Amazing article! I'm showing this to my lab colleagues. We might even start a small journal club to hone our data visualization skills.
This isn’t related to your bug bounty, but I did want to point out a factual issue in the post. You wrote: “Unless you are living in 19th-century Japan, you are probably used to reading text horizontally.” This is not quite accurate. I can’t speak for all East or Southeast Asian countries, but readers in China, Japan, and Taiwan are still very familiar with vertical text.
Even today, vertical writing remains common in Japan and Taiwan, where many books are published in vertical layout. In China, modern books are almost always horizontal, but vertical text is still widely used for store signs, poetry, and calligraphy.
Because logosyllabic writing systems (e.g., kanji/Chinese characters) allow characters to stack upright without rotating them, vertical writing is very natural to read. This also avoids many of the orientation issues that arise when rotating horizontally written text in data visualization contexts.
Thank you! Ah, I was thinking of how each language was predominantly written, but you're right and I've also seen vertical calligraphy and some books written in vertical text in Japan. It seems like vertical text is also dominant in Mongolia, but as far as I can tell it's much less common in Taiwan - is that right? I'll rewrite that sentence. For what it's worth, I do consider that a factual error and would be happy to pay out a small reward.
Thank you! I appreciate your willingness to revise and discuss this. I’m from Taiwan, and vertical writing is still very common in books and other printed materials here.
Actually, recent typographic studies show that vertical reading speeds for logographic scripts can match horizontal ones, a unique cognitive flexibility not found in alphabetic systems. In my opinion, Western design "rules" often ignore these cultural strengths, limiting global data accessibility.
I think you may have misunderstood what I meant, which was that people should use horizontal text on charts when they're communicating in languages that use horizontal scripts.
Interesting reading with good examples consistent with what we published in the Royal Statistical Society's Best Practices for Data Visualisation that has R code for most visualizations.
Exceptional depth on the confidence interval vs prediction interval distinction. Most practitioners still conflate them and end up mispresenting uncertainty ranges, exactly as that suicide risk example shows. The tick tock about reproduciblity through GitHub is spot on but still rarely adopted cuz it adds friction to publishing workflows.
Producing concise and high quality data visualisations is truly one of the most under appreciated, discussed and practiced skills in academia AND industry. Awesome stuff!
A great summary — captures the notion that good charts don’t come from following a set of simple instructions, but from questioning whether one’s work is honest, understandable, helpful, etc.
Alberto Cairo says “Visualization design isn’t about applying rules, but about reasoning about possible choices.”
Great article thanks. And also for tips on what software you use. Being in my late 50s and never having coded I am still stuck on Excel for around 25 charts and tables I produce each month. I had already requested some R for absolute beginners books for Christmas!
According to some studies in neuro-aesthetics, the human brain processes complex symmetrical data visualizations significantly faster than asymmetrical ones. In my opinion, while clarity is paramount, we often undervalue the emotional impact that a "beautiful" chart has on long-term information retention.
Superb article. I have not seen anything this good since my three books by Tufte were stollen by another consultant.
Amazing article! I'm showing this to my lab colleagues. We might even start a small journal club to hone our data visualization skills.
This isn’t related to your bug bounty, but I did want to point out a factual issue in the post. You wrote: “Unless you are living in 19th-century Japan, you are probably used to reading text horizontally.” This is not quite accurate. I can’t speak for all East or Southeast Asian countries, but readers in China, Japan, and Taiwan are still very familiar with vertical text.
Even today, vertical writing remains common in Japan and Taiwan, where many books are published in vertical layout. In China, modern books are almost always horizontal, but vertical text is still widely used for store signs, poetry, and calligraphy.
Because logosyllabic writing systems (e.g., kanji/Chinese characters) allow characters to stack upright without rotating them, vertical writing is very natural to read. This also avoids many of the orientation issues that arise when rotating horizontally written text in data visualization contexts.
Thank you! Ah, I was thinking of how each language was predominantly written, but you're right and I've also seen vertical calligraphy and some books written in vertical text in Japan. It seems like vertical text is also dominant in Mongolia, but as far as I can tell it's much less common in Taiwan - is that right? I'll rewrite that sentence. For what it's worth, I do consider that a factual error and would be happy to pay out a small reward.
Thank you! I appreciate your willingness to revise and discuss this. I’m from Taiwan, and vertical writing is still very common in books and other printed materials here.
Actually, recent typographic studies show that vertical reading speeds for logographic scripts can match horizontal ones, a unique cognitive flexibility not found in alphabetic systems. In my opinion, Western design "rules" often ignore these cultural strengths, limiting global data accessibility.
I think you may have misunderstood what I meant, which was that people should use horizontal text on charts when they're communicating in languages that use horizontal scripts.
Interesting reading with good examples consistent with what we published in the Royal Statistical Society's Best Practices for Data Visualisation that has R code for most visualizations.
https://rss.org.uk/news-publication/news-publications/2023/general-news/rss-publishes-new-data-visualisation-guide/
Exceptional depth on the confidence interval vs prediction interval distinction. Most practitioners still conflate them and end up mispresenting uncertainty ranges, exactly as that suicide risk example shows. The tick tock about reproduciblity through GitHub is spot on but still rarely adopted cuz it adds friction to publishing workflows.
Super cool post! Was hoping to see the 'solo' heatmap also make the cut: https://www.scientificdiscovery.dev/p/20-so-many-great-things-you-missed
What an inspiring post!
Producing concise and high quality data visualisations is truly one of the most under appreciated, discussed and practiced skills in academia AND industry. Awesome stuff!
Another book you could point people to: https://clauswilke.com/dataviz/
This is really nice, I hadn't seen it before! Thank you for writing it
A great summary — captures the notion that good charts don’t come from following a set of simple instructions, but from questioning whether one’s work is honest, understandable, helpful, etc.
Alberto Cairo says “Visualization design isn’t about applying rules, but about reasoning about possible choices.”
Great article thanks. And also for tips on what software you use. Being in my late 50s and never having coded I am still stuck on Excel for around 25 charts and tables I produce each month. I had already requested some R for absolute beginners books for Christmas!
Thank you for all you did at Our World in Data and good luck with your new ventures.
Thank you so much :)
So, so good. Thank you!
Thanks Saloni, this makes my heart sing! I remember asking if you could share a word of wisdom on making visuals, and you more than delivered :)
Thanks for reminding me about it! :)
According to some studies in neuro-aesthetics, the human brain processes complex symmetrical data visualizations significantly faster than asymmetrical ones. In my opinion, while clarity is paramount, we often undervalue the emotional impact that a "beautiful" chart has on long-term information retention.
POCA plots are where it's at!