Graphics Reference
In-Depth Information
Introduction
1.1
Over the last decade, researchers have developed many improvements to make sta-
tistical graphics more accessible to the general public. hese improvements include
making statistical summaries more visual and providing more information at the
sametime.Researchinthisareainvolvedconvertingstatisticaltablesintoplots(Carr,
; Carr and Nusser,
),new ways of displaying geographically referenced data
(Carretal.,
),and,inparticular,thedevelopmentoflinkedmicromap(LM)plots,
oten simply called micromaps (Carr and Pierson,
;Carr et al.,
,
a).LM
plots, initially called map row plots as well as linked map-attribute graphics, were
first presented in a poster session sponsored by the American Statistical Association
(ASA) Section on Statistical Graphics at the
Joint Statistical Meetings (Olsen
etal.,
).Moredetails onthe history ofLMplotsand their connection toother re-
search can be found in these early references on micromaps. More recent references
on LM plots (Carr et al.,
b; Carr,
) focused on their use for communicating
summary data from health and environmental studies.
he basic idea behind LM plots is to link geographic region names and their val-
ues, as shown in quality statistical graphics such as row-labeled dotplots, with their
locations, as shown in a sequence of small maps, called micromaps. his provides
the opportunity to see patterns in a geospatial context as well as in the traditional
statistical graphics context. Figure
.
shows a simple example of LM plots. his fig-
ure shows the
US states with the highest white female lung and bronchus cancer
mortality rates for
. he states are sorted in descending order by the mortality
rates and partitioned into groups of five to promote focused comparisons. he let-
handcolumnconsists oflinked micromapswithonemicromapforeachgroupoffive
states. he top micromap has five regions with black outlines. Within each group of
five states, theshadeofgreyfillforaregionlinks totheshadeofgreyinthedotbeside
the region's name and to the shade of grey in the dot indicating the region's mortal-
ityrate.hesamefiveshadesofgreyordistincthuesincolorplotsarelinkswithin
each group of five states. helinking isbasically horizontal within groupsof five. he
right column of Fig.
.
has familiar dotplot panels showing US state mortality rates
estimates and
% confidence intervals. he data are from the US National Cancer
Institute (NCI) Web site http://www.statecancerprofiles.cancer.gov/micromaps, fur-
ther discussed in Sect.
.
.
.
Figure
.
shows a useful variant of linked micromaps that accumulates states out-
lined in black. hat is, states featured in previous groups of five are outlined in black
andshown with awhitefill orwith asixth distinctive hue incolor versions. heblack
outlinebringstheoutlinedstatesintotheforegroundandcreatesacontourcomposed
ofstate polygons. hebottom micromap contour includesstates with values above
deaths per
. While the political boundaries are less than ideal for accurately
communicating geospatial patterns of local mortality rates, the progression of con-
tours and the visible clusters in the bottom micromap are much more informative
geospatially than the values in a table or a dotplot.
