Why do I need a voter model, and which is right for me?
When folks started creating targeted communication, sophisticated voter selects were the beginning of microtargeting. Voter modeling was soon to follow, and the age of advanced microtargeting began. These days, voter models are everywhere. No matter the campaign, you have access to voter modeling, but many folks don’t know what to use or how to use it. We’re here to help!
What is a voter model?
Put simply, a voter model is a sample of voters who are asked a question or a few questions, and the results are scored and applied to a voter file. Typical voter models include turnout models, candidate support models, party models, ballot measure support models, issues models, and advocacy models.
Why do I need a voter model?
Voter models can be beneficial, and you are likely using them already. Many voter models are incorporated into common voter file data. No matter which voter file you are pulling from, there is likely model data that you are using. You need model data because it is impossible to reach enough people to come up with core issues or support conclusions alone.
What are the benefits?
Modeled data can help you focus and prioritize who to talk to. It can also help you remove people who are not your priority. Accessing a ready-made model is fast, and you can find folks who care about your issue without having to have a list of people, which your campaign may not initially have, and it gives you a quick approximation.
A model is not a list or a strategy!
Folks often make the mistake of thinking that a model is a list of voters, but that is not the case. There are also actual lists of voters that are very helpful for targeting, whether action takers or donors, and you can use those to target or build models from. Actual lists are more accurate than models. These take time and patience to build, but in some cases, it may be more efficient to build a list instead of a model, at least at first.
Voter models should be thought of as a tool, but not a strategy. Using models and advanced microtargeting methods and voter data is not a strategy—it is a tool to help drive your strategy. Using models and data effectively can help you win, but just using data alone is not a strategy and is likely not a singular key to victory.
Complete a data audit
At the beginning of a campaign, it is best to do an audit of what data exists on your voter file, including lists and models. Think about your goals and what data you would need for success. Here are some examples of the types of models you may want to access:
-
National models: These can help, especially when you don’t have the resources to build your own, and they are also very helpful when it comes to specific issues. This can be a head start in building a base or engaging a specific universe.
-
State-based models: These are usually better at applying data in specific local demos and usually outperform national models, but it can be expensive to create your own state and local model.
-
Turnout models: Predicting turnout has been a part of voter modeling for a long time, but not all turnout is the same. Presidential turnout versus primary turnout versus local election turnout can vary greatly, so make sure you understand what you’re using the scores for and that the scores have been built for what you’re using it for.
-
Candidate support models: These have been built for presidential and senate races, but they are now being built for congressional, municipal, and even smaller races. They model who is likely to support a candidate or oppose one.
-
Party models: As politics becomes more fractured, folks have used party models to help define support for a specific political party, but not every model is the same, and some are old and have not been updated.
-
Ballot measure support models: They are unique in the electoral landscape, as they cross party lines and make people declare a specific decision on a custom issue in their area. Not every region has ballot measures, but those that do allow this direct democracy and allow direct change and approval on the ballot. Models have done a good job of defining predictive support because that language is so specific. But just like with other models, every model is not created the same, and how the models are created can lead to varying results. Are a great place for a custom model. A local measure will not likely have the resources for a custom model unless it is a big city or county.
-
Issue models: These are a great tool for garnering engagement. They are helpful when you can’t pay for a custom model, especially when you are working on smaller local ballot measures, candidate campaigns, or advocacy.
What are the best use cases for voter models?
-
For candidate campaigns: Whether it is about turnout, score issues, models, or candidate support scores, there are lots of models that can be helpful in a candidate campaign. If you have a specific issue or focus, if you have an incumbent you’re running against, or if you are the incumbent, all these models can really help. They can provide a head start in finding people who will likely support you, may be against your opponent, and/or people who share an issue with you. All these things can be a quick way to organize, but it doesn’t necessarily mean these folks are locked in as voters yet.
-
For advocacy: Issue advocacy matches up quite well with issue support models. They can help you prioritize folks who may be great advocates on a specific issue and deprioritize folks who will not be.
Who makes models?
There are great companies that make models and do a good job of documentation. Please know this list is not complete. Some freelance modelers build models directly as well.
What is the cost of building a model?
The cost of a custom model can range, but is usually between $25,000 and $50,000.
What is the cost to use a model?
Sometimes the model cost is included in your access to the file. Sometimes there is an additional upcharge to use specific models.
What are the drawbacks?
Models are not perfect, and there are many potential drawbacks to using models. Understanding how models work and what their limitations are can make you a better user of models.
-
Unscored voters: Not every model can score 100% of a file, and details are often buried in the model notes about likely unscored voters and who the model will not be able to be applied to.
-
Technical disparities: Models are created by different companies, and therefore, the different specs of each model are going to be different. Since the technical details are different, the way they score data will also be different. There can be a vast difference between different models depending on who created them and how well they perform.
-
Model notes make a big difference in understanding what the model can and cannot do. If you’re using a model, take the time to read the notes, so you really understand how it was made and what the goal of the model is.
-
Look for transparency. Some models have a lot of transparency when it comes to notes and how they’re created, but others do not, so making sure you understand the documentation and the models you’re using is key to making sure the models will perform in the way they are intended.
-
Models take time to build. Building a model from scratch takes time, and you must make sure you’re asking the right questions about your goal for the model. Be sure you understand the right questions of voters, and then make sure that it is created properly, so that it will match a voter file. This is a process that usually takes a couple of months.
Have questions or want to learn more about voter models? Click below to get in touch or check out the rest of our blog for related content.