Replies to Msg. #750907
.
 Msg. #  Subject Posted by    Board    Date   
10013 Re: The MI poll
   Yup (on flier polls). I haven't used the huff tools much, and do appr...
DigSpace   ALEA   13 Sep 2012
7:43 PM
10011 Re: The MI poll
   re mi - i think it shows romney weakening but not obama strengtheni...
Cactus Flower   ALEA   13 Sep 2012
7:33 PM

The above list shows replies to the following message:

Re: The MI poll

By: Cactus Flower in ALEA
Thu, 13 Sep 12 7:21 PM
Msg. 10010 of 54959
(This msg. is a reply to 10009 by DigSpace)
Jump to msg. #  

Hi dig,

haven't used the rcp map. but the nice thing about the pollster version is you can remove poor quality pollsters. there are some who look like they are deliberately polling to move the dial for everyone (eg purple strategies, gravis). there are others who just can't get a consistent result (eg mitchell).

when i exclude these guys from ohio, it changes from a chart that looks like a narrowing race:

http://elections.huffingtonpost.com/pollster/2012-ohio-president-romney-vs-obama?gem

to this, in which I think a shorter trendline regression interval would show Obama's numbers bending upwards:

http://elections.huffingtonpost.com/pollster/2012-ohio-president-romney-vs-obama?gem#!hiddenpollsters=gravis-marketing,purple-strategies

the latest ppp poll looks on the money to me: 50:45:5.

removing pollsters really shouldn't change anything. yet it does. which tells you something.

note - i leave rasmussen in, because the fact that they lean to the right is less important than the fact they seem to derive results consistent with the pattern of other pollsters.

so i have ohio in the obama column. i'm with nate silver on this. i don't treat all pollsters as equal. some are more equal than others.

silver shows 78.1% probability of ohio for obama, by the way.