Minh hoạ vịt

I’m watching Calling Bullshit 6.4: Dataviz Ducks - YouTube, and my impression is that the presenter calls bullshit to all dataviz duck. I don’t think that’s a fair call. Yes, I’m aware that the decorations are distraction or even skewing the read, but that’s because we prioritize reading accuracy over catching attention. People who want to understand the research correctly should read the original paper. In the context where dataviz ducks are used, perhaps catching attention is more important. I don’t say that we are allowed to do that at all cost, but I think sacrificing a little reading accuracy doesn’t leave much harm. The most audience of them doesn’t want to validate the data, and most details will be forgotten quickly.

I do think that there are harmful sacrifices, though:

• Truncating Y axis on bar charts
• Uneven range bins
• Viewing the chart from an angle
• Comparing circles by radii, not areas

So I think that as long as the visualization cite the original research properly, and avoid those harmful designs, using ducks for non-researchers may have more benefit. Am I missing something?

The trade-off of the bar chart in the form of circle chart or tree chart to represent hierarchy is already acceptable.