Will These Skills Survive Contact With the Job? What the Transfer-of-Training Research Means for Course Evaluation

Bottom line: Universities increasingly ask course evaluations to speak to employability—did this module build skills graduates will use at work? The transfer-of-training research is sobering here: a skill demonstrably learned in a classroom often fails to appear on the job, especially when the workplace context differs from the learning context ("far transfer"). This means self-reported skill gains on an evaluation form are a weak proxy for employability, and the popular claim that "only 10% of training transfers" is itself a myth you should not repeat. What evaluation can credibly measure are the conditions that make transfer more likely—and that is a more honest, more useful target than a satisfaction score dressed up as a jobs outcome.

The question every programme director is now asked

Accreditation bodies, ministries, and prospective students all want the same thing from higher education: evidence that a degree builds skills that matter after graduation. So course evaluations have quietly acquired a new job. Alongside "was the lecturer clear?" they now ask "did this course develop skills you will use in your career?"—and the answers get rolled up into employability narratives.

The problem is that learning a skill and using it at work are separated by one of the best-studied and least-appreciated gaps in the education and training literature: transfer.

Near transfer, far transfer, and why the difference is everything

Transfer is the degree to which something learned in one setting is applied in another. Researchers distinguish near transfer—applying a skill in a context very similar to where it was learned—from far transfer, applying it in a context that differs in surface features, tools, stakes, and social setting. A university module is, almost by definition, a far-transfer challenge: the seminar room, the assessment, and the supportive lecturer are nothing like the messy, time-pressured, politically complicated workplace where the skill is supposed to reappear.

The comprehensive meta-analysis by Blume, Ford, Baldwin, and Huang (Transfer of Training: A Meta-Analytic Review, Journal of Management, 2010), which synthesised 89 empirical studies, makes the pattern precise. Transfer is reliably predicted by trainee characteristics (cognitive ability, conscientiousness, motivation) and—crucially—by a supportive environment on the other side. They also draw a sharp line between closed skills (one correct way to perform, applied by rule) and open skills (no single right answer, requiring judgement). Open skills—precisely the "critical thinking," "communication," and "problem-solving" that employability rhetoric prizes—transfer less predictably, because there is no fixed procedure to reproduce and the workplace context does most of the shaping.

The uncomfortable implication: a course can teach an open skill superbly, students can genuinely gain it, and it can still fail to show up at work because the destination environment does not support it. That failure is not the course's fault, and it is invisible to any end-of-module survey.

The 10% myth—please stop citing it

There is a widely-repeated statistic that "only 10% of training transfers to the job." It is worth dismantling, because it appears in employability strategy documents that ought to know better. As documented in work reviewed by Work-Learning Research and in Fitzpatrick's and Saks's methodological critiques, the figure traces to a 1982 article by Georgenson that offered "10%" as a rhetorical aside, with no data and no citation. It was never a research finding. Decades of repetition laundered a speculation into a fact, and it is now frequently mis-stated as "10% of training spend transfers," which the original never claimed.

The honest position is that there is no single credible transfer percentage—transfer varies enormously by skill type, learner, and context—and that quoting "10%" is, in Fitzpatrick's phrase, bad science. We raise it here precisely because a thought-leadership piece on employability that invented a tidy statistic would be committing the same sin the evaluation field keeps warning about.

What this means for course evaluation

Three conclusions follow, and they are more demanding than the usual "add an employability question" advice:

1. Self-reported skill gain is not evidence of transfer. "I feel this course improved my teamwork" is a claim about the learning context, made before the far-transfer test has even been attempted. It is a legitimate formative signal, but presenting it as an employability outcome overstates what the data can bear.
2. Evaluate the conditions that predict transfer, not just the feeling of learning. The meta-analytic predictors—authentic practice, application to realistic tasks, opportunities to perform, alignment with workplace demands—are things a course controls and a good evaluation can ask about. "Did you get to apply this to a realistic problem?" is far more diagnostic than "rate your skill gain."
3. Close the loop with destination evidence. The only real test of transfer is downstream: graduate and employer feedback about whether skills actually appeared at work. Course evaluation cannot see that alone; it needs to be linked to tracer and employer data.

But surely asking students about skills is better than nothing?

Yes—and this is the fair counterargument. Self-reported skill development is not worthless. It captures students' perceived preparation, which affects confidence and job-seeking behaviour, and it is cheap to collect at scale. The mistake is not asking; it is over-claiming. When a self-reported gain is reported as an employability result, it quietly promises far-transfer evidence it cannot deliver. The correct framing is modest: evaluation measures the learning-side conditions for transfer and students' perceptions of them; the workplace supplies the verdict. Stated that way, the student voice is valuable and honest at once.

The employability claim, stated honestly

Put the pieces together and a defensible position emerges. A course can raise a student's capability and their confidence; it can create authentic opportunities to practise; it can align tasks with the demands of real work. Those are genuine achievements, and a good evaluation should capture them. What a course cannot do is guarantee transfer, because transfer is co-produced by an environment the university does not control—the manager who does or does not let a graduate use the skill, the team that does or does not value it, the job that does or does not resemble the training. This is why employability is a shared outcome, and why any single actor claiming sole credit—or accepting sole blame—is misreading the evidence. The implication for evaluation is liberating rather than deflating: stop asking a course to prove something only the labour market can confirm, and start asking it to demonstrate the things that reliably raise the odds of transfer. A programme that can show authentic practice, realistic application, and alignment with graduate roles has made the strongest honest employability case available to it—and one that will survive scrutiny from an accreditor who knows the transfer literature.

Where Koji fits

Koji for Education is designed to collect the transfer-relevant signal that a Likert skill-rating throws away. Its AI-moderated conversational interview does not stop at "rate your problem-solving 1–5"; it probes for the conditions that predict transfer—"Where did you get to apply this to a realistic task? What would have made it more like the work you expect to do?"—turning a vague self-rating into concrete evidence about authentic practice and alignment.

Its six structured question types let a programme measure perceived skill gain and the presence of application opportunities as distinct items, so the two are never conflated. Automatic thematic analysis reads open text for whether students describe genuine, contextualised application or only classroom exercises—the very near-versus-far distinction the transfer literature turns on. And because Koji reports at programme and institution level, it can be linked to graduate tracer and employer feedback loops, so the learning-side signal meets the destination evidence that alone can confirm transfer. Koji's closing-the-loop action tracking then records what a programme changed in response.

Universities that run graduate and employer research more broadly use the same AI interview engine on the main Koji platform; the education product focuses it on the course-to-career question.

Transfer is where employability claims live or die. A course evaluation that promises to measure it with a single skill-gain score is promising something the science says it cannot deliver. Measure the conditions for transfer honestly, link them to what actually happens at work, and you get an employability story that survives contact with the job.

Want evaluations that measure the conditions for transfer, not just satisfaction? Explore Koji for Education.