Well... there was talk on this thread of spending money on software to analyze statistical data such as the PR-forums could produce.
My general point is simply: Personally, I am with Cort in thinking that the PR-forums could produce very interesting data about a large group of patients with ME/CFS and it's probably very worthwile to analyze it ourselves, if we can. And I think we can in principle, and certainly to some extent if only in good spreadsheets (Gnumeric is good, open source, and free and has tools for ANOVA and a lot more), and that it is wise to first look around what's available on the internet for such purposes, and what programming and statistical knowledge members have and can and are willing to use sometimes.
As to data: "a first step would be to decide what data we want" is a good point, and e.g. Cort may want to invest some thought in various questionnaires including privacy-issues.
As to: "As long as our data is available in a tabular or separated format": It actually doesn't matter as long as datablocks are somehow reliably separated and in ASCII text-format. If they are reliably separated - by beginning and endmarkers - and in txt it is easy to recook them to any standard a statistical package accepts, which generally is of the form tab-, comma-, or line-separated between blocks. (So actually is easier than to understand my written explanation, and spreadsheets do these things also for one ;).
Anyway: As I said thinking about various questionnaires including privacy-issues that could be done on the PR forums seems required, and I'll do some too, but here too my recommendation is not to act hastily and to investigate and experiment and discuss the possibilities for scientific statistical research that the PR Forums - presently > 1800 members, mostly with ME/CFS - offer.
And only later commit ourselves to anything or any software or spend money on software before we have tolerably clear ideas about the possibilities, of which there are many.