![]()
23-April-2006. The zipped setup file file is
18.3Mb in size and is named HDA-1.zip. It can be downloaded
here. This program attempts to solve a general
problem of determining "diversity" in variable magnitudes amongst members of
predefined groups. It achieves this by defining diversity as "variability", such
that those groups whose members are more heterogenous in their magnitude vectors
across the variables of interest are deemed to show more "diversity" than those
groups whose members show greater homogeneity of magnitude vectors. In the
language of biological ecology, the groups may be species, members of the groups
the samples of such species, and the variables may be associated with those in
the habitat within such species exist. Therefore, in looking at habitat
diversity amongst species, the program enables an investigator to rank species
in terms of their diversity of use of habitat variables. Indeed, the program was
specifically written for an investigation in marine biology. However, it has application wherever the concern is to
compare predefined groups of individuals in terms of the degree to which score
or magnitude vectors on variables vary between members within groups. The
program works by comparing each member of a species/group to all other members
of the group, using three kinds of distance/discrepancy-expressed-as-similarity
functions scaled between 0 and 1. Thus, the raw data of the within group
comparison consists of (N2 - N)/2 comparison coefficients, where N = the number
of members within a species. Rather than using the conventional summary
point-estimate approach (using moments such as the mode, mean, or
variance/standard deviation, or an assumed hypothetical sampling distribution
for a test statistic based upon these coefficient distributions, the program
instead relies upon the use of the median and interquartile range (IQR), with
its results presented as a range plot (plotting the median and IQR for all
species in one comparative graph). Obviously, the program is computationally
intensive when species/group numbers may be measured in thousands. The
coefficient distributions per species/group may be output for further analysis;
the program also provides a binned frequency distribution for each species/group
for the convenience of the user.
Download and Installation Information
You can run the setup file from the web-page download dialog, or save it to some
convenient location on your own disk and simply run the setup file. The software can be installed in any directory of your choosing.
The installation includes a very large online
help file (7.9Mb: lots of screen graphics to support the "How to" and Results
sections) as well as a printable 54-page pdf manual (5.6Mb) - which is why the
installation is so large.
27th April, 2006 -
For Windows ME users only (if there are any
left out there!) - the installation requires a DLL (GDIPLUS.DLL) to be installed
in your c:\Windows\System subdirectory. I just noticed this whilst testing the
installation on an old machine. You can download this file from
here (0.9Mb) - within a zipped archive.
![]()