Related to the share of voters who are white: have you or any of the other bloggers ever done an analysis of "malleability" of voters by race/ethnic group? Not only are hispanics (and blacks) very small voting blocs, they are so consistent in their voting patterns from election to election that it makes it especially not worth it to try to make inroads. If you are a politician, you have to go where the pickings are easiest and I suspect that is the whites.This is the mental framework I operate under, specifically due to how reliably one can count on the overwhelming majority of black votes to consistently go to Democrats. On the other hand, the GOP is the de facto white party in part because it regularly garners more white votes than the Democratic party does.
If this is true, you could almost say that white voters are the most diverse in their views but we all know whites are never diverse. Ever.
To test this assumption, let's look at exit polling results by race for Presidential elections from 1992 (as far back as they are easily accessible) through 2008. For a rough measure of malleability, I take the Democrat-Republican voter gap by race (expressed in percentages) for each presidential election and the presidential election preceding it (with the exception of 1992), determine the absolute difference in the gaps between each of the two consecutive elections, and average these over the time period under consideration. The higher the figure, the more variation there is in voting behavior in presidential elections according to exit polling data* (which aren't always reliable):
It looks like that assumption needs some qualification and refinement. The 'old stock' whites and blacks haven't bounced around as much as the more colorful yellows and browns have over the last several presidential elections. Asians actually supported the GOP when the first Bush and Dole were its standard bearers.
What I described as a measure of malleability might more accurately be conceived of as a measure of fickleness. That whites appear to be less malleable than non-whites is probably attributable in varying degrees to the following list of non-exhaustive factors: Major party platforms are, despite the constant braying about need to appeal to Hispanics at all costs, are most attuned to the sensibilities of whites since that's where most of the votes are; exit polling data on non-whites, especially Asians and Hispanics, are less reliable than the data on whites are due to small sample sizes and, more generally, the smaller the group, the larger the margin of error is going to be; Ross Perot was a significant force in the 1992 and 1996 elections, and most of the votes he received likely would've gone to Bush and Dole if he hadn't run; white concerns tend to be more abstract and philosophical than NAM concerns are; and the ethnic compositions of the Hispanic and Asian categories have changed more over the last couple of decades than the compositions have changed among blacks and whites.
It should also be noted that a 5 point swing in the white vote still represents a larger absolute change in the total count (which is, after all, what really matters) than a 17 point swing in the Hispanic vote does. Indeed, even today, with fewer than half of births in the US to non-Hispanic whites, a 5 point swing among whites is worth more than a 7 point swing among blacks, a 17 point swing among Hispanics, and a 17 point swing among Asians combined.
* For data sources of each election cycle, click on the corresponding year: 1992, 1996, 2000, 2004, 2008.