Thank you very much for the very interesting study. The truth is that your findings are actually fairly correct. The problem is, however, that the vast majority of those who reply, do not use the hashtag. They just don't. There are usually more, much more replies that the ones with the hashtag, although relatively speaking it's still not what you probably would expect.

A more accurate average number of replies can be estimated if you consider how many users participate in a poll:

http://twitter.com/#!/smashingmag/status/62126045498839040

This poll was published on a low traffic weekend day which is clearly reflected by the number of participants.

Polls published on weekdays have of coruse a better response ratio:

http://twitter.com/#!/smashingmag/status/55762312174370817

http://twitter.com/#!/smashingmag/status/54897184239398912

http://twitter.com/#!/smashingmag/status/54221240826732544

We never get over 1000 replies to a single tweet, but 300-500 replies is a very reasonable number, especially if it's accumulated over a couple of days.