LA Times poll thread

[The Vorlon]

you defended the SurveyUSA poll against the complaint that it was not weight well for party ID and that you panned the LA Times poll using the fact that it was poorly weighted for party ID.

Maybe I have been unclear on party ID.

Party ID is a useful VALIDITY CHECK.

I DO check it very carefully.

It's like male/female and young/old.

It "should" be 51/49 women/men

If it's 50/50 or 52/48 it's no big deal, if it's 57/43 your poll is deeply &*&ed.

Party ID "should" be "fairly" close to an even split.  If its 38/33 Dem/Rep nobody worries, if it's 35/37 the other way nobody worries.

When it's 38/25 something is just deeply wrong...

Again...

You don't need to be "perfect" you need to be "in the game"

For example, the new PEW (Bush +4/+2) looks at first blush like it has about 2% (+/-) more Republicans than Democrats in it.  Possibly (likely?) a "reagan effect".

This is a few % too GOP friendly a sample - but it is close.  The poll is likely still pretty valid.

The last Ipsos/AP (Bush +1) had 7% more Democrats.  A bit too Dem friendly, but again "in the game" - still a valid sample.

I view Survey USA and LA Times as being VERY different in this matter for the following reasons:

1) Survey USA was (barely) in the margin of error, and on a weekend when Reagan died - a bit of a shift to the GOP was entirely a reasonable expectation.

1A)  SUSA clearly and several times flagged that this Survey was done while the Reagan mourning was going on, and that it likely impacted the Survey.

2) The LA Times was well outside the margin of error (Much bigger sample)

3) SUSA clearly provided the information as to party ID in their sample.

4) The LA Times went out of their way to hide it - Nowhere in a 30 page .pdf release did the even mention it.

5) Survey USA randomly misses BOTH ways (Kerry +10 in Michigan, Bush -1 in California)

6) LA Times ALWAYS misses left. -The AVERAGE +11 to the Dem side on party ID - clearly just wrong.

I view the ethical conduct of the two organizations to be very different.- I suspect you too see a difference, if I am not mistaken.

BTW - the "Slander" was a "joke" hence the "just kidding" but I hereby officially apologize

(flogs him self with rope, says 4 hail mary's, says mea-culpea)

[Bandit3 the Worker]

I did a House breakout about a week ago, and when I put on my GOP rose colored glassed (The ones where it looks like Bush has a chance in Vermont) - I found 21 potentially vulnerable Democratic seats.

Bush has NO chance in Vermont. None.

[Bandit3 the Worker]

The two are NOT mutually exclusive.  You do not throw out peopel if you are close to a reasonable split in party ID, say a 5% Dem advantage.  Might be a bit off, but not so far teh poll needs tweaking.

Nationwide, Democrats outnumber Republicans by probably about 1.5 to 1.

[bullmoose88]

The two are NOT mutually exclusive.  You do not throw out peopel if you are close to a reasonable split in party ID, say a 5% Dem advantage.  Might be a bit off, but not so far teh poll needs tweaking.

Nationwide, Democrats outnumber Republicans by probably about 1.5 to 1.

Registered Democrats/Republicans right?


Aren't there a bunch of states that don't require registration?

[Beet]

Please don't "grovel and apologize"... how embarassing! :x

Of course you are probably right about the likely quality of these two polls. So, you use party ID weight as a validity check, and it was a matter of degree between these two polls. The SurveyUSA poll diverged less from the norm, had a larger margin of error, and they were more honest about what factors influenced their result. So the differing party weight offset was relatively less significant, hence the contrasting treatments. Thank you for explaining that.

The LA Times is probably slanted consistently, no disagreement there.

[agcatter]

A lot of southern states don't have formal party registration.  My state for example.  I can, for example vote in the Democratic primary if I so choose.  

I believe the exit polls from Texas had substantially more self identified Republicans than Dems - just nor formally registered as such.