Miscellaneous notes and resources on learning GIS for beginners
Adam Storeygard
October 29, 2007
Disclaimer: This is all assuming you are using ESRI products, and especially ArcGIS (Arcmap, ArcCatalog, ArcToolbox), not because I think they are better than anything else, but simply because they are the industry standard and I know very little about alternative software.
Please email me with corrections as some problems eventually get fixed.

Training resources
S4 has a good list of GIS training resources, from online cookbooks (CSISS) and courses (ESRI; free for students at some institutions including Brown), to in-person courses at Brown, ranging in length from two hours to two weeks to a semester:
http://www.s4.brown.edu/S4/Activities_Courses.htm

Handy analysis tools
Xtools – for example, you can delete multiple fields in a table at once.  This sounds like it should be easy but it’s not.  The basic version is free.
http://www.xtoolspro.com/
Hawth's Analysis Tools for ArcGIS
http://www.spatialecology.com/htools/

Notes
These are some things that I have seen cause problems in the past:

1. Note that a “Map document” (.mxd file) does not include the data it displays.  It is just a pointer to those data, plus a bunch of display and formatting information.  If you lose the data, your map document will no longer be very useful.  If you move the mxd or the data (or the relative path between the two, if you are using relative references), a red exclamation point will appear next to the data’s entry in the Table of contents to the left or your map.  You will have to tell the mxd how to find the data again
2. If you must “Connect to Folder” to get ArcGIS to find your data in a network folder, a USB datastick or external hard drive, make sure that you are connecting to a folder, not a dataset.  Try connecting to the outermost folder you can – then you will only have to do this once.  This dialog box sends you back into the Windows file structure, as opposed to the ArcGIS file structure, which takes some getting used to (see below).
3. Until you have a good understanding of the ArcGIS file structure, use ArcCatalog for all file management needs.  If you use Windows to move certain kinds of files, you can corrupt them irrecoverably.  If you absolutely must move them outside ArcCatalog, (e.g. emailing them), zip something that you are sure is actually a folder (i.e. _not_ a grid that windows thinks is a folder) first. 
4. Don’t use spaces or any characters other than letters, numbers, underscores (_) or hyphens (-) in filenames or even folders into which you put data.  This is less of an issue than it used to be in older versions of Arc, but it can still cause problems.
5. If you can’t use a feature using a limited version of ArcGIS, such as the student version, try it in a computer lab with a more comprehensive (i.e. expensive) version.
6. Many social scientists want to use GIS purely to get summary statistics (mean, median, standard deviation, etc.) of a spatially varying variable of interest (elevation, rainfall, etc.) for well defined regions (countries, states, etc.).
1. If the regions you care about are countries, this resource has some variables of interest: http://sedac.ciesin.org/place/.
2. If you need to get your hands dirty with GIS, the class of procedures you want are called zonal statistics.  The variable of interest has to be a raster dataset (i.e. a grid), known in this context in ArcGIS lingo as the “value raster”.  ArcGIS will allow the regions (”zone dataset”) to be raster or vector (in this case, polygons, though in general vector data also include points, lines, and networks).  However, using vector zones can give you bad answers – see ESRI’s article about this problem here.  Converting your zones to raster before running zonal statistics allows you to specify how the conversion is done, and see the results.  This should lead to a better zonal calculation, with more potential for error-checking afterwards.  If none of this makes sense to you, you probably aren’t ready to use zonal statistics yet.  If on the other hand I have misinterpreted the ESRI article (and the knowledge of real GIS experts that I am trying to pass on with this advice), please let me know so I can stop giving bad advice!