Designing for an Audience - Data Points: Visualization That Means Something - page 611

You might want to see the specifics of each distribution though, so instead use
histograms stacked on top of each other to compare, as shown in Figure 6-26
Again, you can see that as a whole, obesity rates increased. Each distribution
looks bell curve-ish, and the peak shifts right as you scan from top to bottom.
Figure 6-27 is specific and shows a time series line for each county. Some trans-
parency is used so that areas in which there are many lines appear darker. So
you see the distribution around the 20 to 30 range and an increasing trend.
The black line shows the annual county median.
Because you do have geographic data, it might be useful to see the data
in map form, as shown in Figure 6-28. Darker shades of red indicate higher
obesity rates. Are there any geographic patterns or clustering? It looks like
the southeast has higher obesity rates than the rest of the country, and that
pattern is more pronounced in later years.
However, the nationwide increase isn't especially obvious year-to-year. For
example, the map for 2004 and 2005 look similar, except of course the 2005
map is a little darker. There are two things that get in the way of making
the differences more clear: county borders and a color scheme that doesn't
provide enough contrast. The former makes the map look lighter, especially
on the east coast where counties are smaller, and the latter makes it harder
to see differences between counties and by year. Figure 6-29 provides more
contrast and removes county borders for a clearer look.
Are there counties that improved, as in decreased obesity rate, during the
years? There must be. Figure 6-30 shows such counties highlighted in blue,
but there aren't any regional patterns. The blue looks randomly scattered
across the country.
The key with weight of course is about the long term. There were only 51 coun-
ties out of 3,138 that had lower obesity rates in 2009 than in 2004. Figure 6-31
highlights the improved counties in blue, and the results aren't impressive.
It looks like it's best to focus on the increase of obesity rate across the country.
There wasn't much decrease. However, even with the color scheme that has
more contrast and removed borders, the maps shown in Figure 6-30 don't
make the annual changes that obvious. There's change each year, but not
that much. However, compare the map for 2004 against the one for 2009. The
differences are much easier to see. What if the final graphic uses the map from
2004 and the one from 2009? That seems to make for a better comparison,
so go with that.