Poll (Aggregator) Bias in Dem Primaries To Date

I’ve averaged pollster and poll aggregator bias for all polling firms and for 538 and RCP for the three Democratic primaries to date. I have not included firms that only had one poll. The best of them all is CBS/YouGov with only a +2.1 bias in favor of former Secretary of State Hillary Clinton (HC). Only one of the ten has a pro-Senator Bernie Sanders (BS) bias: CNN at +5.43 BS.

Excluding CNN, the Clinton bias, to date, averaging the biases for all polling firms plus 538’s polls plus and RCP is +5.37.

This will look much more interesting after South Carolina for two reasons: 1) only two results for several of these firms means a less accurate bias measurement; they all have polls out ahead of tomorrow 2) there is a massive gap between RCP averages and 538’s polls plus predictions for tomorrow. Both may change before final vote; I’ll update Sunday or Monday accordingly.

Gravis: +5.9 HC
Emerson: +9.35 HC
538 Polls+:  +2.33 HC
RCP Avg: +3.27 HC
ARG: 4.1 HC
NBC/WSJ: +2.6 HC
CBS/YouGov: +2.1 HC
Monmouth: +9.6 HC
Fox: +9.1 HC
CNN: +5.43 BS

Updated and Final: South Carolina Prediction (based on weighting best poll and poll aggregators to date, weighted for bias, then averaged)

Updated and Final: South Carolina Prediction (based on weighting best poll and poll aggregators to date, weighted for bias, then averaged)

For all pollsters and aggregators with a less than 5.5 bias, I’ve subtracted that number from their South Carolina predictions for Saturday, then averaged them for a prediction that Saturday’s outcome, if this is an accurate method, should be 27.05% in favor of Clinton. If instead we put CNN (with bias) into a cage match with the other nine (with bias), the result would be a 25.4% advantage for Clinton. I’ve taken all of these factors and many more, including especially Ann Selzer’s prediction as the best pollster in the land, and my final prediction personally is Clinton +23.47.


538’s Polls Only and Weighted Polls Only Forecast As of 12:01am Friday Feb. 26, 2016

Interestingly, before what I am going to call the 538-polls+ Effect set in over the course of yesterday, 538’s polls only average for South Carolina was 25 points. It has now jumped to 33.2 unweighted and 31.7 when weighted. 538’s final Polls Plus call is Clinton +38.3, giving them a whopping 16.3% difference from their favorite pollster (Selzer).



n.b. I could have included PPP as it technically meets the criteria, but have eliminated them for now as their New Hampshire poll was nearly two months  ahead of time. They would accordingly have a 16.4 bias.

Updated Saturday 3:30 pm eastern to reflect final numbers from 538 and RCP, to correct minor math errors in the “South Carolina Prediction,” and to peg my final forecast.


One comment

  1. brian · October 30

    I mean they were right


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s