To Knock Or Not To Knock

In the press coverage surrounding political campaigns, we are inundated with stories on what campaigns should do to win. Often times, this coverage takes the form of observing the machinations of a particular campaign, comparing it to a study or set of studies, and then asking the question “Why aren’t these guys doing what the study says gets you more votes?”

For example, many reporters have recently discovered the work of political scientists who are quantifying the effect of campaign tactics. This branch of academia, started by Don Green and Alan Gerber, has shed light on what works and what doesn’t work in campaigns. At Øptimus, the firm I work at, the books and papers emanating from this branch of political science are required reading. To put it lightly, we are big fans. That being said, you have to be careful in how you interpret and apply the findings from this branch of work.

Take for example, what the literature tells us about the efficacy of volunteer operations. We often ask volunteers to knock on doors and deliver messages that increase the likelihood that a voter will turn out to vote. Academic research has repeatedly demonstrated that competently run door knock operations can generate a measurable increase in turnout. At my firm, we have put this theory to the test in live-fire targeted races, and verified this finding. To be specific, we generally observe that you can expect about a 2-4% lift-in-turnout over a control group (a group you did nothing to) from volunteer door knocking operations. To translate that into English, if you knock on 1000 doors and attempt to have good-quality interactions with voters, you will get 20-40 people to turn out who otherwise would not have.

So what do we take away from this? Some who cover campaigns see this as a solution to winning campaigns. They see the simple and obvious prescription for any campaign to execute if it wants to win. They say door knocks need field offices, and so if we look at who has field offices, we know who is running a smart campaign, and who isn’t. The problem is, this is declaring a cookie-cutter solution to winning campaigns that fails to acknowledge a lot of complexity that a campaign must face. In ways, this is no better than if one were to read about positive effects of fish oil on heart health, self-medicate without speaking with doctors, and then never understand that while fish oil is indeed helpful, exercise, diet, and statins are pretty important too.

Let’s consider a hypothetical case study. First, let’s accept as a given you can get a 4% increase in turnout on those you door knock if you can place that door knock within 7 days of the election date. Now let’s start bringing that assumption into a real-world situation a campaign might face. Consider the imaginary case of a contested primary race in a midwest state with a field of 10 candidates. Further, consider that the likely voter universe (those who have some chance of turning out to the election) is 150,000 people. In this hypothetical case, let’s assume the front runner has 30% of the vote. Further, let’s assume you work for a candidate with 20% of the vote. What this means is that you have 30,000 supporters within the total universe of people who will vote. While you know that these 30,000 must exist, you don’t know exactly who they are. So you do lots of phone calls to ask people who they will vote for (in campaign terms, this is known as “voter ID calls”). You make calls for months and attempt to contact everyone to find your 30,000, but at the end of the day you discover that some people just aren’t reachable. This results in you being able to find and confirm 20,000 supporters.

Now, let’s say you run some tests and find out that among your base of supporters, you can get 2% of them to do volunteer work at your campaign offices. This means inside your supporter base of 20,000, you’ve got 400 folks who could work for you as volunteer at some point in the campaign. Let’s also say from past races that you have learned that on average you can get a volunteer to perform a 3-hour shift of door knocking in the 7 day period leading up to an election (some people will do 5 shifts, many people will never show up for a shift in the final 7 day period, but average will be about 1). Let’s say you also run the math on the geography of this electorate living in this very rural state, and surmise that on average you can hit 30 doors with one volunteer hour (including driving to door knock site, walking between houses, taking coffee breaks, etc.).

If you crunch all the numbers, you find out you’ve got 1200 man hours, which can produce 36,000 door knocks. You start to make door-walk lists, only to realize that since this is a very rural state, 2,000 of the 20,000 supporters you have identified are not going to be reachable. So you’ve got 18,000 voters you can actually reach, and the ability to hit 36,000 doors. So you try and hit everyone twice. Now because you hit everyone twice, let’s say you achieve a 5% increase in turnout (instead of the 4% we might expect with only one attempt). For your work, you’ve generated 900 votes (5% times 18,000). Let’s assume that this state actually turns out 142,000 voters in its election. This 900 votes you generated means you got an additional 0.6% (six tenths of a percent) of the vote for your candidate.

To be sure, 0.6% of the vote is nothing to sneeze at, and many elections are won by far less. The problem is you’ve missed the elephant in the room. Your candidate only has 20% of the vote, where the front runner has 30%. You’ve got a 10 point gap you need to close. While getting 0.6% of the vote produced via extra turnout is important, you are not even in the game until you have closed the 10% gap that must be closed via persuasion.

Take another case, this one we’ll base on reality. Let’s say you are Barack Obama’s campaign in 2012. You know the race in a virtually 50/50 country is going to be very tight. Further, let’s say you can recruit 5% of your supporters to be volunteers, and on average they will work 2 shifts in the final 10 days. Further, let’s assume you can ID 80% of your supporters over 9 months leading into the election. Further, let’s assume that you can knock on 60 doors per hour, because your voters don’t live in rural farmlands but in dense cities. Suddenly, the productivity of door knocking changes in a drastic and favorable way, and the produced results are projected to be of a scale that is critical to the strategic problem the campaign is facing (winning a 50/50 election by a hair).

All this is not to say that door knocking doesn’t work. We know it does. But it like every other tactic must only be applied in support of a strategy designed to the realities that each unique campaign must face.