Evaluating Physiotherapy and Allied Health Programmes: Why a Satisfaction Survey Cannot See a Placement

Bottom line up front: In physiotherapy and the allied health professions, the most consequential learning happens on placement, supervised by a practice educator your university does not employ, in a service your quality office does not control. A generic end-of-module student satisfaction survey measures the wrong construct, in the wrong place, at the wrong grain of detail. To evaluate an allied health programme honestly you need evaluation that is placement-aware, practice-educator-aware, and robust to very small cohorts. This is where legacy Likert-form tools fail structurally, not just cosmetically.

The placement is the programme

A UK physiotherapy degree accredited by the Chartered Society of Physiotherapy must include a minimum of 1,000 hours of supervised clinical practice, spread across settings such as musculoskeletal, neurological and cardiorespiratory care (Chartered Society of Physiotherapy). Comparable practice-based-learning requirements run through the professions regulated by the UK Health and Care Professions Council (HCPC) — physiotherapists, occupational therapists, paramedics, diagnostic and therapeutic radiographers, dietitians, speech and language therapists and more — each governed by the HCPC Standards of Education and Training, which require practice-based learning to be integral to the programme and appropriately supervised. Across Europe, networks such as the European Network of Physiotherapy in Higher Education (ENPHE) coordinate exactly because placement models, hour thresholds and supervisor standards vary by country while the pedagogy is fundamentally the same: competence is built at the bedside, not in the lecture theatre.

If a third or more of the programme is placement, and placement is where professional identity, clinical reasoning and safety judgement are actually formed, then an evaluation instrument that asks students to rate "the module" on a 1-to-5 scale is quietly conceding that it cannot see the part that matters most.

Three ways a standard SET misreads allied health

1. It attributes placement variance to the university. When a student rates their experience low, the standard survey routes that signal to the module or the academic lead. But the dominant source of variance is often the individual practice educator and the clinical service — caseload, staffing, whether the educator had time to teach. Attributing supervisor variability to academic staff is a category error that produces unfair comparisons and unactionable dashboards. It is the same attribution problem we describe for team-taught and multi-instructor courses, amplified because the "instructor" here is off-site and not on your payroll.

2. It drowns in small numbers. Allied health cohorts, and especially individual placement sites, are small. A ward might take two students a year. As we set out in the statistics of small-class evaluation, a mean built from three or four responses carries enormous uncertainty, and a single unhappy respondent can swing a placement from "green" to "red" on a traffic-light dashboard. Ranking placements on raw means is close to reading tea leaves.

3. It measures reaction, not readiness. Kirkpatrick Level 1 — did you enjoy it — is weakly related to whether a student can now safely mobilise a post-operative patient or take a defensible radiograph. Satisfaction and competence are different constructs, and in professional education the gap between them is the whole point. A programme can be popular and still send students onto their first rotation under-prepared; it can be demanding, uncomfortable and superbly effective. Averaging enjoyment tells you almost nothing about either.

"But we already survey our placements"

This is the strongest objection, and it is fair: most allied health programmes do run a placement evaluation, often a bespoke form covering induction, supervision, learning opportunities and safety. The problem is not that placement feedback does not exist — it is that it is thin, retrospective and mute.

It is thin because a fixed form asks the same closed questions of every setting, so a stroke rehabilitation ward and a paediatric respiratory team are scored on identical items that fit neither well. It is retrospective because it arrives weeks after the placement ends, when a strong student has already normalised a poor experience — the "mum effect" of reluctance to report negatively is powerful when a struggling student fears their assessor holds their pass in their hands. And it is mute because the richest signal — the free-text box — is read by one overworked placement coordinator, if at all, and never aggregated into a pattern across sites. A recurring safety concern voiced by six students across six different wards will sit in six separate comment boxes and never be seen as a system.

What better evaluation actually looks like

Evaluating allied health well means changing when, what and how you collect.

• When: move part of the collection into the placement, not after it. Formative, mid-placement check-ins let a programme intervene while a student is still on the ward — the difference between rescuing a placement and writing it up in a post-mortem.
• What: ask about the mechanisms of practice learning — supervision quality, feedback, graded exposure to complexity, sense of safety to ask questions — rather than global satisfaction. Separate the practice educator, the service and the university so each signal reaches whoever can act.
• How: treat open text as primary data, not decoration, and analyse it systematically across sites so patterns and safeguarding concerns surface as themes rather than isolated anecdotes.

Who evaluates, and when

Two design choices quietly determine whether allied health evaluation is trustworthy. The first is who is asked. Placement evaluation suffers acute non-response and selection effects: the students who had the hardest placements — the ones whose signal matters most — are often the least likely to complete a retrospective form, whether from exhaustion, disengagement, or fear that criticism will follow them into a small professional community where everyone knows everyone. A programme that hears only from its comfortable placements is flattering itself. The second is when. Feedback gathered after a final assessment, once grades and pass decisions are settled, is safer for students to give honestly but too late to help the cohort that gave it; feedback gathered mid-placement can change an outcome but raises real fears about anonymity and reprisal from an assessor who holds the student’s progression. There is no perfect answer, but the worst answer is the common one: a single, late, low-response survey that manages to be both unrepresentative and unactionable. Designing evaluation that collects at more than one point, protects anonymity credibly, and reaches the students who struggled is not a nicety in allied health — it is the difference between an evidence base you can defend to a regulator and one that merely decorates a committee paper.

How Koji fits

This is the gap Koji for Education is built for. Instead of a static form, Koji runs an AI-moderated conversational interview that adapts to what a student says — probing "you mentioned you did not feel confident escalating concerns; can you say more about that?" rather than leaving a Likert item to stand in for a paragraph. Its automatic thematic analysis reads every open-text response across every placement and surfaces recurring issues — a supervisor consistently unavailable, a site with genuine safety concerns — as aggregated themes rather than comments lost in a spreadsheet. Because moderation is standardised and bias-aware, a shy student on a high-stakes final placement is prompted as fairly as a confident one, mitigating (never eliminating) the reluctance to speak plainly. Formative, mid-cycle collection means problems are caught while the placement is live, and programme- and institution-level reporting lets a head of division see across dozens of sites at once. For very small placement cohorts, conversational depth compensates for what statistics cannot deliver: four rich interviews tell you more than four numbers ever will.

None of this "eliminates bias" or replaces the professional judgement of practice educators and academic assessors — it surfaces better evidence for those humans to act on, and it closes the loop by tracking what the programme changed in response. Many teaching-and-learning teams also run wider student and staff research on the shared AI interview engine behind koji.so.

If you evaluate physiotherapy, occupational therapy, paramedic, radiography or any practice-based health programme and your evidence stops at the classroom door, see how Koji for Education evaluates the placement itself — where your graduates are actually made.