Report Card – National Polls – 10/1/2012

Today’s Notes – 10/2/2012 -

CNN / Opinion Research Poll Shows Essentially No Movement since last poll

Polls put out by CNN / Opinion Research are fairly low quality and not very dependable. Their less than 1000 sample size (783) and poor sampling model along with a failure to release full demographics – no race or sex information is included although the Cross Tabs show they did gather this information – hardly makes the results dependable. But if we look at the previous poll they did from earlier in September, it would appear that Romney is gaining on Obama. But the truth is, these polls are not “apples to apples” because of the differences in samples and the fact that the previous poll was Registered Voters as opposed to Likely.

The skew of the first poll was Democrats +9 and Obama had a 6 point lead. This latest poll skews only +6 Democrat and Obama has a 3 point advantage. Judging two poorly conducted polls from the same group with different sample variances is not an exact science.

Note – For the calculations and a discussion of each poll, please click on the poll name. All of the figures we used in the calculation as well as the raw data are either shown or linked. The last column, Normalized Result create a standard benchmark to which the polls are “zeroed” to eliminate discrepancies due to polling more Democrats or Republicans as well as improper consideration of Independents. The normalized results should track movement without the skew from improper sample mix of political affiliation.

(Today’s RCP Snapshot 10/9/2012)

Click on the name of the poll in the table below to see the analysis and breakdown of the polls.

(Note: Rasmussen seems to be having a problem posting the latest cross tabs.This section will be updated when they correct the problem.)

Grades and Skew Table

Background

Real Clear Politics has become a site that is popular with the media because it is not a single poll, rather they “average” all of the polls which, at least in theory will tend to smooth over the polls that tend to lean too far left or too far right. And there was a time when averaging produced meaningful results. The problem is, as the Mainstream Media has relied more and more on the RCP Averages, political operatives have started manipulating the polls to skew the averages and make it look like a candidate is doing better than he really is. They do this by oversampling Democrats to make the poll outcome more favorable to Democrats and the same is true of Republicans.

And Democrats seem far more guilty of this than Republicans. Some of these polls favor Democrats by 10% or more. And despite the fact that America is pretty evenly divided between Democrats and Republicans, it is extremely rare to see a poll that favors Republicans at all.

And Real Clear Politics has no standards to keep out these intentionally skewed polls. They pretty much take them all no matter the sample size, political leaning, or the methods used.

Political Poll Check was established to perform two tasks. Grade the polls and adjust the results to accurately reflect the makeup of the country (or the state if it is a state poll).

And we will toss out the polls that do not pass. Anything “F’s” will not be used to calculate the real average.

Real Clear Politics had an opportunity to insist on quality polls in order to be included in their average and they have, instead, allowed the polls to be manipulated without so much as a footnote to their readers. We believe this gives a false reading.

Garbage in, garbage out.

This chart shows how badly the polls over sample Democrats. These are all polls that passed with a “D” or better.

This chart shows the “Skew Factor” of the last 5 polls. The skew factor takes into account the deviation from the polled normal average of  Democrats, Republicans and Independents determined by Rasmussen. The over sample rate does not always tell the complete story.

(Note: The Blue Watermark is the spot all 3 political affiliations should fall. 31% to 36%.)