This initial exploration of the
appointment backlogs and characteristics of the VA medical centers
followed a progression beginning with univariate graphs followed by
bivariate and multivariate visualizations. While it is tempting to
jump right in and look for relationships between variables, it is
important to first visualize each variable individually. Not only
does it focus the analyst on the observed magnitude and variation
for each variable, it allows outliers to be identified. Outliers
may be unusual observations or data errors that warrant further investigation
and in the case of data errors correction or removal.
Visualizations are an
effective way to familiarize both the analyst and stakeholders with
the data available for analysis and the variations between the VA
hospitals. Plotting this information on a map adds geographic context.
The exploratory analysis allows us to form initial impressions about
the relationships between variables and develop hypotheses. However,
it does not allow us to definitively answer questions such as “Has
there been a significant change in the appointment backlogs from 2015
and 2016?” In the next case, we will make use of statistical
tests of hypothesis to address such questions.