Avoiding Manipulation by Data Visualisation: A Quick Guide

With fake news and consumer manipulation on the rise, it's important for the public to have an eye for misleading data. This post gives a few brief tips on what trends to look out for when identifying manipulative data visualisations.

Misleading Axes

One of the easiest ways to assess a graphic’s credibility is by analysing its axes. Axis manipulation involves blowing out the scale of a graph to minimize or maximize a change. If a 0 value is not represented on a chart’s axes, this should be a warning sign. Compare the charts below:

Despite depicting the same data, the axis in the first chart is skewed to amplify the perceived differences in the variables under analysis.

Going Against Conventions

When dealing with visualisations that do not incorporate a numerical scale or other objective differentiators, caution should be exercised. For example, in the case of a colour-coded map conveying population density, one would assume that the darker regions display areas of greater population density. However, if no colour legend is present, there is no way to be sure of this. The graph below highlights this issue:

Intuitively, from the graph on the left, you might think that states such as Montana and the Dakotas in the North are the most densely populated, while Texas and California are the least densely populated. However, upon inspection of the colour-legend, you will see that the opposite is true.

Correlation Does Not Mean Causation

The golden rule of data analysis. Data is often represented so as to imply that a causal relationship exists between two variables. This is, however, not the case. While one thing may cause another, it is impossible to know for sure. So, if you come across a graph that claims an absolute state of causality between two variables based off data, be wary!

Does the increasing incidence of pool drownings coincide with Nicolas Cage appearing in more films? Yes. Is that the extent of the relationship between these two variables? Probably.

If you have any questions, please leave them in the comments below.

Interested to learn more about Data Visualisation?

Then join us here 👉 https://www.datasoc.co/

Follow us on Medium 👉 https://medium.com/search/publications?q=datasoc