Graphics Reference
In-Depth Information
in Sect.
.
.
will only show counties within a given state, and then only a few coun-
ties are on the screen at one time. Texas has so many counties that the vast majority
of them are out of sight in any view. Refreshing human memory about a big visual
space via a scrollbar that reveals little pieces is not very effective. Second, the par-
titioning sliders in CCmaps are fun to use. If a search starts to get tedious, there is
a built-in option to find initial slider settings with relatively good fits with respect to
the R-squared value. hird,theCCmaps sotware isuseful aseducational sotware as
itemphasizes two-way comparisons and weighted averages, and the sliderseven lead
to discussions about the di
culties related to long-tailed univariate distributions.
Trellis Graphics in S-Plus provide another approach to the display of data. Since
Trellis Graphics are programmable, discussion here focuseson what isrelatively easy
and what is harder to do with Trellis Graphics. he perceptual grouping in LM plots
can easily be fed to Trellis Graphics as an ordered categorical variable for the control
of panel production. Showing dotplots with confidence bounds is not hard, nor is
showingreferencevalues.However,TrellisGraphicsareprimarilydesignedtohandle
a single dependent variable. hey are poorly suited for showing multiple dependent
variables in side-by-side columns such as the lung and bronchus cancer mortality
and the percent of current smokers as in Fig.
.
. Trellis Graphics were not designed
to provide a variety of options such as geographic drill-down into subregions and
blinking of linked symbols that are built into the Web-based sotware that produced
Fig.
.
.houghTrellisGraphics can providesomealternative views that maybevery
useful, they are not ideal for producing LM plots.
Acknowledgement.
JürgenSymanzik'sworkwassupportedinpartbytheNSF“DigitalGov-
ernment” (NSF
-
) grant #EIA-
and by a New Faculty Research Grant from the
Vice President for Research O
ce from Utah State University. Dan Carr's work was also sup-
portedin part by NSF grant #EIA-
.
We would like to thank our many coauthors of newsletter articles, conference papers, and
journal articles underlying this work. Much of the material presented in this paper goes back
to collaborations with colleagues in federal and state agencies, projects with colleagues from
severaluniversities,and,mostimportantly,workwithourstudentswhocontributedtothede-
velopment of LM plots over the last
years. Additional thanks are due to Naomi B. Robbins
forherhelpfulcommentsonanearlyandalatedratofthischapterandtoSamson Gebreabfor
his helpwith the preparation of some of the figures. Finally, thanks are also due to D. Andrew
Carr who worked on the grouping options as part of the Visual Basic front end to LM plots at
the BLS.
References
Becker, R.A., Chambers, J.M. and Wilks, A.R. (
). he New S Language -
A Programming Environment for Data Analysis and Graphics,Wadsworthand
Brooks/Cole, Pacific Grove, CA.
Boyer, R. and Savageau, D. (
). Places Rated Almanac, Rand McNally, Chicago, IL.
Brewer,C.A.(
).SpectralSchemes:ControversialColorUseonMaps,Cartogra-
phy and Geographic Information Systems
(
):
-
.
