One reason is something over which spatial analysts and GIS people have little or no direct control. Take a look at recent issues of the "Annals" or "Transactions". Assuming that the contents of these journals accurately reflect contemporary scholarship in geography, it is obvious that many of our colleagues are simply not concerned with explicitly spatial issues whose examination would be enriched by sophisticated spatial analysis. The most that GIS can contribute to such work is to provide a convenient means of data storage and mapping capabilities. In view of this, it becomes imperative that we communicate effectively with those, both inside geography and outside, who still have an interest in the spatial analytical paradigm. How can we do this?
First of all we need to demonstrate more convincingly that GIS based SA techniques can be used to address important substantive problems and that such techniques are capable of providing meaningful insight which cannot be obtained otherwise. Why is it, for example, that despite the development over the past twenty years of diagnostics for spatial dependency, we still routinely encounter published papers which feature mis-specified regression models? Are the authors, reviewers, and editors of such papers unaware of our material or have we failed to convince them of its value? Fortunately, there are also positive signs. For example, interest in the Getis statistic has been increasing in image analysis in remote sensing as it has been shown to capture in a single value much of the same information that required the calculation of a battery of statistics based on co-occurrence matrices. Few individuals combine GIS and SA expertise and even fewer combine this with a topical expertise and so it is not surprising that the illustrations which accompany our presentations of new techniques often appear inconsequential or to reflect more the availability of a convenient data set. I think we need to do much more collaborative work with topical specialists. Not only will this make our techniques more visible in applied arenas, more importantly, it will also help us to identify meaningful spatial questions for which new techniques need to be developed.
However, it is one thing to perk the interest of a few topical specialists,
it is another matter to enable our techniques to become part of the everyday
tool kits of the majority of their fellow practitioners. This
requires that not only should such individuals be aware of our analytical
procedures, they must also be able to implement them with minimum effort.
The former requires good, practical texts which illustrate both the conceptual
underpinnings of our procedures and clear guidance on how to use them.
The latter requires user-friendly, integrated software. The optimum
would be a combination of the two. As far as I am aware, there is
only one example of this (but a good one), Bailey and Gatrell's INFO-MAP.
I think that there is a great need to develop software and accompanying
texts in the genre the SPSS* manuals. There is no doubt in my mind
that the ease of use of SPSS* has contributed to non-spatial statistical
procedures becoming commonplace throughout the social sciences. In
this regard, I feel we should be a little less elitist and a little more
tolerant in the way we present our material. For example, there are some
very good spatial statistical texts around but they are not easily accessible
to those without formal statistical training (and without either the time
or the desire to obtain it). I'm not advocating that we lower our
standards but that we change our emphasis, at least as far as textbooks
are concerned. We have to recognize that the cultures of other disciplines
may differ considerably from our own. The majority of practitioners
in other disciplines who can benefit from our materials will only use them
if they are packaged in accessible and readily usable forms. There is a
large number of potential users for our materials, we cannot afford to
ignore them.
* SPSS is a registered trademark of SPSS Inc.