Your retention data tells you what already happened. Persistence data tells you what is about to.
Albert Bandura's concept of self-efficacy — a person's belief in their own ability to succeed at a specific task — is one of the most replicated findings in educational psychology. It predicts academic persistence better than GPA, better than SAT scores, and better than socioeconomic status.
Here is the problem: self-efficacy is invisible in your data system. You cannot query it. You cannot pull a report. It lives in the beliefs students carry into every interaction they have with your institution — and it is either being reinforced or eroded by those interactions, whether or not anyone is paying attention.
The students most at risk are not the ones who do not try. They are the ones who tried, received ambiguous feedback, and quietly concluded that they were not capable of succeeding here. That conclusion typically forms 6 to 10 weeks before any measurable outcome appears.
"Retention is not a crisis response system. It is a climate. The question is not what you do when a student is failing — it is what your institution communicates to every student, every day, about whether they belong here."
The institutions that retain the most students are not the ones with the most sophisticated data systems. They are the ones that have built cultures where advisors have time to notice, capacity to act, and frameworks to guide their response...
This Content is for Members
Subscribe to The Praxis Brief for full access to every issue, the complete archive, practitioner toolkits, and monthly deep-dive PDFs.
The 5 Behavioral Signals That Predict Departure 8 Weeks Out
These signals are not in your early alert system. They require advisor attention and a culture that treats advisor conversations as data.
- 1Reduced Advisor CommunicationA student who was responsive to email stops replying — not because they are busy, but because they have begun to disengage from institutional relationships. Response latency increasing by 72 or more hours is a signal.
- 2Withdrawal from Peer Learning CommunitiesStudy group attendance drops. Collaborative project contributions become minimal. Peer withdrawal predicts attrition with 74% accuracy in students who were previously engaged.
- 3Decreased Use of Support ResourcesThe student who came to tutoring every week stops. They do not tell anyone. They have concluded — often incorrectly — that the support is not working.
- 4Increased Surface ComplianceWork gets submitted but lacks depth or effort. Assignments are technically complete. The student is present but no longer invested. This is different from academic struggle — it is academic withdrawal.
- 5Changes in Self-Referential LanguageIn advising conversations, the student shifts from "when I graduate" to "if I graduate" or stops referencing future plans entirely. This linguistic shift is among the most reliable behavioral predictors available and is almost never systematically captured.
Building a Persistence Index: What to Measure and How
A Persistence Index is a composite score built from leading indicators, not lagging outcomes. Here is a practical framework for institutions without a dedicated research team:
- 1Academic Engagement ScoreCombine LMS login frequency, assignment submission timeliness, and time-on-task data available from your existing SIS. Weight timeliness more heavily than completion — consistent late submission is more predictive than occasional non-submission.
- 2Relational Contact IndexCount the number of distinct institutional contacts initiated by or with the student in the past 30 days. A student with zero contacts in 30 days is three times more likely to depart than one with three or more.
- 3Self-Efficacy PulseA single validated question — "How confident are you that you will succeed in your program this term?" on a 1 to 10 scale — administered monthly. Trend direction matters more than absolute score.
- 4Resource Utilization RateTrack use of financial aid, counseling, tutoring, and food pantry services as a percentage of demonstrated need. Students who are eligible for but not using resources are at higher risk than those using them consistently.
Advisor Persistence Monitoring Checklist
Use during advising conversations at weeks 4, 8, and 12 of the semester. If fewer than 5 of 8 are checked, schedule proactive outreach within 5 business days.
- Student initiated contact with me at least once this month unprompted
- Student can name at least one peer on campus they consider a genuine connection
- Student references future plans within their program
- Student attended at least one non-required campus event or support service
- Student's language about their ability to succeed has remained consistent or improved
- Student has not expressed intent to reduce enrollment, even informally
- Student's assignment submission pattern is consistent with the first three weeks of term
- I could describe this student's non academic life in at least one sentence
Next issue: how AI is being deployed wrong as an intervention tool — and what actually works.