With all the Coronavirus stories in the news right now, you’re seeing a lot of graphs. Although numbers are objective, the way they are presented often isn’t.
Tricks like suppressing zeros can make small changes look gigantic. Choosing one type of graph over another can make trends clearer or hide them.
I was inspired by a graph in my analysis of the Homestead-Miami race to look deeper into a particular question. You’ll see the results of that effort on Friday here, and at 4pm on Dave Moody’s SiriusXM NASCAR radio show.
In working on my blog, I ended up with a really interesting graph. I plotted it a couple of different ways because I knew what the numbers said. Not all the graphs made the point clearly. Some ways of graphic it made the noise look more important than the feature.
Here they are:
I knew there was a distinct downward trend, but the first graph makes that really hard to see. By connecting lines between data points with a lot of scatter, it gives more emphasis to the scatter.
The third graph, the scatter plot, shows the new downward trend much more clearly. It’s not as visually pretty, I know, but if your purpose is to clearly communicate what the data say, it works just fine.
This is one of the most important things scientists have to do: They have to constantly question themselves to ensure that they’re not making the data say what they think it should say.
In fact, while I was writing this, the following popped up on my twitter feed from my friend Jim Kakalios, whose book The Physics of Superheroes you should read if you haven’t.
Note the last sentence. That’s why everyone is talking so much about peer review. Scientists are sticklers for proof.