The recent advances in predictive modeling have now made it possible to understand how teacher candidates move through the licensure testing experience, where they struggle, and where they persist.  

Using over a decade of data from Praxis Elementary Education: Multiple Subjects (5001) assessments, ETS researchers Caitlin Tenison, Geoffrey Phelps, and Teresa Ober applied survival analysis, joint longitudinal modeling, and mixture cure models to estimate: 

• the likelihood that a candidate will reengage after an unsuccessful attempt; 
• how quickly they’re likely to return; and 
• the probability and timing of eventual passage. 

Unlike static pass rates, which offer a single snapshot at the group level, these models provide individualized projections that evolve over time with each testing attempt. For state agencies, educator preparation programs (EPPs), and candidates themselves, this shift from static averages to dynamic, person-level insights creates the potential for more personalized, transparent, and fair licensure pathways without lowering standards. 

What the data show 

Testing and retake patterns vary widely across candidates, revealing important insights into how they move through the licensure process.  

• Most candidates move through testing quickly: more than half pass all four Praxis 5001 subtests on their first attempt (51%), and another 14% pass on their second. Nearly one in four follows a longer, less certain route, and about 8% disengage without finishing the licensure requirement. 
• Candidates who pass more subtests on their first attempt are more likely to stay engaged. Among those who retake, the ones whose scores are closest to the passing threshold tend to retake sooner. Those further below passing often take longer, using that time for additional preparation or reconsidering whether to continue. 
• The farther a candidate is from the passing score, the greater the likelihood that they walk away.  

What this means for candidates, EPPs, districts, and policymakers 

Aggregated across cohorts, these predictions offer a clearer picture of what the near‑term teacher supply may look like, where bottlenecks in licensure pathways are emerging, which subject or grade areas may face shortages, and how preparation programs might be strengthened.  

At the candidate level, the models identify where and when support could help shorten time to completion. Personalized outreach should be aligned to candidate performance or paired to specific testing structures to make progress more attainable. This shifts guidance from a one‑size‑fits‑all approach to support that reflects each person’s progress. 

States and EPPs can use this information to adjust policies and support systems based on expected completion timelines, making the licensure system both rigorous and more responsive to candidates’ needs. 

How Praxis is leveraging this research as part of practical, candidate-centered solutions 

Several Praxis innovations—some already in place and others launching soon—reflect what the models show about how candidates move through the licensure process. 

• Praxis Steps, expected to launch in summer 2026, allows candidates to retake smaller content sections, called Steps instead of the entire test, reflecting research findings that timely retesting is most effective when preparation is focused and success is likely. By breaking retakes into more targeted components, Praxis Steps enables candidates to concentrate on specific content areas, increasing the probability of success on the next attempt while reducing unnecessary delays in the licensure process. This modular approach maintains standards while helping candidates make quicker progress toward completion. 
• Praxis Bridge provides learning‑oriented alternatives for candidates who are near the passing threshold. Research examined in this article shows that in multi‑subtest assessments like Praxis 5001, progress on one subtest can accelerate completion of others. Praxis Bridge is especially valuable in this context because it enables candidates to demonstrate readiness through a blend of learning and assessment, freeing their time to concentrate on areas that require deeper development.  
• Free After Three feature removes fees after multiple retakes, reducing the financial strain that can discourage candidates from continuing.  

Together, these efforts translate the research into practical supports that align with real patterns in candidate behavior. 

A more transparent, supportive future for licensure 

When licensure systems adapt intelligently, more qualified candidates reach the classroom. Predictive modeling strengthens licensure standards by showing where the system can better support the people moving through it, benefiting schools, states, and, most importantly, students. By using data to guide individual candidates and inform state-level decisions, we can maintain rigor while building pathways that are fair, equitable, and aligned with real-world candidate experiences. 

Looking ahead, ETS will continue expanding this research and exploring partnerships that bring predictive insights directly into candidate support systems.  

State and EPP leaders interested in how these data driven approaches could strengthen the teacher pipeline or improve candidate support can connect directly with their Director of Educational Partnerships.