Getting out the vote (or "GOTV") involves three basic steps: (1) creating a list of registered voters who are likely to be for your candidate or persuadable, (2) asking those voters who they intend to vote for, and (3) getting the confirmed supporters to the polls on election day.
If you're interested in learning how the Obama campaign collects and processes the GOTV data, there's a long explanation after the jump.
The first step is compiling the "GOTV universe", all registered voters who we want to talk to. It's not something volunteers get involved in, but from what I understand we're targeting (a) registered Democrats who have voted infrequently in the past two presidential cycles, (b) independents who voted in prior years in precincts when Democrats did well, and (c) newly registered Democrats and independents.
Four years ago, I was a precinct leader for MoveOn, getting out the vote for John Kerry. MoveOn staff identified the voters we wanted to reach, which were compiled in what are called "walk lists" or "call lists". MoveOn had its volunteers pull from the call list online, make calls from home, and then enter the results online. For the walk lists, precinct leaders could print them out. We then turned to the teams of volunteers each of us had created and went out and knocked on doors in the weeks leading up to election day. The walkers would record the results of the canvass on the paper walk lists, which came back to the team leaders for data entry on the web so that the next time the walkers went out, they would only try to reach the voters who we hadn't contacted yet. At the time, it seemed like magic, even though the online data entry task was a bear.
The Obama campaign has taken the process further and made the data-entry portion quicker and easier. Here's a (heavily redacted) copy of a page from a walk list:
I've edited the identifying information for the first four voters, moved stuff around and blurred out everyone else on the page, but as you can see, there are twelve voters on this page, the page is 3 of 35, and it was printed last Saturday. You can't see it as readily, but I can tell you that of the twelve, only three were successfully surveyed. Of the rest, six were not home, two did not speak a language the canvasser spoke, and one had moved. The overwhelming number of times we knock on a door, there will be no answer, which explains how one canvasser can cover a list of 420 names in a single shift. It also means we have to make many attempts via phone and doorknocking to reach any given voter.
Here's a closeup of the top left of the page:
Starting from the top: The list is identified as Philadelphia, ward [blank], division 19, [blank] St., odd. This identifying information is on every page so you can pull the list apart and not lose track of where the page belongs. The designation "odd" means this list is for odd-numbered addresses on [blank] Street in that ward and division. Why just the odd numbers? So you can walk straight down a sidewalk and knock on the doors on one side of the street. Another page or another list has the even numbers for that street. Saves time.
The next line gives the campaign's numeric code for the precinct (called a division, in Philadelphia) and the location of the polling place. The reason for that is obvious: The number one question most voters have about election day is, "Where do I vote?" This can change before election day, but usually doesn't.
Then the actual list starts. Each record has the Voter Activation Network (VAN) number for that person in numeric and barcode form. Beneath that is the name of the "FO", the field organizer, an Obama staffer for that area. The next column again has the voter's city, ward and division, and then gives the volunteer team leader assigned to the canvass. The next column gives the voter's name, address, age, sex, and party. The name and address is bolded to help the canvasser find the right person. The other details serve the same purpose: You might be surprised how many addresses have two voters of the same name - such as a senior and junior - and they vote in completely different ways. You really don't want to canvass the wrong person.
The first three voters are all at the same address, house number 121. Then there is a heavy line, and the next voter is at number 129. Visually, the canvasser knows how many people she's looking for at any given address before she knocks on the door.
Here's the closeup of the top right of the page:
The first three voters were not home, so the canvasser circled "NH." (I'll explain the other codes later.) However, the daughter of one of the voters was home, and even though she is ineligible to vote at age 16, she volunteered to work on election day, which the canvasser duly noted. When the lists go back to the field office, the FO copies all the notations and makes sure they are followed up on. Then the lists go to Philly HQ to be data-entered.
But I am getting ahead of myself. Look at the line that says "ID" in bold: This is the information we need: The voter's current preference for President. The numbers mean: 1 for support Obama, 2 for lean Obama, 3 for undecided, 4 for lean McCain, 5 for support McCain, 6 for support Nader, 7 for support Barr, and 8 for other. The first voter has a notation: "Current ID: 2." We probably got that information somehow over the summer. It's stale now, but still helpful.
The last column is where the canvasser should record the response to the question she should ask all the 1's, the support Obama voters, "Will you volunteer between now and election day? If yes, when?"
The fourth voter, at number 129, was home, but unfortunately the canvasser did not speak her language. A volunteer who does will contact that voter.
So the completed walklist is reviewed by the FO and then sent to headquarters for data entry. This is where a simple technological improvement like bar codes really helps. Four years ago, I had to open each voter record and check the right boxes to update the walklist. It took at least 30 seconds each. At Obama headquarters, I scan each voter's bar code on the walklist using a reader attached to the computer, the voter record comes up on the screen, then I scan the canvassing results using the barcodes on a list kept next to each station:
The "Canvass Status" with the notation "Call" next to it on the left and the "Canvass Status" with the notation "Walk" on the right are nearly the same. The codes on the left are used for phonebanking and the ones on the right for the door-knocking. When you contact a voter, you either are successful in getting their information (who they are voting for, what they care about, and if they want to volunteer), or you aren't. If you're not, for the phonebank canvass the possible categories are busy, call back, deceased, left message, not home, (speaks only) other language, refused, (speaks only) Spanish, and wrong number. For the door-knocking canvass, the responses are deceased ("DC"), inaccessible ("IN", meaning canvasser can't get to their door or buzzer), moved ("MV"), non-citizen ("NC" - hmm, how are they registered, then?), not home ("NH"), speaks only a language other than English or Spanish ("LG"), refused ("REF"), or speaks only Spanish ("SP").
Using the bar codes, I can record a "Not Home" - the most common response - in about 3 seconds on average. For those we do reach, it's no more than 5 seconds each to record the result and wait for the screen to refresh. I timed it.
In practical terms, what this means is that Sunday night a team of twelve of us burned through all of the data-entry for that day's canvass results for the entire city while watching the Phillies on somebody's slingbox hooked up to a projector. (The walklist page above is the one I was working on when Ryan Howard hit his three-run shot.)
If a voter is successfully canvassed, or if the record is marked deceased, moved, non-citizen, inaccessible or refused, that record is removed from the next walklist. When a canvasser is successful in speaking to a voter, the most common answer to the presidential preference question is "1", support Obama. Those people should all be asked to volunteer, but many canvassers fail to ask the question. If the person is lean McCain or support McCain, they are also removed.
If the person is not home or there is a language barrier, they go back on the list to be contacted again. When we come across undecideds or lean Obama voters, the data entry volunteers say, "Those poor people." We sympathize because we know that those voters are going to suddenly be the focus of a lot of attention from the Obama campaign, whether it's wanted or not. The undecideds we'll try to persuade, and the leaners will be offered every assistance in getting to the polls, because they are the people who are most likely not to show up.
That's GOTV as I understand it up until election day. This process ends the moment the polls open, at which point the entire effort becomes about turnout.
COMMONWEALTH ELECTORAL ACT 1918 - SECT 245
Compulsory voting
(1) It shall be the duty of every elector to vote at each election.
I know it raises American hackles, but it does a huge amount to equalise the parties, as does requiring the independent electoral commission to ensure that every elector is registered and providing a simple and fast method of appeal to deal with errors on the roll. The penalty for not voting is AU$20.00 (US$12.80).
Posted by: Alan | October 29, 2008 at 09:43 AM
Is it enforced? What are the exceptions? Is early voting or voting by mail allowed?
Posted by: Mithras | October 29, 2008 at 09:51 AM
Hey, this is a great description of the nuts and bolts process. Very interesting.
Posted by: Andrew | October 30, 2008 at 08:55 AM