Thanks, Glenn. In fact, I think they are equally important, and require different messaging. The Bill White campaign, for example, appears to have hit on a simple resonant theme with "Perry's in it for himself," hammering that home with the rental mansion spot. This message exists in a simple frame that activates the strong throw-the-bastards-out feeling among low interest voters.

The ambivalents, though, have been the key to gains over the past two cycles, and offer exceptional opportunity again this year. If we get the right message delivered, Democrats can expect to win at least 55% of these voters. Will that be enough? Certainly enough to make it close, but to win statewide a Democrat will not only need to win a majority of these groups, but will also need to see Latino participation rise to about 20% of the total vote.

In short, it's a complex formula--gain persuasion, stop-loss persuasion and effective, targeted GOTV.

Exceptionally insightful, as we've come to expect from George Lakoff. Of particular importance is his identification of a voter group he calls "bi-conceptuals." By whatever term (in my work I call them "ambivalent" voters), they are a crucial part of the electorate, often contributing 15% to 20% to the final vote total. These voters made the Democratic comeback happen beginning in 2004, as they grew increasingly uncomfortable with the Bush/Perry frame, even when they received little focused attention from Democrats. This year they are crucial to Democratic candidates, who have substantial opportunities for gain if they can make their message resonate inside voter frames.

A couple of comments:

The polling issue relates as much to the "instant" nature of polling as to survey questions. Most pollsters do not match respondents to the voter file, and most don't consider respondents' measured or interpolated voting behaviors over time. Considering a persons current response in the context of a time series is, I believe, the quickest path to ambivalent voter identification. Although they are not the only voters with behavioral changes, they are easily distinguished from impulsive voters, the other big group in this category.

Another necessary consideration is the interest level of individual voters (and the household where they live), which I measure by turnout probability. Ambivalent voters exist almost entirely within the top probability households--very few people spend the energy required for ambivalence on matters that don't hold great interest. Most of the remaining voters in these top-probability households have taken sides already, and typically fewer than 10% are persuadable in the sense of responding through simple frames.

As voter interest wanes, however, pure persuadables increasingly dominate. In the 2010 General Election, I expect 15% of the total vote to come from households where the best voter is less than 33% likely to vote, putting low-interest voters on par with "bi-conceptuals" or ambivalents to the outcome of the election. These voters pursue a shallow understanding that is quickly within the grasp of their ready frame.

Lakoff has given us some persuasive advice to avoid the small message and hammer home the large themes that resonate. We'd all do well to heed this advice.

Leland