It conducts two dot-plot user studies to quantify grouping tendencies under nominal x-axes and real-data framing, then builds transparent predictive models using interpretable features (co-linearity, convex hull overlap, centroid metrics, and axis separations) that achieve hold-out $F1$ scores around $0.97$–$0.99$. The work demonstrates two practical applications: a diagnostic tool to flag likely data-induced groups and a redesign tool that searches axis permutations to emphasize desired groups while reducing violations. The findings highlight the predominance of co-linearity and clustering cues in grouping behavior and offer design-time methods to mitigate misleading patterns, with implications for improved visualization reliability and user understanding.

Abstract

Making sense of a visualization requires the reader to consider both the visualization design and the underlying data values. Existing work in the visualization community has largely considered affordances driven by visualization design elements, such as color or chart type, but how visual design interacts with data values to impact interpretation and reasoning has remained under-explored. Dot plots and bar graphs are commonly used to help users identify groups of points that form trends and clusters, but are liable to manifest groupings that are artifacts of spatial arrangement rather than inherent patterns in the data itself. These ``Data-induced Groups'' can drive suboptimal data comparisons and potentially lead the user to incorrect conclusions. We conduct two user studies using dot plots as a case study to understand the prevalence of data-induced groupings. We find that users rely on data-induced groupings in both conditions despite the fact that trend-based groupings are irrelevant in nominal data. Based on the study results, we build a model to predict whether users are likely to perceive a given set of dot plot points as a group. We discuss two use cases illustrating how the model can assist visualization designers by both diagnosing potential user-perceived groupings in dot plots and offering redesigns that better accentuate desired groupings through data rearrangement.

Figures (14)

• Figure 1: Nominal data showing test scores of seven students. One could see the three highest performing students as a group and the rest as another (a), or see a 'growing trend' in the left five data points with the two points on the right as outliers (b). However, order along the nomial x axis is not meaningful. Rearranging the x-axis may encourage the viewer to see a 'growing trend' across all seven points (c).
• Figure 2: Task interface. Participants can highlight or lasso groups of points, clear their selections, and add to or remove groups from the gallery on the right. After every 5 charts, we also add a text box so participants can describe what they see in the chart.
• Figure 3: (a) Distribution of group sizes, (b) distribution of the number of groups per chart, and (c) distribution of the number of groups per participant.
• Figure 4: Examples of clustering features except for cluster ratio, which is the ratio between centroid distance and cluster diameter. The red colored dots denote the candidate group $g$.
• Figure 5: Distributions of centroid distance, centroid diameter, error, convex hull overlap, x and y-axis separations. The Constellation study is blue and the Contextual study is red.

Theorems & Definitions (3)

• Example 1
• Example 2
• Example 3