Ideas Les's Explored

Why Test Scores Shouldn’t Define Teacher Performance

Rethinking how we evaluate educators in a data-driven age

by Les Huysmans

Test Scores Miss the Full Picture of Student Learning

In many schools today — from national systems to international contexts — student test scores are increasingly used to evaluate teacher performance. On the surface, it seems logical: if students perform well, their teachers must be doing something right.

But reducing teacher evaluations to standardised test results misunderstands what great teaching looks like — and risks doing long-term damage to both student learning and staff morale.

This isn’t about resisting accountability. It’s about choosing an approach that reflects the real complexity of teaching and the long-term goals of education.


Student Outcomes Depend on More Than Teaching

Test scores provide a snapshot of certain academic outcomes, but they tell us very little about context — the real-world factors that shape student achievement.

A teacher may be working with:

Test-based teacher evaluation often overlooks these complexities. Judging performance without considering context isn’t accountability — it’s distortion.


While standardised testing has been part of education for decades, recent years have seen a shift. With the rise of data-driven school improvement models and pressure from global benchmarking tools like PISA, some schools have begun tying individual teacher performance directly to test outcomes.

The intention is often good: to raise standards and improve transparency. But when data becomes the dominant lens, the human side of teaching — nuance, relationship-building, creativity — gets pushed out of view.

And research supports this: studies show that data is most effective in education when paired with human insight and interaction. For example:


Why Overemphasis on Testing Hurts Real Learning

When teacher evaluations are tied to test scores, it can alter classroom priorities in harmful ways:

Students may leave school with test-taking strategies — but without the curiosity, confidence, or critical thinking that real life demands.

A recent literature review of over 100 papers found that learning analytics and AI work best when students and educators are involved in their design — not just used as data points. Without human-centred systems, even advanced tools risk reinforcing shallow learning2.


Measuring Impact Fairly Across Different Classrooms

Some of the most skilled teachers are working in classrooms with the most complex needs: newly arrived EAL learners, children with behavioural challenges, students who’ve disengaged from learning.

These teachers may not always produce quick score increases — but they’re often creating transformational, lasting change that data doesn’t reflect.

If test results are the main way we measure impact, we risk overlooking the educators doing the most meaningful work.


Balanced, Effective Teacher Evaluation Methods

To truly support teacher growth and student achievement, evaluations should be multi-dimensional. A more balanced system includes:

This kind of mixed-method approach is being adopted in other sectors too — such as healthcare and manufacturing — where data is combined with human context to improve decision-making[^3][^4].


Final Thoughts: Teaching Is More Than a Number

Data can support reflection. But it should never replace human judgment, professional relationships, or the deep understanding that comes from knowing students as individuals.

If we want schools to prepare students for life — not just exams — we need evaluation systems that reflect that broader mission.

Let’s not flatten teaching to a spreadsheet. Let’s evaluate what actually matters.



If this sparked anything — questions, rants, good old curiosity — come say hi via the About Les page.


If this sparked anything — questions, rants, good old curiosity — come say hi via the About Les page.


References

  1. Ekanayaka, E. M. S., Gamage, K. A. A., & Perera, I. (2022). Reflective writing and learning analytics: A study of student feedback combining human and data-driven input. arXiv:2211.08222

  2. Hummel, H. G. K., et al. (2024). Human-centred learning analytics and AI in education: A systematic literature review. International Journal of Educational Technology in Higher Education. ScienceDirect

#education reform #general education thoughts