Greetings all,
There is also a trend that is represented in your posts as well. Those replies that received the highest rep (Re:Some Topic Posts) are also those that are in a Front-Paged thread. Getting on the front page has always seemed to get more ++ votes for a thread. Your other high rep postings are those that contribute code to the community. At least that is the trend as I type this.
Personally I look for a few things when up-voting a node.

1. Clear and well thought out questions (succinct code snippets, illustrative errors and explainations of methodology).
2. Effort (on the part of the OP) to answer the question prior to posting, this includes what research they have done as well as any attempts to identify where the holes in their understanding lie.
3. Replies with code snippets and explainations.
4. Replies with code critiques and suggestions bolstered by either documentation or benchmark findings.
5. Replies with gently guidance to relavent documentation, not just a read the docs link to a module but rather where to look in concert with what to look at/for.
6. Lastly: willingness to admit mistakes, misunderstandings or personal short-comings.

Of course this mainly applies to the Seekers of Perl Wisdom portion of the site, but those are the majority of the postings I read and reply to.
Other than that the effort that goes into code contributions in general I feel warrants up-voting (given that the code actually works and the formatting isn't egregious). Obfuscations never fail to amaze and impress me so most of the ones I read and run I up-vote ;).
You do bring up an interesting question though, that being what the overall community trends are in node voting based on thread content? It might be fun to look at from a purely analytical standpoint.

-InjunJoel