The asterisk* would be the question of how many minorities and students are planning to vote, and for whom, with the question of how many will be thwarted.
This is just the break the McCain… oh.
Not to my knowledge. I was reading something about the demographics a few weeks ago. I’ll see if I can dredge it up. It was thorough.
Age is probably the most variable of the demographics. But to put that in perspective, that means the 18-29 year-old vote has varied from 16.7-18.3% in the last four presidential elections.
Does a 1.6% swing in 18-29 year-old voters affect the outcome? Of course it does. But it’s not like Obama is winning that group 80/20 or something. It’s more like 60/40 on a good day. So we’re talking about tenths of a percent.
By contrast, the minority vote has increased an average of 3% a year since 1992. And Obama wins that group by 70-80%.
So even if the youth vote reverts to 2004, it’s more than offset by the increase in minority voters.
Those are just historical trends, of course, and trends don’t necessarily continue. But is there any real reason to think that the 20-year-long increase in minority voting is gonna flip around this year?
(Voter ID is a different problem, but not really one in the key states yet.)
Of course there is. It makes sense to assume that the opportunity to elect the first African American president would engender a lot more excitement than the chance to reelect him four years later, especially after many will have perceived that electing him the first time did not produce a dramatic change in their lives in the end.
That does not mean that this will actually happen. But it’s certainly a real reason to think so.
[On another note, I don’t know how the quoted Obama people managed to figure out that the Romney people were banking on this possibility, and I rather suspect that it’s just spin.]
I think you should re-read my post. 2008 was not an aberration from the long-term trend. In every election since 1992 the non-white share of the vote has increased, and you can plot a pretty straight line from 1992 through 2008.
I had read your post.
There’s no inherent reason for the increase in minority vote to be linear. So it makes sense to assume that the increase in 2008 would not have been linear were it not for additional excitement of the first AA president.
Of course there is an inherent reason for the increase to be linear. The long-term demographic trend is more non-white eligible voters. The turnout rates are more variable, but the data suggests that the first variable swamps the second, since the actual composition of the voting population has been very steady and predictable since 1992.
Why would you assume that 2008 would have reversed the long-term trend if not for Obama? I see no basis for that odd assumption.
I didn’t say that 2008 might have reversed the trend - only that the increase might not have been linear, and that it ended up being higher than what it otherwise might have been due to Obama.
Minority groups are not a monolithic group growing at a constant rate. They include Blacks, Hispanics and probably others, each of which is growing at a different and changing rate. If the overall increase of the total (as a percentage) has been linear, that would be a coincidence. If you subtract 2008, that leaves only three elections, which is too few to create clear “linear” pattern. (Perhaps you can link to your source data.)
You’re missing the point here by playing with the language in posts like they are contracts or Bill Clinton’s deposition testimony. So I’ll just use your own words to make my point.
What basis do you have for thinking that “it ended up being higher than what it otherwise might have been due to Obama.” Is this just a circular argument in which you assert that Obama increased the share because he is Black, and your proof is that he increased the share because he is Black? Or do you have some actual reason for thinking the share would not have kept up with the trend?
Linear or nonlinear isn’t the point at all. The point is whether it is increasing or decreasing, and whether each year’s increase swamps the variability of turnout – as it has for two decades. If your position is that the non-white share of voters is increasing more slowly now than it was in the last 20 years, do you have any support for your position?
Here’s a good source for exit poll data: http://www.dimpledchad.info
There’s nothing circular and you’re confused.
I’ve not claimed to have proof of anything. Only that there is “real reason” to believe that the first african-american candidate might have engendered more excitement in the AA community than would be present for round 2. I’ve not tried to prove this in any way, circular or otherwise. It makes sense to me (as a possibility). YMMV.
Linear or non-linear is exactly the point. You are attempting to prove that minority turnout in 2008 was not any higher than what it otherwise would have been by pointing to some alleged linear pattern. I’m saying that this linear pattern is a coincidence at best, and that we don’t know what the minority turnout would have been in 2008 absent the unique circumstances of that race. And, absent knowledge, there is “real reason” to believe that it might have been lower.
Thank you. What that source shows is that your claim is incorrect. Here are the number for “Distribution of Voters in National Exit Poll by Race, where Race = “White”, Presidential Election Years”
1984 87.7%
1988 85.1%
1992 87.4%
1996 83.0%
2000 82.3%
2004 79.0%
2008 75.7%
As can be seen, the minority share has been steadily increasing, but the increase has not been linear. Mostly in that from 1996 to 2000 it increased by 0.7%, versus 4.4% compared to the prior election and 3.3% in each of the subsequent ones. (Which is leaving aside that the 1992 start year was apparently chosen to eliminate the fact that minority share decreased that year.)
In sum, there is no reason to discount the possibility that the minority percentage of the vote in 2008 might have been lower than 24.3% by some unknown amount.
And, to be clear, your reason for thinking that this enthusiasm translated into a significant departure from the overall turnout levels is purely speculation that is contradicted by the trendline.
That’s not an argument for why it matters if it’s linear or not.
This is basic math. If you plot the points, you see a trendline showing that the share tends to increase. If you extrapolate using any reasonable method, your likeliest scenario for predicting the 2008 vote based on the preceding elections is going to be a modest increase in non-white turnout. Which is exactly what happened. It is also the case that your likeliest scenario for predicting the 2012 vote based on the preceding elections is going to be a modest increase in non-white turnout.
Now, the exact nature of your extrapolation is going to alter the likelihood you assign and the variance of your prediction. But any reasonable method is going to conclude that the likeliest outcome was a modest increase.
Since I apparently cannot paraphrase your wording without being accused of misrepresenting it, I’ll just leave that paragraph there its original wording for all to contemplate.
Do you have any sort of math background?
Brother, the problem here is English, not math.
If your position is that there is sufficient variability in the turnout numbers that the 2008 result was just as likely to show a decrease as against 2004 as a modest increase, then we can proceed with the actual math to test your proposition.
But something tells me you’ll disavow that position once I’ve stated it.
I’m not sure. For example, I think you may be unfamiliar with math terms like “linear increase”, which in retrospect you appear to have used to mean “consistent increase”. This type of thing causes confusion, since I assumed you used the term correctly and responded accordingly.
Plus, if you’re not used to math concepts, you will have a harder time understanding them, even if the English is OK.
I’ve never said this.
My point was that even if minority voting % would have in any event been higher in 2008 than it was in 2004, it could have been bumped higher yet by the Obama factor, and thus might revert back to form in 2012.
Here’s an illustration of the concept. Suppose absent any additional excitement in 2008, the numbers for 2004/2008/2012 might have been 21%/22.5%/24% (for example). Instead, the number went to 24.3% in 2008, but might revert back to 24% in 2012. (The numbers are illustrative.)
I don’t see how your position, as you now explain, isn’t trivial. Of course it’s possible that Obama increased the non-white turnout in 2008. Anything’s possible. The interesting and relevant questions are:
(1) Did 2008 diverge from what we would have extrapolated the result to be based on the pre-2008 data?
(2) How robust is that extrapolation for predicting 2012?
I won’t hazard guesses at your position on those two questions, but I’d be curious to hear them. I promise I’ll try to keep my mathematical terminology accurate so you don’t get confused about my point again.
That’s relevant and non-trivial, because if he increased it in 2008, it’s possible that some of that increase might not be present in 2012.
Probably.
No idea.
But I don’t think it’s necessarily irrational to assume that an extrapolation based on the pre-2008 data would yield a lower result in 2012 than the actual 2008 numbers, which would produce a decline in 2012 versus 2008.
It doesn’t seem especially likely to me, having seen some actual numbers, but it’s possible and in any event has some rational basis.
My point is that saying that it’s possible doesn’t say much. It’s possible that the electoral composition of 2008 would have been 65% white except that white women love Joe Biden’s blue eyes. Saying that some unknown counterfactual is possible doesn’t tell us anything about what is likely, or about what we should expect for 2012.
Why?
What result would you extrapolate from the series 2.6, -2.3, 4.4, .7, and 3.3?
What is the most probable next number, and why? Are you saying there’s just not enough information to say anything more robust than that 3.3 was just as likely as 2 or -1?
Obviously, when you only have a limited data set, there’s a whole bunch of *rational *extrapolations, in the sense that they fit the data and we can come up with some kind of ad hoc justification for them.
But the far more interesting question is not what’s rational but what’s probable.
Do we think the weird curves that would fit the data but that would show a 2012 decline are likely to reflect reality?
Alright, here ya go. From the data posted up-thread with a linear fit:
Using the entire data set gives a fit of y = -0.4759x + 1032.8 where x is the year and y is the percent of white vote. The R-squared is 0.8781.
Using the set excluding 2008 gives: y = -0.4021x + 885.96 with R-squared of 0.8187.
So, plugging in 2012 in each we get a predicted white vote of 75.3% for the “all-years” fit and 76.9% for the “excluding 2008” fit. As an aside, the “excluding 2008” model would have given a non-white turnout of 78.5% in 2012 (rather than the 75.7% actually found).
Again, that’s just using a rudimentary Excel linear best bit, but it at least somewhat backs up the possibility that 2008 was out of line with the previous trend. Either way the expected range for the 2012 white turnout should probably be between 75% and 77%.
As usual, fast-and-dirty math may be wrong.
I would add one thing that’s pretty clear from the data - the trends-lines are much more robust if 1992 is thrown out as the “outlier”. The R-squared for each fit goes up to 0.93 and 0.94 respectively by removing that data-point.
The new projections, however, remain roughly the same (between 75.3% and 76.8% white vote share).
Thanks Jas09.
What result for 2012 if you include in the model the expected result for 2008 (i.e., 78.5 or whatever it would have predicted)?
(I think neither equation captures that result since applying the 2008-free equation would be different from applying an equation that assumes this result for 2008, but I could be wrong.)