Framing, Damn Framing, and Statistics

I am not the least bit opposed to framing in principle. However, framing gurus should take care not to feed the (fairly common) opinion that framing equals paltering. Matt Nisbet fails that task with this graph:


Nisbet is showing this to support his thesis that there are "two Americas" and that only the one that agrees with Al Gore politically has become more concerned about climate change in recent years. But that is not a valid reading of the graph unless the relative numbers of the Democrats and Republicans have been stable over the period, and data from the same source suggest that they haven't. There are fewer Republicans now than a few years ago. If we assume that most of the party-switchers are moderates and that most moderates are concerned about climate change - and both assumptions seem reasonable - then the remaining loyal Republicans are, on average, more partisan and more dogmatic than before, and the divergence of the two lines in the graph just illustrates that corollary.

There are more conclusions that can be drawn from that graph, but none of them is about Al Gore's ineffectiveness.