How Learners Sketch Data Stories

TL;DR

This paper analyzes 101 data sketches created by learners, revealing common practices and preferences in visual encodings and storytelling methods to inform educational tools and programs.

Contribution

It introduces a classification of learner-created data sketches and provides insights into their common visual and structural features, bridging a gap in data storytelling education.

Findings

Preference for positional and shape-based encodings

Frequent use of symbolic and textual representations

High prevalence of stories comparing data subsets

Abstract

Learning data storytelling involves a complex web of skills. Professional and academic educational offerings typically focus on the computational literacies required, but professionals in the field employ many non-technical methods; sketching by hand on paper is a common practice. This paper introduces and classifies a corpus of 101 data sketches produced by participants as part of a guided learning activity in informal and formal settings. We manually code each sketch against 12 metrics related to visual encodings, representations, and story structure. We find evidence for preferential use of positional and shape-based encodings, frequent use of symbolic and textual representations, and a high prevalence of stories comparing subsets of data. These findings contribute to our understanding of how learners sketch with data. This case study can inform tool design for learners, and help create educational programs that introduce novices to sketching practices used by experts.