Database Reference
In-Depth Information
Figure 9-8. Dual-axis line chart with y-axes formatted
What can we glean from this dual-axis line chart? We can see right away that teams added
pitchers starting in 1871 until about 1940, and then held their rosters steady at about 15
pitchers per team until the early 1980s, when they began adding pitchers once again. Thus,
the turnaround in average strikeouts per game correlates with the point in time when teams
began adding more pitchers to their rosters. Correlation does not imply causation, but we
could come up with a causal theory to test:
more pitchers per team means each batter is likely
Let's consider another way of showing how these two variables are correlated: the connected
scatterplot.
The Connected Scatterplot
We haven't yet strayed from the rule of thumb that time should be encoded on the x-axis in-
creasing from left to right. An innovative way to encode time is to make a scatterplot of the
two variables in question and connect dots that represent successive years. Let's create one to
see what it would look like.
To start with, we can create a standard scatterplot as we did in
Chapter 8
with pitchers per
team on the x-axis (or Column shelf) and SO, or average strikeouts per game, on the y-axis
(or Rows shelf). If we do so and then add a trend line as before, we will find that the vari-
