Nate Silver vs. Sam Wang

elections, Obama Administration

In 2012, as I remember, Silver and Wang’s election forecasts remained close to each other. But right now they are considerably apart. Silver says Republicans have a 63.8 percent chance of winning a majority. Wu says Dems have a 79 percent chance of keeping the Senate.

Where do they differ? Silver thinks Pat Roberts of Kansas will keep his Senate seat; Wu does not. The pair of prognosticators also split over North Carolina. They must disagree on some other races but I cannot tell which ones.

Share Button


  1. LongHairedWeirdo  •  Sep 9, 2014 @3:20 pm

    They don’t actually differ by all that much, it looks like. Silver is calling 54% chance of Roberts keeping his seat. That’s a tossup – but maybe within a margin of error, it’s a 49% chance. Similarly in NC (55%). And Silver does random “plays” of elections, running many simulations to see who wins most random elections, which is a good method, but there may be superior methods, or superior models for the random elections.

    Obviously, I hope Silver is wrong (both to hold the Senate *and* because statistics will learn something new because he will by-golly figure out how to improve his model if he’s wrong).

  2. c u n d gulag  •  Sep 9, 2014 @4:25 pm

    Silver may be the “Golden Boy,” but, if I remember right, Wang actually out-performed him in the last election – maybe two.
    Of course, with my memory…
    I used to have a mind like a steel trap. Now, it’s more like a… a… a… a spaghetti-strainer!

  3. buddhacrone  •  Sep 9, 2014 @5:38 pm

    I have a mind like a steel trap–rusty and illegal in 47 states. The current strainer is an improvement……I think.

  4. maha  •  Sep 10, 2014 @6:47 am

    I have a mind like a steel trap–rusty and illegal in 47 states.


  5. goatherd  •  Sep 10, 2014 @8:00 am

    It’s probably too old to be funny, but, we use the “mind like a steel sieve” variant.

  6. James F. Epperson  •  Sep 10, 2014 @11:52 am

    My understanding is that Silver uses “fundamentals” to augment his models when there isn’t much polling data, i.e., stuff like “Kansas is a deep red state.” If more polls say Roberts is in trouble, Silver’s model may begin to reflect that.

  7. JDM  •  Sep 10, 2014 @3:11 pm

    I’d go with Wang if I were betting. Silver has shown, particularly with his climate change BS (which is at the very least borderline denialism, and brain dead in his hiring of Pielke), that his belief systems can and do get in the way of his numbers sense.

1 Trackback