Every initial teacher training programme has to assess trainees on technology. Most programmes use TPACK, or a derivative of it, as the framework underneath the assessment. TPACK is a beautiful piece of theory and the standard reference in the field since Mishra and Koehler in 2006. It is also the wrong framework to use as the spine of trainee assessment, because it is not built to be observed.
This piece argues that activity-classification frameworks (PICRAT being the cleanest example) belong in the assessment role TPACK has been bearing for two decades. TPACK should still sit on the reading list. The roles are different, and the difference matters more in ITT than anywhere else.
What TPACK is, and what it is for
TPACK names the three knowledge bases a teacher draws on (Content, Pedagogy, Technology) and the four overlap regions (PCK, TCK, TPK, and the central TPACK). The implicit claim is that good technology integration sits in the centre, where all three knowledge bases combine. The framework is widely cited because it captures a real intuition: a teacher needs all three kinds of knowledge to make any given lesson work, and their interplay matters.
That intuition is correct. The framework is also useful for what it was designed for: programme architecture, course descriptions, and theoretical conversations about what a teacher needs to know. TPACK puts vocabulary on a complex problem and gives course designers a shared map.
The honest read is that TPACK was never designed as an assessment instrument. The original Mishra and Koehler papers describe it as a framework for teacher knowledge, and most empirical work using TPACK measures it via self-report surveys: teachers rate their own confidence in each region. Self-report is a useful research method for many things. Trainee teacher assessment is a place where it falls short.
What ITT assessment actually needs
An assessment framework for trainee teachers has a specific job. It has to let a mentor in the room form a defensible judgement about a trainee's lesson, let a course leader compare across trainees and across mentors, let a moderator audit a sample of judgements without sitting in the lesson, and let the trainee themselves see what they need to work on without it feeling like an opinion.
That job has three properties any framework used for it should have. The properties are: reliability across raters, validity against later classroom behaviour, and transfer into the trainee's own planning. Activity-classification frameworks have all three. Knowledge models struggle on all three. That is a feature of how knowledge models were designed, since they were built to describe what teachers know.
Reliability: activity-coding holds up across raters
If three mentors observe the same lesson, the question "what cell does this lesson live in?" produces high agreement. The cell is determined by what the students did with the technology and what the technology did for the lesson. Both are observable in the room. PICRAT classification by trained raters lands in the same cell most of the time, with most disagreements being one cell adjacent.
Asking the same three mentors "what TPACK region does this lesson demonstrate?" produces lower agreement. TPACK regions describe the trainee's knowledge, which is inferred from the lesson but not visible in it. A lesson that looks like TPK to one rater can look like TCK to another, with no fact of the matter that resolves the dispute. Lower inter-rater reliability means harder moderation and more contested grades.
Validity: behaviour predicts behaviour
The thing an ITT programme is trying to predict is how the trainee will teach in their first year as a qualified teacher. Empirically, the best predictor of next year's classroom behaviour is this year's classroom behaviour. A trainee who is consistently planning IA lessons in their final placement is overwhelmingly likely to plan IA lessons in their NQT or ECT year.
A trainee who self-rates strongly on TPACK in a survey may or may not. Self-rated confidence on any pedagogical instrument has a documented weak relationship with later observed behaviour, partly because trainees genuinely improve at predicting their confidence faster than they improve at the practice itself. Activity placements are observed; observed behaviour predicts observed behaviour.
Transfer: trainees use what they were assessed on
A trainee assessed on TPACK regions during ITT does not, in the wild, sit at home of an evening and ask whether her next lesson is more TCK or TPK. The framework rarely transfers into the trainee's own planning behaviour, because the categories are theoretically rich but practically heavy.
A trainee assessed on PICRAT does sit at home and ask which cell next week's lesson should live in. The two-question placement test is light enough to use during planning, and the categories map to design choices the trainee actually makes (what students will do; what the technology will do). Transfer is the property an ITT assessment framework most needs. TPACK was built for other jobs.
What TPACK should keep doing
None of this is an argument against TPACK. TPACK is a strong framework for the jobs it was designed for. Three of those jobs deserve naming.
The first is theoretical literacy. ITT trainees should know the framework. It is in the literature, it shapes much of the field, and a trainee who reads it once will recognise it on the reading list of any postgraduate course they take later. Naming the three knowledge bases and their overlaps is a useful intellectual move.
The second is course design. A PGCE programme that wants to balance content tutoring, subject pedagogy and digital practice can use TPACK as a curriculum lens. It maps cleanly to the parts of a programme that need investment, and to the gaps that show up when one of the bases is under-served.
The third is teacher development at experienced-practitioner level. An established teacher engaged in CPD can usefully think about which of their TPACK regions is least developed and target it. The framework rewards reflection of a kind that ITT does not yet have time for.
What an ITT programme should add
Two changes, both small.
First, name PICRAT in the assessment instrument. Replace "use TPACK to evaluate the trainee's technology integration" with "place each observed lesson on the PICRAT grid and discuss". The trainee's portfolio carries the placements; the moderation conversation is about the distribution. Each lesson's intent stays in the lesson plan, where it was always going to live.
Second, train mentors on placement. A single 90-minute session is enough for most mentors to reach reasonable inter-rater reliability, because the framework is small. The training transfers immediately into the next post-lesson conversation. This is the single highest-leverage CPD an ITT programme can run for its school-based partners.
The change to the rubric itself is small. One section of one form, replaced with two prompts the mentor can answer from the room.
TPACK stays on the reading list. The grid lands in the lesson plan. The two together describe a trainee's technology integration more honestly than either alone.
Internal frameworks are for thinking. Observable frameworks are for assessing.
What to do tomorrow
If you lead an ITT programme, audit a recent batch of lesson observations. How many used a knowledge model in the assessment? How many used an activity classification? Where mentors disagreed, what was the disagreement about? The answer to that last question usually points at whichever framework was bearing more weight than its design intends.
If you mentor trainees, place the next observed lesson on the grid before writing the report. The placement is the spine of the report. Add the smallest move and the conversation has a structure the trainee can take into next week.
If you want a faster start on the placement habit, drop a sample lesson plan into Coach. The tool reads the lesson, places it on the grid, and walks through the smallest move. The output is the start of the conversation a mentor and trainee should be having anyway.
Andy Perryer is the head of digital learning at a group of international schools and the creator of PICRAT Suite. The PICRAT framework was developed by Royce Kimmons, Charles Graham and Richard West in their 2020 paper in the CITE Journal. The TPACK framework was developed by Punya Mishra and Matthew Koehler, with the foundational paper published in 2006.