I think it's wonderful that you're trying to get more feedback in a systematic way, but you're wayyy of the mark when it comes to assessing both the reliability of generalizatiblity of your data. I'm not sure how many Quama users you have, but unless it's *very* close to your N of 102, then the confidence interval (margin of error) you're left with is enormous, and that's even assuming you had a random sample, which you don't. Trying make claims about the "general blogsphere" from a non-random survey of Quamana users--drawing solely from that user base itself itself an enormous selection bias--with an already inadequate sized and nonrandom sample is... well, bad idea theater. Again, the desire to get evidence-based learning is fantastic. And there are some limited but I bet helpful lessons you can draw from your data. BUT, it's important to remember that the only thing than being unsure of anything is mistakingly thinking your sure of something you're actually entirely wrong about.