Only a few days out from the Iowa Caucuses, we still don’t know what to expect. The polls look more like darts tossed by a drunk than a discernible pattern. Worse still, we don’t really even know what they are measuring.

There is one critical factor in a caucus system that makes it different from a primary. Unlike a primary that may have early voting and absentee voting and where the voters straggle in over the course of a 12-hour day, in a caucus you actually have to get your voters to a caucus site within a relatively small envelop of time. Historically, this has meant that a successful candidate had to have a strong ground game composed of door-to-door canvassers, phone banks, and local influencers who could get people out of their houses on a cold and perhaps snowy night in Iowa and get them to mingle with idiots people they hate barely know.

Consultants are telling us that Trump has basically ignored the labor intensive GOTV effort in favor of paid and earned media. Will this work? Who knows. It could. Trump has pulled off a lot of stuff so far that the consultants called crazy. And just because his GOTV effort has gone unnoticed by the political class doesn’t mean it doesn’t exist.

The second question that we have no way of answering beforehand is who are Trump voters, that is, are they traditional Iowa caucus-goers or is Trump relying upon a previously disengaged cohort of voters that he has energized? If the latter is the case, will they turn out to vote.

We have a hint to answers for these questions with voter registration numbers in Iowa. These are the number of registered voters, by party, in Iowa.

Jan. 2015

• Dem: 603,469
• GOP: 623,465

Jan. 2016 (one month out)

• Dem: 584,111
• GOP: 612,112

Now compare that with the last two cycles…

Jan. 2007

• Dem: 609,633
• GOP: 590,187

Feb. 2008 (after Jan ’08 caucuses)

• Dem: 664,658
• GOP: 583,192

Jan. 2011

• Dem: 664,588
• GOP: 622,042

Feb. 2012 (after the ’12 caucuses)

• Dem: 633,747
• GOP: 629,269

Note that registration declined by 7K in the year leading up to the 2008 caucus. Registration gained 7K in the year leading up to 2012 caucus. Registration has dropped by 11K since January 2015. This does not imply that there are a lot of new voters being generated. To the contrary, it indicates that the caucus population will be people who have attended caucuses in the past.

To add further context, let’s take a look at GOP caucus participation:

January 2012 Republican turnout:               121,503

January 2012 Republican registration:        614,913

2012 Republican caucus turnout rate:           19.76%

January 2008 Republican turnout:               119,200

January 2008 Republican registration:        576,231

2008 Republican caucus turnout rate:           20.69%

Of course, registered voters before the caucus aren’t everything. Anyone can register to vote at the caucus site and you can change your party affiliation at the caucus site. This leads us to the interesting nugget obscured in the above tables that permits us make a solid guess at what day-of-caucus registration looks like. One of the sets of voter registration figures is from after the Iowa caucus. The other, lower number, was from no later than the day of the caucus. A solid hypothesis would be that most of the delta between those numbers represents people who registered to vote at the caucus.

In 2008, there were 7K more GOP registered voters after the caucus than before it. Whoa, what happened in Iowa in 2008 that could have caused that?

Huckabee and Romney significantly outperformed in the polls. What did they have in common? A dedicated force of volunteers to get out the vote. In particular, Huckabee’s state chair was Bob Vander Platts (this would be the guy that Donald Trump accused of being a crook. an accusation that was proven to be a lie).

In 2012, there was a healthy 14K more GOP voters after the caucus. Again, what happened?

Those new voters, we can imagine, account for the strong showing of Rick Santorum in the caucuses which catapulted him from being 5 points behind Romney in the last poll taken to winning it.  And Santorum’s turnout, it must be remembered was driven by the Evangelical organization headed by Bob Vander Plaats, who has endorsed Ted Cruz.

Based on this we know the number of registered GOP voters has declined over the past year. There has been no up-tick based on the primary campaigns. From that, we can infer that the great majority of caucus goers who are currently registered will vote not-Trump. That leaves us with same-day registrants. In 2012, about 14,000 people registered the same day. We can make an intelligent guess that most of those were brought out by the old fashioned method of family, friends, and neighbors encouraging them to caucus. The moving force behind that was the Evangelical community. In short, in a turn-out driven caucus, Cruz has control of an experienced GOTV effort that has won both of its last caucuses despite being behind in the polls.

The polls are all over the place in Iowa. But some things are fixed. As Iowa radio talker and Ted Cruz supporter Steve Deace has noted, the polls giving Trump the largest leads are based turnout models that have 300K Iowans attending a caucus. This would translate into close to 200,000 Republicans, or, to put it in perspective you would have more people voting in the GOP caucus as there were votes cast in the Senate primary in 2014 by both parties.

Erick has gone out on a limb and predicted a Cruz victory. Anyone who has followed me here for the past decade knows that a forgotted DailyKos diarist was 100% correct when he described me as someone who would rather pull his own head off than admit he was wrong. But, at the risk of a very messy and painful death, I think I’ll join Erick in predicting that Ted Cruz wins the Iowa Caucuses.