I’ve always wondered why the Zogby poll numbers tend to be so peculiar, and why they don’t match up with other national polls. Maybe others knew the answer — I didn’t. Nate Silver at 538.com explains it.
The national tracking polls are actually in pretty good agreement with one another, with IBD/TIPP, Research 2000, Gallup, and Hotline all settling in the [Obama +]5-7 point range. Zogby is the outlier at Obama +2.7, and that’s because Zogby has the odd practice of fixing his poll’s party identification weights based on what they were in the last presidential election. In Zogby’s world, then, it’s still 2004, when there were roughly as many Republicans as Democrats. Although Zogby’s trendlines may be worth looking at, his topline numbers are basically unusable.
Glad that’s cleared up. In other polling news, Nate confirms that WV really is a “lean red” rather than a tossup. Meanwhile, pollster has moved MT into “tossup” status, taking 3 more EVs out of the McCain column.
I had understood this statitical issue with Zogby but I’m not sure that means it’s now “cleared up.”
<
p>The other polls are relying on voter registration (and, I think, primary turnout numbers) to predict that there will be more Democrats voting than in 2004. It’s unclear to me what the party idenification numbbers look like for the GOP after a lackluster primary but post-Palin.
<
p>I hope Zobgyby is wrong but you would agree that the actual turnout could end up being closer to the Zogby numbers than not? If a lot of newly registered Dems and/or Dems who voted in the primary don’t actually show up in the general Zogby wil be closer to the actual outcome. Another facotr is predicting Republican turnout which may or may not be better than it would likely have been pre-Palin.
<
p>This is the problem with polls. The vast majority of people don’t understand that ther methodology of a poll involves quite a bit of guesswork.
… party identification numbers have shifted toward Democratic from 2004. If so, it would follow that relying on 2004 as a basis for designing your sample would skew the results (unless you were predicting a 2004 election). Of course, the ratio shift toward self identified Dems results in many polls weighing their sample criteria toward with a similar ratio. This has resulted in a GOP talking point that these polls are ‘inaccurate’ because they sample more Dems than Republicans.
<
p>Point taken on Palin though.
The jury is still out on whether the new wieghts and party identification ratios accurately reflect those who will actual go out and cast a vote. Potential problems with this theory (and potential reasons why Zogby may be right) include:
<
p>1. In 2008, many voters may have voted in the hotly contested Democratic primary (in some states this required switching their “party identification”) but may either stay home (committed Clinton voters) or “switch back” to the GOP and vote for McCain now (“independent” and Reublican Clinton voters).
<
p>2. The new Democratic party advantage has been built upon massive new voter registration drives. These folks may have voted in the primary and may say they are “likely voters” but they haven’t actually voted in the general yet. All evidence is that newly registered voters tend to vote in lower numbers than longtime voters.
<
p>3. The new polls rely on a predicition of lackluster GOP turnout. I hope this will be the case but Palin (or something in the next two weeks) could change that and bring GOP voters out more. Conversly, Democratic voters (particualry new ones) could be lulled in to complacency, give up because of long lines, be deterred by bad weather or otherwise not turn out in the big numbers expected.
<
p>4. Party ID is fluid. Adjusting it in the middle of a Presidential election conflates the issue with the survety respondant’s preffered presidential choice. To avoid this issue, Zogby uses a relatively constant weight by taking last two presidential exit polls as an indicator. His polls in 2004 used a party weight of 39% Democrat, 35% Republican and 26% Independent as indicated by the average exit polls from the 1996 and 2000 presidential elections. I believe this has changed and there truly are more Democrats than 4 or 8 years ago but Zogby’s numbers at least don’t flucaute with the news of the day.
<
p>That’s true, but what is also true is that newly registered voters tend to vote in higher numbers than all but the most regular clique of voters. Newly registered voters will turn out. The open question is: will they be successfully caged?
<
p>
<
p>Zogby’s use of 4 and 8 year old party ID belies any understanding of or appreciation for the trends of party ID. It’s a reasonable first stab at 2008 party ID in that it’s a conservative estimate void of any personal bias. That doesn’t make it a particularly useful estimate when trying to quantify 2008 outcomes though.