I also added some questions out of my own curiosity, partly from trying to figure out how similar or dissimilar my situation was from other fans. Questions about who introduced anime to the fan, how many other fans the person had regular face-to-face contact with, were two examples of such.
As time went on, more discussions provided fodder for more questions, BUT responders to each questionnaire presented interesting subjects to add, too.
Every time a suggestion or topic presented itself, I usually ended up breaking it down into multiple questions, or thought of additional related questions. As a consequence, the questionnaire grew by about 20 questions a year, and the result is a fairly massive beast.
If you want to publish a copy of the survey, please don't do so without contacting me and getting my permission. Of course, quoting two or three questions at a time, and their responses, is fine, but, again, be sure to attribute fully.
Tossing a few numbers from the survey around in a discussion? Go ahead, that's what it's for. Just make sure folks know where the numbers came from, or they'll accuse you of pulling numbers out of the air (this way, they accuse me of pulling numbers out of the air, not you).
Well, that depends on what you mean by "data".
If you mean the raw data, as in what I get when a person hits the "Submit" button, absolutely not. In my privacy statement (on the questionnaire) I state individual information will not be published or made available to anyone. I take that as a promise and a point of trust.
If you mean the numbers I get as a result of processing the raw data, what you see on my results pages is what I use. There is no intermediate database or spreadsheet. The questionnaire varies from year to year, so the first data format has to be the flat ascii files you see, just so I can check that the data extraction worked properly, and to make it easier to see what the significance of the numbers are. Once in that format the results files are easy enough to parse that I find I don't need to go the extra step of entering them into a database or spreadsheet, and simply process those results files directly (e.g. for my trend plots). The bottom line is that if you want to do some processing of your own, the results pages are the closest data I've got to the raw data.
In 2001, I had to cut back to an annual survey because of the time constraints, but also because the trends I was most interested in have pretty much slowed down. I also cut the duration from 4 weeks to 3 weeks.
2002 was when I made the questionnaire available as a web form, in addition to my standard text posting to the rec.arts.anime.* newsgroups. As you can imagine, the widgets on a web form are a lot easier to fill out, and from my end, there are fewer user mistakes to deal with and I get more compact returns.
From 2003 and on, Anime News Network has been kind enough to post an announcement for this survey, with an excellent response, as you can see from my results page.
As the questionnaire got longer, folks have been indirectly hinting (yes, that's two layers of indirection) at a bit of intelligence to mitigate the length of the questionnaire. In particular, after reading some accounts of heroic lengths to complete it, I was concerned with what a person would do if they were interrupted, either by circumstance or by an error in submission. The typical time to complete the questionnaire is in the one hour range, and it TOTALLY SUCKS to loose the form data because you forgot to fill a necessary text field, or because your significant other requires you at the dinner table NOW (I have to beta test the thing, so believe me, I know, more than anyone). The solution was to use cookies to store which inputs the user entered.
I hated doing it, because I am suspicious of cookies. But this was the best way I could come up with to at least recover from a submission error. It also allowed the user to do the survey offline, and with a few "Save Progress" buttons sprinked about, it could also mitigate the effects of a crash or similar, again without relying on an external server. But let me know if you think my using cookies is a bad idea.
Not long after I started using the web form, I started getting requests to add some "intelligence" to my selectors, in particular, to automatically skip questions when the answer to previous question obviously made them irrelevant. That pretty much required JavaScript. I opted to "simply" hide those questions rather than disabling their inputs and moving the focus to the next question mainly to see if I could do it. That actually doesn't seem to eat up much CPU power, though.
The "killer" is the other bit of intelligence that folks have been indirectly hinting at, a way to mitigate the length of the questionnaire. The solution was to use cookies (see the question on Cookies above), but the data that fills the cookies is generated by JavaScript code, and, boy, that does bog down.