FiveThirtyEight's "Facebook primary" numbers were a pretty reliable predictor of Sanders' performance by county in New Hampshire. If you deduct 20 percentage points from his two-way share of FB likes with Clinton, and you're within ~2 points of his actual result in every county.
Whether this dynamic will hold up in later primary states that don't get the same kind of treatment remains to be seen, and there's no accounting for movement in the weeks just before the vote, obviously. Nevertheless, here's what we get if we apply the same rule to NYS:
Color categories represent 10-point intervals. Sanders counties are green, Clinton counties are red.As impressive as this looks for Sanders (and it is, to be clear, an over-optimistic portrait), over two-thirds of ballots cast in the 2008 Democratic primary were in New York City, Long Island, or Westchester County. Excluding the counties containing the three largest cities - Buffalo, Rochester, and Syracuse - Upstate New York was responsible for less than one-fifth of Democratic primary votes. The result, based on county-level turnout during the 2008 primaries, would be:
52.1% Clinton (~985k votes)
47.9% Sanders (~905k votes)
That would be a shocking number, and "the machine" will undoubtedly keep Clinton well above 55% at a minimum. It's also worth keeping in mind that in New York State we only have "elections" in the loosest sense of the word. (e.g. In 2008, votes for Obama were literally
not even recorded in many majority-black precincts. This isn't a Clinton thing, it's a New York State thing).
