A new poll says Angus King and Obama are winning in Maine. Most people would say, “Duh. I live here. I know that.”
Naran says the poll is a fraud because.. well… because she imagines the mathematical field of statistics is bound by the laws of Naran’s Common Sense Kitchen Table Whatever I Think Off The Top Of My Head Sounds Good worldview.
She quotes the Press Herald and then adds her own little commentary (in italics):
The company called 615 voters around the state from June 20-25. It used random landlines and cellphones and conducted personal interviews.
The results were statistically weighted to reflect the demographics of the state’s voting population.
The poll has a margin of error of 4 percentage points for results based on the entire sample, with larger margins for subgroups such as independent voters or older voters.
An additional 100 women were polled to provide additional data on women’s perspectives.
So – they poll an additional 100 women, which, in a 600 person poll, constitutes a large percentage. Sounds like they cherry-picked what they wanted, and then “weighted” it, and then got the results they wanted. Imagine that.
My God! They did not just poll 100 additional women and add them to the pool, thus unbalancing the poll with an overabundance of women. They did not do that. No, Naran, hush. They didn’t do that. Listen, I’m just saying- no, that’s not it. That’s not what they did. Really. Please listen. Oh my god. I want to kill myself.
No, Naran, they took a sample of the state, which was 600 people. Only about half were women. When you consider answers the woman-only demographic gave you, it has a much larger statistical error because there are only half as many respondents. So you poll some more women, and use their answers in interpreting the responses by the woman’s demographic. You weight the woman’s responses by their percentage in the original poll, or some reasonable approximation of likely voters if you have that data. This would not significantly affect the percentages for the prior answer. If it did, it would only do so by probably giving a more accurate approximation of the woman’s vote because your sample size is larger.
And for crying out loud – Angus is winning huge. The poll said he had more than 50% of the vote among both men and women. So, yeah.
ME SO ANGRY.
I look forward to Naran’s next submission for the Journal Of I’m Not So Good With The Math Thingy.