In this case we have explored the use of heat maps as
an effective means to visualize geographic data. The impressions
drawn from the heat maps of cessation expense and cessation expense
adjusted for population size are quite different. The choice of which
visualization is most appropriate depends on the problem statement
and the important messages to be conveyed to your audience. In the
smoking cessation class, where only a few graphs and statistics would
be desired, the heat maps showing total health care expense and total
cessation expense communicates the nationwide magnitude of the health
care costs associated with smoking and the efforts being made to prevent
smoking. The average total health care cost per capita of $459 per
person is another powerful statistic in illustrating smoking related
costs.
Correlation analysis
quantifies the relationship between continuous variables both in terms
of direction and strength of association. Examining correlations
can also help identify when adjusted values should be reported. For
example, the strong relationship between total health care expense
and population size suggests that when comparing these expenses between
states, a value adjusted for population size should be reported.
Always bear in mind that a correlation established from data does
not imply that one variable causes another. Causality must be determined
from other evidence found in the problem domain. Interpreting correlations
between adjusted and unadjusted variables can be difficult. Finally,
only a few of the possible analyses have been presented in this case.
There are many more relationships that can be explored from this
relatively small data set. Considerable time is required to fully
explore the relationships between the relatively small number of variables
and to examine the various ways in which the data can be summarized
and visualized. It is important to choose only those data visualizations
and statistics that will resonate with the audience and communicate
the intended message in the time allotted.