Campaigns employ two primary methods to target voters with persuasion and get-out-the-vote (GOTV) messaging. Campaigns use microtargeting to deliver specialized messages to voters via various communication methods. Canvassing is an old-school version of microtargeting to speak directly with voters on their door steps.
In this essay I briefly describe these two approaches for campaigns to communicate with prospective voters. I then step through the basic process of how to devise a canvassing program for a campaign.
A modern method of targeting voters is known as microtargeting. Microtargeting is beyond the scope of this assignment, but it is worthwhile to understand what it is.
Microtargeting uses information about individuals from sources such as voter registration files or from companies to tailor personalized messages to individuals. In the modern Information Age, companies regularly target advertising to individuals based on what they buy, what they post on social media, and what they browse on the internet. Campaigns similarly leverage individuals’ information found on voter registration files to target individuals with campaign messages. Voter registration information includes the past elections a registrant participated in, age, gender, and sometimes information such as party registration and race. Voter registration records also contain names and addresses, which allows campaign consultants to cross reference voter registration records with the same information as companies use to sell products to consumers. Such databases are powerful tools to empower campaigns to develop specific voter profiles, such as “soccer moms,” i.e., married women with grade-school children.
Campaigns analyze surveys to deliver targeted messages to prospective voters within various profiles. Soccer moms might receive messages about K12 education policies, for example. In my previous life as a campaign consultant, a campaign used my microtargeting software to identify single Republican males over age 65 – what the campaign dubbed the “barfly target” – to send a message opposing a ban on smoking in bars.
Campaigns may deliver these messages via campaign mailers, phone calls, in-person canvassing of neighborhoods, email, and various online platforms including social media and streaming services. We as a society have grown so use to companies’ and campaigns’ ubiquitous microtargeting that we are barely aware of the extent narrow messages are being delivered to encourage us to buy everything from toothpaste to a candidate.
In the days before large-scale databases that enable microtargeting, campaigns would – and still do – send volunteers and the candidate to canvass neighborhoods. In the days of political machines, the ward boss was known to manage these operations. Today, campaigns have a staffing position known as the Field Coordinator that fills a similar role.
Canvassing is knocking on doors to speak directly with voters. Canvassing is still widely used among candidates for local offices that do not have the resources to contact voters through media. Campaigns often provide canvassers with lists of voters gleaned from voter registration files to speak with (a blend of canvassing and microtargeting). However, with the candidate’s and volunteers’ time being precious resources, campaigns need to make strategic decisions on where to focus their canvassing activities. Canvassing focuses door-to-door activities in specific neighborhoods where persuadable voters may be found to explain why they should support the candidate and where strong supporters of the candidate need to be encouraged to vote.
Why do campaigns canvass? Former Democratic House Speaker Tip O’Neill is famous for his saying “all politics is local.” He told a story of talking to a friend and neighbor who he was surprised to learn that had never voted for him. When asked why, the woman replied, “because you never asked me.” There’s truth to this anecdote. Subsequent research by campaigns and academics finds that a person is more likely to support and vote for a candidate if they have a favorable interaction with a trusted source: a family member, friend, neighbor, or even the candidate. Canvassing is thus about making person-to-person contacts to persuade people to vote for a candidate, and to encourage them to carry through with the act of voting once they are identified as a candidate’s supporter.
There are two types of canvassing activities, persuasion and get-out-the-vote, also known by the acronym GOTV. Persuasion targeting generally occurs in the early portion of the election as the campaign attempts to identify enough people as supporters of the candidate so that they may win. GOTV occurs once voting begins and encourages the identified supporters to vote for the candidate. As early voting has grown, persuasion and GOTV canvassing activities have become more inter-mixed. Still, a campaign will rarely put little effort into persuasion when Election Day arrives, favoring more GOTV activities instead.
The Field Coordinator for the campaign is often charged with developing a canvassing program. For small campaigns, staffers may wear many hats such that the field coordinator is also the campaign manager or even the candidate. The candidate is the campaign’s most valuable resource, so it is a waste to have a candidate manage the daily operations of a campaign, if that can be avoided.
There are three primary components to a canvassing program:
These three steps are discussed in turn.
The first step a canvassing plan – indeed, a campaign plan – is to determine is how many votes are required to win the election. All campaign resources are dedicated to this goal.
The campaign cannot know how many votes will be cast before an election is conducted, so an educated guess is in order.
A key source of information to make a good guess is turnout in past comparable elections. The key word is comparable, not just any election. It is well-known that turnout varies in federal elections, with higher turnout in presidential elections and lower turnout in midterm elections. Primary and local elections almost always experience even lower turnout than midterm elections.
To identify which past elections are most comparable to the one your canvassing plan is designed for, consider these questions:
For example, consider these scenarios:
If the office is for a lower office, such as US Congress or state legislature, there is a phenomenon that occurs called “roll-off” whereby voters who cast votes for the top of the ticket office fail to vote – or roll-off – in lower offices. For lower ballot races, it is often useful to account for the roll-off dynamic by computing turnout by examining past election results for the particular office the candidate is running for. There is a potential difficulty in that the particular office may not have been particularly competitive or even contested in the prior elections. Competition and contestedness are well-known to affect turnout. For example, if the last election for a state House had only a popular incumbent for one major party running, it may not be a good comparable election to one where the incumbent retired and two major party candidates are running for the seat.
The next question to consider:
The past two most recent elections for the office the candidate is running for are typically sufficient to estimate turnout in the upcoming election. There can be judgment call when the these past two elections are not entirely comparable. A Field Coordinator may consider elections from other years, in other states, or in other districts. Indeed, most campiagns analyze more than two elections, but for the purposes of this project two elections are sufficient.
Once these three questions have been addressed, it is time to get data to estimate turnout for the upcoming election.
There are two important pieces of information to collect to forecast turnout.
Typically, this information is available from state election offices. These election offices are typically run by Secretaries of State, State Election Boards, or in the case of Utah, the Lt. Governor.
A simple internet search for “<insert state here> election results” should lead to the proper state election website.
Once on the site, there will often be a link or tab for election results.
To illustrate, let’s design a canvassing plan for the 2024 Montana US Senate election. The Montana Secretary of State’s website is found here.
The 2024 US Senate election will be held during a presidential election due to the six-years between Senate elections. For forecasting turnout, the best past two comparable elections will be the 2020 and 2016 presidential elections.
I searched within the Montana’s Secretary of State’s website for the election results archive, here. In this case, the results are reported in a document called the “canvass” which is the a report of the final certified election results. Canvass reports are often great sources of information for developing a canvassing program since they will often not only report election results, but also voter registration statistics. Such is the case for Montana. This may not always be the case. Election results and voter registration reports may sometimes be found on different parts of an election website. Unfortunately, there is no standard format for election websites, so some searching may be necessary.
The average of these two voter registration turnout rates will serve as the turnout rate forecast: (81.33 + 74.44)/2 = 77.89%.
Montana’s population is growing, and along with it the number of people registered to vote. The final step is to calculate the to number of votes needed to win. To do so, we apply this forecast turnout rate of 77.89% to the current voter registration total. Montana has a report for the current voter registration total here. As of February 2024, there are 744,265 registered voters. surprisingly, this is lower than the 752,538 in the 2020 presidential election – some registered voters have been removed as a part of election officials’ usual voter registration list cleaning procedures. For the purposes of this project, please use the most recent voter registration count. (One can project a voter registration estimate forward to the 2024 general election, if necessary because current voter registration totals are unavailable.)
Multiplying 744,265 by a turnout rate of 77.89% yields an expected 579,708 votes (rounded to the nearest whole number). As a check on the work, note that this turnout estimate is higher than the 2016 total number of votes and less than the 2020 total number of votes.
We’re almost there! The goal of the campaign is to win 50% + 1 of these votes. 579,708/2 + 1 = 289,854 votes are needed to win.
The goal of Montana’s 2024 US Senate campaigns is to win 287,545 votes.
The next step of the canvassing plan to is devise a strategy to get 287,545 votes for your candidate. It doesn’t matter for the persuasion target if you write a memo from either perspective of the two major political party candidates. The persuasion target will be the same for both candidates! What will be different is the GOTV target, discussed in the next section.
The first step in identifying a persuasion target is to determine where voters change their votes for candidates most often from one election to the next. For the purposes of this project, the geographies we will concentrate on are counties within a state. Why? Election results for counties are often the most easy to obtain. For district and local campaigns, the field coordinator will usually rely on data for precincts. These data are available but are often more difficult to work with.
Senate elections are a little tricky because of the six-years between elections. Usually the best elections to analyze are elections to the same office the campaign is running for. In this case, there was an election for the other Montana US Senate seat held in the 2020 general election and the current incumbent, Senator Tester, ran in the 2018 general election. These two elections will serve for our analysis. For elections on a more regular calendar, such as a presidential or governor’s election, elections for these offices are held every four years and identifying good candidates for comparable elections is more straightforward.
The Montana canvass reports includes election results in addition to the voter registration totals for each election. With a little data wrangling, I created a spreadsheet of the 2020 and 2018 uS Senate election results by county using a spreadsheet program (Excel).
For the purposes of this project, I analyze candidates for the two major political parties. I ignore minor parties, such as Libertarian Rick Breckenridge who ran for Montana’s US Senate seat in 2018.
For the purposes of showing the logic behind identifying the counties where the most persuadable voters are located, I’ve created the following table.
There are two key columns of data generated from the election results (from the Democratic perspective, although for the persuasion target the two campaigns will have the same target counties):
These columns are calculated again for the 2018 election.
Dem Vote Difference is the key column that identifies the potential persuasion target counties where the field coordinator will send volunteers to canvass neighborhoods with messages to persuade voters to support their candidate. This measure represents how much a county swung between candidates of the two parties between the two elections. These are the persuadable voters because they were willing to change their vote! Where the swing is the greatest is where the most persuadable voters may be found.
Sort the counties by Dem Vote Difference from the highest value to the lowest value. The top ten counties are where there was the most change in the vote shares between the two elections.
Since we only analyze the two major parties, this method works regardless if you calculate for Democrats or Republicans. (Why?) However, if you’re doing these calculations for a Republican campaign, you would be wise to persent your statistics as if they were computed for the Republican candidates.
Identifying these top ten persuadable counties is a good start, but there is something more to consider: how many voters are there in each county? In this example, Treasure County had only 464 votes in the 2020 US Senate election. While the county ranks highest in persuasion, there are few voters in the county to persuade. Sorting the counties by the total vote and ruling out smaller counties is a way to refine your persuasion target.
The GOTV target is where late in the election the campaign deploy volunteers to encourage supporters to vote for their candidate. The focus for a campaign’s GOTV activities are counties that have high concentrations of supporters for the candidate. The GOTV target is the counties that have the highest vote shares for the candidate’s party.
The key column to calculate is:
Sort the counties from largest to smallest, and the top ten counties are the GOTV target. As with the persuasion target, you may wish to exclude a county if there are relatively few voters in it.
Write a memo for your assigned campaign. The memo should have the following elements:
Along with the memo, submit a spreadsheet with your calculations.
Consider this a work product for a campaign manager. Indeed, students have shown this assignment to prospective campaign managers who hired them to be field coordinators and one student even went on to be a congressional staffer when their candidate won! As such, write a professional memo. Assume the campaign manager does not know anything about your methods. Do not simply produce a memo with some calculated numbers and lists of counties. Provide some words and context to help the campaign manager understand what it is that you are doing.
The memo should be about three to four pages, double-spaced. You only need to list the counties in your persuasion and GOTV targets. Data for all counties will be provided in the accompanying spreadsheet.
The project is intended to illustrate the mechanics of how to create a canvassing program. It is not what campaign professionals actually do. They typically rely on a blend on individual level voter registration data, survey data, and election results within precincts to fine-tune which homes to send volunteers to. They may engage in voter registration drives and target places where there are high numbers of unregistered voters. They further consider housing density. The goal is to talk to as many of the persuasion or GOTV target as possible. It is inefficient to knock on doors in rural areas where homes are miles apart.