I’m fairly new to the blogosphere and Twitter and most of the other forms of social media so I hope you’ll forgive me for arriving late to the party and digging up relatively old posts. I read a recent post by James Fee on his Spatially Adjusted blog that referred back to a 2009 blog post arguing that “Spatial is not Special“. Sorry to be 2.5 years late to comment but if no one else has formed an opposition to this opinion, please allow me to be the first.
If James had titled his post “GIS is not Special”, I wouldn’t have felt compelled to debate, because GIS software is not special, but “spatial” is indeed special. Here’s why:
Spatial dependence and statistical inference. If you’re working with spatial data and not accounting for spatial dependence or measuring spatial autocorrelation then you are probably making some major mistakes that could have a significant impact on the accuracy of your results. There are scads of studies, publications and analyses, academic and otherwise, that probably need to be completely redone and rewritten because the statistical techniques applied assume iid where spatial dependence may have biased the results in an important way.
Spatial error. Everyone who has been trained to use GIS software should be painfully aware that geocoding is often inaccurate, sometimes wildly inaccurate. In addition, there are a number of sources of error faced by those who work with spatial data include the modifiable areal unit problem, ecological fallacy, locational fallacy and other problems involving geographic scale that, in my opinion, are uniquely spatial and make spatial analysis a unique branch of data analysis.
There are other factors that could be identified here but I don’t think it’s necessary to list them all in order to effectively argue that spatial is in fact special. For those who can stomach academic research articles, I would recommend this paper by Luc Anselin if you want to read more.
The key issue may be perspective. If you’re a programmer or software developer type who works with GIS software components, you’ll probably find at some point that GIS components aren’t really different than other software components and that it’s just another piece of the software development landscape. This may lead you to incorrectly conclude that there’s nothing special about spatial in general because there’s nothing completely different about GIS software in terms of how it can be used to develop software applications. However, for those who are *analyzing* spatial data, you had better recognize that spatial is indeed special or you’ll be making rookie mistakes.
I like Don Meltz’s post on this entitled GIS is Dead – Long Live GIS. Don equates GIS software to word processing software and it’s a good analogy. GIS software will evolve and become easier and easier to you use. But just like word processing didn’t make good writers obsolete, easy to use GIS and mapping software won’t make good geospatial analysts obsolete either.
So if you’re considering a GIS career, you might want to reframe the question. Instead of learning ArcGIS and getting a GISP certificate, I would recommend that you choose one of 4 paths: (1) geospatial or geostatistical analyst, (2) cartographer or visualization design expert, (3) software developer or (4) Geographic Information Scientist. Or some combination of the four (and an application area of interest wouldn’t hurt either). If you only train yourself to be a GIS user you’ll be in the same boat as a professional word processor before too long.
To learn more, read my new post entitled: Spatial Career Guide for Undergrads Currently Studying GIS – Curriculum Suggestions for 6 Geospatial Career Paths