If you followed the Dog's advice from last week's diary, you have some notion of what turnout is like in your race, and you have calculated your win number. The next question that needs to be answered is "which voters do I contact to find my supporters?" Like all good Kossacks, you know that not everyone is registered to vote, and not all registered voters show up at the polls. You need to determine who are the likely voters on election day. Defining your target audience is an arcane skill, blending art, science, and more than a touch of the dark arts. Choosing the right voters can make or break your entire campaign.
Determining likely voters requires access to the voter lists for your district. This is because you need to know the voting history of your constituency. In the ideal world, your local registrar of voters or county clerk has included voting history on their computerized voter rolls and can be persuaded to give you the information you need, or you know somebody who owns the voter list and can run the numbers for you. In the less than ideal world, you will have to purchase the data and run the numbers yourself. Hopefully, you won't have to manually add voter history for each name on the roster. For the rest of this lesson, I am assuming you have a database that includes voter history for recent elections. (Definition: by "voter history", I mean whether or not an individual voter cast a ballot in a given election. My home county tracks back for the last 20 elections; yours may vary.)
The best indicator of whether someone will vote in your election is they have voted in similar elections in the past. Typically, you will draw from the last three to five similar elections in analyzing voter history. What counts as a similar or comparison election depends on your election. If you are running for local office in an off-year (no state or Federal races), you will use recent local elections and possibly low-turnout state primary elections as comparisons. If, on the other hand, you are running for state legislature in a primary, you will use recent primary elections as your guide, possibly adding the 2006 general election into the mix.
In most cases, a likely voter is somebody who has voted in a majority (three of five, two of three, three of four) of comparison elections. It is typical to add new voters to your pool as well: people who have registered since the last comparison election. Fine-tuning these criteria is as much art as science, and might also reflect your budget constraints. A more rigorous criteria (such as three of four or even four of five) means fewer voters to contact, which is cheaper and increases the odds of reaching definite voters, but allows less room for getting to the win number. The final number of likely voters should roughly approximate expected turnout, but may vary by 10-15% from that figure.
In a nonpartisan local office, you may choose to stop refining your target voter pool at this point and communicate to every likely voter. However, in most elections you will want to focus your efforts even more. Obviously, in a primary election, you will target only the Democratic voters, while in a general election you would normally include independents and minor liberal parties such as the Greens. If you live in a GOP-leaning district, you may need to persuade Republicans to cross over in order to win. At this point, I advise aiming at segments of the Republican electorate, rather than trying to win them all. If your campaign emphasizes education, perhaps target married Republicans between 25 and 40. If veterans issues are high on your list, target men over 60. Obviously, this requires the input of experienced campaigners (and perhaps competent pollsters) in your area who can tell you which messages have the best crossover appeal and which Republicans are most likely to switch.
In some situations, you may further refine your targeting by dividing the electorate into regular or always voters, occasional voters, and marginal or infrequent voters. If you are using five elections as your basis of comparison, always voters will have voted 5 out of 5 elections, while regular usually refers to those who voted 4 or 5 times. Occasional voters participated in 3 of the last 5 elections, while the marginal or infrequent voters showed up only once or twice. However, it is just as common to classify voters by number alone, so you may hear a campaign manager say something like, "In Lincoln County, we're targeting 2s and 3s among Democrats, and 3 through 5 among independents." Determining which subgroup of voters you will target depends largely on the circumstances unique to your campaign.
As I said at the beginning, targeting is a skill that is as much art and science, and even within a single election, the targeting for different campaigns will vary. Last November's campaign is a perfect example. In my office, we had four active campaigns operating, each with different objectives and targeted voters, but all four were operating under the (correct) assumption of massive turnout and a landslide for Obama. The Assembly race was a certain win; we wanted to boost our own numbers purely for internal Assembly politics. Our own targeting aimed at marginal Democrats and occasional voting independents -- people who would show up for Obama but don't pay attention to politics generally. Our companion state Senate candidate was leading in a competitive race. Their campaign assumed that regular voting Democrats were with them, so they focused on occasional voters among Democrats as well as both occasional and regular voting independents. Our county supervisor candidate chose to run as a Democrat in a nominally nonpartisan race; he bypassed Republicans and focused his campaign on regular & occasional Democrats and independents. Finally, one of our city council candidates chose the nonpartisan approach, which meant he was reaching out to regular voting Republicans as well as Dems and independents. In practice, since we were coordinating field operations while running separate mail programs, the field operation ran on the county candidate's list in his district, the Senate race outside, and the city council candidate set up a separate precinct program aimed at Republican households that ran parallel to the coordinated effort.
In comparison, the primary election for the Assembly race was projected to have very low turnout. This was the first year California split the Presidential and other primaries, so we anticipated (correctly) turnout for the Assembly primary would be less than half that for the Presidential. Our targeting in that race went to the other extreme: 5-of-5 Democrats and independents only. (In California, Democrats allow independent voters to vote in their primary.) In both elections, resources were targeted where they would do the most good, not wasted on non-voters (i.e., occasional voters in the primary) or guaranteed supporters (regular voting Democrats in the general).
Choosing the right voter will make your campaign much more effective. Instead of talking to 100,000 registered voters, or 50,000 Democrats, you will focus your message on the 17,000 or so who will be voting and need persuading to choose your candidate. For a few hours of analysis, you've cut your workload by 5/6 -- not a bad deal!