The [Hillary Clinton] campaign relied far too heavily on something that campaign technicians call “data analytics.” This refers to the use of models built from a database of the country’s 200 million–voters, including turnout history and demographic and consumer information, updated daily by an automated poll asking for vote preference to project the election result. But when campaign developments overtake the model’s assumptions, you get surprised by the voters—and this happened repeatedly.
Astonishingly, the 2016 Clinton campaign conducted no state polls in the final three weeks of the general election and relied primarily on data analytics to project turnout and the state vote. They paid little attention to qualitative focus groups or feedback from the field, and their brief daily analytics poll didn’t measure which candidate was defining the election or getting people engaged.
This was supposed to be a re-run of the Obama campaign approach, repeating its success.
I'm transcribing this passage from a physical book, The Shield of Achilles, by Philip Bobbitt, quoting Edward M. House's Philip Dru (p. 393).* So you're welcome. For the 1916 re-election campaign, President Wilson's close advisor and strategist House believed there were:
twelve states that were debatable , and upon whose votes the election would turn. he divided each of these states into ["the smallest possible units that could be arranged with available campaign funds"] and began by eliminating all states he knew the opposition party would certainly carry ... He also ignored the states where his side was sure to win. In this way he was free to five his entire thoughts to [a very few districts]. Of [a unit of] five thousand, he roughly calculate there would be two thousand voters that no kind of persuasion could turn from the national party and two thousand that could not be changed from the opposition. This would leave one thousand doubtful ones to win over. So he had a careful poll made in each unit, and eliminated the strictly unpersuadable party men, and got down to a complete analysis of the debatable one thousand. Information was obtained as to their race, religion, occupation, and former political predilection.
The campaign focused on just the key districts in the key states they needed to add to their base electoral vote tally. It worked.
And it was done before computers.
The passage leapt out at me, and seemed so out of place and time that I'd almost suspect a time traveler with a laptop doing the data crunching to return President Wilson to the White House for some purpose we will never know.
Although to be fair, Clinton's data crunching was really model crunching. Model outputs aren't data.
*Although the work quoted is fictional and pre-dates the campaign, Bobbitt takes it as reflecting what House believed and followed in the actual campaign.
