I was doing a bit of research, and I came across a bit of interesting analysis that Nate Silver did a few years ago.

In looking at very early primary polling (polls conducted the year prior to primaries, like where we’re at now), he attempted to compensate for name recognition among candidates by taking their polling and dividing it by their name recognition.


Candidate A is polling at 10%. But only 50% of people have heard of him. So, we attempt to anticipate how he’ll poll when 100% of people have heard of him by taking his 10% (0.1) and dividing it by 50 (0.5). That gives us 0.2, or 20%. It’s not a perfect system by any means, but allows us to deemphasize people who are polling well in large part due to major name recognition (Biden, Sanders), but don’t have much room to grow because pretty much everybody knows who they are, versus people with lower name recognition (Booker, Gillibrand, Harris, etc.) who will become more well known as time goes on, and gain supporters.

So, I dug up current name recognition data and the current average polling numbers I’ve been generating and performed the operation described above (and then make an adjustment to make sure everything was out of a potential total of 100%), and produced polling numbers adjusted for name recognition. And, uh… it didn’t change a hell of a lot.

  • Biden: 29.8%
  • Sanders: 21.6%
  • Harris: 14.4%
  • O’Rourke: 8.3%
  • Warren: 8.1%
  • Booker: 6.8%
  • Klobuchar: 4.5%
  • Brown: 2.2%
  • Everyone Else: < 2.0%

The main effect was that it strengthened Harris, O’Rourke, and Booker a bit, and revealed how weak Warren is compared to her name recognition.