Shuichiro Ogawa
日本語

Notes · updated 2026-09-22

How Data Visualization and Data Art Have Been Treated in Creativity Education

Data education runs in two streams. One teaches students to read figures correctly and measures that ability with instruments (VLAT, CALVI, Mini-VLAT, a multidimensional scale), and it has accumulated critiques of those very instruments.

Contents (10)
  1. Two educational traditions that start from different questions
  2. Visualization education has instruments for measuring reading
  3. Data art education speaks in case studies and has no scale
  4. The making-to-learn lineage bridges the two
  5. Bringing art in creates tension with accuracy
  6. Effects are only confirmed where they are measured
  7. In the Japanese-language literature, the topic is not an educational subject
  8. Side by side, the difference lies on the education side
  9. No review crosses the two yet
  10. What is still unresolved

Data education runs in two streams. One teaches students to read figures correctly. The other brings the making of data works into the classroom.

Both work from the same material, yet the two grew up in separate research traditions. One built psychological instruments. The other piled up course cases. Placed side by side under creativity education, where do they overlap and where do they part?

This note reads across 77 sources: 20 on the theoretical frames of creativity and arts education, 20 on visualization and data literacy education, 19 on data art, physicalization, and creative coding in teaching, and 18 on Japanese-language educational research.

Two educational traditions that start from different questions

The theoretical footing of visualization education sits in the psychology of graph comprehension. Shah and Hoeffner organized how comprehension depends on the task, the display format, and the reader’s prior knowledge. Splitting reading outcomes into what the reader brings and what the design does set the premise for the education research that followed.

Data art education has no such psychological premise. What it has instead is Papert’s constructionism. Children do not receive knowledge as instruction; they build knowledge while making artifacts they can share with others. Making is not a means of learning but the form learning takes.

The difference shows in what each side calls correct. In visualization education a misreading is an error. In data art education the spread of readings is the content of the work, not a failure.

Visualization education has instruments for measuring reading

Measurement of visualization literacy reached its reference point with VLAT. It built an objective test scored by correct answers and reported item difficulty and discrimination, giving later scale work a basis for comparison. The finer details of its item structure do not agree across secondary sources. [primary check pending]

CALVI changed what the instrument claims to measure. It defines visualization literacy as the ability to see through misleading visualizations and asks test takers to solve figures designed to deceive. It reports that accuracy on misleading items falls far below accuracy on normal items, and argues that existing scales miss an ability that exists. The reported numbers are [primary check pending].

Shorter and adaptive forms exist for practical use. Mini-VLAT reduces VLAT to twelve items so that short surveys can carry it. Cui and colleagues converted the scales into an IRT-based adaptive test to reach comparable measurement with fewer items. The reliability figures for both are [primary check pending].

There are also attempts to widen what is measured. The multidimensional scale by Saske and colleagues adds dimensions that performance tests do not reach, such as aesthetic evaluation, critique, and contextualization. In a survey using climate data figures, simple reading still produced many errors, aesthetic evaluation tracked with trust, and confidence did not accompany the ability to articulate critical improvements. The dimensions and the numbers are [primary check pending].

Critique of measurement itself has become a standard part of this field. Ge and colleagues frame visualization literacy measurement as a wicked problem and, from interviews with test users and designers, sort threats to validity into conceptual, operational, and methodological layers. They recommend widening the scope, improving ecological validity, and working across disciplines.

The education side also has agenda-setting work. Bach and colleagues sorted practitioners’ knowledge into challenges and themes and gave assessment its own theme. Assessment tools for student work have been developed as well. The Graph Rubric by Angra and Gardner scores student-made graphs on mechanics, choice, and communication, and gathers content, face, and construct validity evidence from consultation with students, instructors, and science education researchers, plus external trials.

Börner and colleagues define data visualization literacy as reading and making together, and give a taxonomy and a procedure usable for curriculum and assessment design. They state that a large-scale teaching practice revised and validated the framework, which puts the work at the joint between teaching and measurement.

Data art education speaks in case studies and has no scale

Research on teaching data art centers on accounts of course design and delivery. Roberts ran a course for three years in which students made data art from self-chosen datasets and exhibited it publicly. The brief was intentionally undefined, and the exhibition alone supplied the constraint, which the author argues kept creativity and gradeability compatible. Assessment weights four criteria evenly: design and presentation, clarity of data communication, implementation and results, and final report and reflection. [primary check pending]

Few examples go as far into designing assessment. Jiménez Garcia proposes a teaching model that uses data physicalization as both medium and assessment instrument. The studio sequence builds in an instructor-made “meta-physicalization” that turns audience engagement into indicators of depth of reflection, emotional resonance, and perceived impact, and tracks how audience responses change across three public exhibitions.

The material of making varies widely. Wang and colleagues taught data comics in a workshop and identified three difficulties: introducing visualization alongside journalistic narrative, structuring the story, and iterating on drafts. Çay and colleagues compared three master’s-level physicalization courses and showed that data serves several roles at once: a knowledge base, a source of ideas, a ground for material choices, and a driver of narrative. van Koningsbruggen and colleagues designed “Data Diaries” to teach physicalization of personal data and reported that physicalization shifts attention away from accuracy and efficiency toward the story of the data. Nelson and colleagues framed embroidery as an embodied and reflective data practice and ran workshops in which personal data was stitched onto cloth pieces and assembled into a collective Data Quilt. Strantz reported that playing with low-cost materials contributed to creative exploration even when it never became a finished visualization, and Li and colleagues compared textile workshops across different participant groups.

What these studies share is that outcomes are described in the vocabulary of each case. A standard scale for data art education comparable to visualization literacy did not appear in this search.

The making-to-learn lineage bridges the two

What separates reading education from making education is the choice of method, not the object. Having children make figures appears on the visualization side as well. Bae and colleagues built a kit of paper, cardboard, and mirrors with which children assemble bar, line, and pie charts, and observed the process. Bishop and colleagues developed a tablet tool that allows free-form mapping by children’s own ideas, co-designed with teachers and children. Both use the word constructionist explicitly.

The theoretical starting point is constructive visualization. Huron and colleagues formalized assembling data by hand through tokens, a token grammar, the act of assembly, and change over time. Assembling is analysis and meaning-making at once. That formulation is the bridge between visualization education and data art education.

Teaching data physicalization lines up on that bridge. Zhu and colleagues analyzed how children physicalize waste data and extracted strategies such as mapping quantity to size and proportional unitizing. They report that bodily making supports spatial thinking while craft skill constrains ideas and feedback opportunities are scarce. Bae and colleagues’ study was found independently from both the visualization side and the physicalization side. Work on children’s visualization education and work on learning by making meet in the same paper.

Bringing art in creates tension with accuracy

Practice that integrates artistic expression into data education reports the same tension repeatedly. Matuk and colleagues co-designed units with middle-school art and mathematics teachers and showed that bodily expression, narrative, and visual imagery support reasoning that uses data as evidence. They also report that tension arises between artistic creativity and the accuracy of data. Bertling and colleagues studied professional development for science, mathematics, and art teachers and confirmed gains in student confidence and engagement while naming three challenges: balancing artistic creativity against data accuracy, constraints of time and pacing, and the need for staged, long-term professional development. Galbraith and colleagues describe a curriculum in which students take the roles of story finder and storyteller, and sort out the competences involved in finding and communicating data stories.

There is also a view that critiques this integration from outside education. Woods and colleagues analyzed arts-integrated data science practice qualitatively and then posed an ontological critique of aestheticization in education. Bringing aesthetic representation of data into teaching enriches learning while carrying a risk of obscuring the political nature and incompleteness of data.

Informal settings handle the tension differently. Peppler and colleagues studied a real-time data exhibit in a science museum and derived four ways visitors read it: finding their own record, noticing patterns not yet represented, doubting measurement error, and correcting a reading of the distribution through bodily experience. Learning here is not receiving a correct reading but noticing that data is incomplete.

Effects are only confirmed where they are measured

Effect measurement is a shared weak point of creativity education. van de Kamp and colleagues ran a quasi-experiment with secondary visual arts students to test whether explicit instruction in metacognition raises divergent thinking. Fluency improved; originality did not. The expectation that instruction design alone moves originality is not supported by that result.

Reviews of STEAM education place similar reservations. Aguilera and Ortiz-Revilla collected STEAM intervention studies systematically and concluded that while positive effects are reported, there is no consensus that STEAM develops creativity. They also note that creativity is disproportionately measured with Likert-type questionnaires.

Negative results are reported as well. Huynh and colleagues tested in a between-subjects experiment whether narrative games raise children’s visualization literacy, and found that engagement and enjoyment improved significantly while graph comprehension did not. Enjoyment and learning are measured as separate things.

Reviews of arts integration state both positive conclusions and limits. Phelan and colleagues scoping-reviewed research connecting the arts to data literacy, concluded there is practical educational value, and noted that most published studies remain short-term projects. Arts education itself has causal evidence of a different kind. Bowen and Kisida ran a randomized controlled trial in 42 elementary and middle schools and found that assignment to arts opportunities was associated with fewer disciplinary infractions, higher writing scores, and greater empathy. The outcomes measured there are not creativity, and that matters.

In the Japanese-language literature, the topic is not an educational subject

The clearest result of the Japanese-language search is an absence. Research using data art in education was not found in CiNii Research or J-STAGE. CiNii returns two records for “データアート 教育”, both in clinical pharmacology education and home-economics cooking, unrelated to the topic. J-STAGE’s literature search returned no results for “データアート” and none for “クリエイティブコーディング” either. The only confirmed Japanese-language academic record with data art as its subject is a system for turning vital-sign data into art, and it is not about education.

Adjacent fields fill the gap. In information design education, Ozawa had students photograph and present everyday objects that are hard to use, and suggested the practice encourages critical thinking about familiar tools, while Sugimoto named the algorithmic generation of visuals from data “data-driven design” and built a teaching program around it. On making and critique, Kinoshita and colleagues compared classes with and without a review session and reported educational effects on satisfaction and active participation. On statistical graphs, Aoyama argued that graphs should be repositioned as a means of expression rather than only of reading and drawing, and Sasaki examined how bubble charts are interpreted. Sahara, studying US media art education, notes that securing budgets through vocational education frames has pushed media art toward a vocational character.

The second absence is measurement. Japanese-language work on measuring or assessing visualization literacy, comparable to the English-language scale research, was not found. What appears instead is classroom practice and curriculum development.

Side by side, the difference lies on the education side

The two share the material, the act of making, the makers, and the venues. Yet inside education they operate as different things. What makes the difference is not the nature of the object but the design of goals and assessment.

The goal of visualization education is to read correctly. Attainment is measured by scales and a misreading counts as a defect. The goal of data art education is to express something about data and to reflect on it. Attainment is discussed through exhibitions, critiques, and rubrics, and a spread of readings becomes the content of the work.

This asymmetry shows in how research accumulates. Visualization education holds comparable measurements and can argue about intervention effects. Data art education holds depth of cases and almost no measurements. Current research has no instrument to compare the two on whether creativity grew.

An instrument that spans both already exists. It is the critique. Research on design education analyzes the studio critique as communication between teacher and student and shows wide variation between teachers. Weinstein replaced the spoken critique in an information design class with an anonymous questionnaire and had students visualize the response data itself. The site of criticism becomes the material of visualization.

No review crosses the two yet

Reviews exist around this topic. An agenda for visualization education, a scoping review of K-12 data literacy education, a scoping review connecting the arts to data literacy, and a review of constructionist approaches to critical data literacy. Each leans to one side.

A review that crosses both data visualization and data art under the frame of creativity education did not appear in this search. The asymmetry of measurement, the tension between artistic creativity and data accuracy, and the absence in the Japanese-language literature are the open questions such a review would take up.

What is still unresolved

  • Many item counts, reliability coefficients, and sample sizes for the visualization literacy scales are known only through search summaries (VLAT, CALVI, Mini-VLAT, the adaptive test, the multidimensional scale, the children’s tools, the role-playing game experiment, the inoculation experiment). [primary check pending]
  • Most numbers in the data art education studies (assessment weights, workshop participant counts, observation counts, sample sizes) come from abstracts. [primary check pending]
  • The peer-review status of several Japanese-language venues is unconfirmed (デザイン学研究, 日本情報科教育学会誌, 学校教育実践学研究, 日本創造学会論文誌, 芸術工学会誌, 数学教育学会誌). [primary check pending]
  • Some sources were confirmed bibliographically but not obtained in full text (five in the Japanese set, one in the data art set). Their descriptions rest on abstracts and indexes. [primary check pending]
  • The Crossref record for Studio Thinking 3 gives 2010 as the year, which does not match 2022 at ERIC. [primary check pending]
  • Aoyama’s statistical graph paper has a mismatch between the volume year and the J-STAGE publication date. [primary check pending]
  • The visualization education paper by Bach et al. is dated 2023 on the basis of an early-access record. Its final issue may be 30(1), 2024. [primary check pending]
  • The Korean convergence education paper has no registered DOI; it was confirmed only through the KCI page. [primary check pending]
  • Weinstein’s paper has no DOI registered with Crossref. [primary check pending]
  • Lodi and Martini’s paper carries a correction record in Crossref. The correction notice (10.1007/s11191-021-00268-1, Science & Education 31(2), 561-562, 2022) could not be retrieved, so whether it affects the claims is unconfirmed. [primary check pending]
  • ACM DL, Springer, ScienceDirect, and JSTOR refused direct access, so methodology sections (sampling, analysis procedures) were not verified first-hand. [primary check pending]

References

These are the full bibliographic entries of the corpus this note relies on (source/review/data-viz-data-art-creativity-education/papers.md, 77 items).

Author: Shuichiro Ogawa (Design Researcher / Consultant) About me →