Local campaign veterans know the drill: examine precinct voting records to map base, swing, and hostile precincts, then canvass in the swing precincts and GOTV with your base. But how many votes did you miss? (More)
The Victory Lab, Part I: Who Votes How, Where?
This week Morning Feature discusses Sasha Issenberg’s The Victory Lab: The Secret Science of Winning Campaigns. Today we look at the familiar, precinct-based analysis and planning. Tomorrow we’ll see the rise of data-mining and campaign plans that target individual voters. Saturday we’ll ask whether micro-targeted voter contacts provided the Obama campaign’s decisive edge.
Sasha Issenberg writes the Victory Lab column for Slate and is the Washington correspondent for Monocle, where he covers politics, business, diplomacy, and culture. He covered the 2008 presidential election for the Boston Globe, and has also written for New York, The New York Times Magazine, The Washington Monthly, The Atlantic, and other magazines.
Still the Standard: Precinct Analysis
Last July I attended and wrote about the Democracy for America Campaign Academy. In that course, we learned the kind of precinct-level analysis that was developed by the National Committee for an Effective Congress and is still widely used by Democratic candidates and local party groups across the country. It focuses on three measurements for each precinct:
- Democratic Performance Index (DPI) – The average percentage vote over the past three similar elections for viable Democratic candidates.
- Persuadable Voter Index (PVI) – The average number of voters who switch between Democratic and Republican candidates, over the past three similar elections.
- Voter Turnout Index (VTI) – The mean and standard deviation for turnout over the past three similar elections.
Based on that data, candidates and party groups can prioritize both persuasion and GOTV efforts. For example, in base precincts (DPI over 65) you needn’t spend much time convincing voters to vote for Democrats. Instead, you should make persuasion contacts in swing precincts (DPIs between 35 and 65) and focus first on precincts with the most persuadable voters (highest PVIs).
As election day nears and you shift to GOTV, your efforts should begin in base precincts – where most voters will support you if they vote – and focus first on the precincts with lower average turnout but higher standard deviations. Your GOTV efforts then shift to swing precincts, focusing on voters that looked promising in persuasion contacts.
The Benefits of Precinct Analysis
This approach to campaign planning is still widely used, and it has benefits. For starters, it’s inexpensive and fairly easy. The data are public information, usually kept by local elections officials. They may charge a nominal fee to get the precinct-by-precinct, race-by-race vote data in spreadsheet form, and tinkering to turn raw vote totals into each precinct’s DPI, PVI, and VTI scores may take you a few hours the first time, but that tinkering is less difficult than tedious. And for federal races, your campaign or party group can also get precinct DPI, PVI, and VTI scores directly from the NCEC.
Precinct analysis also helps you narrow your target universe. It can be daunting to see you need 135,219 votes to win a county commission race in a midterm election. To see that broken out to 1256 votes for Precinct 19, 1183 for Precinct 22, etc. feels easier, especially if you then prioritize your precincts for both persuasion and GOTV. The vast task of amassing 135,219 votes now has a specific and manageable first step: recruiting volunteers to make persuasion calls and canvassing in Precinct 53.
If you get the volunteers and work your plan all the way through to election day, you should have a reasonable shot … as reasonable as the electorate that year, anyway. If a blue wave is carrying you, you may ride it to victory. But if a red tide is pushing against you, even your best efforts may fall short.
The Limits of Precinct Analysis
But did you get every vote you could have? There are usually at least some Democratic supporters even in hostile precincts (DPIs under 35). How many didn’t vote for you or your candidates because no one in your campaign or party did GOTV work there? How many voters did you turn out for opponents in your base precincts because your GOVT efforts boosted turnout overall and didn’t distinguish supporters from non-supporters?
And if it you lost by only 552 votes, was it really ‘okay’ to skip persuasion calls in those 12 precincts, totaling 14,281 voters, whose PVIs averaged only 8%? Had you reached and won half of those 1143 voters, those 571 votes would have put you over the top. But what else would you have had to cut back to commit your and your volunteers’ time to identifying and persuading those 1143 swing voters? Would chasing them have cost you more than the 571 votes you might have gained?
When precinct-level data are all you have, those questions can be as maddening as they are impossible to answer. You can’t know how persuadable those 1143 voters were. Maybe you’d have won more than half, with the right message. Or maybe they were merely a statistical mirage, and every volunteer-hour you spent looking for them would have been wasted.
Even if those 1143 voters existed and you found them, what would the “right message” have been? Would they have responded to your basic persuasion script, or would you have needed to emphasize some points and downplay others to reach them?
Come to think of it, should you have used the same persuasion script countywide? Might you have gained a few more votes here, and lost a few less there, if you’d tailored your script to the issues and solutions that mattered to those voters? How could you know?
Political consultants and media types you worked with all have their answers. Oddly, the consultants blame the media types, and vice versa. Each consultant and each media type has a different story, there’s only one thing they all agree on. Each insists your loss was not his or her fault.
In fact, about all you’re certain of is that you now know every sound your house makes at 2am as you lie there, staring at the ceiling, pondering What Might Have Been.
Tomorrow we’ll see how advances in data-mining help campaigns find those 1143 voters, and others, and know what messages are most likely to resonate with each.