10/09 Zogby Kerry+2, Rasmussen Bush+1.4; WP Bush+1; BOT state track

[Pollwatch99]

Zogby          :  Kerry 46 Bush 45 Nader 1
Rasmussen :   Bush 50 Kerry 46 (95% samples pre-debate)
WP              :   Bush 50 Kerry 46 Nader 1

BOT daily tracking with leanners included

Bush Leading(10/09)
    FL    Bush 52 Kerry 46  (stable)
    OH   Bush 50 Kerry 47 (very slight drift towards Kerry)
   
Kerry Leading(10/09)
    MI   Kerry 50  Bush 48 (stable)
    MN  Kerry 49  Bush 47 (stable)
    PA   Kerry 49  Bush 48 (almost no movement in this state for 10+ days)

[Giant Saguaro]

Interesting. First time all year anyone has been at 50% in the RR. JA at 53%.

[The Vorlon]

Interesting. First time all year anyone has been at 50% in the RR. JA at 53%.

No "blips" in the sample either - the last three samples have all been pretty similar actually..The party ID distribution has been constant within 0.5% and the differential between the two candidates is pretty similar too.

I sense a massively strange sample coming

[Pollwatch99]

The date the trend changed was 10/05. Interesting after the VP debate.  I still think Cheney laid out the administration case for the war on terror and IRAQ for all to see.  It could just be a random date that caused the trend to change but I for one don't think so.  Cheney as a minimum stabilized the situation and Bush last night helped.

[Reds4]

Vorlon, Does the fact that three samples have been quite close to the same look at least like Bush has stopped the momentum of Kerry for the time being? When you say you see a very strange sample coming do you mean you think rasmussen will have a strong Kerry sample in order to even it back or what you mean by that?

[Pollwatch99]

Vorlon, Does the fact that three samples have been quite close to the same look at least like Bush has stopped the momentum of Kerry for the time being? When you say you see a very strange sample coming do you mean you think rasmussen will have a strong Kerry sample in order to even it back or what you mean by that?

Vorlon, is the best on this board.  Here is my take for what it's worth.  Kerry gained about +2-+3 after the first debate.  if you believe that Bush went into that debate with +4 to +5, that matches current believes that Bush is now slightly leading by about +2. 

Kerry's sample seems pretty consistent at +2 to +3.  The sample that is all over the place is Bush.  Big drop down followed by a recovery back to pre-debate levels which I attribute to Cheney restoring the base.   

I actually do not expect to see a suprising sample but Vorlon is pretty sharp so we will see what he says.

[ATFFL]

Vorlon, Does the fact that three samples have been quite close to the same look at least like Bush has stopped the momentum of Kerry for the time being? When you say you see a very strange sample coming do you mean you think rasmussen will have a strong Kerry sample in order to even it back or what you mean by that?

Vorlon, is the best on this board.  Here is my take for what it's worth.  Kerry gained about +2-+3 after the first debate.  if you believe that Bush went into that debate with +4 to +5, that matches current believes that Bush is now slightly leading by about +2. 

Kerry's sample seems pretty consistent at +2 to +3.  The sample that is all over the place is Bush.  Big drop down followed by a recovery back to pre-debate levels which I attribute to Cheney restoring the base.   

I actually do not expect to see a suprising sample but Vorlon is pretty sharp so we will see what he says.

We have some very sharp people on these forums, not just Vorlon, but he is clearly the man on polls here.

We will see blips in each poll.  It is inevitable.  5% of all polls will end up outside the margin of error.  Personally, I feel Rasmussen's style of bot polling lowers the confidence in the poll so he is more prone to blips.

[The Vorlon]

When you say you see a very strange sample coming do you mean you think rasmussen will have a strong Kerry sample in order to even it back or what you mean by that?

I was actually just kidding around a bit.  Rasmussen's internals have been just chaos for the last week and I had rather a harder time than normal sorting them all out.

Rasmussen has a bunch of heavy weights in his poll - party ID, gender, race, etc and he had a few strange samples which made things bounce around a fair bit.

The last three samples have all been very nornal and I have all his internals sorted out now.

Thus, according to Murphy's law, the next sample MUST be strange and bizarre

I have nooooo idea what tonights sample will look like, other than to say I expect Bush will do ok. 

As Tedrick says... 1 sample in 20 blows up... you just never know when...

Re Cheney and race stabilization:

Both ABC and Rasmussen have show a stable race for 3 days in a row now, an assessment I agree with.

Zogby shows a move to Kerry.. with Zogby.. who knows?  He releases so little information I have no opinion one way or the other on his poll.

The "Time" poll BTW was really just a bad sample.  It was only 57% married people, versus 65% in the 2000 exit polls. (Racial breakout was also quite a bit out)

Bush won married folks by 9% in 2000, while he lost singles by 19%, compensate for this is Bush was up 4ish in Time also.

You can argue weighting by party ID, you cannot argue that the racial and marital status data should be right.  The time poll was just wrong in that respect.

[ATFFL]

When you say you see a very strange sample coming do you mean you think rasmussen will have a strong Kerry sample in order to even it back or what you mean by that?

As Tedrick says... 1 sample in 20 blows up... you just never know when...

Is it so hard to rememer there are 2 "r"s in my handle?

Really, I might have to make a deal with that nice Mr. Morden to force people to remember.

[The Vorlon]

When you say you see a very strange sample coming do you mean you think rasmussen will have a strong Kerry sample in order to even it back or what you mean by that?

As Tedrick says... 1 sample in 20 blows up... you just never know when...

Is it so hard to rememer there are 2 "r"s in my handle?

Really, I might have to make a deal with that nice Mr. Morden to force people to remember.

Point taken Mr. Tedrrick Tredrick

Keep those nasty shadows away for a while.. ok

[Reds4]

I see, because things have become too predictable you assume we're due for a crazy sample. I'm assuming the last 3 nights have all been pretty close to 49.5 to 45.8 or something so unless tonight's sample is crazy the poll should look fairly similar tommorrow, but that could be wrong too. We'll see. Haven't seen a Tipp/IBD poll in a while, wonder when it will come out. I think Battleground will release another poll late this weekend and I'd bet CBS tries something soon, but I dont even consider them halfway reliable.

[Pollwatch99]

tracking updated in 1st post;, including WP

[Reds4]

Thanks for the updating the numbers pollwatch, interesting state polls from the bot too. Do you have rasmussen premium service in order to get those state numbers?

[Pollwatch99]

Thanks for the updating the numbers pollwatch, interesting state polls from the bot too. Do you have rasmussen premium service in order to get those state numbers?

Yes; to get the state numbers you need premum service.  These numbers (unlike his daily bot) are a 7 day rolling average.

[muon2]

I see, because things have become too predictable you assume we're due for a crazy sample. I'm assuming the last 3 nights have all been pretty close to 49.5 to 45.8 or something so unless tonight's sample is crazy the poll should look fairly similar tommorrow, but that could be wrong too. We'll see. Haven't seen a Tipp/IBD poll in a while, wonder when it will come out. I think Battleground will release another poll late this weekend and I'd bet CBS tries something soon, but I dont even consider them halfway reliable.

It's due to the statistics. In any given set of statistical samples of a population there should be some fluctuations in both directions beyond the margin of error. That's what a margin of error means.

Rasmussen removes many of the fluctuations through the use of rolling averages, but the one-day samples that the Vorlon breaks out should show the expected amount of fluctuation. If too many samples are too close compared to the margin of error, it "feels" wrong. OTOH, statistics predicts that there can also be a run of nearly identical results every once in a while.

Unfortunately, statistics doesn't predict which way the next sample will fluctuate. For that we leave to our hunches.