Sunday, October 16, 2005

New polling on Referenda C and D put issues behind. Ritter (D) favorite for Govs race.

(updated 10/18/2005)

A new Denver Post poll out today disagrees with a Rocky Mountain News poll released two days ago.

Here are the results of the Denver Post poll:

Mason-Dixon Oct 11-13
Referendum C Yes: 47% No: 44% Undecided: 9% MOE:4%
Referendum D Yes: 44% No: 45% Undecided: 11% MOE:4%

These results are interesting because they contrast somewhat with the results from the Rocky Mountain News poll. The RMN poll put support for C at 49% and D at 50%. The results of the Denver Post poll are within the margin of error of the earlier RMN poll, but they pull the average significantly down.

As a result we are further modifying our election projections:

Referendum C is Too Close To Call Yes: 49.4% No: 50.6% Probability: See Footnote.

Referendum D is Too Close To Call Yes: 48.7% No: 51.3% Probability: See Footnote.

The poll also asked who you would vote for in the 2006 Governor's race:

Mason-Dixon Oct 11-13
Ritter (D) vs. Beauprez (R) Ritter: 42% Beauprez: 36% Undecided: 22% MOE:4%
Ritter (D) vs. Holtzman (R) Ritter: 41% Holtzman: 30% Undecided: 29% MOE:4%

Update: We've updated our Referendum C and D homepage so that you can look at the polling history of the campaign and see how the campaign has evolved over the last few months.

Our last headline said that Referendum D was dead, that was a typo. Sorry for any confusion.


The reason why the races are too close to call is because the predicted vote percentages are the mean points of a zone of probability. For each predicted value (excuse the mathematics) the 95% zone of probability is -/+ 2.5.

For example, our current projection on Referendum C is that it will get 49.4% of the vote. However we can only predict to 95% accuracy that the percentage of the vote that Referendum C gets will be between 46.9% and 51.9%. The same spread applies for Referendum D. A good rule of thumb then would be that if the predicted result falls below 47.5% then we will be able to be 95% confident that the issues will fail (and vice versa if the predicted result is above 52.5%.

As for the predicted probabilities, both measures have a slightly higher than 50% probability of failing.

1 comment:

  1. Inspiring blog. I love finding blogs this good on
    the internet, when I have the time. I'm going to go
    back to it!
    I beg of you, just check out my va virginia house cleaning blog.