Module 2: How to Critically Analyze and Interpret Data Visualizations
Module 2: How to Critically Analyze and Interpret Data Visualizations
Misleading Data Visualizations
Now that we know how to analyze and break down a data visualization, let’s go through a few examples of design choices (and mistakes!) that can create confusion.
Using the Wrong Type of Data Visualization
As we learned in Module 1, some types of data visualizations work well for communicating specific types of information, but not others. For example, pie charts are good for making comparisons between a few different categories, but are not great for identifying patterns or showing data over time.
Data visualizations can be confusing and misleading when the designer has picked a format that isn’t well suited to the data they are analyzing.
Review Figure 2.9[1] below, a pie chart of Ontario television viewing in 2004. There are 12 categories of television and similar colours used in the graph, as well as white font over the bright colours, making this hard to read.
False Causation
Correlation does not imply causation.
If you’ve ever taken a statistics or data analysis course, you have almost certainly come across this common phrase. It means that, just because two trends seem to fluctuate alongside each other, it doesn’t prove that one causes the other or that they are related in a meaningful way.
Review Figure 2.10[2][3] below, which shows a line graph of the decrease of Canadian automotive apprenticeship registrations and nectarine production. What do these two things have to do with each other? They are unrelated quantities that appear to decrease at the same rate over a similar time period.
Inconsistent or Manipulated Scale
It’s important to examine the scales of a data visualization carefully. Compressing or expanding the scale of a graph can make the changes between data points seem either more or less significant than they really are.
Review Figure 2.11[4] below, which shows the cost of sugar in Canada from January to July 2021. Because of the expanded scale on the line graph, there does not appear to be much fluctuation in the cost of sugar in Canada. This makes the data appear less significant than it could really be (see Figure 2.12 below for a more compressed scale).
Cherry-picking or Omitting Data
The term “cherry-picking” refers to only presenting the best data, and omitting data points which are less favourable, in order to reinforce a particular narrative. This can create a false impression of the data. For example, showing an upward sales trend over the first few months of a year, while omitting the data that showed sales declined for the rest of the year.
Review Figure 2.13[5] below, which shows a downward trend on gasoline prices in Canada from May 2019 to February 2020. Because of the carefully selected timeframe (i.e., short timeframe), it appears that the gasoline prices in Canada are decreasing.
Now review Figure 2.14[6] below, which shows an overall upward trend on gasoline prices in Canada from May 2019 to November 2021. Whe