Wednesday, April 6, 2016

Fuck Politics. It's Numbers Time

Anyone who knows me well, knows I love talking about politics. I hate that by its very nature, it's divisive and tears friendships apart, but only in as much as it means I can't talk about it all the time, with everyone I meet. So here's how I can talk about politics, without actually getting into the politics. Anyone who actually reads my blog knows I love to screw around with numbers. I am not a statistician, and a real statistician may look at my methods and lose their damn mind, but this is how I've chosen to look at the race.

The big questions are:

What does a state win really mean?
Which are the big, important races, and which don't matter much?
How can we compare state results?



What the news likes to talk about is who "won" each state and by what percentage of the vote. Both of these terms are pretty useless when what really matters is who gets the most delegates. What got me on this train of thought was the night of the Michigan primary. Yes, it was very impressive that he won, but I didn't quite believe it was the turning point people were saying it was. That night, the numbers got very confusing. Bernie won Michigan with 49.8% of the vote, in a state with 130 pledged delegates, while Hillary won Mississippi, a state with 36 pledged delegates, with 82.6% of the vote. Surely, Michigan matters more because it has more delegates, and surely Hillary's 82.6% would be balanced out by Bernie's 86.1% in Vermont. Is your brain swirling yet, like two trains leaving Chicago and New York, headed to each other at different velocities?

But if you throw out who "won" and by what percentage, you're left with net pledged delegates (since I'm not talking about super delegates, from here on, I'm going to use "delegates" to mean "pledged delegates"). Keep in mind none of the democratic contests are winner take all. Of Michigan's 130 delegates, Bernie won 66, and Hillary won 64, meaning Bernie only had a gain of +2 delegates. And even though Bernie won all the delegates in Vermont, that's only 16 delegates, to the net gain +25 delegates Hillary won in Mississippi. That means even Michigan + Vermont is still less than Hillary's win in Mississippi.

In terms of only net delegates, it doesn't matter that Hillary won 20 contests, while Bernie only won 16. What matters is how big the margin of delegates was. Hillary's biggest victories were Texas (+73), Florida (+58), Georgia (+44), and Alabama (+35), while Bernie's were Washington (+46*), Utah (+21), Vermont (+16), and Minnesota (+15). You could flip the total number of wins, with Bernie winning 20 contests an Hillary winning 16, and she could still be ahead by a sizable margin. But since she is both ahead in number of contests and largest margins, Bernie has a long road ahead.

* Washington still has 17 unpledged delegates, so Bernie's total in Washington could potentially jump as high as +63. I don't know their exact rules, but I think the most likely landing spot is around +55.

But what does that even mean? It may look like I'm just skewing these numbers toward Hillary, but let's look at where we are today:

Hillary Clinton: 1280 pledged delegates
Bernie Sanders: 1030 pledged delegates

Bernie had a great week March 22-26. Before that, he was behind by about 345 delegates, with 28 contests left to go. That means he needed to gain an average of 12.3 delegates per contest. As of   right now, with several delegates still left to be allocated, Sanders walked out of that week with an average of +15.8 delegates per contest, his best week of the campaign.

For comparison, Hillary has been averaging +19.9 delegates per contest, obviously skewed by her bigger victories, while Sanders is averaging +11.9, even after his big week.


As of today, all that matters is whether or not Bernie can close the 250 delegate gap remaining. It would be really easy for either side to see that number as insurmountable or minuscule. Starting with Wyoming, there are 21 contests left. That means Bernie has to win all 21 of the remaining contests by 11.9 delegates, and every loss makes that number go higher. So let's imagine a best case scenario for Bernie, he wins all 21 contests, and he can maintain his current average.

Unfortunately, Guam and the US Virgin Islands only have 7 delegates. That means if we take Bernie's current total and say he wins 100% in those territories, that then means he instead has to close a gap of 236 in 19 contests, which means in those 19 contests, he needs to win an average of 12.4 net delegates. To be clear, if he wins all 19 by no more than 12 delegates, he still loses.

To connect to what I said above, Bernie could potentially win 47 contests to Hillary's 20, and still lose. On the other hand, one big win, in one big state could close that gap considerably. So what I'm going to do is translate the remaining contests back into terms of "wins" and percentages, starting from the idea that only net delegate gain counts. Right off the bat, I'm throwing out the complicated delegate allocation rules. In some states, even a small win can result in a rather sizable bonus (see Washington above).

To maintain the idea that this is an optimistic look at Bernie's chances, I'm doing all the math based on +12, not +13, as he only needs 13 in less than half of the contests. Let's just assume he's going to clear the bar by a lot in some of these. So for the remaining contests, these are the percentage margins Bernie needs to clear to win the election (assuming Super Delegates will, as they usually do, flip to support the holder of the majority of pledged delegates).

STATETOTALGOAL%
Wyoming141392.8%
New York24713052.6%
Connecticut553461.8%
Delaware211780.9%
Maryland955456.8%
Pennsylvania18910153.4%
Rhode Island241875%
Indiana834857.8%
West Virginia292172.4%
Kentucky553461.8%
Oregon613760.7%
Puerto Rico603760%
California47524451.4%
Montana211781%
New Jersey1266954.8%
New Mexico342367.6%
North Dakota181583.3%
South Dakota201680%
D.C.201680%


In case you feel like checking my math, keep in mind, states with odd numbers of delegates, I rounded up to +13 because +11 isn't good enough, and you can't split a delegate in half. And there are still a few delegates in North Carolina, Illinois, Idaho, and most notably Washington left to be assigned, so this whole thing is within a margin of error. This is just to paint a picture.

Bernie is highly favored in smaller states, but to really win those smaller states, he needs to win in the 80% range. That is by no means impossible, but it also means he needs to be pulling resources from the states where he's the strongest. A full court press by the Sanders campaign could result in major wins, but a 65% showing in half of these contests, while impressive in a news story, won't be enough to win.

That means Bernie has to lower the margins in his strong states. He needs to focus on the bigger states to start evening the field. Right now, Hillary is polling at 53% in New York, according to CBS. Fivethirtyeight has Clinton up 53% in Pennsylvania. LA Times has Hillary ahead with 47% in California. Using Wisconsin, Michigan, and Minnesota as models, Sanders could win, but it has to be by enough. A win of +24 is like winning two states, but keep in mind, he has only topped +24 once. A win of +18 is like a state and a half, but he has only topped +18 twice.

But let's be super optimistic for Bernie. Let's say he wins Wyoming by +10, based on similarly sized caucus states. And let's say he ekes out a narrow win against Hillary in her home state, but it's tight, so he only wins by +15. Based on comparative size, let's say he also wins Pennsylvania by +58, as she did in Florida. That's a gap of 153 delegates in the remaining 16 contests, for an average of 9.5 per contest. No problem, right?

Let's swing the pendulum back a little. Between Pennsylvania and California, there are 9 other races. Coming off a huge win in Pennsylvania, he'd have a lot of momentum, but these are mostly Northeastern and a few Southern states, where she's favored. Let's say they split these races, which is still only giving her four small to mid-level wins in this model. And let's say he wins the ninth, Oregon, by +20, as he does well in the Northwest. That's 133 delegates in 7 remaining races, for an average of +19.9. That means he not only needs to win New York, Pennsylvania, AND California to win, he also needs to bring in the kind of numbers he has only so far achieved twice.

This scenario is a pretty optimistic one for him, and it still doesn't look good.

So to Bernie supporters, I am not saying lose hope. Fighting insurmountable odds is what America is all about. I am not saying give up.

But if you're going to say he still has a chance, at least know what that means. And to Hillary supporters, take heed. It's a lot closer than you think. But, again, to Bernie supporters, if you're wondering why the media is treating her like she has it in the bag, maybe it's this, not a conspiracy. But yeah, Hillary supporters, Bernie is for real, and you need to be worried.

UPDATE 4/9/16: Sanders won Wyoming with55.7% of the vote, but since they both walk away with 7 additional delegates, his net delegate gain is zero. That means he still has a 250 delegate deficit, and only 20 races to come back (18 sans Guam and the Virgin Islands). That means he must now hit 13.1 delegates per contest. Based on how I had to round the last table, this new average only affects the following races.

STATETOTALGOAL%
Wyoming14755.7%
Rhode Island241979.1%
New Jersey1267055.6%
New Mexico342470.5%
North Dakota181688.9%
South Dakota201785%
D.C.201785%

UPDATE 4/17/16: Going into the NY primary, just want to make sure my numbers are accurate. Some results were still coming in, still are actually, so Bernie's lead is actually around 200. I'm saying around because I can't find multiple sources that agree. Nate Silver says 206, NuffPo says 194, so 200 sounds safe. With 18 contests remaining, that's 11 delegates per contest. That means Bernie can stay on pace with 129 delegates, not 130, so he needs 52.2%, not 52.6%.  I'll do a for real whole new count after Tuesday.

Labels: ,

0 Comments:

Post a Comment

Subscribe to Post Comments [Atom]

<< Home

Newer Posts Older Posts