302 
cause GeoMed incorporates the methods of 
Stat!); but GeoMed goes beyond Stat!, pro 
viding a broader range of spatial statistics in 
cluding 
• Cuzick and Edwards’s test 
• Grimson’s Method 
• Ripley’s K-function 
• Besag and Newell’s Test 
• Local Indicators of Spatial Autocorrelation 
(LISA) 
• Turnbull’s Test 
• Kulldorff’s Spatial Scan Statistic 
• Bithell’s Test 
• Diggle’s method 
• Score test of Lawson and Waller 
For each of the many tests and methods above 
we provide an on-line Information Frame, or 
succinct textual and graphical description of 
the test: this takes a user from description of 
the statistic, to the null and alternative hy 
potheses, to an example analysis using the 
statistic. These tests are also linked together 
by the statistical advisor, which serves as an 
aid to a researcher in choosing the appropriate 
test given their particular statistical problem. 
Ultimately GeoMed software will be available 
for download from our web site; in the mean 
time other statistical software (e.g. Stat! and 
Gamma) is available in demo versions from 
the web site. 
5 Our Course 
We now describe our course in some detail. 
This will serve to illustrate how one might use 
the resources which are currently in place for 
the development of a course of one’s own. 
The University of Michigan is involved in 
the GeoMed project because members of the 
University wish to see their students well- 
educated in some of the latest techniques for 
spatial analysis: we thus strive to see that 
these student needs are met. We focus on the 
course because it serves as the center around 
which we continue to build the web site; but 
we consider that the web site has broader ob 
jectives than the course (we see our course as a 
specific imlementation of the materials avail 
able at the web site). 
5.1 Objectives and Audience 
The course objective is as follows: To pro 
vide students with the knowledge, theory, and 
methodological skills for analyzing and inter 
preting the spatial patterns of various diseases 
in order to elucidate underlying exposure pro 
cesses giving rise to the observed patterns. The 
students referred to are expected to have a 
special interest in Public Health, and recruit 
ment for this new course especially targets 
students in the School of Public Health. This 
includes especially students from the depart 
ments of Biostatistics and Epidemiology, at 
both the PhD level and the Masters level. 
Students are expected to have had introduc 
tory probability theory and to have some basic 
knowledge of biostatistics. 
Another objective is to exercise certain 
BioMedware software products, including one 
established product (Stat! [5]), a product soon 
to be released (Gamma [1]), and especially the 
new software designed for this project (which 
we call GeoMed [2]). This last piece of soft 
ware will serve as the keystone for much of 
the spatial statistical analysis we do in our 
course in the future. Several computer labs 
have been structured around these pieces of 
software, each of which is designed for per 
forming spatial statistical tests. The course 
allows us to both alpha/beta test the soft 
ware, and to generate insight from the stu 
dents about ways in which the software can 
be improved. 
5.2 Modular Design 
From a logistical standpoint, we have 24 hours 
of lecture time, divided into two-hour blocks 
in which to meet our objectives, along with 
another 24 hours of lab (again in two-hour 
blocks). We focus on seven basic ideas: 
• Introductory epidemiological concepts (1