I'm fairly certain now that an average is calculated for every site which is then compared to the average for your niche and/or type of site, which produces your Panda score (which is then used to adjust rankings produced by the main algo).
I think the negative effect on Page A of linking to Page B can be due to how the link text affects the content on Page A. If it introduces words that result in Page A gaining new badly targeted traffic then that harms the metrics of Page A and can result in a drop in rankings in between Panda's (i.e. it's nothing to do with Panda). I don't think it's connected to the quality of Page B. If Page A is a high traffic page however, the negative effect on metrics could impact the whole site (meaning less traffic at the next Panda).
I went through a process of removing non-selling products from categories and saw rankings improve within 1-2 weeks which, at the time, I thought was because we'd reduced the amount of low quality pages accessible from that category page, but now I realise it's because it produced a more focused page which removed some badly targeted traffic that was producing bad metrics for that page. Basically, bounce rate improved because it left us with more tightly targeted traffic, so rankings improved.
We've gained 40% traffic since the last Panda, focusing on improving metrics on our highest traffic pages. A few things are telling me this is a much less ominous issue than I first thought.