I wrote that changes in the Hispanic vote wouldn’t have had much effect on the 2008 election and, with the exception of Florida, might not be as important as people think in 2012 either.

Yair responds:

I don’t know if I agree with your conclusions. It’s true that no state truly “flipped,” but many states became a lot more competitive and some become statistically indistinguishable from 50/50. Some specifics on this:

1. In the first scatterplot (excluding hispanics), you mention that NM, FL, IN, and NC come within 1% of switching and that changes to the model might make them flip. My interpretation is that this is a big change … it would significantly alter the strategies of the campaigns, the media coverage, and everything else. It would also likely have an effect on other groups in the state, but this is tough to determine statistically and in fairness you assume everything else is held constant.

2. In the second scatterplot (halving hispanic McCain vote), TX AZ and GA all become competitive (as you mention), but also NM and to some extent FL become uncompetitive. Considering the number of electoral votes in TX and FL, this is potentially a massive change … imagine if Democrats no longer had to worry about FL and Republicans started having to put a ton of money to keep TX’s 34 electoral votes. This becomes even more pronounced in the final scatter (increasing 20% and halving McCain).

The point is, even though no states truly flipped, the playing field would hugely shift and have a ton of consequences on the campaign. In the post you use the assumption that nothing else would change, but my guess is that a less academic audience (i.e. 538) wouldn’t like this assumption (I haven’t read the comments yet). Because of this and the fact that there is modeling error, my interpretation is that these changes are more meaningful than described.

Good points; see also my other thoughts here.