It's great!
Sean Trende and David Byler have done yeoman's work in not only putting this thing together, but in putting it out there for public consumption. Having put a couple of delegate allocation models together during primary season in 2012, I can tell you that it is, at best, an imperfect science.1 Juggling all of the various factors embedded in the patchwork of delegate allocation rules across the entire country is no easy task. But that doesn't mean that we shouldn't try.
However, it does mean that one has to make assumptions about certain factors of the system to model it properly/accurately (see for instance the variation in delegate rounding). And while those assumptions can potentially be fodder for criticism, FHQ would rather take the tool for what it is: an instrument to help us better understand the complexities of the rules, how they are differently implemented across states, and maybe what that means for the 2016 race.
[NOTE: And FHQ really cannot stress enough how open Sean and David are to any and all comments, feedback, perspectives and alternate approaches/assumptions about this thing. This is a first pass. And having spoken with them while they began putting this together, they are intent on improving it into and through primary season.]
--
But it is that maybe above that is of some overall concern to FHQ. The reactions to this thing have been positive from what I have seen, but I do wonder how people will approach this thing. My fear is that it ends up overemphasizing or artificially inflating the odds (in terms of perceptions on the individual level) of the various contested convention scenarios out there (see for instance this). That overemphasis tends to be on the outcome rather than the input. By input I mean plugging in poll numbers that traditionally have not been predictive at this stage of past races and more importantly the role of winnowing in the process. And that latter option is available, but I wonder how often users will account for those effects. It is relatively easy to produce a contested convention outcome if you carry 14 candidates through the process or even five. But is that the likely path? FHQ does not know to be honest. However, past results point toward a sequential process -- like the one that is still in place for 2016 -- gradually winnowing the field like kids in Willy Wonka's chocolate factory.
This is FHQ's way of saying that this delegate allocation interactive tool is a very powerful one and its utility will be derived from how it is used. So rather than be critical, let's put the gadget to use in testing a hypothesis.
FHQ likes to talk about rules. Perhaps you have noticed. And there have been some changes to the delegate allocation rules employed by the Republican National Committee between 2012 and 2016. The gadget give us the chance to -- in a very rough way -- estimate the impact of those rules changes. The premise is simple: Take the 2012 primary results, plug them into the tool and see how the path/outcome differs.
There are a lot of questions that come out of this. Does the smaller proportionality window slow down Romney's march to the nomination? Does the tighter definition of proportionality come to the aid of his opponents? Do the winner-take-all states clustered on March 15-22 push Romney over the top? Are the SEC primary states winnowers and the northeastern states coalesced in late April the deciders as Trende and Byler hypothesize?
Before FHQ digs into that, a few notes. First, this, too, is an imperfect approach. The rules are different and so is the sequence of contests. That means that late 2012 states that are early in 2016 have lopsided results that favor Romney (see Arkansas, Kentucky, Texas and Utah). Similarly there are early 2012 states that are later in 2016 where results are likely imprecisely competitive during a likely less competitive part of the calendar (see Washington). There are also some of the warts in the tool's code in this early stage. For instance, the gadget allocates delegates from states like Colorado, North Dakota and Wyoming, where it will remain unknown for quite a while -- perhaps up to the convention -- just how many delegates are actually bound to what candidates (in the eyes of the RNC).
Still, there is some utility to be gained in gaming this out despite that.
Here's how this ends up looking:
Please note that Trump = Romney, Carson = Santorum, Rubio = Gingrich, Cruz = Ron Paul, Rand Paul = Perry and Bush = Hunstman.
So Romney still wins. The former Republican nominee would have received 1527 delegates given the caveats detailed above, but also without counting any of the 135 RNC/automatic delegates that are mostly left unbound through the RCP gadget. There are actually 168 of those automatic delegates. Approximately 40% of those delegates are actually bound based on primary or caucus results. The remainder would be unbound. But even without some of those bound automatic delegates, Romney still wins comfortably. His nearest competitor -- Santorum -- ends up with just a quarter of Romney's delegate total. Furthermore, Romney's total competition only amasses a little less than half of his delegate total together.
Well, sure, if a candidate wins 42 out of 56 contests, that candidate is probably going to win the nomination. There's no big surprise there.
The map is nice, but how does this look if we examine the pace with which the candidates accrue delegates over the course of primary season? As it was in 2012, Romney used a win in winner-take-all Florida to establish a lead in the delegate count that widened to roughly a 3:1 ratio after Super Tuesday on March 6, was more firmly established toward the end of March when 50% of the delegates had been allocated, and was solidified by the northeastern primaries in late April.
By the time 75% of the delegates had been allocated -- the week prior to the Texas primary -- Romney all but had the nomination clinched. Texas on May 29 put him over the top. The southern winnowers/northern deciders hypothesis Trende and Byler proffer is basically the 50-75 rule, but perhaps a less precise one, FHQ would argue. But that was basically what the system produced in 2012 under the 2012 rules.
How does this change when the 2016 rules (and calendar) are inserted and combined with the 2012 primary results?
The quick answer is not much. The longer version is it changes but only in a very subtle fashion. With no winner-take-all Florida, Romney would not have broken away from the pack in the same way. Sure, he had the advantage, pulling away from everyone else individually after what would have been the SEC primary. But if we shift our focus to the contested convention scenario, all of Romney competition combined were still neck and neck with Romney after the hypothetical March 1 contests.
That would change after the proportionality window closes and winner-take-all contests are introduced on March 15. Notice how Romney -- already apart from all the other candidates -- separates from even the combined "Anti-Romney" line (in purple). The former Republican nominee's lead only increases from there (after the 50% allocation point), jumping after the northeastern series of contests on April 19-26, but not clinching the nomination until the Oregon primary on May 17. That is about two weeks ahead of where Romney clinched in 2012 under the 2012 rules. Additionally, he does not arrive at that threshold until after the 75% allocation point that is crossed in the northeastern states.
Again, there are caveats to this, but the change in rules from 2012 to 2016 do not bring significant changes to either the outcome of the 2012 Republican nomination race or how the process arrived at its conclusion. What we can say is that the rules changes did not result in a contested convention.
But different inputs in 2016 may alter things. Still, use those winnowing buttons at RCP, folks.
--
1 You can find more details on those models here and here (WSJ).