Lies, Damned Lies . . . and Pollsters?

I am certain of an Obama victory come this fall; I am just as certain that the pollsters are either unable or unwilling to accurately gauge the electorate this year. Their inability to poll accurately became legendary throughout the Democratic Primary process and the erratic nature of the polls since only tells us that they have not gotten a handle on what to do.

Regardless of their intentions, their daily, weekly and monthly polls do have an effect on the voting populace. In a very real sense, they do not just report on how the electorate is feeling - they shape how the electorate feels.

They know (this) http://books.google.com/books?id=DdSPsxHX5BQC&dq.

This tells us that their actions are in furtherance of whatever their goals are. While we are not privy to their internal discussions, we can know this much for certain: their goal - stated or unstated - seeks to reduce the enthusiasm of the Democrats and increase the enthusiasm of the Republicans, in a pathetic attempt to prevent the change we as a nation seek to make from coming to pass.

It will not work.

They will not succeed.

Their attempts are growing more patently obvious by the day.

But all of that is mere words and in truth, purely preamble. Here is the analysis that proves the above hypothetical statements true.

From the latest AP Poll (http://www.ap-gfkpoll.com/), we see the following top-line numbers:

  • McCain - 48%
  • Obama - 44%
Looks bad for our hero, but those Hollywood serials would not have kept us on the edge of our seats if we could tell from jump that the good guy was going to win.

The first lie these numbers tell us is they are based on "likely voters". Now, every pollster has their own "secret sauce" for how they determine "likely voters", but suffice it to say for the external viewer that it consists of a formula based on previous voting experience; demographic data; and enthusiasm. Translated more directly (and we are family here, no?), that means newly registered voters have their responses discounted; impact of white voters is raised - due to their historically higher voting rate compared to other demographics; and adding Sarah Palin to the ticket has boosted Republican enthusiasm - which was beyond lackluster for John McCaain. Three debatable assumptions - as they look backward and not forward - that pollsters treat as gospel.

But there is more wrong with this poll:

View Picture 10 and you will see that the Democratic voters total out to 33% of the "likely voters" polled and Republican voters totaled out to 31% of the same population. Nationally, Democrats have a more than 11M - (according to the AP) http://ap.google.com/article/ALeqM5iyCjg56QEYy3r8Gz2X09TnpaWOMwD9317JU00 - registered voter advantage over Republicans - and that is in states where registration also capture party affiliation. Other news sources (USA Today from 2004) http://www.usatoday.com/news/opinion/columnist/neuharth/2004-01-22-neuharth_x.htm say that the Democratic advantage is almost 20M - and that lead is only growing with each passing day, as the Obama campaign continues to press for new registrations. And as these folks are registering just for this campaign, they have to be considered more likely to vote! Any poll that does not show a Democratic lead in the poll responders, it flat out false.

But there is even more wrong with this poll:

View Picture 11 and you will see that of the likely voters questioned in this poll, a full 10% more voted for George Bush in 2004 than John Kerry! Now, we are all sad that Kerry lost, but he did not lose by 10%; in fact, his margin of defeat was a mere 2.4%. This means that what this poll is actually showing, is that some folks who voted for George Bush in 2004 are planning on voting for Barack Obama in 2008; I bet you did not see that in any headlines.

But there is still more wrong with this poll:

View Picture 12 and you will see that the pollster is once again forecasting 17% of the electorate will be in the under-30 voter group - which is what we saw in 2004. Ask yourself if you - the average observer of this year's political season - thinks that more, less or about the same number of voters in the under-30 demographic will hit the polls this November? You can take as much time as you like in answering this question and remember - this will count as 25% of your final score!

But there is yet one more thing wrong with this poll - and I am not even a statistician (although I did stay at a Holiday Inn Express last night):

View Picture 13; here you will see that the pollster is once again forecasting 7% of the electorate will be Hispanic/Latino, as it was in 2004. This almost goes without saying but the primary season alone showed a huge increase in the turnout of Hispanic voters. To believe that they have all decided to vote in the primaries - only to return to previous levels in the fall - is an assertion without evidence in support of it (others might call it an out and out lie, but I am too charitable for such strong language).

I hope this has shown that this latest poll from AP/GFK is more cotton candy than high cotton; I encourage everyone to go to the poll themselves and see just how many others holes are in this thing. All snapshots of poll data taken from the AP source data here: http://www.ap-gfkpoll.com/pdf/AP-GfK_Poll_91208_Topline_findings_final.pdf.

For comparison, you can find details on the 2004 election here: http://en.wikipedia.org/wiki/United_States_presidential_election,_2004.

Sphere: Related Content

No comments: