Highly Accurate and Cost-Effective
The ACT Predictive Modeling Service offers accurate college enrollment predictions at a fraction of the cost of other services. Because it is based on student-level information, ACT predictive modeling data are more accurate and less expensive than predictions based primarily on geodemographic data.
The ACT Predictive Modeling Service uses a statistical analysis based on student-level variables, including:
- Student characteristics and demographics
- Student enrollment preferences
- Student academic ability
- Student choice set (mix of campuses to which students send ACT scores)
Four Separate Models
Through the Predictive Modeling Service, ACT builds four different custom predictive models:
- A model for ACT-tested score senders to the institution
- An inquiry pool model that allows an institution's entire inquiry pool to be scored "on the fly"
- A model for names purchased through the ACT Educational Opportunity Service (EOS)
- A model that predicts first- to second-year retention
