Statistical Modeling in IEA began in 2018 to more holistically understand student success outside of group identities. A wide variety of institutional and external data are used in these models including information on student academic preparedness, student demographics, academic progress at OHIO, socio-economic background, and financial aid information. Beyond student success models, other university phenomena are being modeled to aid in strategic planning. These models are econometric and revolve around resource consumption.
Analytics & Research
Analytics & Research
-
Statistical Modeling
Course Success Modeling
The A&R team have been developing and producing course success models since 2019. From single course logistic models predicting success to multilevel logistic models across all courses predicting grades these models have excellent explained variance, and can be customized for understanding effects at many scales: course, department, college, university. This offers insight into which students are likely to struggle and where.
Use Cases
Measuring correlations • Identifying course outliers • Testing pre-requisite effects • Testing intervention efficacy • Predicting successful course completions • Prescribing interventions
Retention and Student Lifecycle Modeling
These models were initially developed to more accurately forecast retention for the Budget Office. Many policy decisions have been analyzed through this retention modeling, including test optional, tuition increases, financial aid changes, and modality changes. These models have evolved from simple logistic models with Markov chains to multinomial-multilevel models for all students at every campus and career level.
Use Cases
Forecasting revenues • Testing intervention efficacy • Estimating policy effects • Predicting likelihood to retain, transfer or graduate • Identifying potential areas of student challenges • Prescribing Interventions
Course Demand Modeling
More than 60% of our undergraduate enrollments on Athens campus can be forecast a year in advance relying on multilevel Poisson models. These predictions were used by the course planning committee for the first time for Fall and Spring 2025. The recent addition of OGP Plans and template data for students with estimated terms of enrollment has enabled reliable prediction of course demand. Additional campuses and career levels will be incorporated for Fall 2026.
Use Cases
Estimating seats 1-year out • Planning faculty lines • Planning GenEd course offerings • Monitoring transfer credit and College Credit Plus credit impacts
Econometric Modeling for Revenue
Using machine learning approaches combined with traditional stochastic approaches the A&R team has developed forecasting models. These time-series models help estimate µÛÍõ»áËù’s position in the context of economic conditions and the performance of other µÛÍõ»áËù public institutions. These methods can also be applied to university enrollment, revenue and expense to provide long-term forecasts.
Use Cases
Forecasting revenue • Understanding economic events (i.e. changes in inflation, unemployment, etc.) • Estimating Program Opportunities
Intervention Modeling
Partnering with various programs around campus has enabled quasi-experimental research designs to understand the impact various academic and student support programs are having on the students they serve. Helping better understand the value of resource deployment and identifying areas of opportunity.
Use Cases
Estimating Return on Investment • Program marketing • Prescribing interventions