Was the Iowa Caucus outcome determined by a coin flip?
We have seen several reports that Hillary Clinton won the Iowa Caucus by a coin toss, or by six coin tosses. Or some other number. We've also seen reports that six delegates were awarded to Clinton on the basis of coin tosses, implying that of the 44 delegates determined on Monday in Iowa a large percentage were chosen by the toss of an unfair coin, that somehow the Clinton campaign controlled the coin tosses causing them to all come out in her favor. And so on.
After a barrage of these reports, we are now seeing a small number of reports trying to describe what actually happened, which was very different. Some of these reports are somewhat accurate, but most leave the reader not fully understanding what really went down. Here, I want to 'splain this one more time in a way that I hope makes sense. Warning: This is not simple. Which, really, is the point I want to make.
Then, I want to make a couple of meta comments about what this all means.
Most of the difficulty in understanding what happened in Iowa arises from the fact that Iowa uses a caucus system, not a primary. So does Minnesota and over a dozen other states and territories. This is important because quite a few convention delegates are determined by this system, and those involved in the political process need to understand how a caucus works in order to effectively engage.
I'm going to generalize here, mainly from my Minnesota experience; your caucus may vary. But there are some general principles that seem to apply widely.
How does a caucus work?
The exact rules of how a caucus works vary from state to state, but also, from year to year. The rules are determined by the party officials through an arcane process that anyone is welcome to join in but few do. The rules are handed down and enforced by those leading local meetings. The people running local meetings are always experienced dedicated party officers and volunteers. Except when they are not because they ran out of party officers or experienced volunteers, then the meetings are run by confused and frightened citizens who do their best. This is the first thing that makes a caucus system different from voting. Voting is easy. Caucusing is harder, and how well it goes at a given site can vary.
The next thing you need to know is that the caucus system often addresses many issues, not just a presidential nominee. There may be local candidates, state wide candidates, and congressional candidates chosen by the caucus process.
Normally a caucus system, therefore, happens at near or at the lowest geographical level for a state. This then leads to a second caucus later at a higher geographical level. Then, at the state level. For instance, Minnesota will caucus on Super Tuesday (March 1st this year). But we will have another caucus in April at the State Senate district level. Eventually there will a meeting at the Congressional District level. Then a statewide convention. Depending on the state or year, the national political party convention delegates may be chosen somewhere along the line. In Iowa (and in Minnesota, and may be everywhere) number of national delegates representing each presidential candidate, are chosen at the first caucus, at the lowest level. Which, of course, makes no sense until you understand the entire system. (If it makes sense to you then, please let your local Democratic party officials know because they need you.)
There may be delegates at each level. Indeed, everybody who shows up and is a legal voter in the precinct or district might be considered a delegate. So there can be many many thousands of delegates involved in the caucus system, depending on how it is defined. But, at the lowest level (a precinct, in Iowa) these people use a system to pick a subset of themselves to be the delegates that advance to the next level. Typically, these delegates are committed to a candidate. In the case of Iowa, they all had the potential to become Clinton, Sanders, or O'Malley delegates. And there were thousands of them.
These delegates then advance to the next level at which a subset of them is selected to go on to yet another level, etc. Eventually, in Iowa, 44 delegates are chosen. Except the actual delegates do not winnow down to 44. Rather, the distribution of the thousands of delegates by preferred candidate is used to determine the distribution across the candidates among those 44, and who those individuals actually will be -- who gets to go to the national convention and party, er, represent -- is a whole other system not addressed here.
The initial selection of the lowest level of thousands of delegates is chosen at each local site using one or more methods. The methods, as I said, follow the rules handed down by the party. I don't know the exact rules used this year in Iowa, but there are two methods that are generally used. For small groups, it is not uncommon (if the rules allow) for the group to sit down and talk in a room then vote for how many delegates of each type they will put forward. In some cases, such as when a party has an incumbent president, they don't even do that, but rather, use a simple ballot system to chose one of two "candidates," the sitting president or "other" (or write-in). The point is, the process can look like a mini election that results in some paperwork and that's it.
But the more traditional and more fun way is to use a Walking Caucus. Here is a typical framework for Walking Caucus (but, again, the rules vary).
Everybody who is properly signed in gets in one part of a big room. The number of people involved is known because they checked in, so there is a number, we will call N, of individuals. Using duct tape or some other means, the rest of the room is divided off and no one who is not a delegate can cross that line. Party officials are usually standing on a stage or table nearby directing things.
Some people have signs they hold up with the name of a candidate or issue on them, like "Clinton, Climate Change, and Jobs" or whatever. They are all yelling things at each other. Over time people move into clusters where each cluster is a group of people who favor a given candidate or set of issues.
Variant: Sometimes people are not running around and shouting, but rather, sitting around and talking, and not all caucuses address issues. But it amounts to the same thing, people will move to physical proximity to each other by candidate or issue or something.
Either way, you get a bunch of clusters of people. We will call these clusters "sub caucuses" because that is what they are sometimes called.
It might look like this:
There is a time limit on this process, and eventually the party officials "freeze" the caucus. At this point, or some other later point, delegates are told they can't return if they leave the room or cross the line. Some delegates will leave because they are tired, need to go to the bathroom, are annoyed, or have something to do. These become "missing delegates." Sometimes they are let back in later, sometimes not.
Meanwhile, the party officials have taken two numbers, D, which is the number of delegates that can advance from this caucus, and N, the number of people supposedly in the room, and applied some basic integer math. They need to use integer math if they can't advance proportions of delegates (i.e., they must use whole numbers) and because they are not allowed to saw the actual delegates into bits even if they sometimes want to. This integer math results in a Viability Number. If you are allowed to put ten delegates forward, and you have 100 participants, the viability number is 10. Any group of people fewer than 10 does NOT get to advance a delegate.
Now, back to the frozen caucus. The party officials demand a count of each cluster of people. Each sub caucus is then determined to be "viable" (the number in the group is at or above the viability number) or "not viable." At this point, the "not viable" sub caucuses are dispersed, not with dogs or firehoses or anything, just told that they are not viable and should break up and find somewhere else to go.
This is when the shouting starts again, as the non-viable people are invited to join viable caucuses.
Eventually the walking caucus is frozen again, and counted again.
A lot of things can happen at this point in time. Ideally, no one has left the room, and everybody has divided themselves into groups that are exact integer multiples of the Viability number. In this case, each sub caucus is simply counted (how many people in it) and then that number is divided by the Viability Number. The result is the number of delegates that sub caucus can advance. Those delegates are typically committed to support the candidate the sub caucus they were in represents, if the caucus is choosing state level or congressional candidates.
For the national convention, this simply translates into a number that is passed on via a form or electronic device, and the people standing in the room will have little to do with what happens next. The party will figure out who gets to go to the convention, and the number sent representing each candidate (if proportional representation is used) will be decided later. But the actual delegates that are actually advanced from each sub caucus will, in theory, go on to the next level (state Senate district, or US Congressional district, or whatever). So, typically, there will be a tiny little election within each sub caucus to elect their moving-on delegates. In my experience, this is often the most important moment in the caucus, because you have to send someone on you can trust, but you don't know these people and suddenly you are faced with choosing them.
In Minnesota, we require gender proportional representation (male and female only at the present time) so if your sub caucus has two delegates one has to be a boy and one has to be a girl. The party may reserve the right to remove and add delegates later if they need to to make sure the gender proportionality is 50-50.
Now, imagine the following scenario. The exact number of people who ended up on the floor exactly matches the number of people who are properly signed in. Nobody leaves for any reason. The number of people who showed up happens to be an integer multiple of the number of delegates your caucus gets to put forward. There is no confusion. As the people make their sub caucus, they all, each and every one, understand the above described system or its applicable variant, and have thus formed perfect sub caucuses, each divisible by the viability number. So, when the sub caucuses are polled as to how many delegates they represent and for which candidate, absolutely nothing goes wrong.
Now, imagine this scenario. It is late. The party officials have burned off half the scheduled caucus time dealing with esoteric party decisions. Half the delegates are retired older people with weak bladders. Some are parents with a time limit on their baby sitter. Nobody really knows how to do math these days. The psychic unity of humankind has failed to make sure the number of people who showed up is an exact multiple of the viability number. And so on.
So, you end up with a set of sub caucuses that does not perfectly produce the exact number of delegates required.
This will be, typically, off by one. It really can't be off by two, because if the number of people in the room changes that much during the process, the actual viability number is simply changed. This should happen before the walk, and when the viability number changes, there is a lot more movement between sub caucuses. In any event, a little adjustment there, a bit more shouting and cajoling and walking around, and counting and recounting, and with luck the number of delegates that is required from that meeting can be assigned.
Or, one will be orphaned. Because of all of these sources of error, it is possible via many possible sequences of events to end up not being able to fairly assign one of the delegates to a candidate.
There are various ways this can be dealt with, but a common and accepted method is to flip a coin. If there are two candidates, each has a 50-50 chance of getting a delegate. This is interesting because it gives candidates that have a low representation a small chance of getting one delegate more than they deserved, which is considered both harmless and polite. If the split is very close, it could determine whether or not a single meeting location sends one more or one less than the other on to the next level.
What has not happened is that the outcome of that caucus, at that location, was determined by a coin flip. In Iowa, the average precinct has about 7 or 8 delegates, I believe. The number of delegates that are passed on because of a coin flip should usually be zero, but now and then, one. One out of seven or eight. Not all seven or eight. One.
In Iowa, over 11,000 delegates were chosen with the caucus method. About a dozen (according to reports, though I suspect a few more) of those 11,065 delegates were the result of a coin toss.
Iowa will send, using this process, 44 delegates to the national convention. They determined the number that would represent each candidate by using integer arithmetic to divide up the 11,065 precinct level delegates into 44 national delegates. So if there was a coin toss for a dozen precinct level delegates, then one tenth of one percent of that decision was influenced by random chance. The rest of the allocation was determined by the totally sane non-random process described above.
There is more, of course. The actual delegates and the actual numbers depend not only who shows up on caucus night, but who shows up later. Given this number of delegates, and the fact that they are regular citizens, means that many individuals will abandon the process along the way. As noted, there may be issues of allocation by gender, or other factors.
Because of the possibility of little things going wrong along the way, I think it is typical for party officials to have a lot of leeway in who ends up being a delegate. During off years, the number of people who show up at the first meeting during which people are recruited to be delegates is often so small that anybody who calls up the precinct captain (or some other official) later on can probably become a delegate. During presidential or, even, midterm years, there may be a lot more public involvement, but it might be highly variable across the party's geographical space, so the number of people involved in selecting a given delegate may vary. This may also be a difference between states (such as Iowa vs. Minnesota). The point is, if you want to find randomness, capriciousness, or arbitrariness in the system, there is plenty.
But not the coin toss. The coin toss is an effective and fair way to allocate the occasional orphan delegate. It will favor low-number candidates slightly, and mean nothing in a close race.
What can we learn from coinflipgate?
Coinflipgate got legs because the people who initiated the meme, or later spread it, were ignorant, willfully or otherwise, of how the caucus system works.
Coinflipgate was a political ploy exploited inappropriately by anti-Clinton or pro-Sanders activists for the purpose of affecting people's attitudes about the Clinton campaign.
Coinflipgate, because of the thinness of its veneer and its inherent absurdity, was not a good political strategy.
An important aspect of the caucus system is that it is complex, confusing, and difficult. Contrast the following two scenarios.
Scenario A: Dozens, maybe close to 100, people show up at a caucus site because they feel strongly about a candidate, but have never been to a caucus before and have no clue as to what to do. Party officials are barely able to manage the ensuing fray. The caucus happens, but inefficiently. Delegates are advanced, but several people leave the room realizing they were in the wrong sub caucus, or an orphan delegate emerges and is randomly assigned.
Scenario B: The candidates being considered at a caucus recruit experienced caucus experts to attend a caucus and help guide their supporters to form efficient sub caucuses, and to effectively persuade the undecided to join them, or occasionally, to turn a delegate to their side. These activists may not even be voters in that precinct or district, and in fact may even be from out of state (because you don't want to use individuals who should actually be caucusing). Each candidate also has several supporters among the delegates, and those supporters have met once or twice with an expert who has trained them in how to caucus.
One could easily imagine one candidate following scenario A and a different candidate following scenario B at the same caucus. If those two candidates are roughly even in their support, the candidate that runs scenario B at most of the caucus sites will come away with more delegates. This is very different from voting. In short, the "ground game" wins the day.
This may seem rather capricious but it is not, for one of the reasons that the caucus is a good system. The candidate that can run a good ground game in a caucus is the candidate that can manage everything better, run a better campaign, have a better chance of winning in other contexts, in other caucus states, in non-caucus states, and in the general election. This is the candidate that has good people working for them. All else being equal, you want to support the candidate that can do this better. Ability to manage a caucus is not the only criterion on which we chose our candidates or leaders, but it is one criterion that is meaningful and a valid test of skill, level of organization, commitment of supporters, and so on.
That is certainly not the only goal of a caucus. The main goal of a caucus is to get people together to make an important decision in a way that goes beyond merely showing up and checking off a candidate's name in a box.
My description of how a caucus works is based mainly on my experience in Minnesota and some reading and conversation about how Iowa works. If you are an experienced Iowa caucuser, feel free to add or correct details in the comments below. But the basic idea is there, and similar across cases.
People who are repeating the absurd idea that national delegates were chosen with a coin flip, and people who are calling for a "recount" of the caucus, are either not understanding how the system works, or are exploiting the fact that most people don't know how this system works, to toss some mud and cast some doubt. That is not helpful, no matter which candidate one supports, because it is a less than honest and, in the end, very ineffective, likely to backfire tactic.
And now, a word from the Minnesota DFL
Good! Bloody! *GODS!* Good thing the process is so logical, intelligent, stream-lined, and efficient....
I don't object to the coin toss: I think it is a fair and logical solution.
Some people reported that there were "about 12" coin tosses, six going to Ms. Clinton and "about six" to Sanders.
Greg, are the caucuses open to the non-voting, out-of-state public? This sounds very peculiar to me, and I want to see it in practice!
FWIW, I have described here
how to get a fair result despite using a biased coin :-)
I think you have missed some details that makes it a bit closer.
I could be wrong about this as I can't find details anywhere.
Premise 1: It looks like 2 national delegates were awarded to Hillary for winning the state. I have seen reports of 24-21 and can't reconcile that with the 44 total. If it is 23-21, then this would mean 'winning' changed it from 23-21 Bernie to 23-21 Hillary.
Premise 2: The state delegate equivalents reported is 701-697, and breaks down to about 700.7 to 697.3.
So 700-698 is easy to see based on coin flips.
Premise 3: The coin flips are not 6-0 Hillary, as there is video of Sanders winning at least 1. So the odds have gone from 1 in 64 to one in 16, and talk of rigged coin flips are ridiculous. I haven't seen anything about 12 flips, or what the split is.
Premise 4: Other voting irregularities occurred, costing Bernie county delegates. There is reportedly one precinct caucus that had just one person, who supported Sanders. The precinct result was one delegate for Hillary. There are other cases like that.
Premise 5: It is possible for the other county delegate discrepancies to add up to a result that would be 699-699 tie.
Premise 6: Bernie leads the popular vote in Iowa that has not been reported. I saw CNN giving county vote totals on Election Day, so I'm not sure why they don't just add them up. It is possible these numbers represented something else.