Benchmark Politics recent tweets on CA primary

make of them what you will

 In response to Mark Murray tweeting about new poll from NBC/WSJ/Marist poll showing Clinton up only 2

x

I don't anticipate the Benchmarks to go down too far though for Clinton. Polling is only 25% of explained variance. https://t.co/Hm2zNMhQzq

— Benchmark Politics (@benchmarkpol) June 1, 2016

Now this is noteworthy

x

Among those who already voted, Clinton up +17%... so that's money in the bank for Clinton. Means poll swings may not be AS important.

— Benchmark Politics (@benchmarkpol) June 1, 2016

 and IIRC demonstrates a trend that has been consistent, that Clinton does better with early voters, and Sanders tries to match that with later voters he turns out by his rallies.

x

New California Benchmarks: Clinton 54%- Sanders 46%. pic.twitter.com/lXViD6KQyv

— Benchmark Politics (@benchmarkpol) June 1, 2016 x

Nate Cohn and 538 and we all almost have identical numbers. Interesting. Models probably all converging. https://t.co/RTl0B6sbcd

— Benchmark Politics (@benchmarkpol) June 1, 2016

and in response to Nate Cohn tweeting that the poll shows Clinton winning among whites and losing among non-white:

x

This is probably not likely. But its interesting that two polls found it. https://t.co/yS8zEn5aXR

— Benchmark Politics (@benchmarkpol) June 1, 2016

So remember, Benchmark depends primarily on its demographic model, with only some tweaking based on polling

Also remember that an average of polls is usually far more dependable than relying on any single poll, although a single poll could be absolutely on target.