My notes on John Burn-Murdoch’s presentation on Reporting and visualisation

John Burn-Murdoch’s, and FT’s charts on twitter has always impressed me. I first became aware of him during the first pandemic lock down. To me also, back then it seems the whole world (or at least the UK public) was also going thru’ a data visualisation and data science awakening.

It was John’s and many other charts during those uncertain times that eventally convinced me to use Power BI instead of Excel in my reporting at work. Today, I’m still learning how powerful telling a good story, and the curiosity for this area has not yet subside.

Video below is from Posit PBC. Notes further below are my interpretation and summary for future references, all original credits to John, FY and the community. Conversely, any mistakes are mine.

On visualisation regonition and recall, there was an experiment that tracks the eye movement and recall accuracy of readers when presented a chart for a short duration.

https://pubmed.ncbi.nlm.nih.gov/26390488/

https://youtu.be/L5_4kuoiiKU

Ironically for data visualizers, Readers seems to only recall the “texts” a lot better than stunning charts.

Readers then do the following when presented….

  1. eyes explores the title FIRST

  2. (then if there’s still time…) it then head towards annotations

  3. (and if have more time…..then only read across the chart contents

John’s Lessons & key notes

John Burn-Murdoch | https://youtu.be/L5_4kuoiiKU?t=817

John Burn-Murdoch | https://youtu.be/L5_4kuoiiKU?t=817

John Burn-Murdoch | https://youtu.be/L5_4kuoiiKU?t=1438

John Burn-Murdoch | https://youtu.be/L5_4kuoiiKU?t=1630

Example: “less noisy” spline charts to better tell a story

https://youtu.be/L5_4kuoiiKU?t=2088

The true data

When plotted the true dat is noisy and jaggy

Rolling Average

This could be 1 solution, and less noisy, and deemed “accurate” method by most

Spline

The spline method was also a good choise, very smooth, clean.

they went with Spline (S) instead of “moving average” (R) , and did an analysis on the “accuracy” on data on last 7 days - and used spline because it appears “less noisy” and is more accurate for “Recent data”

JBM’s Summary

  • Text is critical

  • confused reader = BAD chart, anticipate and explain in situ

  • emotional and political context (aim to be neutral in these 2 senses)

  • stick around, don’t publish and vanish

  • ease of understanding comes first

  • use animation to build and release tension

Previous
Previous

Start of Story: Jaw hung open

Next
Next

Exploring Bing AI Chat for cost-saving and lazy reasons