Onsight/Keating Research in Colorado: Obama 48%, Romney 44%

8 thoughts on “Onsight/Keating Research in Colorado: Obama 48%, Romney 44%”

1. Now, let’s make sure those Dems actually get out to vote.

   Log in to Reply

   1. From a Democrat polling shop. Once again, if Rasmussen gets no respect here, I’m not giving Keating any.

      Log in to Reply

      1. Speak English, not Republican.  

         Sorry I set the bar so high.  

         Log in to Reply

2. follows both the VP announcements (in Ryan’s case, unusually small) and the conventions (still to be seen) Those bumps fade and small ones fade faster and more completely. Also poll averages are much more reliable than individual polls.  So lets all try to stay on an even keel for the next few weeks.  There will be better days and worse days all round.  Hold the dancing.  Hold the hair tearing. No hyperventilating. At least until the R and D bumps have a little time to fade and we can
   
   see more clearly where we really are. This  looks like it’s going to be a nail biter and we don’t all want to be wrecks by November.  

   Log in to Reply

   1. To prevent Ryan from seeing the “bump” he should have. But Democrats won’t be able to keep the outrage ginned up until November. Once the dust settles, the American people will be more receptive. With Paul’s help, the GOP will turn the Medicare debate back on the Democrats.

      We are the custodians of a responsible future.

      Log in to Reply

      1. That was supposed to be a secret, but you glommed onto our program.

         The Dems prevented the expected bump.  What a fucking joke you are.  You should not be allowed to vote.

         And the Pubs are going to save Medicare?  Pubs have always hated our great social programs and now suddenly you care?

         Gimme an effin break, moron.  

         Log in to Reply

3. The real key to understanding these polls in a race like this (incumbent running for reelection) is not so much where the candidates are, but where is the incumbent.  This is because undecided voters tend to vote against the incumbent.

   Look at the release and you’ll see that Obama and Romney are roughly equally popular (Favorable to unfavorable ratio is virtually the same).  So then how will those 6% undecided go?  History says most will go to the challenger.

   Also, the other key in looking at these results is their underlying demographic mix.  The NTY/CBS News/Quinnipiac poll from a few weeks ago that showed Romney leading, was a terrible piece of research despite being from respected organizations.  They said 38% of the electorate was 65 or older.  In 2008, exit polls show those 65+ were only 13% of the electorate, and were the only age group Obama lost.  In 2010, only 22% of CO voters were 65 or older.  So 38% is just awful.

   Looking at the demos in this poll suggest it might be a little more favorable toward Obama than the overall electorate, or at least compared to the 2008 electorate from exit polls.

   First, it has 53% women, but the exit polls from 2008 say the electorate was 50% women (http://www.cnn.com/ELECTION/2008/results/polls/#COP00p1).

   Second, this poll has a relatively weak screen for likely voters, boosting the pool of those less likely to vote.  If you look at the screening question, they don’t just rely on voters listed on the voter file.  If the known registered voter is not available, the pollsters ask for the youngest person in the household who claims to be registered to vote (past studies have shown people lie about their likelihood to vote), and then it keeps people who say they have a 50-50 chance of voting.  

   Of course, there are no crosstabs available to further analyze the data to see if those 50-50 voters benefit Obama or Romney.  That’s perfectly common, but unfortunate.  Polls used to be based on random sampling methodology, but that hasn’t been an effective form of research in over 10 years because of the explosion in the use of cell phones and its concentration among young people.

   Third, the party registration numbers are a couple points more Democratic than active registration numbers suggest they should be.  The latest Sec of State registration data show Democrats account for 31.5% of active registrants, whereas Republicans account for 36.5% of active registrants.  That’s a 5 point GOP advantage.  This poll shows 33% are Democrats, 36% are Republicans — only a 3 point GOP advantage.

   My firm’s poll from April (http://www.peakcampaigns.com/content/Results/Results.asp) showed Obama in a similar position – leading but shy of 50% support) but included a 5 point GOP registration advantage.

   Finally, the poll results released don’t say what the geographic distribution of interviews looks like.  Unaffiliated voters from Denver are very different than those from the Western Slope or Eastern Plains.

   Again, that’s where having crosstabs of the demographic and geographic makeup of the sample would be helpful in telling how pro-Obama or Romney this poll would be.

   I highlight these issues not because Keating does bad research.  I know Chris Keating, and he is a talented, quality pollster.

   Instead, I think we need to move the debate over and analysis of these polls beyond margin of error (which is not a meaning metric these days) and house effect.  If pollsters release more data about their samples, we can more accurately judge their public results.

   Log in to Reply
   1. Log in to Reply