Get yer polls, bets, numbers & pretty graphs! Elections 2008

Sure!

That info would sure help confirm/erase a lot of my unease and suspicions. But it would probably be very difficult to obtain for a few more months since many of the states haven't even finished counting indys or absentee ballots.

Thanks for the kind words re: job, guys! Deadline to apply was today so hopefully will find out something soon. It's super short-term -- just until March 5th -- but that suits me fine.

Really interesting analysis, nimh. My preliminary reaction (as in, I'd want to go back and re-read more carefully and that might change) is that Obama is getting safer and safer. He has a reasonable claim to front-runner status. He won the last 8 primaries. He does better than Hillary in head-to-head match-ups against McCain. He's getting more endorsements from "safe" people -- and of course there's a trickle-down there, too, not just in terms of information. (Ted Kennedy endorses Obama, which frees up a congressperson to endorse him, maybe (that's something I read a lot about at the time of the Kennedy endorsement, how it gave the signal that you can be a good Democrat and endorse Obama), which in turn leads to a person who is influential in that congressional district to endorse Obama, which leads to a church leader saying good things about Obama... etc.)

That's all somewhat tenuous though. If Obama wins WI and HI that could really add to the snowball. A WI loss (win for Hillary) would have a pretty big negative impact I think.

Ohio and Texas will be telling. After those, it should be pretty clear if we're in a situation where one or the other is the people's nominee, don't y'all think?

This is pretty interesting. Whether Obama or Clinton end up the nominee, medical services/insurance reform will be a key, if not the key, platform element up to november. This graph suggests that maybe, just maybe, the term "socialized" no longer has the connotation of marx and satan...


http://matthewyglesias.theatlantic.com/
original data from
http://www.hsph.harvard.edu/news/press-releases/2008-releases/poll-americans-split-by-political-party-over-socialized-medicine.html

For a latest update on how the Rasmussen and Gallup daily tracking polls compare: both now have had Obama in the lead for 5 days, and they currently agree he's leading by about 5%. Here's the graph:

Nimh, I find your analysis over the last two pages or so to be personally inspiring.

Cycloptichorn

I am in the red bar, so I have a question, why don't the blue bar just go form their own program, and choose to be taxed to support it, and leave the rest of us alone. After all, I thought this was a free country?

Okie, do you support the idea of paying tax dollars for public schools independant of having children in those schools?

T
K
O

Socialized medicine is an excellent idea. Everyone should see Moore's film Sicko.

On the other hand; forced insurance is pretty friggin stupid if you ask me. What is the point of the middleman?

Second that.

T
K
O

Trivia 08: Looking back at Rudy's demise

I've been tracking these match-up polls, by state but also nationally. Polls that ask respondents to choose in a hypothetical match-up between one of the Democratic frontrunners and one of the Republican frontrunners. They have shown some interesting patterns.

For example, all the way through to mid-December, Giuliani was the Republicans who matched up best against the three Democratic frontrunners. Most of the time he managed to keep their lead under 5%, and in the earliest months (up through May), he actually had a lead on Hillary. (Obama and Edwards were doing somewhat better than her in those early months, but from June through to December there wasnt much difference between them anymore.)

But as Rudy's campaign started unravelling when the actual primaries approached, so did his match-up numbers against the Democrats. Look at how these numbers evolved until he finally dropped out after the Florida primaries:



(Depending on what your screen size is you might have to click the image twice to get it right.)

Graph manual: For each poll that asked "match-up questions," bars go up (or down) to represent the lead (or deficit) the Democrat had on Giuliani: the yellow bars are results for John Edwards, brown is Obama and blue Clinton. Of course some polls only asked about Hillary, or Hillary and Obama. There were also polls that included match-up questions, but not about Giuliani, in which case there's no bars at all for a bit here.

The trendlines are running averages - the average of the last 10 polls out (or 15 for Edwards). The last 10 (or 15) polls that included any kind of match-up, mind, including ones that didnt include Giuliani. So if the last 10 polls that included match-ups only included 4 that asked about Giuliani, then it's actually just an average of those four (hence why I set the bar higher for Edwards, because many fewer polls asked about him).

Cool...

Hey nimh, have you seen any Hawaii numbers? Everyone seems to be assuming it'll go for Obama but I couldn't find any actual numbers. Pollster didn't have any, and a quick Google search just turned up references to Hawaii and then polls for WI, OH, TX, etc.

Matt Yglesies over at The Atlantic.com has the same complaint. Apparently they aren't big on polls in Hawaii.

Obama opens up electability gap

Tracking the same national match-up polls as discussed above, those that pit either or both Hillary Clinton and Barack Obama against John McCain are currently showing a notable development. Look for yourself:



(See the post above on how to read the graph.)

As in the match-ups against Rudy, both John Edwards (not included in this graph) and Barack Obama initially did better than Hillary Clinton. Back in April through June of last year, both men on average led McCain by 5, while Hillary on average was more or less tied with him.

Come summer, a new dynamic was established. The number of match-up polls that included McCain dwindled, as his campaign was judged to have gone south. In the polls that were still conducted, all Democrats did about equally well: all tended to beat McCain by an average of 5 points.

This phase lasted from about July all the way to November. But come December, McCain's come-back, initially driven purely by New Hampshire, became apparent. He was included in the match-up polls more often, and started doing better in them.

By the time the New Hampshire primaries came round in early January, he was leading Hillary by 5 points on average, and Obama by 1 point. Even Edwards, who had still gotten three solid 5+% leads against McCain in polls in November-December even as the other two were dropping, suddenly was 8 points down in the last poll that was to include him.

By late January, McCain's immediate post-NH bounce waned somewhat, and both Obama and Hillary trailed him in match-up polls by one or two points on average. And since then, it's been all good news for Obama. In thirteen successive polls now, Obama has matched up better against McCain than Hillary.

In the first few (NBC/WSJ, Rasmussen, NPR, all conducted in the last 10 days of January), this just meant merely tying McCain when Hillary trailed him, or trailing him by a percentage point or two less. But it's been different this month.

There have been six match-up polls in February: ABC/WaPo, Cook, CNN, Time, AP/Ipsos, and USA Today/Gallup. All have seen Obama matching up 5-7 points better than Clinton, with Clinton's results ranging from trailing McCain by four points to leading him by three, while Obama's results ranged from leading him by two to leading him by seven.

In addition, there are now also the the Rasmussen daily tracking match-up polls. It's a four-day tracking poll, so in my graph above I only include every fourth edition. But you can see for yourself how the numbers have evolved: Obama's has consistently had a 2-6 point lead on McCain, while Hillary consistently trails by 1-8 points.

Fabulous.

I was just trying to piece this together -- I've been saying that Obama matches up much better against McCain than Hillary does, but was largely relying on an article somewhere that had the last 7 polls (OK, that may have been you :-), I don't remember). And I knew there had been more polls since then so was just going through Pollster and Polling Report trying to figure out what was up. Then this lays it out beautifully! Thanks.

Geez, I wish every voter in WI, HI, OH, TX and PN (PN? Pennsylvania) could see this...

Yeah, neither keeps track of the match-ups systematically, I think.. But RealClearPolitics does, and Presidentelectionpolls does, too.

Thanks!

I guess we'll get the poll to end all polls tomorrow. This one is making me pretty nervous, especially Wisconsin. If Obama loses it's not the end of the world, but so much better for him if he wins and especially if he wins big.

Hawaii makes me nervous because there's nothing but assumptions going on so far as I can tell. There must be internal (campaign) polling or something...?

Thanks for those two match-up links, nimh. RealClearPolitics I knew about (until I forgot about it that is ;-)) but Presidentelection polls is new to me.

A few dozen pages ago, nimh made an interesting analysis on the different stregth of Clinton and Obama on "blue" and "red" states, stressing the importance -for the Democrats- of winning the "purple" ones.

On the links, one is too general and tells us nothing new: Obama is doing better than Hillary overall.

The other one had some beef: a couple dozen state polls of potential presidential match-ups, and Obama is doing better, also. But the data doesn't seem sound. Like Obama losing Massachusets and winning Nevada.

Any other source for state polls of potential presidential match-ups?

Well, it's a site that collects all state-level polls - kind of like I do as well, just in a different way.

Some polls are better than others, of course -- but to be fair, it would be hard for a site collecting these data, like this one, to put a value on individual polls in any systematic way. Especially since many pollsters provide no or little insight into their methodologies.

The best one can do is to highlight it when a poll seems an outlier, or analyse the crosstabs at least in as far as they are available - like pollster.com does on its blog when it comes to primary polls and the like. (But unfortunately pollster.com doesnt cover match-up polls.)

Presidentelectionpolls.com doesnt have anything like that, probably because it doesnt seem to be run by the kind of academic super-authorities that are behind pollster.com. Aside from an occasional, seemingly rather off-the-cuff post, it just collects the numbers. But that in itself has a value, especially because some of those state-level polls can be hard to find.

It does mean, though, that it relays any dodgy-looking poll just like it would relay a poll that's right in line with others. But as far as I've seen, it's as reliable as any other site in passing on the correct polling numbers. For the record, for example, the poll that had Obama leading McCain in Nevada was a recent Rasmussen poll, while Obama has not just once, but twice trailed McCain in Massachusetts in Survey USA polls (in November and January). In both cases, there are no numbers from other pollsters to put these results in perspective - Rasmussen's Nevada poll is the only one that has polled an Obama vs McCain match-up in the last four, five months, and SUSA is the only pollster that's done match-ups for Massachusetts..

The reason I dont use the presidentelectionpolls site myself much, except for the occasional check, is that I dont find the design and navigation user-friendly. So instead I will go directly to the Survey USA site, the Quinnipiac site and the Rasmussen site, and then only "mop up" state match-up polls by other pollsters from the TPM Election Central poll tracker, the RCP site, or this one, presidentelectionpolls.com.

Anyhow, no need for you to do any of that, because I will have an update on how Obama and Hillary are matching up against McCain and Huckabee on a state-by-state level myself here soon - should be tomorrow!

I already made the graph files, just need to add a little write-up...