Data Visualization: Cognition, Perception and Anscombe’s Quartet
‘In short, good visualization methods offer extremely valuable tools that we should use in the process of exploring, understanding, and explaining data. But they are not a magical means of seeing the world as it really is. They will not stop you from trying to fool other people if that is what you want.’- Kieran Healy.
Data visualization through graphs is a relatively quick and efficient way to understand a dataset. There is not a set of rules to generate good graphs that apply to all circumstances. It is preferable to think about your data and the perceptual features of your graphics. It will help to develop an ability to make good taste-based judgments. We have to consider before start writing codes the aspects of graph construction, perception, and interpretation that matter for the code we will choose to write. The tools you use can help you live up to the right standards. But they cannot make you do the right thing. We will begin by asking why we should bother to look at pictures of data in the first place, instead of relying on tables or numerical summaries. It is important as tasteful and well-constructed graphics can mislead us.
The purpose is to make the graph more easy to understand as the cognitive aspects of data visualization make some kind of graph reliably harder for people to interpret. That is why cognition and perception are important. When we look at graphs, our brains sometimes make up connections between things we see, even if those connections aren’t there. This happens because of how the points and lines are arranged on the graph. It can help us understand the data better, but sometimes it tricks us into seeing things that aren’t true. So, we need to be careful when relying too much on what we think we see in graphs.
Anscombe’s Quartet is four sets of data that seem similar when you look at basic numbers like averages and spreads, but when you plot them on a graph, they look very different. It shows that just looking at numbers can hide important differences in the data. Consider the following dataset:
Dataset: Comparison of two parameters
| X | Y |
|---|---|
| 10 | 8.04 |
| 8 | 6.95 |
| 13 | 7.58 |
| 9 | 8.81 |
| 11 | 8.33 |
| 14 | 9.96 |
| 6 | 7.24 |
| 4 | 4.26 |
| 12 | 10.84 |
| 7 | 4.82 |
| 5 | 5.68 |
At first glance, if you just look at the averages and other summary statistics, this dataset might seem similar to others in the quartet. However, when you plot it on a graph, it could reveal a different story.
