Brad DeLong wrote an interesting article a few weeks ago, Josh Barro Makes Me Aware of Niall Ferguson. But as an aside, he noted, “[Hillary Clinton] is behind by roughly 22 percentage points among southern whites — who are, increasingly, acting like a very separate ethnicity (and 40 percentage points behind, perhaps, among southern white males — suggesting white females are close).” I immediately thought, “Ah, a chance to do some math and nail down those southern white females!”
It should surprise no one that DeLong’s offhand guess turns out to be roughly right. But I still think it is interesting to go through it. Why? Because math is fun. And yes, this is not pure math; it is applied math. And it is only pure math that gets me really excited. But nonetheless, I’m pretty excited. Just how likely are southern white females to vote for Hillary Clinton?
Let’s Get Mathematical
Now, I don’t know where he gets the statistics — maybe from this CNN/ORC poll (PDF). And I’m not terribly happy with him throwing in “perhaps” when discussing Trump being ahead by 40 percentage points among southern white males. But it all sounds about right. And I’m more interested in the math anyway.
Math nerds will immediately notice that we can exactly calculate the percentage of southern white females who will vote for Hillary Clinton. In order to do this, we need to define a few variables:
- F: percentage of southern white females who will vote for Clinton
- M: percentage of southern white males who will vote for Clinton
- T: percentage of all southern whites who will vote for Clinton
- P: percentage of all southern white voters who are southern white females.
The Equation Enters
Thus we can set up the following equation:
P×F + (1-P)×M = T
If Hillary Clinton is down by 22 percentage points among all southern whites, that means she is getting 39% of the vote. This is the value of T. Similarly, if she is down by 40 percentage points among southern white males, she is getting 30% of the vote. This is the value of M. For the moment, let’s assume that men and women vote in equal numbers. So P = 50%. Thus, the equation becomes:
F = (T – (1-P)×M)/P = (0.39 – (1-0.5)×0.30)/0.5) = 0.48 = 48%
That looks good for DeLong’s prediction. But we aren’t done.
The Plot Thickens (But Not Much)
It turns out that in presidential elections, men and women do not vote (PDF) at the same rate. In 2012, voter turnout for men was 59.8%. For women, it was 63.7%. In addition, more women are registered to vote. In 2012, only 71.4 million men were registered to vote, while 81.7 million women were. I see no reason why things should be different for southern white females.
Doing a little simple arithmetic, we see that in 2012, 43 million men and 52 million women voted. (Note: this does not add up to the 127 million who voted. I assume this is due to the sampling and that the percentages are still correct.) This means that women represent 55% of all voters. If we assume the same is true of southern white females, we get the slightly more complicated equation:
F = (T – (1-P)×M)/P = (0.39 – (1-0.55)×0.30)/0.55) = 0.46 = 46%
Clinton Doesn’t Need Southern White Females
Clearly, Hillary Clinton could stand to do a bit better with southern white females. But it was never a question but that she was going to the lose the white vote overall — much less in the south. And note that this really isn’t about Hillary Clinton or Donald Trump. Earlier today, I saw a list of polls pairing Clinton vs Trump over the last year. The polls really haven’t changed that much. There are blips when Trump said something outrageous or when the media made a big deal out of any number of Clinton non-scandals. But it’s been pretty consistent.
In the end, Hillary Clinton will do about as well as any other Democrat would have. And as amazing as it may seem, Trump will do about as well as any other Republican. Because when Trump manages to even win southern white females, you know it isn’t about him. They’re just voting their tribe — like most people.
Still: math is fun!