Spotting Differences - Data Points: Visualization That Means Something - page 268

More concretely, you can use a boxplot, which shows quartiles in a distri-
bution. Boxplots generated in R with the boxplot() function can automati-
cally highlight points that are more than 1.5 times more or less than the
upper and lower quartiles, respectively (Figure 7-36).

P A quartile is one
of three points in
a dataset, which
marks quarter
spots. The middle
quartile is the me-
dian or the halfway
point; the upper
quartile marks the
spot where 25 per-
cent of the data
is greater than
that value; and
the lower quartile
marks the bottom
25 percent.

FIGurE 7-36 Boxplot showing
distribution of subscriber counts

If I had a small number of subscribers in the single digits, then sure, such
a big percentage-wise decrease could be possible, but it's unlikely that
I said something so offensive to compel tens of thousands of readers to
defect (and then come back a couple of days later). It's much more likely
that Feedburner, the feed delivery service I use, made a reporting error.

These outliers in these datasets are obvious because you know a little
bit about the data. It could be less obvious if you use datasets you're not
familiar with. When that happens, it can be helpful to go directly to the
source and just ask whoever is in charge. The person or group curating
the data is usually happy that you're making use of it and will offer some