For Democrats, the results of the November elections were hard to immediately understand. The party retook the White House, with Joe Biden winning more votes than any other candidate in history, while it simultaneously lost at least 10 net House seats to Republicans and failed to win Senate seats in states like Maine that had appeared to be easy pickups. It was a confounding mix of success and disappointment for a party that had aimed to capture a mandate for unified government.
That confusion didn’t stop prominent Democrats from quickly offering their theories of the case. In the weeks since the election, they’ve sorted themselves into predictable camps. Centrist Democrats like Abigail Spanberger are arguing that Democrats lost seats because of leftist policies like “Defund the Police” and the embrace of “Democratic Socialism.” Rep. Alexandria Ocasio-Cortez and others towards the left of the party blamed the Democratic National Committee, an overreliance on poll-tested messaging, and traditional campaign tactics that focused on TV at the expense of digital ads and social media.
Both sides are equally adamant, and neither side is really sure what it’s talking about. They can’t be. We quite literally do not know who voted yet.
I don’t mean there’s something dubious going on. It’s just that during this postelection period, when opinions about what happened in the election are at their fieriest and in the greatest demand, the nation’s secretaries of state are still in the process of updating each state’s voter rolls, which are in many cases amalgamations of county voter rolls. If past years are any guide, most states won’t have their voter rolls ready to release to the public before the new year, and some particularly slow states won’t be available until March or even April.
The voter rolls, commonly called voter files, are the material the entire political industry is built on, essential tools for political campaigns and the network of consultants and companies they employ. At their most basic level, voter files contain a list of every registered voter in the area, including their name, address, and a record of which elections they have voted in—but not who they voted for—along with their method of voting (in person, absentee, etc). These lists are packaged together into national voter files by partisan firms like The Data Trust on the Republican side or TargetSmart on the Democratic side (disclosure: I used to work at TargetSmart). These firms merge together the various state files, adding useful information sourced from voter contacts, commercial data purchased from companies like Acxiom, and statistical models along the way.
Before an election happens, modern campaigns use the files for everything from telling field organizers which doors to knock on to helping large agencies plan and target their ad campaigns. Afterward, the updated files are the main tools data professionals use to figure out what did and didn’t work: They are the data you turn to if you want to say that a campaign tactic caused a specific segment of voters to turn out, or if you want to contact actual voters to scientifically survey them about whether a message changed their minds.
When I worked at Bully Pulpit Interactive, a national consulting firm, part of my job was to figure out how to measure the impact of our work for the Hillary Clinton campaign and other Democratic organizations. While we routinely used surveys and other methods to improve our ad campaigns, we couldn’t prove our impact on the only metric that really matters, ballots cast, until well after the election when voter files became available.
Throughout the course of a campaign, political operatives generate vast troves of data on which potential voters they talked to, which voters they tried to reach with advertising, what voters said during telephone interviews, and more. But it’s not until the voter files come out that we know exactly which ones of them participated in the election.
Without a voter file, pundits and politicos are left to conduct analyses with a limited, and often deeply flawed, set of tools. Typically these include pre-election surveys that make assumptions about who will turn out in an election based on turnout in prior years, exit polls (conducted at the polls with voters and over the phone for voters who have already cast ballots), and initial returns as precincts fully report.
These methods failed in 2016 and are likely failing now. In 2016, reliance on non–voter file methods produced the widespread claim that Trump had won an outright majority of white women, giving the President and his supporters a powerful talking point—one that had already become conventional wisdom by the time verified voter surveys from Pew and others, using the voter files, debunked it. The exit polls also significantly underestimated the number of non-college voters that turned out.
Waiting to speak, or speaking with less certainty before solid data was available, would have prevented these errors. When voter files become available next year, the DNC’s data team will be able to analyze the specific impact of their work in the elections they contested. AOC will be able to show how her work with the congressional campaigns she supported did or did not increase their margins.
Until then, though, these factions are largely relying on exit polling and anecdotal evidence. The Democratic party spent large parts of the 2020 campaign calling on Americans to use data and science to inform their decisions. They should take their own advice and wait for the data to become available before drawing conclusions about what those voters did.
from Slate Magazine https://ift.tt/3m9OS1H
via IFTTT
沒有留言:
張貼留言