I’ve been meaning to write this for a while, as the following theme has been repeated at a few different points during this election cycle, most prominently in many articles written in late August and early September: this will be the last presidential election where Republicans try and win with only the white vote. See, for example, Brownstein in the National Journal, Chait in New York Magazine, and Atkins here at Daily Kos. There were many other articles written on this subject, and they all seemed to be based on the comment by a Republican operative to Brownstein that “This is the last time anyone will try to do this.” The general conclusion from these articles is that as the white vote shrinks in terms of the percentage of the overall vote, Republicans will no longer be able to win with only the white vote. From the title of this post, you can see I have a slightly different take. The evidence suggests Democrats are not far from winning the electoral college with ONLY the white vote.
Looking at CNN exit poll numbers from the previous two elections (2004 and 2008), and assigning states based solely on the white vote (lets call these white EVs), Kerry won 65.5 white EVs in 2004 (Kerry and Bush tied in terms of the white percentage in OR, and although one of them obviously won the white vote, I just gave them each half of the OR EVs), while Obama won 222 white EVs in 2008. Obama lost the white vote in NJ by 1%, in MD by 2%, and PA by 3%. Thus, he was a hair from having 268 white EVs. Moving forward Republicans will certainly have to significantly improve their standing in the minority vote if they expect to be competitive, but the reason isn’t solely because the white vote is shrinking in terms of percentage of the overall vote. The fact that the white vote alone is close to delivering an electoral college win for Democrats is an equally large (perhaps larger) cause for Republican concern. This may not be the election where the Democrat wins the electoral college with only white voters, but the current trajectory suggests that we will get there soon.
This post is related to a few themes that are more current. According to the CNN exit polls, Kerry lost the white vote 58-41, while Obama lost it 55-43. Now a +5 movement towards team blue is always nice to see, but the 55-43 drubbing Obama suffered among white voters should not have resulted in a 316-222 white EV win for McCain – the difference should have been much larger. Compare this to the real election, where Obama won by a much smaller 53-46, yet won the electoral college by a much larger 365-173. This illustrates two important points that are relevant to this year: (1) the white vote is far from homogenous throughout the country; (2) the white vote at the state level does not move in concert across all states – in fact, far from it. The LA white vote went from +51R in 2004 to +70R in 2008 – a +19R movement. The AL white vote went from +61R to +78R – a +17R shift. States like these two helped keep Obama’s white vote margin relatively close to Kerry’s margin, even though he made gains among whites in most states compared to Kerry (whites in 39 states + DC shifted D in 2008 compared to 2004, whites in three states stayed the same, and whites in 8 states shifted R). What about the other confederate states? VA, NC, and FL don’t seem to fit the confederate picture anymore – at least based on their white vote. NC white voters were +29R in 2008, VA whites were +21R, and FL whites were +14R. FL stayed virtually the same as it was in 2004, but NC and VA whites shifted significantly blue by +17D and +15D respectively. SC, TX, and GA were +47R, +47R, and +53R, respectively, in terms of white vote in 2008. The white vote in TX and GA changed very little from 2004 to 2008, while the SC white vote actually shifted +9D. AR was +38R in 2008, and this was a shift of +11R from 2004. This leaves MS, where the white vote was already +71R in 2004, and moved to +77R in 2008. I noted that VA, NC, and FL diverged from the other confederate states in 2008, and this is because white voters in states like OK, UT, WY, AK, ID, and TN all voted more Republican. The top 8 Republican white vote states in 2008 were AL, MS, LA, GA, TX, SC, OK, and AR. Only OK was not part of the confederacy (not a state in the 1860s), and AR was almost tied with OK among white voters.
It is interesting to note that in 2004 MS was the only over +70R white vote state, but LA and AL white voters joined them in 2008. AL isn’t that surprising since they were already close at +61R, but LA went from being almost identical to GA, TX, and SC to being almost identical to MS and AL, in terms of the white vote. What if the white vote in states like SC, TX, GA, and even AR, all become more like their southern brethren in LA, AL, and MS – over +70R? This wouldn’t change the electoral college vote at all, but it would certainly change the national percentages – especially if Obama doesn’t increase his standing among whites in the non-confederate states. It wouldn’t surprise me if the closer national polls we are seeing this year, compared to 2008, are partially due to the whites in states like SC, TX, GA, and AR moving to +70R or more.
Last night, in his polling wrap, Steve Singiser noted that we might be heading to a rare split between the electoral college vote and the popular vote. Many other polling experts are saying the same thing (Silver, Cook, and Blumenthal have all written on this topic). But are we realizing how rare an event we could be heading towards? This could be an election where one candidate wins by a point or two nationally, and it might even be Romney (but probably not), yet Obama delivers a convincing electoral college win of 332 EVs (everything he won in 2008 minus IN). Furthermore, there is a good argument that this might constitute the new normal, at least for the next few presidential election cycles (the Southern Strategy failed outside the deep south, but at its strongest point ever - even getting stronger - within the deep south). Such an outcome has the possibility of rending national polling moot - what might look like a close race in national polls would in fact be a comfortable Democratic lead at the state polling level.