How do you avoid selection bias, particularly in regards to frequency of comments? For example, a source of technical bias could be that it is a simpler process to log into a Disqus pseudonym than to authenticate with Facebook when posting. Usability is incredibly important in attracting repeat users, so if it is easier to post under a disqus username then it is not surprising that pseudonymous users post more frequently. Another source of obvious selection bias is the disqus branding: I expect people who have Disqus accounts are both more likely to frequent blogs, and are more likely to recognize that they can easily comment on a blog with a Disqus comment section. By way of contrast, Facebook users may not even recognize that they have the credentials to easily comment, and may be reluctant to share their credentials with your third party service except in cases where they have a strong opinion about a piece of content.

I can see the data on comment quality being more objective and interesting, but even here there is probably selection bias: if most of the indicators of quality come from Disqus accounts' appraisals, then you may be measuring whatever bias Disqus users have as regards comment quality from Facebook commenters.