Princeton Election Consortium

A first draft of electoral history. Since 2004

Meta-Margins for control: House D+1.0% Senate R+4.2% Find key elections near you!

Sharp cutbacks at HuffPost Pollster

June 15th, 2017, 2:21pm by Sam Wang

This is terrible news. As part of cutbacks at the Huffington Post, the tracking of polls will be curtailed dramatically. As reported by HuffPollster’s Ariel Edwards-Levy, they will only be tracking Trump’s favorability/unfavorability. All the other charts are now frozen. Evidently that includes the
generic House ballot, which I think is quite important.

This is a tremendous loss. The Princeton Election Consortium relies on their feed. They curate data, they apply their judgment, and they generate a structured API for dozens of races every year. It’s been a wonderful resource.

Personally, I would be willing to pay for such a feed. However, I’ve never been asked. Also, it’s not obvious whether there are enough subscribers to sustain their work.

I hope they re-expand operations in the future. In the meantime, I thank the many people who make HuffPost Pollster possible: Ariel Edwards-Levy, Janie Velencia, and before that Natalie Jackson, and before that Mark Blumenthal, and too many others to list here.

Here’s one of my favorite charts of theirs: the generic Congressional ballot. It’s for aficionados…but I do love it.

Tags: Politics

7 Comments so far ↓

  • LondonYoung

    HuffPost staff – unionized just a few months ago … I suspect the poll data will be back, but the new jobs to support it will be outside the union and outside the US

  • Amitabh Lath

    I know the Huffpost pollster provided a convenient place to get all the poll results but couldn’t a smart undergrad write a chron job that you run hourly to scour the top dozen or so pollster sites to scour the most recent results?

    Or did they also provide some subjective filtering based on the expertise of Blumenthal et al?

    • Sam Wang

      They did filter by reputable pollster…but actually, this became a sticking point for me. They started omitting landline-only polls, which I thought was excessively stringent. When I expressed concern, I was blown off dismissively.

      So…yes, I agree with your recommendation. Probably the right approach is to start using a combination of FiveThirtyEight and RealClearPolitics data. RCP doesn’t provide a feed – it’s their business model to make people go to their site.

    • Amitabh Lath

      Why not go direct to,, etc?

    • Sam Wang

      There are so many polling organizations, and a lot of scraping to do. Local-race polls come from unexpected places – for example, neither Gallup nor Pew conducts such polls. With limited time and no budget, I don’t want to reproduce the work of others.

  • Lorem

    (Mostly off-topic) Sam gets a mention primarily for eating a bug in a pretty reasonable and informative article about polling by the Economist:

  • Colin McAuliffe

    The number of undecideds seems a lot higher than I would have guessed in our polarized political environment. Do pollsters have methods for trying to determine if someone has their mind more or less made up despite reporting that they are undecided?

Leave a Comment