Reporters like a good story. In theory, they like a good story based on facts. When an event happens, they like to be able to say what it means. The rules of the nomination process, however, are inconsistent with the way that reporters normally operate. Particularly in caucus states, the event that reporters want to treat as the election is merely the first step in the process. So the reporters make estimates and report those estimates as if they are fact. In some cases, these estimates are good. In others, the problem is readily apparent up front.
The problem used to be worse. In previous cycles, while Republican caucuses reported a presidential preference poll, that preference poll was just a beauty contest and what really mattered was the unstated preferences of the individuals elected as delegates to the next round of the process. So when the media treated those preference polls as an estimate of the delegates from the state that estimate had nothing to do with reality. The Republicans have changed the rules for this cycle. If a state has a preference poll, that poll binds the delegates (with three major exceptions — Pennsylvania, Illinois, and West Virginia — which elect, at least, congressional district delegates directly). For the three states that directly elect delegates, the reporters are likely to get the story right and look at the pledges of the delegate candidates in Illinois and West Virginia. They might screw up Pennsylvania in which none of the delegates are technically bound to any candidate (but might do the legwork to find out who the delegates actually support). (On the Democrat side, the delegates are bound based on the presidential preference vote.)
Some states, however, are not having a preference poll which will make things harder for reporters. In Colorado (March 1), Wyoming (March 1), Guam (March 12), American Samoa (March 22), North Dakota (April 1), there are no preference polls. In Colorado and Wyoming, if a delegate candidate declares a preference, that delegate is bound by that preference if elected. In American Samoa, the convention will choose whether to bind the delegation by resolution. In North Dakota, the convention can decide on an apportionment formula.
On the Democratic side, the issue is the caucus states. Basically, there are two types of caucus states on the Democratic side — those that bind delegates by a straw poll at the caucus (sometimes referred to as a “firehouse” primary) and those that merely use the preference vote at the first meeting to elect delegates to a subsequent meeting. The Democratic caucuses in Minnesota (March 1), Kansas (March 5), Utah (March 22), Hawaii (March 26), Washington (March 26), Puerto Rico (June 5), and North Dakota (June 8) will pledge all delegates based on the results of the straw poll at the initial meeting (although some allocate the state-wide delegates in proportion to the district delegates won instead of the raw vote). The Democratic caucuses in Idaho and Wyoming are hybrid states with the district delegates allocated based on the straw poll at the initial caucus but the state-wide delegates allocated based on the results at the state convention. Finally, Iowa (February 1), Nevada (February 20), Colorado (March 1), Nebraska (March 5) , Maine (March 6), and Alaska (March 26) only use the preference vote at the first meeting to allocate the delegates for the next meeting.
For this last group of states, while the results of the first meeting are a good estimate for the final delegate allocation, that estimate will almost certainly be wrong at the margins. The biggest changes are likely to occur in Iowa for two reasons. First, the Iowa results are reported as “state delegate equivalents.” Thus, the number of delegates elected from a precinct to a county convention is multiplied by the ratio of (county delegates to the state convention divided by delegates to the county convention) . This results in fractional delegates. The county convention does not elect fractional delegates; so those fractional delegates in each county will be rounded away. To use Adair County as an example (only because it is at the top of the list), Clinton won 28 county delegates to Sanders’s 22 in a county that elects 3 state convention delegates. At the county convention, Clinton should win 2 delegates to Sanders’s 1, but the count of state delegate equivalents is 1.72 to 1.28. In theory, it is unlikely that one candidate will have the larger remainder in all of the counties (and thus round up their fraction in all of the counties), but it is likely that the rounding will change the numbers in some fashion. Second, because Iowa is earlier in the process, candidates who have dropped out of the race are likely to have some delegates in the initial result, but most of those delegates will switch to other candidates.
Based on the numbers reported from the precinct meetings in Iowa, it seems like even accounting for the rounding issues and the potential influence of uncommitted and O’Malley delegates at the county convention, that the district level delegates are unlikely to be impacted by any changes. The three districts with eight delegates are effectively tied so it is unlikely that any change would create a large enough margin to change the split from 4-4 to 5-3. In the remaining district, Secretary Clinton appears to be far enough ahead that the changes are unlikely to swing that district from its current 4-3 split in favor of Secretary Clinton. Similarly, the state-wide race is close enough that the PLEO delegates should stay at 3-3. The one potential change is for the state-wide at-large delegates. Current estimates use a 4-3 split in favor of Secretary Clinton based on her narrow lead after the precinct meetings. There are enough O’Malley and uncommitted delegates to the county convention that Senator Sanders could gain the lead after the county conventions (if enough of the O’Malley and uncommitted delegates switch to Sanders). That would change the delegate breakdown in Iowa from 23-21 in favor of Clinton to 22-22. We should have a firmer handle on Iowa after the county conventions on March 12. (For anyone interested in calculating what might happen on March 12, the full breakdown of precinct delegates won by each campaign can be found here. None of the congressional districts cross county lines.)
Because it is now a two-candidate race, the chance of a swing in the remaining states will be primarily to the rounding issue at any second-tier convention — an issue in Colorado, Nevada, and Nebraska. In Nevada, the state-wide results appear to be clear enough that rounding will not be an issue (and there are not enough uncommitted delegates at any county convention to make a difference of more than one state convention delegate). Additionally, two of the four districts are entirely inside Clark County, so there is only one delegate count (the delegates in Clark County from that district to the county convention) to round. Currently (both because the numbers are not quite final and because of the issue of converting county convention delegates to state convention delegates), the only district that might be effected is the Fourth District. Secretary Clinton is just over the threshold needed to win 4 of the 6 delegates. Depending upon what happens in the county conventions, those numbers might change just enough to move to a 3-3 split. (Some of the media outlets recognize how close this count is and are reporting the count as 19-15 with 1 delegate undecided rather than as 20-15.) Aw with Iowa, the picture should be clear after the county conventions meet (on April 22).
A good example of a county where things might be interesting on March 12 is Crawford County. According to the precinct results, there are 60 delegates to the county convention who will elect 5 delegates to the State/Congressional District Conventions. Assuming no switches, there will be 27 Clinton delegates, 25 Sanders delegates, and 8 O’Malley delegates. It takes 9 delegates to be viable. Do one of the Clinton or Sanders delegates join O’Malley to force a 2-2-1 split or do the former O’Malley delegates get to decide who gets the fifth delegate between Clinton and Sanders. Given the current state-wide projection of Clinton up by 3 delegates with O’Malley having 8 delegates, it is the counties like Crawford where O’Malley has 0.8 delegate equivalents that will decide who ultimately gets the extra national convention delegate at the Iowa state convention.