We're putting together information to help you move to virtual advising, deploy nudging, get answers on data science questions, and use your solutions during COVID-19. ACCESS RESOURCES HERE. 

Support is Moving! You can submit a new ticket at support@civitaslearning.com

Open navigation

Persistence Predictions

Persistence is successful retention: the outcome of a student re-enrolling for a future term and staying enrolled past the census date. Persistence predictions are key to finding ways to take action to boost retention.

What is ‘Persistence’?

Understanding what persistence means here is key to interpreting the data presented in Administrative Analytics. Persistence has two parts. To persist, as student must achieve two things:

  1. Have enrolled for a future term at your institution 

  2. Have stayed enrolled in that term past the census date (usually 14 days after the start of term)

Persistence is calculated based on how your institution defines program continuity. Here are three common examples that institutions use:


  • A student currently enrolled in fall persists to the next spring OR

  • A student currently enrolled in spring persists to the next fall

Term to Term

  • A student currently enrolled in fall persists to the next term (winter or spring) OR

  • A student currently enrolled in spring persists to the next term (summer)

Fall to Fall

  • A student currently enrolled in fall persists to the next fall

Persistence Prediction chart

The Persistence Prediction chart shows the expected persistence of two entire groups. It offers a quick comparison of your active filters either with the entire population of active students or with a selected filter set: 

  • Active Filters: The percentage of currently enrolled students meeting the selected filter criteria predicted to enroll in the future term and stay enrolled.

  • Overall Population: The percentage of all active students predicted to enroll in a future specified term at your institution and stay enrolled past the census date, typically 14 days after the start of term.

Tip: Use the drop-down list compare your active filters to one of your saved filter sets.

Prediction Distribution chart

The predictive models for your institution deliver an individual persistence prediction for each of your students (which you can see as a column in Student Lists). View the distribution of these individual scores in the Prediction Distribution pie chart, to the right of the overall persistence predictions.

The Prediction Distribution includes only those students who meet the active filter criteria. When no filters have been applied, the Prediction Distribution represents the full set of active (currently enrolled) students. 

Review the left column to see how many of your active students fall into each persistence prediction bucket: Very Low, Low, Moderate, High, and Very High. Select the help pop-up to see if the buckets for your institution are set up differently from the default ranges, shown here:

All students — Look to the center of the pie chart to see the total number of students as currently filtered. Click this number to open the Student List for all of those students.

Students by bucket — Hover over any prediction bucket to update the pie chart with the number of students in that bucket. To open the Student List for that bucket, click on the bucket name.

Powerful Predictors

Each Powerful Predictor is a predictive factor or behavior for the filtered student group. These variables are specific to your institution, based on which indicators are proven most helpful in predicting persistence for your unique student population. They appear in order of predictive power, and you can select them to open the specific predictors (data columns) that make up the Powerful Predictor: