# Clinton/Sanders expose American innumeracy

So Clinton and Sanders were in pretty much a dead heat in the Iowa caucuses yesterday, with Clinton edging out Sanders by a few votes. Now I’m watching people freak out over a couple of stupid facts.

1. A few of the caucuses were settled by flipping a coin. Yes? So? The votes were tied. The rules require representatives to be selected. A coin flip is a fair way to settle which candidate will be represented, when there is a tie. I have no problem with using a chance distribution to decide, but some people are just horrified at this ‘primitive’ way of making a decision. How else do you propose to do it? Trial by combat?

2. Most annoying are the people who are shocked that there were 6 precincts where there was a tie, and in all 6 the coin flip favored Clinton. Again, that’s to be expected, that occasionally you’ll get a run of consecutive identical results. It would be curious if you never got a run of 6. But now we’ve got all these reporters earnestly explaining trivial outcomes.

The Iowa Democratic caucus vote count was so close last night that at least 6 precincts were decided by flipping a coin — an obscure procedure in the Iowa caucus bylaws. And in all 6 instances, the last remaining county delegate went to Hillary Clinton. Winning 6 coin tosses in a row is extraordinarily rare, and only has a 1.6 percent probability of occurring. As journalist Ben Norton explained, that’s broken down by calculating (1/2)^6, which is 1/64 — or 1.6 percent.

This needed explaining? No wonder casinos and lotteries do so well — and no, that’s not an extraordinarily rare frequency.

Maybe this is why so many people have difficulty grasping evolution. Small probabilities with many trials adds up to a significant likelihood.

3. It is absurd to try and spin this into a ‘win’ for either side.

If Bernie Sanders had won half of the coin flips and split the six county delegates three and three with Clinton, he would have finished at 698.49 delegates to Clinton’s 696.57, effectively giving him an Iowa victory. According to a live map of all Iowa precincts, Clinton has a razor-thin 0.3 percent lead over the Vermont U.S. Senator with 99.9 percent of precincts reporting.

The race was close enough that it could wobble one way or the other on the basis of a couple of coin flips. As far as I’m concerned, it was a statistical tie, and I don’t care how many ifs you concatenate to shift it into an imaginary win for your favorite candidate.

Also, it was Iowa, and a primary. Who cares? We’ve got two Democratic candidates who are equally appealing to motivated voters in one state. This does not settle the election, and a good case could be made that it is completely irrelevant to the final outcome, which is going to hinge on far more variables.

I swear, watching pundits overanalyze this one result is almost as infuriating as the Republican slate. Stop it.

Jebus. Look at this headline at The Blaze: Hillary Clinton Has The Most Statistically Improbable Coin-Toss Luck Ever. NO IT’S NOT.

1. qwints says

If Bernie Sanders had won half of the coin flips and split the six county delegates three and three with Clinton, he would have finished at 698.49 delegates to Clinton’s 696.57,
</blockquote

This is incorrect. MSNBC originally screwed this up last night, but corrected it. The 699 are "state-equivalent delegates" not county-delegates. The delegates awarded by coin flip are on the order of 1/10th of a state-equivalent delegate.

2. says

It is absurd to try and spin this into a ‘win’ for either side.

It shows that the democratic voters aren’t thrilled to death with either of the candidates.
Wouldn’t the interesting outcome not be “won by a hair” but rather “crushed the opponent”?

3. qwints says

*fixed*

If Bernie Sanders had won half of the coin flips and split the six county delegates three and three with Clinton, he would have finished at 698.49 delegates to Clinton’s 696.57,

This is incorrect. MSNBC originally screwed this up last night, but corrected it. The 699 are “state-equivalent delegates” not county-delegates. The delegates awarded by coin flip are on the order of 1/10th of a state-equivalent delegate.

4. How else do you propose to do it? Trial by combat?

So, like a political Pon Farr?

5. On the Today show this morning, I overheard a pundit characterize it something like this…

“Clinton won but they’re acting like it was a loss. Sanders lost but are treating it like a win.”

6. says

Don’t care. That people are fussing over results to 2 decimal places in such an intrinsically fuzzy process is simply absurd.

It was a tie. That’s the lesson.

7. Nerd of Redhead, Dances OM Trolls says

The news, of course, must have a winner. But I see it as a tie. Maybe it will still be competitive when the IL primary occurs 3/15, making it meaningful for a change.

8. UnknownEric the Apostate says

“Clinton won but they’re acting like it was a loss. Sanders lost but are treating it like a win.”

And Martin O’Malley is wondering if he should have taken his shirt off more.

9. petesh says

@8: Well, as long as he doesn’t tempt any of the others so to do.

10. says

I had to turn off the news. Basically told all the analysts to shut the fuck up. I am calling it a tie, which is what it is.

Discussion of the Iowa caucus circus begins here on the Moments of Political Madness thread, and it continues for about twenty comments so far. (Engaging in our own tribute to fuzzy math just for the hell of it.)

Winning 6 coin tosses in a row is extraordinarily rare, and only has a 1.6 percent probability of occurring

This bit of misleading math bothers me. The probability of each and every possible arrangement of wins is also 1.6%. What’s that? Side A came up on the first, fourth, and sixth time, and side B the other three? Wow, so amazing! Only a 1.6% chance of that happening! Someone must have tampered with it!

12. @YOB – Ye Olde Blacksmith

On the Today show this morning, I overheard a pundit characterize it something like this…
“Clinton won but they’re acting like it was a loss. Sanders lost but are treating it like a win.”

Well, the problem is that the only real things Clinton has going for her (which don’t also apply to Sanders), according to pretty much everything I’ve read by Clinton supporters, are:
1. People in her own party think she’s a centrist and so she will be able to accomplish things as a result, and
2. She was able to convince people that her nomination was inevitable.

#1 is nonsense; she’s pretty far to the right, and she has started to campaign as being against doing anything at all in response to Sanders actually, you know, wanting to accomplish things. (And, of course, the Republicans think any Democrat is a screaming Marxist anarchist terrorist raving salivating radical, so the idea that a Democrat will be able to get anything out of the Republicans, no matter how much they bend over backwards, is ludicrous right from the beginning. Look at how Obama kept getting the ACA watered down to try and attract Republican votes, and ended up not getting any at all — the only result of all the watering down was that the bill which eventually passed was much weaker than it otherwise might have been. Clinton will, if she wins, be more of the same.

So she was relying on #2: the inevitability of her nomination. And in a state which conforms pretty heavily to the demographic which goes for her within the party (aging, white, right-leaning, scared stiff of change) she still couldn’t pull off a decisive win.

One wonders what will happen once the Republican machine lines up behind Cruz and the Democrats realize (hopefully not too late) that Hillary is projected to lose against Cruz…

13. Rasalhague says

In the only sense that really matters, the result was a tie: both Clinton and Sanders walk away with the same number of convention delegates. The rest is just noise.

14. says

I was gonna dispute your math there, but the more I thought about it the more I realized I was playing into the gambler’s fallacy. You’re right, these are unconnected coin tosses. They’re each singular events with no relationship between their outcomes. So, it’s exactly 50% for every coin toss. Damn that was a good catch. I also agree with your analysis at the end of the day. Hillary has a lead, but it’s a statistically insignificant one. Might as well treat Iowa as if it didn’t happen unless that razor thin margin actually starts to matter somewhere down the road.

Even on the other side, 28%, 23%, and 22%. In terms of raw delegates, they’re all pretty close to the same numbers.

I don’t think Iowa even matters this election.

15. doublereed says

No one cares about the order. They only care about how many wins and losses there are.

16. Bill Buckner says

This bit of misleading math bothers me. The probability of each and every possible arrangement of wins is also 1.6%. What’s that? Side A came up on the first, fourth, and sixth time, and side B the other three? Wow, so amazing! Only a 1.6% chance of that happening! Someone must have tampered with it!

That is true (every outcome has equal probability) but your conclusion is false. If there were 30 coin tosses instead of 6, and she won them all (one out of a billion), surely you would not dismiss it by saying “it is just as likely as any other outcome of 30 tosses” even though that would be true. In fact, if she won 30/30, I’d bet the farm she cheated. Now we’re talking about entropy.

17. This whole thing is fuck annoying
Americans, get over and done with this shit. Seriously, the rest of the world simply can’t just close up for business for a year or so until you’ve figured that presidency out.
+++

This bit of misleading math bothers me. The probability of each and every possible arrangement of wins is also 1.6%. What’s that? Side A came up on the first, fourth, and sixth time, and side B the other three? Wow, so amazing! Only a 1.6% chance of that happening! Someone must have tampered with it!

I think the problem is comparing apples to oranges.
Sure, 1:A, 2A, 3A,4B, 5B, 6A is just as likely as 1A, 2A, 3A, 4A, 5A, 6A, but people don’t look at it like that. They look at it as All coins A vs all coins B vs a mixture of A and B, where the latter is, of course, much more likely.

18. vereverum says

How did they choose who would be the caller? If it were a random choice, the probability the caller would support a particular candidate is 1/2. Then the call would be 1/2. Then the specific outcome would be (1/4)^6 which is 1/4096 or .02%, which is so remarkable that obviously the Illuminati (Global Elite to the younger crowd) are pulling the strings.
.
Sorry about any spelling errors, but I have to post this quickly ’cause there’s a FEMA bus outside my hous

19. says

“and a primary”

Blasphemy! :o It’s called a “caucus.” It’s totes different. :)

On a more serious note (which doesn’t take much), I must note the difference in turnout. The one apparently had a 61-61 split and the precinct had 5 delegates! About 50 more people were at mine and we only had 4 delegates. A precinct in North Liberty apparently had about 130 more and, yet, only 4 delegates. Whatever, I guess…

20. says

How else do you propose to do it? Trial by combat?

Is that option on the table? Can I vote for it now?

21. MassMomentumEnergy says

Jockeying for Iowa delegates has only started:

http://thehill.com/blogs/ballot-box/267929-sanders-camp-gears-up-to-pick-off-clintons-iowa-delegates

Clinton squeaked out a slight victory on Monday night, winning 700.59 state delegate equivalents to Sanders’s 696.82, according to the state party. Those numbers represent delegates that the candidates will be able to send to the state conventions on their behalf to winnow down the 44 delegates representing the state at the Democratic National Convention.

But those delegates are not pledged, University of Georgia political science professor Joshua Putnam told The Hill.

The top-line totals only translate into the number of loyal delegates each candidate can pick to represent them at the state convention, delegates who may ultimately change their minds. Plus, the almost eight delegates secured by Martin O’Malley, who dropped out of the race Monday night, will also be up for grabs.

The important thing is that in America, tie goes to the underdog, aka the Rocky-rule.

A tie puts a huge crack in Clinton’s inevitability narrative, which is what Bernie needed with Iowa. Just like 2008, huge swaths of democrat voters in later states are pro-hillary just because they think she is going to get the office no matter what they vote, so might as well hitch your wagon to a winner. Americans love an underdog, but hate backing a loser. Now that Bernie has made this look like a real race, it suddenly becomes OK to back him.

22. stwriley says

I’d be alright with people objecting to the coin flip thing if it was about the need for it in the caucus process in general, but not the way that they actually have. I don’t find caucuses to be nearly as democratic a process as primaries, simply because you can have something like these ties that need to be decided by a purely random process like coin flipping. In a primary, no such random mechanism is necessary. If the vote is tied, then it’s tied and the delegate split is equal. If one candidate “wins” by 1 vote, then they win by one vote and even a statistical know-nothing can see the result for what it is: a very close victory that implies a very evenly split electorate. It’s only in a caucus that this kind of result that gets people upset about their candidate losing for other reasons than getting fewer people to vote for them can happen.

23. says

Maybe it does stop momentum. That could work, I suppose. In terms of raw delegates though, check out how little last night meant to the GOP: https://www.gop.com/the-official-guide-to-the-2016-republican-nominating-process/ The number in that chart last night decided was the CD or Committed Delegate category. For Iowa, that’s 12. So, if they’re rounding the way I think, Ted Cruz got exactly the same delegates as Trump and Rubio, 3. If they’re allocating differently, I guess maybe he has a 1 delegate lead. The At Large (AL) delegates who walk into the convention uncommitted either way, have 15 votes. If any of the candidates on that slate make a huge speech gaffe before the convention, (Trump), or ends up being really unpalatable to later states, (Cruz, Rubio), those 15 votes could split to just about any other candidate. If the other primaries and caucuses split out like this, you could have to put up with campaign ads well into May and early June. No candidate has to win, just remain competitive until the convention and have a great debate team for their committed delegate factions.

24. says

#1 is nonsense; she’s pretty far to the right

Not when you look at her senate voting record. There she was definitely on the liberal side of her party.

25. illdoittomorrow says

How else do you propose to do it? Trial by combat?

I’d pay good money to see the Republican nomination process replaced with a no-holds-barred cage match.
Then when it’s over and the winner comes crawling out, have one of the WWE actors body slam him into a table and take away the belt.

26. says

I have seen a certified Six Sigma Black Belt to fall prey to gamblers fallacy (during a course where he was supposed to learn me something) in the form of Martingale system. When someone who us supposed to be good at statistics and learn others how to use it in industry makes such a blunder, it is really no wonder that general public has even worse understanding of statistics and probility.

27. Bill Buckner says

Giliell, professional cynic -Ilk-,

Sure, 1:A, 2A, 3A,4B, 5B, 6A is just as likely as 1A, 2A, 3A, 4A, 5A, 6A, but people don’t look at it like that. They look at it as All coins A vs all coins B vs a mixture of A and B, where the latter is, of course, much more likely.

Exactly. In physics terms, the 16 micro states for 4 tosses (e,g,. hhhh, htth) are all equally likely, but the macro state (2 heads and entropy 1.8) is six times more likely than the macro state (4 heads and entropy 0). What is significant, in determining a cheat, is the probability (or entropy) of the macro state. That is why in poker AH, KH, QH, JH, 10H is more valuable than 2H, 7D, 4C, JC, 3S even though both hands are equally likely. And why you are not likely to be in the unfortunate situation where all the air molecules migrate to the other side of the room.

28. says

Correction:
Ack, I made a poor judgment error @24. Turns out GOP rules do commit those At Large and RNC delegates per party rules finalized in 9/2014. (That’s what I get for trusting wikipedia summaries, instead of inspecting the source material first). So it’s 8 Cruz, 7 Trump, and 7 Rubio. Still as statistically insignificant as the Dem side, though. Iowa really didn’t decide anything at this point.

29. consciousness razor says

The Vicar:

#1 is nonsense; […]

So she was relying on #2: the inevitability of her nomination.

Well, you shouldn’t discount the possibility that she can rely on both types of bullshit, since people can/do believe #1 whether or not it is nonsensical.

One wonders what will happen once the Republican machine lines up behind Cruz and the Democrats realize (hopefully not too late) that Hillary is projected to lose against Cruz…

Quite a few Democrats (like Clinton, I think) genuinely wouldn’t want Sanders to accomplish many of the items in his platform, although they more often say (probably dishonestly) that they merely think those things are impractical in our political environment. (As if that were a coherent reason for a supporter of a policy to vote against it, or against him … having too many good ideas or ones that are too good, which may not all be realized, because “our” opponents will probably resist some of them — so let’s defeat ourselves first so they won’t have a chance to do it!) I figure such people actually do prefer Clinton over Sanders on the substantial issues (or whatever their preferences happen to be about), and for that reason they very well could think the larger risk of losing the election is acceptable. Or they’ll be more optimistic about it than they have reason to be. But most likely, they won’t be thinking about that choice the same way as we do.

30. consciousness razor says

Bah. One little slash is missing, and the whole comment is fucked up.

31. mod prime says

Also, the probability that a pre-specified person would win six coin tosses is 1 in 64. There is also the probability of losing six coin tosses in a row. The chances that two opponents doing six coin flips and one of them winning them all is twice as high. Making it a much more mundane event. Of course this ‘extraordinary’ event is probably obfuscated by some of the opponents being different.

32. @#25: Kristjan Wager

#1 is nonsense; she’s pretty far to the right

Not when you look at her senate voting record. There she was definitely on the liberal side of her party.

Being the tallest of the seven dwarves does not make make you a pro basketball star; Clinton’s era in the Senate was the era during which the DLC (of which she was, of course, a member, her husband having been chair) had stacked the Democratic Party with right-wing “New Democrats” who were basically corporate shills giving occasional lip service to left-of-center ideals. Clinton is pro-war (not just Iraq; she was in favor of every military adventure, planned or actual, of the last twenty years), anti-environment (she’s surprised anyone is against Keystone XL, and is generally pro-fossil-fuels), pro-big-banks, pro-drone-bombing (which translates into “pro-creating-more-terrorists”), pro-war-on-drugs… her “feminism” is shown by talking about how “nobody likes abortion” and trying to make Planned Parenthood responsible for that faked video from Breitbart. Her “support” for universal healthcare consists of saying she’s in favor of it but claiming that any attempt to actually institute it shouldn’t be attempted. (Well, fair enough, the people who worked on it back in the 1990s agree it was her fault we didn’t get it back then, so maybe she’s right insofar as she shouldn’t do anything about it.) Her “support” for universal education likewise consists of denying any actual actions to bring it about… If that’s the “liberal side of her party” then it merely highlights how very far the Democrats have fallen.

33. gmacs says

@PZ

intrinsically fuzzy process is simply absurd.

I was at a caucus location last night. “fuzzy” is a very mild way of describing that clusterfuck. They didn’t have enough forms and stickers. It relies an a damned headcount. The turnout was so great we were sitting for hours in violation of fire code. The chair running it was getting rude and confrontational.

On top of all this, the people running the show didn’t seem to have a clear idea of how the rules of a caucus work. It’s an antiquated system that is unreliable and incomprehensible, and it can be easily gamed by a well-organized campaign. So the real result of the Iowa caucus is a lot of pissed off liberal voters.

34. Bill Buckner says

The chances that two opponents doing six coin flips and one of them winning them all is twice as high.

No, it is not.

35. numerobis says

Bill Buckner@35: the probability of BBBBBB is equal to the probability of CCCCCC.

36. slithey tove (twas brillig (stevem)) says

regardless… Clinton getting such a razor thin victory is, spiritually, a victory for Sanders. (or so I say). Being the very very first contest and its locale being so distant from Sanders home state, it’s gotta at least feel good to Sanders campaign (and the grumpy old man himself). Also, I’m sure Clinton ws expecting to dominate the Dem results in a state so close to her home state, so this has to feel like a loss, with the razor slim victory being a mere consolation prize. IMO, YOMV (y.our o.pinion m.ay v.ary)
hurray Bern.

37. Bill Buckner says

Bill Buckner@35: the probability of BBBBBB is equal to the probability of CCCCCC.

Yes, I grasp the concept. Did you not read my posts? What occurred was CCCCCC, which does indeed have a probability of 1/64. If you or mod prime in #32 are saying anything different, then you are wrong. To me the comment in #32 is difficult to understand, I don’t know what this means:

The chances that two opponents doing six coin flips and one of them winning them all is twice as high.

that’s exactly what happened, one opponent won all six tosses. There is no “twice as high”. It is 1/64. Period.

38. says

No. It’s 1/32.

One way to look at it: the first toss doesn’t count. We’re just asking whether the next 5 flip the same way as the first.

39. Bill Buckner says

#32, #36,
Sorry, now I understand. You are saying that the probability of all tosses coming up all heads or all tails is twice the probability of it coming up all heads. On that I agree. Sorry it took me a while.

40. gmacs says

@38 I think what they are saying is that P(BBBBBB)=P(CCCCCC). #32 was saying that one opponent winning all was twice what was said above: P(BBBBBB or CCCCCC). I took this to mean P(BBBBBB)+P(CCCCCC).

P(BBBBBB)+P(CCCCCC) = P(CCCCCC) + P(CCCCCC) = 2 P(CCCCCC)

41. Bill Buckner says

#39,
I understand (now). But it is not relevant. The probability that Clinton won all six is 1/64. It is true that the probability that Clinton or Sanders wins all six is 1/32. To the question of whether there is a cheat (which no reasonable person believes) the fact that it would be just as unlikely if the other person had the same luck is, again, not relevant. Take the 30 toss case. If Clinton won all 30 it is almost certain she cheated. It would not be relevant that the odds for Sanders were the same, and therefore the odds for either to win 30 in a row were 2/billion rather than 1/billion.

#41
Yes I did not understand what was being argued. My fault.

42. Bill Buckner says

Aside: My favorite introductory probability question to give to students (now dated.)

China has a one child policy, and we hear anecdotes that in rural areas daughters are selectively aborted because of a preference for sons. What if China had a different policy, where every family could keep having children until they had one son, then they had to stop? What, long term, would this policy mean for the distribution of girls and boys in China?

I tend to agree with those who say the human brain, so amazingly wired for pattern recognition, is just as poorly wired for probability.

43. MJP says

A few of the caucuses were settled by flipping a coin. Yes? So? The votes were tied. The rules require representatives to be selected. A coin flip is a fair way to settle which candidate will be represented, when there is a tie. I have no problem with using a chance distribution to decide, but some people are just horrified at this ‘primitive’ way of making a decision. How else do you propose to do it? Trial by combat?

The real question is why they have delegates at all, instead of just using a raw vote count. It’s a completely unnecessary layer of indirection.

44. mod prime says

#42 “The probability that Clinton won all six is 1/64. It is true that the probability that Clinton or Sanders wins all six is 1/32. ”

That was two of the things I originally said. My point wasn’t about cheating, it was about people thinking that 1 in 32 event is extraordinary enough to suspect cheating at all. Clinton winning six in a row is a 1 in 64. *Somebody* winning six in a row is even more likely and when you look at it like that it makes the reaction even more comical.

I mean it really is about as extraordinary event as rolling a pair of 1s on two six sided dice. If that happens occasionally, we wouldn’t expect it to be a news event. Even in a relatively high stakes game.

45. says

MJP #44:

The real question is why they have delegates at all, instead of just using a raw vote count. It’s a completely unnecessary layer of indirection.

+1

46. Crip Dyke, Right Reverend Feminist FuckToy of Death & Her Handmaiden says

Holy Hecate!

You maxed out your d8 longsword damage and your d4+1 flaming damage on the same blow???

Someone call the media!

47. says

This is a caucus. It’s not about fair representation at all. This is a system in which the most motivated activists get together and argue and hash out issues for hours, so it’s already going to exclude most people, for lack of interest.

There’s going to be another level of indirection, too, because next representatives from local precincts will get together at a state convention and assign delegates for the national convention. “Raw vote count” doesn’t even make sense in this process — it’s about building the machine that will spread the word about the party’s ultimate candidate.

We could skip the whole thing, drop the primaries and the caucuses, and just have one great big last minute popular vote next November. Do you think that would work better?

48. numerobis says

By the way, is this a winner-takes-all contest? I thought they sent delegations, not all of whose delegates would vote the same way at the convention.

While I’ve opened that can of worms, I don’t understand winner-takes-the-state at all in the electoral college. I don’t understand why it’s the dominant strategy for a state to choose to send a slate from the winner’s camp, rather than proportional representation. Seems like the latter would mean that a state like California would always get visited, instead of never.

49. slithey tove (twas brillig (stevem)) says

[suppose we] drop the primaries and the caucuses, and just have one great big last minute popular vote next November. Do you think that would work better?

Maybe.
Proposal:: we change the votes into number-sets instead of just check-marks. as in, with N candidates; for one’s vote give each candidate a value of 1 to N. Then the vote-counters add up each candidate’s total and the one with the smallest total wins.
I think some elections are held this way, why not here also (eg 2020) ?

[derail/distraction] gee…I can’t wait to see the common reaction (memes) to fly around when the next election year rolls around. as in, the year will be 2020, anything come to mind? ;-D

50. futurechemist says

I’m a fan of Nate Silver’s analysis and statistics.

Describing winners and losers only makes sense for a winner-takes-all election, not a proportional election. Iowa was a statistical tie for Clinton and Sanders. And not too far off from a 3-way tie for Cruz, Trump, and Rubio. But the media has to have a narrative with winners and losers.

On another, more practical level, a tie is actually a win for Clinton. This is from Nate Silver. Iowa Democrats are considerably whiter and more liberal than Democrats in most other states, similar to New Hampshire. That’s Sanders’ major demographic, since he still seems to be having trouble expanding his popularity with minorities. If he couldn’t pull out a decisive win in a state like Iowa that looks good for him on paper, then he will have a lot of trouble with states like South Carolina or Illinois which are less liberal and more diverse. It also means that a Sanders win in New Hampshire doesn’t necessarily translate to nationwide support.

But that’s a lot of nuanced analysis. The media just wants a headline about who won and who lost.

51. Crip Dyke, Right Reverend Feminist FuckToy of Death & Her Handmaiden says

@MJP #44 & Olav #46:

You do know that the delegates are not bound in any way? Not legally nor even morally – save by personal loyalty to goals that they, presumably, believe one candidate better shares than any other candidate. Look at which way the Iowa delegation votes for the party nominee at the nominating conventions and compare that with the delegate selection on this first day of the caucus process: before 1972 they weren’t often the same unless a sitting president is being renominated at the convention. Even now, there is enough variance (when not renominating a sitting president) that it’s not unusual to see the national delegation differ significantly from the make-up of the delegations chosen at the precinct caucuses.

This is from uspresidentialelectionnews.com:

The caucuses are an event where voters from all of 1,774 Iowa voting precincts meet to elect delegates to the county conventions. From the county convention, of which there are exactly 99, delegates are chosen for the state party convention. Eventually, the state party convention elects delegates to attend the national party convention, where a Presidential nominee is selected. The process is similar for both Democrat and Republican candidates. The winner of the Iowa caucus receives the most delegates elected to the county convention, which then elects delegates for that candidate to the state convention, and eventually, to the national convention.

While amongst other reasons to expect loyalty, demonstrating loyalty to the candidate who selected you as a delegate make you more likely to be wooed in flattering ways (including getting gifts from candidates and their campaign orgs and their “independent” allies) prior to the next caucus. No one wants to spend a lot on wooing you if they can’t trust you to vote for them if they select you.

However, it’s routine for candidates to drop out sometime after precinct caucuses select delegations. For those candidates who have won delegates, these delegates are frequently released (but not always, and even when “released”, meaning an unexpected vote won’t be held strongly against them, there will typically be a recommendation on which of the remaining candidates should benefit from the former candidate’s departure. This “recommendation” can be asserted with more or less strength. A strong enough “recommendation” can color delegates who deviate from it with the same stain of disloyalty as if they voted against the candidate originally selecting them. Moderate recommendations might be seen as a “plus” for those who follow*1, but not a stain on those who don’t. A light or no recommendation frequently results in the delegate voting for whichever candidate is going to win the nomination (if there is a candidate with a majority of the national delegates) even if that candidate lost in Iowa (like Bill Clinton did). This is also seen as “loyal” because it is perceived as strengthening the party’s nominee when there’s no way to vote that would help the delegate’s original selector. The dropout is presumed, of course, to favor any candidate of their own party over any candidate of any other party (even though this isn’t necessarily the case where enmities come into play). So strengthening the nominee is what the dropout would want you to do, right?

And while there’s at least some merit in that position, the way that Iowa determines its national delegates with multiple steps*2 makes it unlikely that the final make-up of the national delegation reflects the percentages of precinct delegates awarded in the original local (precinct) caucuses.

Given that, the idea that the Iowa caucus is a victory for Clinton is ludicrous. There is a strong argument to be made that a “strong showing” is necessary in Iowa – at least top 4, but really top 3 and only a 4th place if 3rd & 4th place have a scant margin of separation. But “strong showing” is about eligibility for big-donor money and about momentum. Clinton already had the big donor money. Big donors were unlikely to give much cash to Sanders both because he wasn’t courting it as heavily and because he wasn’t seen as a serious enough threat to Clinton’s nomination to justify the expenditure.

“Why give money to Sanders if a Clinton nomination is inevitable?” big donors have, reportedly, been thinking/saying. The dead heat makes it impossible at this point for Clinton to credibly claim that her nomination is inevitable…especially given her loss of the nomination to Obama in 2008. It was, frankly, impressive that she’d managed to get back the “frontrunner/inevitable” reputation even temporarily, since she had it before and not only lost the rep but the nomination fight as well. But now big donors will take Sanders seriously. Whether that means they give even more money to Clinton to fight Sanders or whether they hedge their bets or whether they embrace Sanders as the serious threat without which they felt forced to support Clinton – well, those possibilities will play out in varying percentages, and I can’t guess which will garner more billionaires. But you certainly can’t say it was a win for Clinton since Sanders got his “strong showing”. Her real goal was to silence Sanders, to make his seem an inevitable loser. She failed, and the people that dole out US\$ by the million know what her goal was and know she failed to achieve it. That’s not a win.

As for “momentum”, well that’s about exceeding expectations. Sanders did. Clinton didn’t. You might not say Sanders “won” on the previous, “strong showing” criterion (though you might very well say Sanders “won” in Clinton’s fight to paint him as a loser, using the Iowa caucuses as the field of battle). But having exceeded expectations of him when Clinton fell short of the expectations of her, Sanders “won” Iowa on that criterion.

PZ is right that this is a delegate tie. Since delegate counts may even shift (though that’s not very likely given only 2 candidates), you could almost reasonably say that it’s not so much a tie as the result is simply not yet known.

But on any measure other than delegates, Sanders is the clear winner.

as a last comment,

The real question is why they have delegates at all, instead of just using a raw vote count. It’s a completely unnecessary layer of indirection misdirection.

FTFY.
================
*1: The reason for the perception of extra value can be illustrated with this train of thought:

look, that person is really loyal! If I end up dropping out, I can bargain their vote for the VP slot or a cabinet secretary position because the candidates still in the race will know I can deliver them a reliable vote!

*2: each one of which might have one or more dropouts since the last step, with this winnowing dynamic slowly biasing the delegation toward at least the two or three candidates who are really in it for the long haul, and often for the candidate who appears very likely to get the nomination

52. numerobis says

A tie in Iowa between the appointed heir and the would-be usurper is a win for the appointed heir how? I read that at Vox and wondered what they were smoking.

The establishment voices have been united in dismissing Sanders since the start, still are.

Personally I don’t know which I prefer as president. In a primary I’d probably vote Sanders, because I’m young at heart. In a general, I’d vote for a yellow dog before I’d ever vote for anyone presently on the GOP platform.

53. screechymonkey says

numerobis @49:

I don’t understand winner-takes-the-state at all in the electoral college. I don’t understand why it’s the dominant strategy for a state to choose to send a slate from the winner’s camp, rather than proportional representation. Seems like the latter would mean that a state like California would always get visited, instead of never.

It’s a little more complicated than that.

From a “how to get the most attention from presidential candidates” point of view, the best situation is to be a winner-take-all battleground state. The worst is to be a winner-take-all state that one party has locked up. So yeah, California and Texas right now don’t get much attention in presidential general elections, but winner-take-all works pretty well for Florida and Ohio.

But “attention from presidential candidates” — even if you assume that this will also lead to substantive policies that help your state, the way Iowa’s status in the nominating process helps it secure ethanol subsidies — isn’t necessarily the main objective of any state. If, as you suggest, California were to allocate its electoral college votes proportionately, it might indeed get more attention from presidential campaigns — but it would also be a huge blow to the Democratic Party. The average Californian cares more about not throwing the White House to the Republicans than about getting some pro-California concessions from whichever Republican reaps the benefits. So in that sense, going to a proportionate allocation would be against the interests of Californians, broadly defined.

And that’s why proposals at the state level to change that state’s allocation only happen in the odd situations where the party that generally doesn’t carry the state at the presidential level nevertheless has control over the state government — as in Pennsylvania and Michigan in the past decade, where Republicans have toyed with going proportionate. Those proposals usually lead to a backlash, because it’s seen for the gamesmanship it is.

There is a proposal for an interstate compact, pursuant to which the states that sign on would agree to award their Electoral College votes to whoever wins the national popular vote — which by its terms would only take effect once enough states sign on to constitute a majority of EC votes. That’s the most likely avenue for EC reform, as it’s too hard to amend the constitution. National Popular Vote has its own drawbacks: it creates incentives for parties to “run up the margin of victory” in every state, whether by ethical means or otherwise. Imagine a Florida 2000 situation on a nationwide basis….

54. Amphiox says

Re #1, how else to settle a tie.

I propose to take a page from sports and call SUDDEN DEATH OVERTIME!

Voting is restarted for a 6h window the day after the account, and each campaign’s ground game gets that opportunity to bring in as many additional voters as they can.

Then recount, and repeat as necessary!

55. says

Actually, it is not even a primary. They have one of those later. It is a caucus, where we expect an informal local group to convene & talk. Not an international media event.

56. malta says

Giliell, #17:

This whole thing is fuck annoying
Americans, get over and done with this shit. Seriously, the rest of the world simply can’t just close up for business for a year or so until you’ve figured that presidency out.

Ugh, I know. I was sick of the 2016 election before 2016 even started. And we still have nine more months of this shit!

57. says

@4

So, like a political Pon Farr?

“Pon Farr” is periodic male Vulcan drive to mate. The trial by combat, which a female Vulcan may demand of her betrothed, is “Kunat Kalifi.”

Caucuses are great, even if they have all these arcane rules about delegates. A primary just tells you who people’s first preference is, a caucus is a way of taking relative support into account.

Some sort of ranked ballot system might work too, caucuses are nice just because it makes people talk to each other about politics instead of just taking for granted whatever Fox News tells them their neighbors think.

58. Holms says

A few of the caucuses were settled by flipping a coin. Yes? So? The votes were tied. The rules require representatives to be selected. A coin flip is a fair way to settle which candidate will be represented, when there is a tie. I have no problem with using a chance distribution to decide, but some people are just horrified at this ‘primitive’ way of making a decision. How else do you propose to do it? Trial by combat?

Well to be fair, the majority of the US electoral system is primitive.

59. numerobis says

malta@57: I was sick of the 2016 election by December 2012.

60. gmacs says

@55

Because that’s what I would want to do again today: cram into a relatively small space with a bunch of ornery party organizers and my disaffected neighbors for hours on end.

I think a caucus system would work, if it were given more clear cut rules. I hate to say it, but I think the republicans may have figured out how to do it.

61. jackal says

PZ @ 39 is right, if there were 6 coin tosses, it’s (1/2)^5 that they’re all heads or all tails. Some sources are reporting that there were 7 coin tosses, and Clinton won 6 of them. (Snopes hasn’t been able to verify which it was.) That would follow a binomial distribution. B(7, 6, 1/2) simplifies to 7*(1/2)^7 = 0.055, so we’d expect to get those kinds of results about 5.5% of the time, more than 1 in 20.

62. Ragutis says

Iowa does appear to be a better indicator for the eventual Democratic Party nominee than the Republican. So this could be a sign that the race will last a while on that side. However, as PZ and others have noted, it’s more a sign of how motivated the activists are, rather than an accurate reflection of the general electorate. Bernie’s ground game, his grassroots organization, have been his strong point from day 1, so this shouldn’t really be a surprise. He’ll take New Hampshire for sure (Hell, Bill Clinton made it a point to say weeks ago that no candidate from a neighboring state has ever lost the N.H. primary). I expect Hillary to make perfunctory appearances in N.H. to lessen the margin while pumping tons of cash and every political endorser and asset into Nevada and the southern states. NV might be close too, but Hillary has a big advantage with African American voters and that might be enough to get enough ahead in the south and start the type of momentum to all but seal the deal. However, the longer Bernie lasts and the better he performs, will be an ever growing signal to her that a BIG portion of the party is leaning quite a bit further left than she is. She’ll have to shift her message to get all the Bernie supporters to get out to the polls should she get the nomination.

On the Republican side, I don’t know what the fuck is going on. But it doesn’t look good, so that’s nice.

For the record, and just in case anyone cares, I’m way closer to Bernie philosophically, but am thinking that Hillary might be the better executive with world events as they are and little hope of flipping the House or Senate.

63. dianne says

If it’s all coming down to the flip of a coin or the roll of a die anyway, let’s stop fooling around and do it right: Trial by combat…AD&D combat. I’m guessing warrior/cleric for Clinton and rogue/mage for Sanders, but I could be wrong. May the best nerd win!

64. tbtabby says

The reason this result is considered a win for Sanders is because, when he entered the race, everyone said he didn’t have a chance, he was unelectable, that Hillary would win the nomination hands down. Even Jon Stewart said point-blank, and I quote, “She is going to crush him.” Fast forward to the present, not only was Sanders not crushed, the margin was so tight it might as well have been a tie. Bernie Sanders has shattered everyone’s expectations and turned a runaway win for Hillary Clinton into a barn-burner. And without one penny from Wall Street.

65. dianne says

One wonders what will happen once the Republican machine lines up behind Cruz and the Democrats realize (hopefully not too late) that Hillary is projected to lose against Cruz…

Fivethirtyeight keeps pointing out that national polls of hypothetical matches this far in advance mean little and are very poorly predictive of actual outcomes. Also, the Republicans have been running against Clinton for at least the past 4 years. Longer, really: they’ve done all they can to mess with her since Clinton I. They’ve barely started running against Sanders yet and when they do they’ll find plenty of ammo, real and imagined to use against him. The average voter may end up voting for Cruz because they’re convinced that Sanders will take away their right to anything from cheap guns without background checks to deciding what to eat.

I have no problem with voting for Sanders because he’s the person you want for president. I’ll probably do it myself. But voting for him because he’s “more electable” strikes me as a dubious proposition at this point.

Jackal @62 mentioned snopes, which reports a coin-flip for Sanders in a Sioux City precinct. I found out about this video of a coin toss for Sanders in Hardin Township in the slashdot comments on their story about the results.

So that’s at least two for Sanders, and for anyone who didn’t read the snopes page, the precint people weren’t required to report whether a coin flip was used. This should serve as a reminder that the news is kinda shitty at their job when they’re rushing to get a story out.

67. says

Strange, isn’t it? Everyone knows the chances of winning the lottery is way lower than 1.6%, yet it happens all the time.

68. Crip Dyke, Right Reverend Feminist FuckToy of Death & Her Handmaiden says

@neilweightman, #68:

Now you’re just talking crazy talk.

69. robinjohnson says

PZ: How else do you propose to do it? Trial by combat?
How about count up the tied districts at the end of the night, and divide them equally?

70. brucegee1962 says

I have a friend who caucuses in Colorado. He would like to go to the National convention, and intends to vote strategically with whichever group has the fewest highly motivated local politicos, in order to give himself the best odds. (He’d be ok with either candidate winning.)

So yeah, it’s a flawed system.

71. jrkrideau says

We could skip the whole thing, drop the primaries and the caucuses, and just have one great big last minute popular vote next November. Do you think that would work better?

Well who knows. What the devil does a primary do?

@numerois 2 Feb 2016

I have tried to understand the rationale for US Primaries (didn’t realise Caucases were different) for years. I think you have to be born in the USA to grasp whatever they do. Apparently they even have some kind of government finance! (Note I have not managed to integrate sceehymonkey’s post 2 feb. It may actually help my incomprehension.)

72. leerudolph says

slithey tove asks: “as in, the year will be 2020, anything come to mind?”

Yes. George H. W. Bush!

73. jack16 says

Suppose someone had said to you, “P Z , I think those politicians are a bunch of crooks and I will bet you ten dollars that every coin flip for a caucus decision will be in favor of Hilary.” Would you have bet?

74. dianne says

What the devil does a primary do?

Allows voters to directly participate in the decision of who will represent their party in the general election. In principle, the parties run the primaries and they are really nothing much to do with the actual election, but are an internal party matter for any parties that decide to run a candidate for public office (local and state as well as federal.) In fact, in many places the primary is the “real” election and the winner of one party’s primary is effectively elected. See, for example, Philadelphia mayoral elections. It has been suggested that the system could use a little reform here or there.

75. dianne says

Suppose someone had said to you, “P Z , I think those politicians are a bunch of crooks and I will bet you ten dollars that every coin flip for a caucus decision will be in favor of Hilary.” Would you have bet?

Are they giving me 1:64 odds or better? If so, I’d take it.

76. robinjohnson says

There were seven coin tosses, according to snopes: http://www.snopes.com/iowa-caucus-coin-toss/
So the chance of one candidate winning at least six tosses is (6!/5!) / 2^6, or 1 in 8. Unlikelier things happen in every boardgame.

77. jrkrideau says

@ 75 Diane

Thanks, that has to be the most succinct description I have ever seen and it almost makes sense. My problem is that as a non-USAian, I don’t understand why a party convention would not do the same thing.

Does one have to have a paid party membership to the party in question to participate in a primary? If not why should they have a vote?

From my perspective, if you are not a card-carrying member of the party you have no say in who is the party standard-bearer.

78. dianne says

Does one have to have a paid party membership to the party in question to participate in a primary? If not why should they have a vote?

Nope. No dues. When you register to vote you can indicate that you are a member of one party or another and, if you do, you can vote in that party’s primary. You can switch parties at any time, but I think there’s some deadline with respect to voting in a given primary. Not sure when it is, though. Basically, you just have to decide which primary you want to vote in. You don’t have to pay anything or vote for the same party in the general election.

79. What a Maroon, living up to the 'nym says

dianne @79,

It differs by states. Some states require you to register by party (with “Independent” as an option), and you can only vote in a primary if you’re a member of the party; others have open primaries, so you can decide on the day of the election which primary you want to participate in.

jrkrideau @78,

We used to do things that way–state parties would choose their delegates (who were often uncommitted or pledged to a “favorite son”–some state pol who would use the delegates to do some horse trading at the national convention). The national convention would then choose the candidate, usually after several rounds of votes and a lot of shady dealing in the infamous “smoke-filled rooms”–it was a lot like choosing a pope. The primaries and caucuses evolved slowly as attempts to make the process more democratic and less corrupt, but they didn’t really take off until the ’60s/’70s, when pols like Kennedy figured out their value and the national parties started changing their rules.

80. Menyambal says

I think this post and many of the comments are giving the innumerates a little too much grief. Not that they don’t deserve some.

Assuming that there were six independent coin flips, there is only one way to get all six for Clinton – one in all sixty-four possible outcomes. But there are six ways to have one place go for Sanders – any one of the six precincts could have – so that is more likely. Two places going for Sanders is possible about 15 different ways, so that is even more likely still.

It’s late and I am on a tablet, so I can’t work all the odds, but most folks realize that all 6 going the same way is less likely than any of the various mixed outcomes. Not less likely than any one particular mixed outcome, but much less likely than some sort of mixed outcome.

One of the headlines can be read as saying that. The pure run is the least likely result of independent coin flips, compared to all the possible muddles in which nobody cares which particular precinct went which way.

But it isn’t that rare, or all that unlikely, or all that remarkable. It’s certainly not important in the long run. Even the final numbers are practically a tie. The numbers, before the coin flips, really show that Sanders and Clinton were evenly matched – it was a tie. The coins were flipped because of six tie votes. Sanders actually won, if you count the pre-flip precincts with clear wins. But that margin is so slim as to be ignorable.

Counting anything so close as a win, or important, in a race that has so much more shit to slog through is innumerate. But taking a moment to go “huh” at the least likely coinflip isn’t so bad – if that was all they did.

81. treefrogdundee says

Doesn’t matter. The DNC will give the nomination to Hillary (can you say, superdelegates?) despite Sanders being the choice of the overwhelming majority of actual voters. And given the choice between Hillary and whatever the GOP scraps up and nominates, I’ll be sleeping in on election day.

82. some bastard on the internet says

I don’t know if this has been addressed already (I sorta skimmed through some of the later comments), but I really feel the need point something out, regarding this calculation:

Winning 6 coin tosses in a row is extraordinarily rare, and only has a 1.6 percent probability of occurring. As journalist Ben Norton explained, that’s broken down by calculating (1/2)^6, which is 1/64 — or 1.6 percent.

This is only true if we we included specific sequences within an overall result (i.e., CCCBBB vs. CCBCBB).

Since we’re only concerned with overall results, then we can make a small grid to determine the odds:

CCCCCC
CCCCCB
CCCCBB
CCCBBB
CCBBBB
CBBBBB
BBBBBB

Six coin tosses have seven overall possible results, meaning the odds of one person winning them all is actually 1-in-7 (not 1-in-64), or about 14.3%.

Even if you’re betting that the overall result will be a mixture, your odds go from 98.4% (as Mr. Norton’s calculation would imply) to 71.4%. Still good, but hardly guaranteed.

83. Bill Buckner says

Six coin tosses have seven overall possible results, meaning the odds of one person winning them all is actually 1-in-7 (not 1-in-64), or about 14.3%.

Um, no. Let’s be careful. The probability of Clinton winning all six tosses is indeed p = 1/64. The probability of Clinton or Sanders winning all six is 1/32. The odds (odds = 1/(1-p)) of Clinton winning all 6 is 1/63 or expressible as “63 to 1 against” and the odds of Clinton or Sanders winning all 6 is “31 to 1 against”.

If you want to use your odds, I’d like to place a bet!

84. Rich Woods says

Having read through all 83 comments here so far (plus having read assorted news items regarding this and previous presidential primaries going back over the last thirty years) I can honestly say that, from a foreigner’s point of view, it seems to me the USAian concept of a democratic election at such an important level is absolutely fucked. Coin flips on the first day? Fucking hell!

Please, take this as a small piece of advice from someone who has had to live with (and fight against) a Parliamentary system which is fucked up by the governmental placement of grovelling useless fuckwits to the second House: you can do better. So can we, but you expect us to have taken 800 years and multiple civil wars (yes, note the small ‘c’ in civil) to get this far…

85. some bastard on the internet says

@84 Looking back through the comments, you are correct. What I was talking about was more like a game of Yahtzee, where the rolls (or flips, in this case) could be rearranged to fit a particular category. But I failed to consider that the top and bottom outcomes each only cover one subset of flip results (i.e., all Cs can only consist of CCCCCC) whereas the middle outcome covers the most subsets (CCCBBB=CCBCBB=BCCCBB…, etc.).

So yes, all Cs or all Bs would be 1/64.

86. What a Maroon, living up to the 'nym says

@Rich Woods,

The truly fucked thing about the coin tosses is that there’s really no reason to declare a winner in this case. If the county (or whatever the unit is) has 20 delegates or delegate equivalents or whatever they call them, there’s really no reason why they can’t give 10 to Sanders and 10 to Clinton. And hell, even if they’ve got an odd number of delegates, they can go all Solomon and divide one in half. Given the convoluted fucked-up process to get from there to the convention, it really won’t make a difference.

But, y’know, there’s always got to be a winner, right?

87. brucegee1962 says

They only did the flips when there was an odd number of delegates, for the last one.

88. Chloe Carlisle says

I really think you should correct this post, because your point #3 is repeating a fundamental numerical mistake about the delegates. Not good in a post where you’re complaining about Americans’ innumeracy.

The coin tosses were for COUNTY delegates. There are thousands of county delegates; each one is worth about 1/10 of a state delegate (the ratio varies from precinct to precinct). The delegate counts cited in your post are for STATE delegates. People are conflating them, thinking that the six coin tosses equals six state delegates. No. Those were six COUNTY delegates. At most that added up to half a STATE delegate.

So the handful of coin tosses didn’t change the overall outcome, because they were at the county level. And since it turns out that Sanders also won six coin tosses (http://www.cnn.com/2016/02/02/politics/hillary-clinton-coin-flip-iowa-bernie-sanders/), the whole coin-flipping business is a wash.

Clinton won the Iowa caucus. It wasn’t a tie. And she didn’t win because of coin flips.

89. chigau (違う) says

Chloe Carlisle #88
Who are you talking to?

90. chigau (違う) says

#89 that is

91. Ichthyic says

The coin tosses were for COUNTY delegates.

then it’s an issue of innumeracy, instead it’s a category error.

there, now you can breathe a sigh of relief, right?

Clinton won the Iowa caucus

it’s not a winner take all.. so you just made the same kind of error.

92. Ichthyic says

NOT an issue of innumeracy, I meant to type.

93. says

@some bastard on the internet #86 – Please look up central limit theorem. It might interest you.

94. says

According to this article (Coin-toss fact check: No, coin flips did not win Iowa for Hillary Clinton):

It’s been reported that there were as many as six sites where ties were decided by the flip of a coin — and Clinton won every single one. The odds of that happening are 1 in 64, or less than 2 percent. What’s more, that gave her just slightly more than her margin of victory over Sanders — four delegates.

Things that make you go hmm. Indeed.

Here’s proof: Watch one of these coin-toss tiebreakers in this video, taken by Univision reporter Fernando Peinado at a caucus precinct:

Except that doesn’t tell the whole story. In fact, there were at least a dozen tiebreakers — and “Sen. Sanders won at least a handful,” an Iowa Democratic Party official told NPR.

The odds of Clinton winning all coin-tosses, if we only count coin-tosses she won, is 1