An untrained observer, looking at the RCP Average of polling conducted in the Florida Senate race, might conclude that Marco Rubio and Charlie Crist are in a virtual tie.  A fool might look at these numbers and conclude Charlie Crist is more likely to win.  However, a person with an even casual understanding of how polls are conducted will understand that this polling shows Marco Rubio to be a solid favorite in this race, with a lead roughly as comfortable as that of Kelly Ayotte in New Hampshire and Rand Paul in Kentucky. 

The explanation for this, of course, lies in sample of persons polled by the respective organizations polling this race.  From the beginning of this race, Quinnipiac University and the St. Petersburg Times have polled registered voters, a set which purports to winnow out at least those adults not eligible to vote.  Rasmussen and Mason-Dixon, on the other hand, have consistently polled “likely voters,” which purports to be a subset of registered voters who are judged to be more representative of the people who can be counted on to actually cast a vote on Election Day.

Now, screening for likely voters is a tetchy business, and polling companies have varied methods for accomplishing this screening function.  Some ask respondents to self-report likelihood of voting.  Some judge likelihood of voting by objective factors such as knowledge of their polling location, etc. Others do some mix, checking self-reporting with objective facts.  Rasmussen is known to have a very stringent screening process for likely voters.  People who are knowledgeable about polling debate about such things as whether Rasmussen’s screening process is perhaps too stringent; those who are ignorant of polls (or disinterested in the truth) argue that Rasmussen deliberately cooks the books in order to influence the election narrative. Nevertheless, it is clear that screening for likely voters, through whatever method, to some degree eliminates marginal voters and seeks to poll those more “tuned in” and “motivated” to do what the poll is attempting to measure: actually go to the polls and vote.

Over the last six weeks, a clear picture has emerged among the nationally-known and reputable polling organizations that have polled the expected Rubio-Crist-Meek race (Rasmussen, Q-Poll, PPP, Mason-Dixon, and Reuters-Ipsos): organizations polling likely voters showed Rubio faring better, by an increasingly widening margin.  Reuters/Ipsos polled RVs in early July and found Crist ahead by 7. Rasmussen polled LVs July 21st, and found Rubio ahead 2.  Q-Poll polled RVs in late July and found Crist ahead 6.  Rasmussen polled LVs August 9 and found Rubio ahead 5.  Mason-Dixon polled LVs August 9-11 and also found Rubio up 5.  Q-Poll polled RVs August 11-16 and found Crist ahead 7. 

A seasoned poll-watcher seeing this trend would conclude that the overall polling numbers were masking a troubling trend for Orange Charlie: the marginal voters prefer Charlie Crist. The people likely to get out and vote (particularly in a non-Presidential year) increasingly preferred Marco Rubio.  No polling company has illustrated this trend more precisely than Democratic pollster Public Policy Polling, which has by all accounts acquitted itself well thus far this cycle.  In mid-July, PPP polled RVs and found Crist ahead in a 3-way matcup with Meek by 6. Over the weekend, PPP polled LVs and… found Rubio ahead by 8 in the same matchup. In the absence of a universal trend towards Rubio in all polling, this dramatic 15-point shift in a poll conducted by the same organization can only mean one thing: at present, Rubio voters are more tuned in, more motivated, and by far more likely to vote.  This spells DOOM for Charlie Crist; DOOM which will be more manifest as election day nears and all polling organizatoins switch to likely voter models.

At this point, Crist’s main hope is that Obama and the DNC fundraising apparatus go all-in, in an attempt to convince Democrats to bail on CBC member-in-good-standing Kendrick Meek. At this point, that seems doubtful to happen, especially in the absence of an explicit promise from Crist to caucus with the Dems; a promise Crist cannot give without alienating a significant portion of his current voting base.  In the final analysis, then, Marco Rubio should be considered a heavy favorite to win this race, a fact which will become clearer as election day approaches.