Our Models

At i360, the data — not assumptions — drive our predictive models. Utilizing cutting-edge techniques and hundreds of dynamic variables, i360’s predictive modeling provides unprecedented insight into voter behavior. The superior data and analytics that go into our predictive models empower organizations and campaigns to target audiences with precision and communicate their message most effectively.

National predictive models

There are a number of unknowns in politics – through our National Predictive Models, we try to answer those unknowns. Because we maintain a nationwide database and an in-house team of mathematicians and statisticians that use that data to build predictive models, we are positioned to offer those models to you as part of our standard data package. Our National Predictive Models include:

The i360 Issue Cluster Model - What Is It?

For years, i360 has been building and updating issue models for our clients that predict with great accuracy if people support or oppose various issues – such as lowering taxes, Obamacare, and 2nd Amendment Rights, among others. While these models are essential in predicting opinions and beliefs, they fail to tell us one important fact – how much people actually care about these issues as a whole.

If you give them a chance, the majority of people will say they have an opinion on any issue – one way or the other. But, if they oppose an issue that a candidate supports, is it an issue important enough to actually change their vote? To answer this question, our data science team has been hard at work building a new type of model – the issue cluster model – that predicts which issues voters care about most. And therefore, what issues and stances are likely to persuade them to choose one candidate over another.

How Does It Work?

To begin, we identified eight issue buckets and asked voters a series of question to understand where each issue ranked on their list – first, second, or third. By analyzing the responses and overlaying them with our affinity segments, we were able to determine how the issues were related to each other.

How Can I Put it to Use?

This model gives you the knowledge to build a smarter and more effective messaging strategy than ever before. By using the Issue Preference Model when building your lists, you can maximize your efficiency – increasing the likelihood of reaching those voters with an issue that is important to them by up to 67 percent and simultaneously eliminating waste.

The Issue Preference Model is available now to all users in the i360 Portal. Simply select an issue and overlay one or more model score ranges, i.e. .7+ on i360 Affinity Model, and export your list. Curious how many voters care most about their 2nd amendment rights? Use the tool to run the counts! When developing your messaging strategy, this knowledge becomes paramount.

Curious about what issues people care about most? Check out our tool below to see national counts!

Issue

Education
Healthcare
Fiscal
Guns
Social
Energy
Immigration
National Defense

Voter Age

18 to 44
44+

Voter Gender

Female
Male

Voter Type

Swing Voter
Primary Voter
Other Voter
Mouseover to select an issue

Education Level

Likely College Degree
Unlikely College Degree

Voter Interest

Likely Hunting/Fishing
Unlikely Hunting/Fishing

Census Block

Rural
Suburban
Urban

Income Level

<$40k
$40k to $80k
$80k+


Custom models

When a campaign or organization faces a nuanced issue or other challenge, in the absence of hard data, i360’s team of statisticians and data scientists, as well as our investment in data and enterprise modeling software, enable us to build custom predictive models – the most scientific approach to answering your most important questions.

The i360 Modeling Team will work with you and cater data acquisition efforts to retrieve sample sets for model training and validation. This will enable our team to classify and score voters along the criteria you require. In addition, i360 will provide detailed validation reporting to let you know the exact accuracy of each predictive model, by quartile and decile.