« Sorting Out the Latest Horse Race Data | Main | Fun Facts on Religion and Politics »

Battleground States Update

The Annenberg Election Survey released data today that suggest Kerry's ads in the battleground states are having the desired effect of improving public impressions of Kerry in those states. In the May 3-16 period, Kerry's favorability rating in the battleground states was 39 favorable/33 unfavorable. In the May 17-23 period, his rating improved to 44/32. As for Bush, his favorability rating in these states has declined from 48/38 to 44/44.

Other recently-released swing states data are also positive for Kerry. Yesterday, I mentioned the Gallup data which showed Kerry ahead by 5 in the "purple states" (the 16 states where the 2000 winning margin was less than 6 points; note that Annenberg's "battleground states" include all 20 states where the Bush and Kerry campaigns have been running TV advertisements--that means, in addition to the purple 16, Annenberg includes CO, DE, LA and WV). And Zogby has recently released a group of 16 "battleground state" polls(here, battleground states are the same as Gallup's purple states, with the exception of WV being substituted for ME), conducted May 18-23 for WSJ.com. These polls show Kerry ahead in 12 of these 16 battleground states: FL, MI, MN, MO, NV, NH, NM, OH, OR, PA, WA and WI. Bush is only ahead in four: AR, IA, TN and WV.

Note that all these polls included Nader in the trial heat mix, so these results are particularly bad news for the Bush campaign. Note also, that where Kerry leads in 2000 blue states, his leads are all outside the margin of error. In addition, his leads in the 2000 red states of Ohio (+5) and New Hampshire (+10) are also outside the margin of error.

One caveat: the Zogby polls were conducted over the internet with "respondents who agreed to take part. Likely voters from each state followed instructions sent by e-mail that led them to the survey located on Zogby's secure servers in Utica, N.Y". Well, this isn't like an internet "poll" where anyone who wants to can participate, but one still wonders whether this kind of polling might be biased in ways that would throw off the results. I don't know that for sure, but it's a caution that's worth keeping in mind.

Comments

Isn't it more likely for Dems to have internet?

Ya, but Republicans tend to stuff the ballots more than Dems :D

My question is, how does responding to an email make the process more accurate? I mean, unless Zogby has some sort of vetting process...

I imagine that you list your party affiliation and other demographic info, and then John Zogby takes the data, and tweeks the sample based on party affiliation.

For example, say in state A, 75% of the respondats were Dems, 15% were Indys, and 10% were Repubs. However, the actual party registration breakdown in state A is 45% Dem, 40% Repub, and 15% Indy. So, JZ would just do a sample dist of 45% D, 40% R, and 15% I.

The methodology page on the "interactive version" says as much:

"Slight weightings were applied to ensure that the selection of participants accurately reflects characteristics of voting population, including region, party, age, race, religion, and gender."

I am in the Zogby internet polling population though
DEFINITELY not in a battleground state (Kerry +15 in CA..Field Poll)..

I have been doing this for 3 years or so but thought that the project was experimental.

Is this not still the case?
AND

What are the methodolgical probems

I realize that Zogby isn't about to explain the methodolgy but just as with Robo polls the results a can be compared to random sample phone interview polls run contemporaneously.


I receive my email notices during roughly the same period as the Zogby telephone polls

I've participated in Zogby on-line polls for a couple of years. He leads with a few questions to determine past voting preferences, party registration, union membership, and the like before getting to the issue at hand (usually national politics although the last was nano-tech--I passed) and closing with demographic questions.

I'm in GA, which is far from being in play.

Ruy - Ohio isn't outside the MOE. Aren't you supposed to double the reported MOE for two-person contests because the MOE is for each individual number, i.e. +/-3 for Kerry individually and Bush so margin would have to be over 6 to be 95% certainty that Kerry is leading. I've read this in numerous places including an SSRC guide to interpreting polls so assume it must be true. Did you let down your guard here?

You're right angry moderate. I see that happen all over the place, but am a bit suprised to see it here.

xdog - I'm no pollster, but from my understanding of probability, it's not a factor of 2, but of squareroot(2) =~ 1.41

If you check out the National Council on Public Polling's website (www.ncpp.org/poll_perform.htm) you will see that in the 2000 election the closest results were from, believe it or not: Harris Interactive!! Harris Interactive, if I'm reading the results correctly, came closer than any other pollster (including Zogby and CBS) except maybe the regular Harris polling. Maybe there is something to this interactive polling.

I just finished watching Al Gore's speech at NYU on the C-span. Wow! It was a beaut. Gore hit Bush about as hard as anyone has.

I hope Move-on.org which sponsored the event finds a way to play it again and again over the next few weeks. It will really help Kerry. Its the kind of surrogate help that Kerry has needed.

Gore's speech is getting news airtime. MoveOn might want to replay it in its entirety to take advantage of the publicity. CNN played a decent soundbite of it this morning.

Pelosi, Gore. The language of condemnation is getting stronger. What is needed is a Republican voice to speak as strongly along the same lines so as to take the partisan onus off the words. Of course, anyone who did so would be immediately ostracized from the GOP, but then if Zel Miller can shill for the GOP, someone in the GOP can find enough integrity to call Bush out on his ineptitude. I think it would have to be a CA Republican. There's got to be a CA Republican with a liberal enough constituency who could make this move without putting his seat too much on the line.

Like xdog and John Mcc, I'm on Zogby's email polling list.

xdog gave a good description of what we see when we respond.

I'll note, as I have a couple times previously on this board, that the answer options to the Zogby question on party identification don't include the Democratic party. Instead, Zogby lists that mythical dittohead creation, the "Democrat" party.

I've emailed several zogby contact names several times on the error, with no respons. This last time, I included a Google search URL for "Democrat party" on rushlimbaugh.com.

I'll be curious to see if the error is corrected in the next poll. Maybe I accomplished nothing more than getting my email removed from the list.

Isn't it more likely for Dems to have internet?

I doubt it. Isn't the Internet more likely to be used by the more wealthy, and doesn't that mean Republicans?

The only surprise here is Iowa -- for Bush?

It excludes the indigent who don't have internet access or email accounts. Although, most of them are probably more likely to vote for Kerry anyway...

NEWSWEEK/TIME POLLING IS A JOKE.IT MAY AS WELL BEEN DONE BY THE RNC.38%REPUBLICANS WERE POLLED,31%DEMOCRATS AND 31%INDEPENDENT.THAT IS DICTATING AN OUTCOME.