Princeton Election Consortium

A first draft of electoral history. Since 2004

Why follow polls?

August 5th, 2016, 10:00am by Sam Wang


Goal Thermometer - PECBefore the 2016 campaign season, I had reservations about re-starting up this site’s polling analysis. However, there was one big reason in favor of doing it. It has to do with your readership of the site – and how you can best influence the outcome.

The biggest reason not to re-start the site was the market for statistical politics, which looked saturated. In 2004, this strange hobby was made newly possible by an abundance of polls. Now there are many sites, most notably the NYT’s The Upshot, ESPN’s FiveThirtyEight, Electoral-vote.com, and HuffPollster (which provides our polling data). The Princeton Election Consortium may now seem redundant.

Yet that is not the case. Site traffic is the highest it has been since October 2012. Traffic for July 2016 grew over 50-fold – that’s a factor of 50 – over July 2012. Some of this arises from the bizarrity of this year’s campaign. But it feels like there’s something else at work too.

I think it might be the relative purity of PEC’s calculation. I try to make the snapshot relatively transparent; the code and data are open-source. The prediction is based on polls only, and is meant to be accurate and stable. Since the start, the November election win probability has hovered in the range of Clinton 80-85%. It isn’t designed to swing around every day; in my view, a poll-based “prediction” that moves around a lot is no prediction at all. In principle, the probability and electoral Meta-Analysis should be as boring to watch as possible – and of course eventually land on the outcome.

A practical reason to follow polls is that you can learn where to invest time and money. There is value in triaging races because they are in the bag – or because they are hopeless. Instead, donors and activists should invest in races where the outcome is uncertain, i.e. whose win probability is in the 20-80% range for either side.

That’s where the left sidebar comes in.

For Democrats, the ActBlue site goes to a list of five candidates for Senate in close races. They are all women: Ann Kirkpatrick (AZ), Catherine Cortez Masto (NV), Deborah Ross (NC), Maggie Hassan (NH), and Katie McGinty (PA). Current polls put these races within five percentage points. All are within reach for either Democrats or Republicans. Depending on how these races fall, under current polling conditions Democrats would end up with 48 to 53 seats. So control of the Senate is up for grabs.

In addition, control of the House could go either way. So the Democratic Congressional Campaign Committee is on the list.

For Republicans, the logic is exactly the same. These races could go either way, and so I provide a link to the National Republican Senate Committee.

Note the absences of some high-profile Senate races, for example Wisconsin, Florida, and Illinois. For now, these races do not look competitive. I have also omitted downticket races such as the North Carolina’s governor’s race. These are important, but in most cases, data is lacking.

>>>

I will close with a second reason to resist the siren call of this hobby. Work in the lab is going better than ever. We have exciting new projects in understanding the cerebellum’s contributions to cognitive and social function, including the neuroscience of autism. In addition, we are working on some nifty methods that may help researchers record neural activity with better resolution. All of that takes attention and time.

However, even if I am posting less often in the coming weeks, the automated calculation will continue to update multiple times a day!

Tags: 2016 Election · House · President · Senate

128 Comments so far ↓

  • Michael Levinsohn

    Please keep up the good work. Having your numbers — and insights — has prevented many a near-meltdown. It’s also just good to have basic fact in a sea of opinion.

    • Mike Carr

      I am a scientist. By reading and recommending this page I feel loyal to my tribe. Your clear forecasts calm my hippocampus, even in the dismal off presidential years when my gut wants to unskew reality.

  • Bill G.

    Well I for one appreciate you continuing the effort despite your reservations.

    I got interested in polling aggregation during the 2008 election, almost exclusively reading 538, when it was still independent, at the time. I still read 538, although I have grown a bit wary of them since their move to ESPN/Disney which seems to coincide with their move towards more traditional political journalism. There’s a place for that, of course, but I don’t think that mixing it with their data driven work is necessarily a good idea.

    I found you at the tail end of the 2012 election cycle and have been reading ever since. I appreciate that you tend to stick to the data and are open about explaining the methodology behind your system, even if it does go a bit over my head at times. I think that if there is going to be “data journalism,” and I don’t really see it going anywhere anytime soon now, a resource like PEC is important to have, with its more ground, less sensationalistic approach. Of course it’s not your responsibility to provide that outlet, but while you are, I truly appreciate it. Keep up the good work.

  • Kevin Moore

    Please keep going. I find this site to be unique and essential. I especially love the podcasts, and I’m especially looking forward to what you come up with vis a vis the House of Representatives and possible strategies for combatting gerrymandering and the non-man-made problem of rural versus urban versus how the Constitution sets up the districts.

  • dn

    Add me to the chorus. Both your undeniable success as a prognosticator and your forthrightness about how you do it have influenced the way I think about elections for the better – and your unrelenting sanity is tonic.

  • Scott

    I share this site every day – for its purity, transparency, unvarnished presentation of facts and historically accurate results. No doubt others feel the same after having experienced it. Expect continued growth!

  • Kanwaljit Singh

    Dr. Wang, i greatly enjoy reading your work. It is a breath of fresh air in the chaotic rush of media reports with each new poll release. As in, signal in an increasingly noisy scenario. So please keep up your good work. I hope you will not entirely give up on posting your thoughts here.

  • Timothy Strode

    It’s great to have this island of stability in a sea of innuendo and hysteria. I refer friends to it constantly to ease their anxiety. Reading the Bayesian percentage is better than therapy.

  • Apollo R.

    The ActBlue lists four female Senate candidates and the DCCC. Katie McGinty (PA) is not listed. Is there a reason for that?

  • Ed Wittens Cat

    /sigh
    this reads like francis fukuyama’s end of history–
    Datascience and deep learning are gunna transform the world of polls–and its just beginning! N’s google correlate was a perfect example.
    Have u thought–what it means for elections going forward if Project Ivy succeeds?
    It means the demographic timer goes off early.
    real early.

  • Elizabeth Duvert

    I’ve followed your site since 2008, love its quality, and tell everyone I know about it. Thanks, Sam. We need you!

    • Sam Wang

      Welcome back, Elizabeth. Traffic is up about 50-fold from 2008. You were an early adopter!

    • Patricia Raymond

      I too was an early adopter from 2008. I followed electoral-vote.com in 2004. I am constantly telling my friends to come to this site for real science without the blather and reassurance (to the sky is always falling types among them.) Never studied statistics but you explain your methods very clearly to the uninitiated. I also listen to your podcast every week (not this week yet). Sound has improved greatly every time. Thanks for all your hard work, Sam!

  • Dave

    My impression is that the best predictive model would change as little as possible throughout the entire race, yet still provide accurate results at the end. I think your model seems to do that pretty well! I appreciate the stability it shows in both good times and bad.

    538’s model, on the other hand, seems to be designed to do the exact opposite — swinging around wildly as the winds change in order to drive up attention and site visits! What’s the point of a predictive model if you’re not actually doing any predicting?

  • Joel

    ” in my view, a poll-based “prediction” that moves around a lot is no prediction at all.”

    Almost my exact words when I have described to my friends my issues with 538. I have even taken a snapshot of their graph to make my point that either the model is not meant to be predictive, or it’s just not very good.
    And that yours does not even need a graph to describe it, as a good predictive model should.
    Really appreciate the work you do, it’s a safe haven for whenever some of us get a little ocd about checking the daily numbers.

  • David Kline

    “I think it might be the relative purity of PEC’s calculation. ” is definitely part of it for me, and the wonkier parts of your explanations also appeal. I’ve also been following since 2008.

    Remember that some folks who come here for election info also get interested in your day-job work, maybe even buy your books–I did.

  • Anthony

    I have been following you since 2012 and I very much appreciate your work and hope you continue. Although I will be sad if you start posting less often, I would appreciate it if you continue to do the weekly podcast. I find the conversation you and your colleague have to be very insightful.

  • Stephen Hartley

    For what it’s worth I really enjoy and value the work you do and the contribution you are making. I certainly understand if you’ve got bigger fish to fry, however. It must be difficult to be a top expert in so many things. :-)

  • W.Dow Rieder

    Follower since 2012, and I love it! But with the automatic updates running smoothly, fewer posts are fine, because as someone on the autism spectrum I’m interested in your other work as well 8-)

  • Violet

    Dr. Wang, we so appreciate your doing this. I do like that your numbers don’t swing wildly, while 538’s polls only went from 50-50 to 80-20 in about 10 days.

    I don’t understand all the math you describe, but I do understand your devotion to getting it right, which is why this site is the go-to for so many of us. Glad you are sticking with it this election season.

  • Olav Grinde

    It has been an immense learning experience to follow your site since 2004 – and I hope you continue.

    2004 was, of course, the year that George W. Bush ran for re-election, defeating Democratic challenger John Kerry. At the time, I tipped a journalist for the regional Norwegian newspaper about your site, as he was doing a series of feature articles on the American election. He was stunned that you accurately predicted the precise Electoral College result.

    As was I, for I sincerely thought and hoped your prediction would be wrong. But that is another matter!

    Sam, please continue your great work, and keep your excellent posts coming.

  • Jim Kehoe

    I very much appreciate your methods, especially the avoidance of fudge factors in your modelling and the absence of journalistic cherry-picking of outliers. Please keep up the good work.

  • A New Jersey Farmer

    I’ve been here since 2008 as well. You’re my first stop in the morning and the meta-margin is part of my students’ election and mathematics vocabulary.

  • TJ Baker

    Been a long time follower, if not regular commentator. Glad that you have decided to keep the site alive – it’s much needed calm waters after floating around the unsettled oceans of information found elsewhere!

    One thing I might request (and would be happy to help implement, if you wished) – can we please get things in a ‘mobile friendly’ view?

  • Amitabh Lath

    Can you share anything about the exciting new projects? Will there be a seminar in the fall for non-brainiacs? I for one would much rather have Sam Wang figuring out the brain and autism than this election.

    Frankly I’ve lost interest in the 2016 presidential. No suspense left. Of course there will the usual blatherings on (it’s only August/ September/ October…anything could happen…) but we’ve had our 47% incident and then some.

    The senate switching is maybe interesting but state polls do a pretty good job there by themselves.

    • Suvro

      Reminds me of Francis Fukuyama’s 1992 book The End of History.

    • Amitabh Lath

      The title was a bit attention-grabby but Fukuyama’s thesis that there were no real ideologies left to compete with democracy and market controlled economies has been borne out. Even states that are basically autocracies pretend at elections, and nobody really plugs state controlled economies anymore.

    • Sunjay Sood

      Um, China? The most successful country on the planet in the last 25 years is a dirigiste economy with very little democracy? Singapore?

  • Pete H.

    Why is New Hampshire still considered tied, even after the release of the Clinton +15 poll?

    • Froggy

      The last four NH polls (moving back from the most recent): Clinton +15; Trump +9; Tied (with Johnson and Stein included); Clinton +5.

      Before the latest poll the Clinton +5 and the Trump +9 canceled, giving us the median as tied. When the new Clinton +15 hit, the Clinton +5 dropped out of consideration (last week’s worth of polls or latest three), with the Clinton +15 and the Trump +9 canceling each other out, leaving the median as still tied. This sets us up for the next poll to put the median anywhere in the huge range between Trump +9 and Clinton +15.

  • James

    Every time I read a horse-race article, I click over hear to ground myself in the numbers. I’d love to see articles ruthlessly taking down the rival prediction sites’ methods. Also, would you ever affiliate with large media website?

  • Brian C.

    Casual reader/visitor here. And while I come for the political, I’m glad that you have your priorities straight and are focusing your attention on what is definitely the more important, long-lasting, humanity-benefiting work you do. Keep up the great work on both counts! Thanks, Sam!

    • Daniel

      This response made me smile a lot, so thank you for being a nice person.

    • Kari Q

      Exactly what Brian C. said. As much as I value this site and rely on it, if it needed to go to leave your time free for research, I would gladly let it go. May your research live up to your hopes and improve our understanding of how humans function.

  • pechmerle

    I’ve been coming here since the summer of 2008, when PEC contributed greatly to preserving my sanity.

    There is another reason to keep the site going:
    The Act Blue recommendation of close races to consider contributing in is a link that I can send to friends and relatives, suggesting that this is where they – if so inclined – could best make their own contributions to candidates. And I have forwarded the link for that purpose.

    Thanks for all that you do and have done.

  • John M

    PLEASE keep this site active. It keeps me sane!

  • Joseph

    Hey, if all we can be is your “hobby”, we’ll take it! I gladly join the chorus of grateful consumers of PEC in thanking you for all your efforts.

  • Rachel Findley

    I am grateful for the site. Even if you just run the numbers, and spend all your thought time on why and how our human brains work, it’s useful to have a known method presenting results with no thumbs on the scales.
    I do wonder about the the way the Senate prediction switches from Republican mode to Democratic mode, and back again. Is there an (easy) way to put up a Senate map, or a history of the Senate bar chart?

    • Sam Wang

      Yes, the Senate calculation is vexing. Not enough data there. And in many races, the primary is not until August.

      For individual races, click on the links in the right-hand sidebar (first click Senate). We don’t maintain a CSV our interpretation of the feed, which maybe we should do. Medians are found here but require this decoder ring.

  • OldenGoldenDecoy

    Thanks a million there Dr. Wang…

    I don’t know what was in that water back in the late ’60s in Riverside but it should to be bottled under the brand name,

    “Wang Water… It Stimulates the Brain!”

    From the old old duck in North Hollywood.

    ~OGD~

  • SP

    I’ll join the chorus and thank you for keeping our collective sanity intact. It’s no easy feat, what with the media proclaiming on a nearly daily basis that The End Is Near! on the basis of one poll or another.

  • Bruce McLin

    Sam, thank you for the efforts you put into your predictions. I came across, then lost, then found again your sight during the 2012 election. While not quite a 538 refugee, I come to your site to validate or counter their up and down swings. Earlier commentators said what I would have written, so, again, thanks, and good luck on your research. I will continue to faithfully, daily, follow this site.

  • SK

    “I think it might be the relative purity of PEC’s calculation”

    THIS^^^^

    Polls-only makes it boring, but also very reflective of what the race really is.

  • Ravilyn Sanders

    Dr Wang, please keep this site alive. We need someone who is not after clickbaits and page views and ad revenue.

    Every time I look at Nate Silver’s now-cast, I am reminded of the xkcd on extapolation: https://xkcd.com/605/

  • Allan Lindh

    Dr Wang, Extremely grateful. A voice of sanity and intelligence in the wilderness. If I had to pick single best site of the Web, you win.

  • Frank

    Adding to the chorus of thanks to you, Prof. Wang. I’ve also forwarded PEC to others, and have personally learned a lot about statistics through it.

    I foresee that the natural evolution is for PEC to evolve into a nearly fully automated program, with little editorial input. When an update or correction to its code is made by its developer (you), a “what’s updated” list is posted, just like when a new version of an app is released. Otherwise, PEC updates its predictions automatically, doing so without emotion, and providing full disclosure of its inner workings. Like a finely engineered clock, it should run reliably and without comment by the engineer. What it displays is all that should matter.

  • Kristin Larkin

    Thank you for your work. I have a child on the autism spectrum, and I’m appreciative of your research. Your PEC site keeps me from going crazy during election seasons, and I am a long-time follower.

  • Catherine

    I love the openness and honesty of your blog and analysis. It’s fresh, unbiased and provides a scientific take on a somewhat rough-and-tumble process. Keep up the work.

  • Alan Cobo-Lewis

    Is there a bug in the “power of your presidential vote” calculations?

    ME, which has 4 EVs, is currently ranked 2nd, with power of 91.0, even though margin is Clinton +7. Seems very high unless estimated SE is really big. But even if it is, NH, which has same # of electoral votes, is ranked way lower, with power of 45.3. This seems like it should be impossible.

    • Olav Grinde

      @Sam: Would it make sense to add Georgia and Arizona to your “Power of One Vote” list?

      Both states are becoming surprisingly competitive! As Hillary’s MetaMargin increases, the same might be true of other states. (It appears as though even in Utah Trump’s leading is shrinking to the low single-digits.)

    • Sam Wang

      Will investigate. It is supposed to be calculated by shifting all polls by the Meta-Margin, then perturbing one state by a fraction of 1% corresponding to a fixed number of people. The relative effect on win probability is the power.

  • BillSct

    Sam, let me join the chorus of people thanking you for the work on this site. I certainly understand the need to refocus on work. Perhaps the solution for this site in the future is to make it a true consortium with more of the workload distributed to the many capable folks who contribute to comments and the discussion of statistics and polling. I would however, encourage you to continue your efforts on the gerrymander problem. I actually think that is the root cause of the dysfunction in congress and the rise of Trumpism in the Republican party. Far more so then “big money.”

  • Sri Nune

    Thank you Dr. Wang ! You are my morning, afternoon and evening ritual ! This site is amazing !

  • Phoenix Woman

    Thank you, Sam, for doing what you do.

    Please keep this site going. Please. If time is tight, have your students take it over as a gradeable school project, one that you check in on maybe twice a month to make sure it’s not gone off the rails.

    It is the best site of its kind out there – it seems as if every other site out there, with the exception of Electoral Vote, is in it as a money-maker or as a way to please various powerful interests, which means that they have their thumbs on the scales to one degree or another.

  • Keith R

    Let me add to the chorus urging you to continue this work as your no doubt heavy day-job load allows. One thing to consider is the number of people who say this site helps keep them sane, making it, ironically enough, a closer adjunct to your main work than perhaps initially intended.

  • ottovbvs

    Very relieved you did. You are the go to guy Sam for election analysis. Far superior to pale imitations. Buy the genuine article!

  • Dave Kliman

    I come here because you’re accurate. I don’t remember anybody being more correct in their predictions, election after election, than you.

    Why should I look at all the noise out there when I can see what’s going on from right here?

    I appreciate the well educated statistical work that comes from somebody who is a professor of statistics at one of the most prestigious educational institutions in the world. I value being in your good hands, on this subject. Even with all that, you are still very humble, down to earth, and accessible, for example when you did a deep exploration of gerrymandering when some readers suggested it might be interesting to look at.

    All I can say is Thank you for all of your good work.

  • Marlene Snyder

    Sam–I want to add my thanks. I first found that you helped maintain my sanity in Oct of 2012, and you CERTAINLY have done that this year!

    I’ve been a little anxious at the thought that the increased traffic you have noticed was all due to the compulsive checking in to the site that I am doing way too often right now! Can one person’s hyperactivity look like increased traffic?!

  • James

    Sam
    Your’s is only credible site in these matters. All others influenced by personal, corporate or insipid “journalistic” quest for impartiality – an oxymoron especially with Fox, CNN, networks and NPR.

    Followed you closely in 2012; now daily in 2016. Directing my teen daughters to your site.

    Please don’t go away.

    How can we help you monetize this effort and your exemplary efforts?

    Greetings from your old Caltech/Pasadena haunts.

  • Aravind

    Brilliant work, Dr.Wang. Most definitely the very best place to check on the actual state of the race for a political junkie like me, multiple times a day.
    Thank you.

  • Chuck

    Personally, I appreciate the fact that you’re not driven to provide constant updates, whether they are warranted or not—which tends to be a problem for Nate Silver’s site. I think there’s a certain freedom in not having one’s financial welfare tied to the web equivalent of piecework.

  • WildIrish

    I found PEC in 2004, and have been a very appreciative lurker ever since. Not being a stats geek, I really appreciate the clear explanations of methodology, as well as the respectful comments on this site. It’s not about who’s right, but rather whether the methodology is the best it can be. Very refreshing, considering the toxic sludge found in most political comments. I will add my voice to the chorus of appreciation.

Leave a Comment