09/24/2010 12:22 pm ET Updated May 25, 2011

Polling Delaware?...the Missing Horse Race

We are now knee deep in headlines using the all too famous "horse race" poll question; to paraphrase: "if the election were held today, who would you vote for?" The results are broadcast and reported across many different mediums, and the goal is to shed light on the competitiveness of the election in question.

Segue to Delaware.

Since the upstart Christine O'Donnell defeated long time Delaware politico Mike Castle in the Republican primary, pollsters operating in Delaware--including Pulse Opinion Research (for Fox News), Public Policy Polling, Rasmussen, and CNN--have asked the horse race question comparing the Democratic challenger Chris Coons and the Republican O'Donnell. But, the same polls have included horse race questions comparing Coons and Castle.

The Coons versus O'Donnell horse races consistently favor Coons, and the Coons versus Castle horse races consistently favor Castle.

Wait, aren't we missing something? Where is the Castle versus O'Donnell horse race? Isn't this where the real story lies?

We've known since February that Coons is favored over O'Donnell, and Castle is favored over Coons; the important question seems to be whether Castle could lose to O'Donnell twice or stage a dramatic return to power by defeating both Coons and O'Donnell in the general election.

Perhaps the Castle campaign is following the polls to see if his favorable horse race gap against Coons stays the same, or grows, as the campaign goes on. He has not yet endorsed either candidate, and his poll numbers are still strong, at least as they relate to Coons.

CNN's most recent poll (Sept. 17-21) of 703 likely voters in Delaware (margin of error +- 5%) actually shows Castle with a 24% point lead over Coons among Independents, a 13% point lead among moderates, and while he is down by 42% points among Democrats, he still garners 33% points worth of their support. Add this to his 72% point support among Republicans, and his candidacy appears to still be alive and well.

See CNN's results here.

These are very favorable statistics for Castle, and the recent news that he has not ruled out running a write-in campaign has added fuel to the fire that he's not done. The air is also filled with questions about whether a Castle revival would take moderate and Independent votes from Coons, or take Republican votes from O'Donnell. Still, with no post primary data on his competitiveness against O'Donnell we are left to wonder why pollsters are missing this horse race.

There is one interesting nugget from the CNN topline data that may shed light on what demographic segment might make the biggest difference in Castle's decision. Men (58% to 38%) and women (56% to 37%) prefer Castle over Coons, suggesting no gender effect among these two candidates. However, there is a huge gender effect in the O'Donnell versus Coons matchup. Men prefer O'Donnell over Coons (49% to 46%) by only 3% points, but women prefer Coons over O'Donnell (61% to 32%) by 29% points!

Coons' advantage among women is helpful if he faces O'Donnell, but not Castle. The absence of favorable support among women in the Castle-Coons' matchup may speak volumes about what Castle's next move could be. Perhaps in the coming weeks we'll learn more about Castle's intentions, and get some additional data about the missing horse race.