10 not so great data visualisations

Data Visualisation is a great way to covey data to readers but sometimes visualisation can go wrong or be very misleading. If you want to get started but don’t know how, take a look at this article. Once you’ve got to grips with the basics, you’re going to want to know what to avoid, so we’ve scoured the internet to look for examples of visualisations that probably need a rethink.

So here are some examples of what not to do when creating visualisations:

1. NBC US Election Map

The NBC US is a fairly traditional map of America showing the winners of states in their respective party colours. The grey states show the undecided states at the time.

The issue with the map is that at first glance the electoral college votes are not shown and they are only revealed if you hover your mouse over the states. It can’t always be expected that people will hover over the states. If the map is shown as it is, it can be used to give misleading information. Many Republicans used snapshots like the one above to show that they had a lead. A better idea would be to use a cartogram that represents each of the electoral votes in each state, as this shows the situation more representative and doesn’t rely on people actively interacting with the map.

2. UK Government Coronavirus Briefing Map

This is a map from a government coronavirus briefing and shows cases per 100,000 people in local authorities. On the map, the weekly case rates per 100,000 people don’t increase by the same amount each time in the key. There are intervals of 25 for the first two categories, but then intervals of 50 until 200+. Also, the highest interval for case rates in the key is 200+ but in some areas the rate greatly surpassed 200 cases per 100,000 people. Wigan, for example, had 622 cases per 100,000 people.

3. SW Health Nurse Infographic

This infographic from New South Wales is obviously misrepresentative. It shows an increase in the number of nurses that is factually correct but the problem is the scale. Four stick people represent 43,000 nurses but 32 (28 more) stick people are used to represent an increase 3,000 nurses. A 700% increase in stick people is presenting 7% increase in actual nurses.

4. Popsci Android Phone Model Treemap

This treemap from Popsci shows how when an android phone is released there are lots different versions that spin off it. You can see that there are lots of various models that have been released but not much else can be taken from it. The treemap is just way too visually complex and cluttered.

5. Guardian Poverty and School Ofsted rating maps

These maps from the Guardian show levels of poverty on the left and schools that are ‘inadequate’ or ‘require improvement’, according to Ofsted, on the right. But you can’t really see if the worst schools are really in the poorest areas. There are two issues with these maps. Firstly, they don’t group their findings in the same way: the left map shows local authorities and the right shows constituencies. Secondly, the colours used in the scale are far too similar to each other.

6. Arizona Coronavirus Dashboard Bar Graph

These bar graphs show the total daily case counts in Arizona as a whole and the total daily case counts across a particular county: Navajo. Neither chart has a y-axis and the colour gradient is not uniform. Although there are labels with case numbers, visually, it looks like the scale of the case count across the state is the same as in Navajo county alone. The daily case counts in Arizona state alone are in the 1000s, while they haven’t crossed 100 in Navajo county yet.

7. The Economist Brexit Attitudes Line Graph

This line graph from The Economist shows attitudes to the outcome of the EU referendum. The graph seems to show that the opinions are quite erratic. It seems they have connected the actual values of each individual poll instead of plotting the individual polls as a smoothed curve to show a potential trend.

8. NCHA Student Substance use Bar Charts

These bar charts show the results of a survey where participants where asked about their substance use over the past 30 days. The white bars reflect actual substance use, and the red ones show perceived substance use. The problem is clear: all the white bars are the same size and all the red ones are the same size despite representing different percentages.

9. Baby Boomer Survey Infographic

This infographic shows the results of a survey of how baby boomers describe themselves. If you add up the percentages for the different characteristics described in this infographic, you get a 243% baby boomer. It might have been better to show the different characteristics in separate charts.

10. Climate Change denier graph on Twitter

This graph is from a climate change denier on Twitter. This graph seems to show that global surface temperatures haven’t changed much from January 1880 November 2014 (which obviously not true). The problem is that this graph uses the Kelvin scale. The Kelvin Scale is a temperature scale that has an absolute zero, below which temperatures do not exist. Absolute zero, is the temperature at which molecular energy is a minimum, and it corresponds to a temperature of −273.15° on the Celsius temperature scale. So the problem with using the Kelvin scale is the that temperature rarely falls to the absolute zero on Earth. Generally, the Earth is at temperatures between −68 °C (205 K) and 56.7 °C (329.85 K). So, the ‘absolute zero’ scale is quite misplaced here.

Also because Celsius or Fahrenheit scale are more common, using the Kelvin scale can be far more confusing to the average viewer.

This article was originally published on Interhacktives, a website for student journalists at City, University of London.