DVSS - Day 2

Day 2 on the fly notes of MaLGa's Data Visualization Summer School 2025.

• creator of RAWGraphs for sankey diagram construction from a spreadsheet
• hans rosling (gapminder data live explanation)
• non-autonomy principle: vis as a trigger for cognitive processes (generate insights, knowledge is not there)

Data Visualization - a designerly approach

• design as a mean to create connections between disciplines (e.g. design thinking, research through design, ...)
• metaphores: gold, oil, natural resource, raw data
• extract wisdom from data -> build wisdom with data
• 70% of human "sensors" connected to eyes, 50% of brain processes related to visual processing
• pre-attentive processes to reduce cognitive workload

Origins and rationales

• cartesian plane (descartes)
• time series (prestley)
• bar chart, pie chart (playfair)
• data over geographic maps, flowchart (minard)
• willard brinton, "graphic methods for presenting facts"

https://datavizblog.com/2013/05/26/dataviz-history-charles-minards-flow-map-of-napoleons-russian-campaign-of-1812-part-5/

• hack the visual pattern

  1. explore visuals other than the available standard ones
  2. integrate variables and different visuals with each other

• some people started to break the ivory tower and have an impact on decision-makers through data visualization:

  • Florence Nightingale, mid 1800
  • London cholera map of John Snow, mid 1800 (ref. Snow, the movie)
  • treemap (Hewes and Gannett, mid 1800), but it's been included in MS Excel only in 2016

The information design continuum

graph LR
    A(<b>analysis</b><br><i>data visualization</i><br><b>science</b>, <i>expert</i>)
    B(<b>narration</b><br><i>infographics</i><br><b>media</b>, <i>public</i>)
    F(.)
    A --> F
    F --> B
    C(context<br><i>understanding</i>)
    D(message<br><i>communication</i>)
    E(engagement<br><i>motivation</i>)
    F --> C
    C --> D
    D --> E

• business need: fast analytics, Tableau, BI, dashboard (board of primitive cars to protect the driver from dashes)
• Edward Tufte: data-ink ratio
• but: if you have a purpose, violate the data-ink ration is more than possible!
• data art
• different users: data expert, domain expert, policy maker, activist, citizen (ordered by reluctancy wrt. the message being conveyed)
• visualization as translation, a matter of language

Design hybrid experiences

• martini glass narrative model
• data physicalization/sonification (evocative data art)
  • https://www.icad2019.icad.org/wp-content/uploads/2019/06/ICAD_2019_paper_44.pdf
  • https://sonifying.github.io/UNDERSTANDING-CYBER-ATTACKS-ON-WATER-SUPPLY-SYSTEMS/index.html
  • https://ginevraterenghi.github.io/presentazione-prog/index.html
  • http://listen.hatnote.com/#it

Rhetorical Data Visualization

• world -> measured data -> transformed data -> vis -> delivery -> perception data reality gaps - data transformation - visual representation - contextual factors
• vis = efficiency + reduction in uncertainty
• rhetorical -> to build persuasive arguments
• there is always a frame (questions asked, data used, selected, graph choice, ...)
  • different data can be framed in different ways to support different storytelling
  • but some data/vis are more truthful than others! (reliability spectrum)

Objectives

1. mindfulness about data framing
2. assessing validity
3. developing frames w/ integrity

4. common language to disagree constructively

5. develop a systematic approach to map data transformations to narrative frames (designer would benefit to predict what readers will extract, readers will assess the relaibility of a frame)

Variable selection and grouping

Aggregation statistics

• sum/counts -> beware of the "base rate bias"
• average(s)
  • hide distribution shape and variability
  • sensitivity to sample size
• percentages and rates
  • percentage point vs change

Filtering and range effects

• variable selection vs values filtering

Granularity

• binsize