That I leave to my students! I have a research methods class where I like to have datasets like this for just that purpose. Now that I've 'massaged' the data a bit, I can release it to the class for a series of assignments and have them hypothesize spatial relationships with other variables. The tapestry and market segmentation elements in Community Analyst (bought running shoes, watched a marathon on TV, visited a theme park, etc) can be run in regression against the participation index. Spatial auto-correlation, hot spots...there's so much fun to be had with this! When I have new data, I like to hold off on the initial analysis until I do it with the class so that I don't bias their predictions in any way.
When I pulled the data from the results site (
http://results.xacte.com/?kw=WDW) it came with all names and bib numbers. Spatially, that data isn't relevant, and while it's technically public data and could be republished, most people would find it a bit creepy. But you can always use the results site to search by state or city.