Princeton Election Consortium

Innovations in democracy since 2004

Outcome: Biden 306 EV (D+1.2% from toss-up), Senate 50 D (D+1.0%)
Nov 3 polls: Biden 342 EV (D+5.3%), Senate 50-55 D (D+3.9%), House control D+4.6%
Moneyball states: President AZ NE-2 NV, Senate MT ME AK, Legislatures KS TX NC

The 1,000,000th site view, the hundredth flower

October 30th, 2008, 8:00pm by Sam Wang

In the wee hours tonight, one of you will make the 1,000,000th view of this site. That’s on a par with the traffic I received in 2004, when the Meta-Analysis was just a single page of HTML. Compared with 2004, more of you have responded to my suggestion that knife-edge situations are the most efficient use of your resources. I don’t keep figures for the NRSC, but the ActBlue site has raised over $44,000 this year – nearly twice as much as four years ago. Thank you for your readership and support!

The other big change since 2004 (see this Wall Street Journal piece) is the competition on the poll-geek market, as profiled recently in the New York Times and the San Francisco Chronicle. A hundred flowers have bloomed! For primary polling data the leading sites are and RealClearPolitics (R). My favorite is the venerable, by Andy Tanenbaum. A founder figure in computer science, Tanenbaum has done high-quality poll aggregation the longest, his commentary is better than that of most professional pundits, and he does it all for free. In my view, he’s the best of what this hobby has to offer.In the NYT piece, Nate Silver recently said that there is room for three or four sites, by which he presumably meant commercial operations. I hope there’s still a role for all of us down here in the open-source sector. There are a lot of us. Here are just a few: Andrea Moro is an economics professor who also blogs in Italian. (R) is run by a (presumably disappointed) Republican. Raceto270 (D) is run by an activist who’s off in Pennsylvania getting out the vote. Stochastic Democracy is a new site run by a math whiz who’s headed for graduate school at the age of 17. And that’s just scratching the surface. When we include simple poll aggregators who don’t model, 3BlueDudes (D) cites 88 sites, with 18 sites updated in the last day alone. It’s an incredible proliferation from the half-dozen or so sites I found when I started doing this. We’re all part of a holy order of geeks.

This site’s party affiliation, you ask? Although I am (D), the data come from the nonpartisan As far as I know the software, which is automatic, is also nonpartisan.

If I haven’t listed your favorite site, fire away in comments!

Tags: 2008 Election

19 Comments so far ↓

  • Glenn

    The margins seem to be tightening, Sam. True?

  • Steko

    I would also rec — not the most original site name but he let’s the unwashed masses run our own simulations.

  • Independent


    The thing I like about your site is not merely the polling data – I can get that at any of the sites you’ve mentioned. I look forward to your insights, your sense of where things are headed (based on the meta-analysis), and frankly, your own quirky intuition that reinforces my own. So, please, write without inhibitions, adding caveats if you like; but be your colorful, insightful self.

  • ThatSeattleGuy

    Sam, you hit most of my favorites, only leaving out:
    Nice, clean site, minimal commentary, GREAT simulation tool (hit the “interactive election probability calculator” link).
    Neat aggregation of the current Intrade market for each state. Whether you think Intrade has predictive value or not, it’s worth a look.

    Thanks for all the hard work, and get some sleep in the next five days!


  • AJ

    What are your thoughts on Nate Silver?

  • BHS

    I know you’re tired of us pro-Obama worry-warts, but do you care to comment on the Fox poll that has Obama up by only 3 nationally? An article on it suggests that a segment (the wingnuts?) of the GOP wasn’t behind McCain until now. Presumably a couple weeks of “terrorist,” “socialist,” etc. have fired them up?

    Just what *is* the reason for the difference in the national and state polls?

    Thanks for the great site.

  • Sam Wang

    BHS – If you are obsessing over single polls then you’re not really getting the point of poll aggregation sites!

  • Matt

    My hypotheses for the apparent discrepency between the state and national data:

    The tracking polls, which make up the most and most McCain friendly of the national data, have more conservative turnout models on the whole then the pollsters that are doing the majority of the state sampling.

    The state data is for the most part from battleground states where Obama has a better GOTV operation and is seriously outspending McCain on ads. More and more Obama sporatics are being dragged by the ears to early vote by his volunteers (in full disclosure I am one of those volunteers here in NM). We see this reflected in propping up the likely voter models in Obama’s favor a few points in state battleground polls. Also the ads are depressing McCain support and boosting Obama support by a point or two.

    There is only an illusion of a difference between state and national data or a slight two or three day divergence that people are over analyzing.

  • Don Milton

    It would be interesting to see a version of your median EV estimator with the current labels removed, and labeled with the recent economic events — e.g. the Bear Sterns fire sale, Fannie and Freddy renationalization, Lehman collapse, the Paulson Bailout proposal, and the start of capital injections. Then, you could plot the S&P 500 scaled to fit. It would be an more interesting, and I suspect more relevant plot, except for the HRC time point.

  • Observer

    I still chuckle every time I notice:

    NJ 0.020826

  • Sean

    Not to mention that the newest Fox poll has changed its voter makeup. That doesn’t necessarily make it ‘wrong’, but a straight comparison between today’s poll and the previous ones is not very useful at this point.

  • Matt


    I wanted to let you know that I appreciate this column. I’m one of the 88 in the “holy order of geeks.” I was maintaining a projection of my own in 2004 but was not nearly savvy enough to create my own site. With today’s web publishing tools I decided to take a stab at it. My inspiration largely came from this site in 2004. Thank you very much.

  • Michael Wilson

    I can recommend:
    if you want to get updated with the election polls.

    I think you might like it!
    Make a difference, keep on voting!

  • Sam Wang

    Observer – I was just corresponding with a friend in Virginia. Every vote he turns out is worth about 90,000 of my votes. Makes me want to get down there. But then who would do the essential work of blogging?

    All panicked consumers of polls – the median national margin’s still 6%. At this point voters are pretty well committed – even the undecided ones, who are probably committed but will only find out in the voting booth.

    The bigger news is that Senate candidates have slipped a bit. The current expectation is 58-59 votes for Democrats, 41-42 for Republicans. A 60-40 split is starting to look unlikely.

  • David Shor

    Don Milton,

    I’ve tried to explain day to day polling changes using the economy, but econometricly, the relationship doesn’t hold up very well.

    [Shameless plug to my website.]

  • Todd S. Horowitz

    It would be nice to have a plot of the median national margin. I’ve been doing it by hand from the RCP page, but it would be interesting to watch the national polls vs. the meta-margin.

  • Bruce (B)

    I agree regarding most of the sites listed above. really is classic, the standard. 538 is newest, updated more frequently, chock full of data, more processed, worth reading. Then check Sam Wang — (of course) serious, more academic method, 3rd opinion, usable comments. and 3bluedudes are good for their links, and likewise for
    I think this site would also benefit from a blogroll of that kind in the sidebar, probably best broken into categories such as primary poll data, poll aggregation, markets, and combination sites.

    I wish I had known about earlier. I especially like the “Probability of Win by State” graph and the customized simulations.

    My 2 bits:
    I always check IEM, Intrade, and Rasmussen Markets.

    I also like Colley Rankings (Gott and Colley’s Median Poll Statistic):
    Simple methodology (of astrophysicists) is conservative, low-key and powerful. Take the median of all polls for the past month and assign the states according to that.

    I don’t think anyone mentioned
    270toWin has simulations, probabilities, state voting history, and poll data.

    There’s also — a better representation of graphical data from, similar to the youcalc Michael Wilson mentioned earlier.

    Thank you! I hope you keep doing what you’re doing and enjoy it. I look forward to seeing how the many poll-geek sites emerge after the test of the election.

  • Michael

    do you care to comment on the Fox poll that has Obama up by only 3 nationally?

    the newest Fox poll has changed its voter makeup.

    While I think that Sean has largely answered BHS’ question, it is worth noting the difference between pollsters who weight their polling data to obtain demographics that they believe match the overall electorate, and those that simply take the voters they are able to reach as they are. The FOX poll is in the latter group, and they simply polled a lot more republicans this week than last. I’m not suggesting that this was intentional on their part. It just shows that two random samples aren’t always going to match each other, even if they’re large enough to have reasonable margins of error.

    Elections are always about who actually votes. Polling is always about predicting who is going to actually vote.

  • andrea

    Learn how Presidential campaigns are often predicted far in advance by political scientists. Check this out

Leave a Comment