Graphics Reference
In-Depth Information
wasdesignedspecificallytomonitorthisdatasituation.hehorizontalaxisispropor-
tional tothetimescale,thetreatments aredistinguished bycolor,thesample sizesare
identified at each time point, a reference line locates the center of the normal range,
outliers are noted, and extreme outliers are identified.
Microplots
6.7.3
Boxplots capture comparative information better than numbers. hey don't have to
take much space, therefore they can fit into tables of numbers and satisfy both the
regulations for displaying numbers and the legibility of displaying graphs. We call
theplots microplots whentheyaredeliberatelysizedtofitintoaregulation-mandated
table of numbers without interfering with the overall layout. When small plots are
placed into a table of numbers, they can carry the same or more information per
cm
as the numbers themselves. he two examples herehave different objectives and
therefore are arranged in different patterns.
Example - Catalyst Data
6.7.4
With the catalyst data from Montgomery (
) we are interested in comparing the
concentrations of one component of a liquid mixture in the presence of each of four
catalysts. Weinvestigate whetherthecatalysts provideforequal mean concentrations
and then, since this does not appear to be true, we study the extent of differences
among the mean concentrations.
hemicroplot ofsmallparallel boxplots inFig.
.
takes uplittle morepagespace
than the table of summary statistics. Weplaced y=concentration on the vertical axis,
the axis that we have been trained to think of as the appropriate location for a re-
sponsevariable.Undereachboxweplacedthetableofrelevantstatistics:mean,num-
ber of replications, and standard deviation. he F-test shows a significant difference
between the catalyst means. We can see it visually from the microplot or tabularly
from the numbers.
Figure
.
.
ANOVA table, boxplots, and means for each catalyst (File: hh/oway/code/catalystm.s)
