W7 — How Charts Lie: Ch.2-3

yvonne tsu
2 min readOct 7, 2019

To recap, a chart may lie because:

  1. It’s poorly designed
  2. It uses the wrong data
  3. It shows an inappropriate amount of data — either too little or too much
  4. It conceals or confuses uncertainty
  5. It suggests misleading patterns
  6. It panders to our expectations or prejudices

In these two chapters, Professor Cairo explores further what he meant when a chart is poorly designed or uses the wrong data.

A chart is poorly designed when it distorts data using incorrect scales, encodings, or disproportionate perspectives. He explains that even when we pay attention to the scaffoldings (the scale, legend and labels, etc.) of a chart, it may still unconsciously bias our perceptions anyway.

He also underlines the impact of choosing a sensible scales and baselines depending on the nature of information. We should always strive to find an aspect ratio that neither exaggerates nor minimizes the change.

A chart may lie when it displays dubious data. He stresses that we should never trust any publication that doesn’t clearly mention or link to the sources of the stories they publish. And because the internet and social media are powerful tools for creating, finding and disseminating information, anybody can be content creators or curators.

Unfortunately, not everybody is as vigilant as one should be when consuming, creating or sharing information online. The spread of misleading and untruthful contents is hard to stop, because more often than not, people are prone to consuming contents that only confirm their most strongly rooted ideological beliefs and prejudices.

This kind of continues the conversation of society polarization in my blog post last week. I found an article about A Study Finding At Least 70 Countries Have Had Disinformation Campaigns from The New York Times. And it is very alarming.

That’s why fact-checking the sources is the utmost priority when consuming information online and interpreting or designing charts. It’s also important to take different perspectives regarding the data into account.

--

--