Political Blog

Find posts by keyword
Find posts by date


New polls suggest tightening Senate race

New polls suggest tightening Senate race

Rand Paul, Jack Conway

by Joe Arnold


Posted on September 16, 2010 at 12:57 AM

More polls are out this week in Kentucky's U.S. Senate race, which might go down as the most polled election in Kentucky history.

Public Policy Polling for Daily Kos shows a seven point lead for Paul, 49% - 42%:

Interestingly, Kentucky may be one of the few places were Democrats don't suffer from an intensity gap. PPP's Tom Jensen notes that the poll shows a Kentucky electorate that went for McCain by 13 points, when the Republican actually won it by 16. So Rand Paul is motivating Democrats, and given them a three-point intensity gap. However...

The Democrats’ problem in Kentucky isn’t the enthusiasm gap- Rand Paul took care of that for us.  It’s because after showing some wariness earlier in the summer, when we had the race tied, the McCain voters have pretty much all gone home to Paul.  In late June he was getting 70% of their votes and now for all his missteps he’s getting 80% of their votes. In a state that went as strongly Republican last time as Kentucky Conway is going to have to be able to pick off more of their voters than that to have a path to victory.

A new internal poll from the Jack Conway campaign which shows a statistical dead heat, Paul 47% - Conway 45%, got the attention of the Washington Post's Greg Sargent:

Now, obviously you'll want to be cautious about internal polls. But this comes as the NRSC has gone up on the air in Kentucky with a new spot hammering Conway, perhaps a hint that the race is closer than we've been led to believe:

A GOP official told Ben Smith that the ad is not a sign that national Republicans are worried, but that the goal is to "put this race away early" and to try to get national Dems to spend cash on the race that might otherwise be spent on vulnerable incumbents.

 Here is the polling memo from Conway's pollster:

To: Interested Parties
From: Pete Brodnitz
Date: September 10, 2010
Re: New Kentucky statewide poll of likely 2010 voters

New BSG Poll Shows Conway and Paul in Statistical Dead Heat

Difference -2 / Democrat Jack Conway 45 / Republican Rand Paul 47 / Don’t know 8

Our recent September 2010 poll of likely 2010 general election voters
shows the race for Kentucky’s U.S. Senate seat is statistically tied,
with the results within the margin of error.

BSG Poll Is Now Third Poll This Month to Show the KY Senate Race
Statistically Tied

In addition, BSG’s poll marks the third publicly-released poll of
registered voters in the past week that shows the race for Kentucky’s
U.S. Senate seat to be highly competitive:
       A poll of likely general election voters by John Anzalone of
Anzalone-Liszt Research found the race for Senate to be a statistical
dead heat: Paul 48%, Conway 45%.

       A poll of registered Kentucky voters by Opinion Research Corp. on
behalf of CNN and Time and was released this week showed the race tied
at 46%.

With the race continuing to be competitive despite Rand Paul’s
increasingly high public profile, it suggests Conway is well
positioned to win in November because Rand Paul doesn’t understand

If the November election for U.S. Senate was held today and the
candidates were...for whom would you vote?
Sept. 2010
Difference -2 / Democrat Jack Conway 45 / Republican Rand Paul 47 / Don’t know 8

Difference / Jack Conway / Rand Paul / Undecided
BSG: -2 / 45 / 45 / 8
Anzalone: -3 / 47 / 48  / 7
CNN: 0 / 46 / 46 / 8

Conway - Sept 2010 - State of the Race
Why All Polls are Not Alike
The BSG poll was conducted using a sample of registered voters who
were screened to ensure that they are likely to vote on Election Day.

Recently, a series of public polls have been released that either fail
to interview registered voters (Rasmussen and SurveyUSA) or that
interview registered voters but fail to screen to ensure that voters
are likely to vote (CNN). Both approaches are flawed – particularly
this close to Election Day.

In addition, we use live interviewers to ensure that we have a
qualified respondent (i.e. registered voter) on the phone. That is not
the case with polls that use automated interviewing (Rasmussen and

The only recent poll that also called registered voters and screened
to ensure that they are likely to vote was Anzalone-Liszt. Their
results mirror the results of the BSG survey.
The data below shows why a poll in a state such as Kentucky should
draw from a sample of registered voters.

As the chart below shows, the number of adults 18 years or older in
Kentucky as nearly 14% larger than the number of registered voters
(outnumbering them by about 400,000), and is almost double the average
number of voters in the last three federal elections:

There is about a 400,000 person drop between the number of adults
there are in Kentucky and the number of adults who are registered. And
the number of adults who typically vote in Kentucky represents less
than half of the adult population. A poll that surveys adults
therefore has the potential to include a large number of people who
are not really going to vote and whose attitudes maybe different than
those found in the voting population.