Replies

This was my exact call before the election:

http://www.freerepublic.com/focus/chat/2955102/posts?page=117#117

You can read through my posts over the last two weeks to get a better idea of how I did this and also how I address the “voter fraud” nonsense argument....that issue just comes down to occam’s razor

Basically, after Romney won the primary, I started really digging into the demographics and questioned the assumption by the so-called tv, radio and print professionals, both conservative and liberal, regarding the idea that 2008 was special, different or an anomaly vis-a-vis the minority and female voting patterns of that year.

The political meme of both sides was that 2008 was a unique situation given that it was the first black candidate to win a major party primary. Then I started hearing most on the right say that the polls are being weighted improperly toward the democrats (2010 midterms were supposedly evidence of this) and knowing that 2010 was a midterm election which gives republicans an advantage, I tried to quantify this assumption.

That then brought me to the 2010 census where I started pouring over the 2010 census briefs released by the US Census Bureau in 2011. After a lot of comparisons to 2000, I came to a staggering conclusion: a democrat could win with only 38% of the white vote...I was pretty stunned by this, that was staggering. Obama got like 43-44% of the white vote in 2008 which meant he had breathing room to lose more white voters! Here’s the best synopsis of the data I was looking at:

http://2010.census.gov/news/releases/operations/cb11-cn125.html

Basically, as the csmonitor put it: “primacy of white male voters has passed”

So I began with a simple but unverifiable assumption, ie, let’s assume that each race/sex of voter turns out in equal percentages. Meaning if 60% of whites turn out, then 60% of latinos, asians and blacks would turn out. Since Obama’s team targeted latinos with the children of illegals executive action, the black vote was tied up and Obama was making an argument to appeal to manufacturing voters in the rust belt, I figured this assumption was shaping up.

The numbers lined up nicely. If you compare the census numbers in the link above to the actual turnout, it was spooky how much they mirror one another. For example, 72% of the population, according to the census 2010 is white and 72% of voters on Nov. 6th were white. Same for black population and turnout, 13%.

The country was changing at a steady NON-WHITE pace since 2010 and I concluded that based on demographics, the 2008 dem advantage was NOT an anomaly!

Next I looked at the outlier since it was such an easy target, Gallup. They were assuming 78% of the electorate would be white.....and there was the red flag. Next I looked at Rasmussen who was getting dem sample numbers in his surveys and bending the number to match his own assumptions , a really bad idea for any pollster but in light of the new demographic makeup of the US and therefore the electorate, these guys had NO idea what they were doing. Conclusion: throw gallup and Rasmussen polls in the trash.

So I started looking at other national polls who when they sampled 1000 people and 38% said they were dems and 32% said they were republicans, actually kept those numbers in the poll....I have written PAGES on the BS argument of skewing polls based on party ID and how ridiculous that is to anyone that has the slightest idea of how sampling in science, politics, marketing, etc works.

Anyway...I started looking at these national and the state polls that released internal data and formed a nice little model and weighted them according to my own assumptions. Were their demographic data plausible in light of new census data. If they fell within a certain range, I kept them and then weighted according to automated phone, live interviews, cell phones included and if so what percentage, internet poll.

As I posted many times before, I messed up on a few things as evidenced by my Obama +1.8% result and Obama getting 290EV and possibly 303EV. In hindsight I can see that compared to people like Nate Silver, I feel like a 5th grader doing arithmetic while he’s doing calculus.

But keep in mind I was living in the same world as most people on FR. I was being bombarded by pundits, columnists, tv interviews and radio personalities that were making me doubt everything I was doing. I can’t express how difficult it was to not only come to the conclusion I saw on my screen but actually post it here and stick to it. But as a person that had to take graduate level statistics and learn to look at data objectively in very biased circumstances, I was 99% sure that I was right with regard to the data even though people around here were throwing rotten tomatoes at me left and right.

I think if I weren’t in the conservative worldview, I may have allowed myself to believe the Colorado and Florida numbers and been more accurate with my final projection....but being human, I may have threw out or improperly weighed some polls based on my conservative heart...after all, do you have any idea how hard it is to weigh a Daily Kos poll so high as a conservative? Turns out they were dead on, one of the most accurate polls...but it’s a hell of a lot easier to swallow after the fact than before.

I know that’s long but I hope it helps answer your question.