2010/12 polls underestimated Democratic turnout.

[Dawson Leery]

http://www.economyinperspective.com/yahoo_site_admin/assets/docs/Senate_2010.276215848.pdf

[Stallion]

1. Good stuff

Thanks. I like taking a look at RCP because of the presentation but about 80% of their polls have Republican leans. 538 at least attempts to account for the bias

[1StrongBlackMan]

2. This is something that I have been saying ...

since forever ... And I suspect that it will be more pronounced in 2014, than ever.

[pstokely]

3. but the polls may be overestimating Dem turnout this year?

GOTV is the key

[demwing]

4. is that a legit question

or just reckless innuendo?

[Igel]

5. No, it's just a reasonable question.

Polls look to historical turnout to predict the future. That's the assumption--the past predicts the future.

Polls aren't just calling a bunch of people and looking at the numbers. You break the population you call down into demographics--age, sex, race/ethnicity, income, etc., etc. Then you fiddle with the results for each subpopulation to make it match what you expect the electorate to look you. The people you called were 5% black and you think that 10% of the electorate is black, you double that demographic's contribution to the final poll number. Repeat for every subgroup.

If (D) turnout increases, then the assumption's false and the adjustment that pollsters make to the raw data is wrong.

We see that happen in exit polls all the time. They start with one set of assumptions, but as reality provides more and more data about what's actually happening it becomes clear that their assumptions were wrong. So they adjust their assumptions to match reality based on what data they have.

[damnedifIknow]

6. Question

If there is no Democrat on my mail in ballot that is running for a particular office do I have to vote for the Republican in order for my ballot to count?