Adult education and andragogy: the evidence for six assumptions

Adult education and andragogy are treated as one subject in most training material, which is where the trouble starts: one is a field of provision, the other a set of assumptions about learners, and only the second can be checked against evidence. This brief checks it.

The claim being tested

Adult education carries an unusual burden of proof. Almost every other branch of education research argues about method; this one argues about whether its object exists. If adults learn by the same mechanisms as everyone else, then adult education is a field of provision and policy rather than a distinct psychology, and its textbooks have been overclaiming for fifty years. If they do not, the difference should show up in the evidence, repeatedly and in the same direction.

The honest reading is that the mechanisms are shared and the conditions are not. Working memory is not more capacious at forty. What changes is everything around it: how much relevant structure is already in place, whether that structure is correct, how much of the learner's identity is invested in being competent, how little uninterrupted time exists, and whether the course is a step towards something the learner actually wants. Those are not psychological differences. They are differences in the situation, and they are large enough to make adult provision a genuinely distinct design problem.

This is not a demotion of the field. It is a relocation of where its expertise sits: less in a special theory of the adult mind, more in the hard, testable questions of who shows up, what they bring, and whether the workplace lets them use what they learned.

Andragogy's six assumptions, one at a time

Knowles set out six assumptions about adult learners, and they are almost always quoted as a block. Quoting them as a block is what hides the problem, because they are not of the same kind: some are empirical claims that can be checked, one is a design heuristic, and one is a statement of values.

1 · Adults need to know why before they learn something

As a design heuristic this is sound and cheap to act on: stating the purpose of a task costs a sentence and reliably reduces resistance. As an empirical distinction it is weak, because children respond to purpose too. Treat it as good practice rather than as evidence for a separate psychology.

2 · Adults have a self-directed self-concept

This is the assumption that does the most work in course design and holds up least evenly. Adults do generally expect to be treated as capable of running their own learning, and resent being handled otherwise. Whether they can run it well is a different question, and the answer depends on whether they can accurately judge what they already know. Where prior structure is thin, self-direction produces confident detours; where it is solid, it beats any sequence a designer could impose.

3 · Experience is a resource

Supported, with a qualification that the slogan drops: experience is also the main source of the misconceptions that instruction has to overturn. A twenty-year practitioner brings a rich model and, sometimes, a confidently wrong one. Both properties come from the same source, and a course that only treats experience as an asset will be defeated by the half of it that is a liability.

4 · Readiness to learn follows life roles

This holds up well and is the most practically useful of the six. Demand for learning clusters around transitions — a new job, a new responsibility, a qualification that has become a gate. It predicts enrolment better than interest does, and it explains why the same course fills in one year and not the next.

5 · Orientation to learning is problem-centred

Supported. Adults tend to organise learning around a task they need to complete rather than a subject to be covered, which has a direct curricular consequence: sequencing by discipline logic fights the learner's own organisation of the material. This is the assumption most worth designing around.

6 · Motivation is internal

This is where andragogy states a value as though it were a result. Adults enrol for promotions, licences, immigration points, employer mandates and the fear of redundancy at least as often as for growth. Nothing is wrong with an educator preferring to work with intrinsic motivation; the error is presenting that preference as a finding about adults, because a programme designed on it will misread the room.

What the evidence supports, and what is contested

Read this way, adult education and andragogy come apart cleanly: the provision is a policy question, the assumptions are an empirical one. Reduced to its defensible core, andragogy says: state the purpose, expect a task-shaped rather than subject-shaped organisation of the material, treat experience as both resource and obstacle, and time provision to life transitions. That is a real and useful set of principles. What it is not is a demonstration that adults are a psychologically distinct category of learner, and the difference matters as soon as somebody uses the framework to justify withdrawing instructional support.

Self-directed learning is a separate question

Self-directed learning and andragogy are routinely merged, and the merge causes practical mistakes. Allen Tough's work in the early 1970s established something surprising and durable: adults undertake far more deliberate learning projects on their own than institutional enrolment figures suggest. That is a claim about where learning happens, not about who should control a course.

Keeping the two apart lets you ask the useful question. Control over sequence is a design variable you can set anywhere between fully specified and fully open. Self-direction is a disposition learners arrive with in different amounts. A course that hands full control to learners whose disposition is strong but whose domain knowledge is weak has optimised the wrong one of the two.

Transformative learning, and its evidence problem

Mezirow's transformative learning is the other framework adult educators reach for, and it answers yet another question: not how material is acquired, but how a frame of reference changes. A learner who reorganises how they interpret their own experience has done something categorically different from acquiring a skill, and the literature describing it is genuinely valuable.

Its problem is measurement, and it is worth naming plainly. The outcome is a changed frame of reference, evidenced largely by the learner's own account of having changed. That is a real phenomenon and a weak instrument, and it is the reason transformative claims and skills-acquisition claims should never be averaged into one programme evaluation. A course can produce a genuine perspective shift and no measurable competence gain, or the reverse; a single evaluation that mixes them produces a number nobody can act on.

Participation: the barrier is rarely the fee

The most consequential finding in adult education is not psychological at all. Participation is stratified: the adults who take up further learning are disproportionately those who already have the most of it. Provision aimed at reducing inequality tends to be used by the people who need it least, unless it is designed specifically against that gradient.

What actually blocks participation, roughly in the order practitioners report it:

• Time in a usable shape. Not total hours but predictable ones. Shift patterns and care responsibilities defeat fixed weekly evening classes more thoroughly than any fee.
• Previous experience of education as failure. An adult who left school with the message that they were not academic reads a classroom as a place where that will be confirmed again.
• Uncertain return. A qualification whose labour-market value is unclear is a rational thing to skip.
• Logistics. Transport, childcare, and the cost of being somewhere at a fixed time, which is often larger than the tuition that was waived.
• Information. Provision that exists and is not legible to the people it was funded for.

The participation gradient is the same pattern the equity and inclusion hub traces through schooling, and it has the same two halves: who gets provision, and who is treated as belonging in it. Every one of those barriers is a design variable rather than a fact of nature, which is the optimistic part. Modular entry points, recognition of prior learning, and asynchronous delivery each address a specific item on the list, and programmes that move participation usually moved one of them rather than the pedagogy.

Workplace learning and the transfer gap

Most adult learning is paid for by an employer expecting behaviour to change, and this is where the field's least comfortable finding sits: whether practice changes depends heavily on conditions outside the course. The training-transfer literature keeps arriving at the same short list — whether the supervisor supports the new behaviour, whether the work provides an opportunity to perform it, and whether anything in the environment punishes the attempt.

The consequence for course design is uncomfortable but actionable. A well-designed course delivered into a workplace that does not permit the behaviour will produce satisfied evaluations and no change, and the post-course questionnaire cannot tell the difference. Any serious training proposal therefore has to make a claim about the receiving conditions, not only about the content — and if the receiving conditions cannot be influenced, the honest recommendation is often to spend the budget elsewhere.

Recognition of prior learning, and why it stays marginal

If prior structure is the variable that matters most, then assessing what an adult already knows should be the sector's central instrument. In policy it is: recognition of prior learning appears in national qualifications frameworks across Europe, North America and Australasia, usually with a commitment that informal and workplace learning can count towards a formal award. In practice it stays a niche route, and the reasons are structural rather than pedagogical.

Assessing an experienced adult's knowledge against a syllabus is labour-intensive, and it is labour that institutions are rarely funded for. It also reduces the number of taught hours the learner buys, which makes it commercially unattractive to the body that has to carry out the assessment. And it puts an awkward question to the curriculum: if a candidate can demonstrate the outcomes without the course, the course was partly about something other than the outcomes.

The practical lesson for a course designer is narrower than the policy debate. Even where formal recognition is unavailable, a diagnostic that establishes what the group already holds pays for itself immediately — it tells you which parts of the material to compress, which prior models you will have to confront, and how much control over the sequence the room can actually use. That is the same finding as the andragogy section, arriving from the direction of assessment.

What is worth measuring, and what the surveys show

Large-scale surveys of adult skills, most notably the OECD's Survey of Adult Skills, changed the terms of this discussion by measuring literacy and numeracy proficiency directly rather than counting qualifications. Two of their findings are robust across participating countries and uncomfortable for the sector.

First, a substantial share of working-age adults in wealthy countries performs at the lowest proficiency levels — a proportion large enough that it cannot be treated as a residual problem, and one that qualification statistics conceal because a certificate is not a measurement. Second, proficiency and participation in further learning are positively correlated, which is the participation gradient again: the adults who would gain most from provision are the least likely to enter it.

For programme evaluation this suggests replacing two habits. Completion rates borrowed from schooling penalise the flexible, modular provision that participation research recommends, so a modular course looks like a failing course. And satisfaction surveys measure whether the session was enjoyable, which correlates poorly with whether anything changed. What survives scrutiny is narrower and duller: a direct measure of the target competence, taken late enough to be about retention rather than about the room, plus an honest statement of who did not enrol at all.

What adult education provision actually looks like

The term covers institutions with almost nothing in common: further-education colleges, workplace training, trade-union and community education, second-chance basic skills, professional continuing education with mandatory hours, and the university extension sector. They differ in who pays, who sets the curriculum, and what counts as success, which is why national statistics on adult participation are so difficult to compare and why European policy documents in this area spend so much of their length on definitions.

Two structural facts are worth carrying. First, the sector is counter-cyclically funded in most systems: demand rises in a downturn while budgets fall. Second, its outcome measures are usually borrowed from schooling, which is a poor fit for provision whose learners arrive and leave at will; completion rates penalise exactly the flexible, modular design that participation research recommends.

One further structural fact deserves its own line, because it is where most disappointment in this sector originates. Adult provision is usually evaluated on the outcomes of the people who finished, while its purpose is stated in terms of the people who did not start. Those two populations barely overlap. A programme can post excellent completion outcomes and have failed entirely at the thing it was funded for, and the reporting will not show it. Any evaluation worth reading therefore states its denominator: not just what happened to participants, but who the provision was for and what share of them it reached.

What this means for designing an adult course

Six decisions follow from the above, and they are worth stating as decisions rather than as principles:

1. Find out what prior structure is in the room before deciding how much control to hand over. This is the single highest-value diagnostic in adult teaching, and it costs one honest task at the start.
2. Sequence by task, not by discipline, unless the discipline logic is itself the thing being taught.
3. Design experience in as a target, not only as a resource: plan where existing models will be wrong and what will confront them.
4. Time the offer to a transition if you have any control over when it runs.
5. Fix the shape of the time before the fee. Modular, asynchronous or repeated delivery moves participation more than a discount.
6. State the receiving conditions for any workplace programme, and treat their absence as a reason to renegotiate rather than a risk to note.

None of this requires a distinct adult psychology. It requires taking seriously that the same mind, in a different situation, needs a different course.

Works referred to

1. The Adult Learner: A Neglected Species — Malcolm S. Knowles, first edition 1973.
2. The Adult's Learning Projects — Allen Tough, 1971.
3. Transformative Dimensions of Adult Learning — Jack Mezirow, 1991.
4. Understanding and Facilitating Adult Learning — Stephen D. Brookfield, 1986.
5. The Three Dimensions of Learning — Knud Illeris, 2002.
6. Adult Education Policy and the European Union: Theoretical and Methodological Perspectives — Sense Publishers, 2015.
7. Access and Equity in Adult Education — Sense Publishers, 2011.

Questions readers send

What is andragogy, in one sentence?

Andragogy is the set of assumptions about adult learners that Malcolm Knowles put forward from the late 1960s onward, most influentially that adults are self-directed, bring experience that is itself a resource, and learn in order to solve a problem rather than to complete a subject.

Is andragogy a theory?

It is usually described as one and behaves more like a set of design principles with an implicit theory of motivation attached. Stephen Brookfield and others have argued the distinction matters: as principles they are useful and partly supported, while as a theory they make claims about adults as a category that the evidence does not carry.

Do adults really learn more slowly than children?

Not in the way the folk version implies. Some processing speeds decline with age and vocabulary and domain knowledge typically grow, so the net effect depends entirely on the task. For most of what adult education teaches, the size and correctness of existing knowledge dominates any age effect.

Should an adult course let learners choose the order of the material?

Only where their prior structure is strong enough to make the choice well. Learner control helps most when learners can judge what they already know and hurts when they cannot, which is why the same course can produce opposite results in a group of experienced practitioners and a group of newcomers.

Why do free adult-education programmes still reach so few people?

Because cost is rarely the binding constraint. Time, shift patterns, care responsibilities, transport, prior experience of failing at school, and a well-founded doubt that the certificate will change anything all outrank the fee. Removing the fee alone reliably changes participation less than programme designers expect.

What single decision most affects whether workplace training changes practice?

Whether the work allows the new behaviour afterwards. The training-transfer literature keeps returning to the same finding: supervisor support and the opportunity to apply the skill predict change better than most features of the course itself.

Where to go next

1. Learning theories The four families of learning theory compared on the same three questions: what each one can explain, what it cannot, and which teaching decisions it licenses.
2. Curriculum design Backward design, spiral sequencing and constructive alignment explained as a coverage-versus-depth trade, with the cost of each choice stated in teaching hours.
3. Teacher development Which forms of teacher professional development change classroom practice, by how much, and how long the effect lasts before it decays.