Tuesday, October 19, 2010

Informed voters, yes.
Informative reporters, not so much.


So the Ipsos/Reuters poll I responded to has been made public. I always find new numbers interesting, but these are partly troubling, partly heartening and partly too undecided to be useful, especially at this late date. Also, the story gives some information that puts all the numbers in dispute.

The troubling numbers
Boxer's lead over Fiorina has shrunk to 46% to 45%.

The comforting numbers
Brown's lead over Whitman is now 48% to 44%.

The too undecided numbers
Prop. 23, the crap anti-environmental proposal endorsed by two Texas oil companies, is getting shellacked 46% to 35%, but having 19% undecided at this late date is unnerving.

The problem with the reporting is this. Ipsos/Reuters says the poll was done with 601 Registered Voters, of which 448 call themselves likely voters. They say the margin of error on the big sample should be 4.0%, while the smaller sample should have a margin of error of 4.8%.

Hmmmmmm, not so fast.

The probability that the smaller sample has the exact same percentages for Brown, Whitman, Boxer, Fiorina, pro-23 and anti-23 is teeny tiny. There should be some differences, but I can't find them anywhere in the story and they don't link to a raw numbers .pdf.

By the way, if we assume the numbers are for the larger OR the smaller sample, we can do a confidence of victory test for Brown and Boxer, because over 90% of respondents show a preference for one of the two main candidates. With the 81% either pro or con Prop. 23, we are SOL because of the high undecided.

If the numbers are out of n=601
We are about 60% confident this poll shows Boxer will beat Fiorina if the election were held when the poll was taken.
We are about 85% confident this poll shows Brown will beat Whitman if the election were held when the poll was taken.


If the numbers are out of n=448
We are about 60% confident this poll shows Boxer will beat Fiorina if the election were held when the poll was taken.
We are about 80% confident this poll shows Brown will beat Whitman if the election were held when the poll was taken.

Neither election is in the bag, but early voting has already started, so for some people, the election is being held when the poll was taken. There is still uncertainty, but for right thinking people, none of the numbers point to our impending doom.

If you really want impending doom, go to my other blog and read the stories from the Sun.

It's okay to click. Remember, when it comes to my other blog, I'm not silently judging you.

No comments: