Skip to content

Self-Directed Learning

Pattern · All patterns
Evidence · 20 claims (9 for, 9 mixed, 2 against) · 29 studies (9 quant-synthesis, 7 causal, 6 review, 5 theoretical, 2 associational), q1–q4 · 8 of 29 report an effect size · 11 claims rest on one study

Description and scope

A reusable course- or unit-level policy for handing learners control over their own goals, resources, pace and evaluation in stages: probe what the learner can already plan and judge → agree a written plan (goals, if-then plans for obstacles, the evidence that will show progress, dates) → teach the self-direction moves inside the subject work → release choice in steps (pace first, then resources and tasks, then goals and evidence), each step earned by an observed criterion → brief scheduled check-ins against the plan → the learner evaluates the result against stated criteria, an outside check is compared with it, and the next cycle is planned. The intended change is twofold: learning in the subject (achievement, measured on the course's own outcome) and the capability to run the next cycle with less support (self-regulation, observed as plans kept, accurate self-ratings and sensible choices, not only questionnaire scores).

It sits beside three converted pages and does not repeat them. The Self-Directed Learning principle owns the relationship (when learner-directed study helps, and for whom); this page owns the sequence. The Self-Regulated Learning pattern owns the within-task plan–monitor–act prompting and its fading; use it for step 3 below. The Autonomy principle owns the motivational case for choice, and Goal Setting and Monitoring the goal-setting relationship.

No claim in this wiki tests self-directed learning as this page defines it, that is, a course in which control of goals, resources, pace and evaluation is handed to learners, against a teacher-directed course on the same outcome. The nearest tests point in different directions: giving learners control within educational technology showed an overall effect near zero across 18 studies; self-paced mastery courses lost more students than conventional ones; and explicit training in planning, monitoring and evaluating raised outcomes in school meta-analyses. So the sequence below is the earlier page's requirements made concrete, each step labelled with its evidence status, and the staged release of control is an untested design proposal.

Inputs: establish the design brief

Field Record or elicit
Learner and starting observation Before handing over any choice: a short content probe on the course's first topic; a written plan for one sample week; a prediction of the probe score set beside the actual score. Record whether the learner plans unprompted, how far the prediction misses, and prior knowledge of the domain. Preserve unknowns rather than labelling a learner novice or "self-directed".
What can be handed over Which of goals, resources, sequence, pace, products and evaluation criteria the course can actually let learners decide, and which are fixed by an exam, a licence, a standard or an employer.
Context and activities Setting (classroom, online-self-paced, online-instructor-led, workplace-clinical, mixed); single-session or days-weeks; how many minutes a week a facilitator has per learner; whether learners have devices beyond a phone; group size.
Learner-valued goal Ask what the learner wants from the course and what would make it worth their time. Record disagreement with the designer objective; in this pattern the learner's goal often becomes part of the plan, so record where it does and where it cannot.
Designer objective Which matters most: course content mastered, completion, the learner's capability to plan and judge their own work, or a product. Say which supports are still permitted when the target is measured.
Outcome An instrument per objective: the subject test or product rubric; completion by the deadline; calibration (self-rating against scored result); traces (plans kept, choices made, check-ins attended); any questionnaire labelled as self-report. Horizon: end of unit, end of course, delayed.

A request to "make this course self-directed" or "let learners take ownership" specifies none of these. Ask what can be handed over, to whom, and what will show it worked, before choosing a stage or a menu.

Sequence and conditional policy

The default runs over a course of about 8–12 weeks. Durations and release criteria are labelled proposals; where a claim bears on a step it is named, with how far it is carried.

  1. Probe readiness (session 1, 30–40 minutes). Content probe, a plan for a sample week, and a score prediction before the probe is marked; then show the learner the gap. Evidence status: self-assessment accuracy is low without training [~S]: in one randomized experiment 80 secondary students who studied modelling examples of self-assessment rated their genetics problem-solving more accurately than those who did not, and a meta-analysis of 51 higher-education studies found student and teacher marks agreed inconsistently, more closely in advanced courses. Do not rely on a learner's own judgment of readiness. Durations untested.
  2. Agree a written plan, a learning contract (session 1–2, 20–30 minutes, revised every 2–3 weeks). The plan names: the course goal and, for learners new to the content, process goals (the steps or strategies to practise) rather than only a score; two or three if-then plans for the obstacles the learner expects ("if my shift runs late on Tuesday, then I do the 15-minute review on Wednesday morning"); the evidence that will show progress; and dated checkpoints. Evidence status: process goals outperform outcome goals for novices [+S] (84 high-school girls learning a writing-revision skill; a shift to outcome goals once the skill is mastered also suggested); implementation intentions improve goal attainment [+S] (meta-analysis of 94 tests, d = .65, mostly university students; in two randomized online-course experiments with 17,963 learners an eight-minute activity combining mental contrasting with if-then plans raised completion for learners in individualist cultures and left learners in collectivist cultures unaffected). Neither tests a learning contract as a whole.
  3. Teach the self-direction moves inside the subject work (weeks 1–3, about 10–15 minutes of each session). Model, on this course's material, how to set a session goal, choose between two resources, check work against the criterion and choose a next action; use short task-specific planning, monitoring and evaluation prompts at natural pauses, paired with feedback. Run the prompting and fading as in the Self-Regulated Learning pattern. Evidence status: self-regulation strategy instruction improves achievement [+S] (84 school intervention studies, average 0.69 over performance, strategy use and motivation combined, larger when researchers rather than regular teachers delivered it); metacognitive prompts improve learning [+M] (computer-based environments, g = 0.40 on learning outcomes against unprompted controls, varying with feedback, task specificity and adaptivity).
  4. Release control in stages, each earned (untested proposal).
  5. Stage A, weeks 1–3: the teacher sets goals and sequence; the learner chooses pace within weekly deadlines and the order of two or three set resources.
  6. Stage B, weeks 4–6: the learner chooses tasks and resources from a menu (a choice board of four to six options mapped to the same criteria) and sets weekly sub-goals.
  7. Stage C, week 7 on: the learner proposes a goal, resources and the evidence that will show it, within the course outcome; the teacher approves or negotiates.
  8. Move on when, on two consecutive checkpoints, the learner completed what the plan said and the self-rating fell within about 10 percentage points of the scored result (a proposal; set the band locally). Step back a stage after two missed checkpoints in a row. Evidence status: learner control in general showed almost no effect across 18 studies, while learner pacing helped on complex material in one experiment [~S]: a meta-analysis of learner control within educational technology found g = 0.05 overall, near zero across most characteristics of control and context; one experiment with primary-school students found learner-paced animation better only on difficult questions. Control as such is not what helps. Autonomy support increases intrinsic motivation, engagement and persistence [+M] is the motivational case for meaningful choice (and its meta-analysis of 128 experiments found expected tangible rewards undermined intrinsic motivation, d = -0.4); it does not show choice raises achievement.
  9. Keep scheduled check-ins and dated checkpoints throughout (weekly, 5–10 minutes per learner, or an asynchronous progress post answered within two working days). Review the plan against what was done; the learner names the next action. Evidence status: self-paced mastery programmes reduce college course completion [-M]: in 32 evaluations of Keller's self-paced Personalized System of Instruction, 23 found higher completion in the conventional class (h = -0.14 on attrition), though the examination advantage held either way. Self-pacing with no deadlines costs completers; dated checkpoints are the proposed response, untested.
  10. Evaluate and plan the next cycle (end of each 2–3-week cycle, 20–30 minutes). The learner rates the work against the stated criteria first, then an outside check (teacher, rubric-scored product, test) is compared with it; the difference and the next plan are written down. Evidence status: self-assessment accuracy is low without training [~S], above; the cycle length is a proposal.
Observation Discriminating probe Proposed action, with its uncertainty
Plan written, checkpoints missed Ask what got in the way; check whether the planned work was too large, the content too hard, or time was taken by something else. If an obstacle recurs (work, family), add an if-then plan for it and keep the stage. If the content is too hard, return to Stage A for that topic and give a worked route. If time is the constraint, shrink the plan rather than add prompts. Untested.
Self-ratings far above scores Have the learner score an exemplar with the rubric, then rate their own work again; run a short content probe. If calibration improves with the exemplar, keep exemplar-scoring before each self-rating. If the probe is weak, treat it as a knowledge gap and hold the stage. Neither branch is a validated diagnosis.
Chooses only easy or familiar options from the menu Ask what each option would show against the criteria; compare chosen options with the plan's goal. If the learner cannot say what an option shows, map each option to a criterion on the menu. If the choice is deliberate avoidance, negotiate one stretch option per cycle. Untested.
Keeps plans and calibration, subject scores flat Score the topics the learner chose separately from the broad test; compare with the plan's goals. If chosen-topic scores rise, the outcome may not cover what was chosen; agree the outcome first. If not, the self-direction is working on the wrong content; narrow the menu. Untested.
Withdraws, says they would rather be told what to do Ask about time, confidence and prior schooling; offer a fully set week and observe whether work resumes. If work resumes with a set plan, keep Stage A longer and release one choice at a time; preference for guidance is information, not failure. Untested.

Fitting the design to a situation

The three facts that most change the decision:

  • How much the learners already know of the content and how well they judge their own work. Self-ratings are unreliable without training, and process goals helped novices more than outcome goals. If unstated, ask: on a short probe, how far do their predictions miss their scores, and could they plan a week of this subject now?
  • Whether a facilitator will see the learners regularly, and for how long. Self-paced courses lost more students than conventional ones; the check-in in step 5 is what holds the pace. If unstated, ask: how many minutes a week does anyone look at each learner's plan?
  • What can actually be handed over. If the goal and the assessment are fixed, choice is limited to pace, resources and order. If unstated, ask: which of goal, resources, pace, product and criteria can the learner decide?
If the brief says… Then change… Basis
Learners are novice in the content Stay in Stage A for at least the first unit; write process goals, not score goals; give two set resources, not a menu; model a self-check on a worked example before any self-rating. Process goals outperform outcome goals for novices [+S] (high-school writing; carried to other content); self-assessment is low without training [~S]
Learners are advanced or experienced practitioners in the field Start at Stage B or C after the probe; let the learner draft the plan and the criteria, and keep only the checkpoints and the end comparison with an outside check. Untested; carried from the earlier page's "some learners can self-direct immediately"
Learners are child (primary age) Keep the teacher setting goals; hand over pace and the order of two or three tasks, with a daily visible tracker; release one further choice per term. Untested proposal; SRL strategy instruction [+S] included primary pupils, but tested training, not handing over control
Learners with intellectual or learning disabilities, working toward their own goals (transition planning, an IEP) Teach the component skills directly (choice-making, goal setting, self-advocacy) over months, and let the learner lead the planning meeting with a practised script. Self-determination skills instruction for students with disabilities increases self-determination on some measures but not others [+M]: a three-year campus-randomized trial with 371 high-school students raised one of two self-determination measures; outcomes were self-determination, not achievement
online-self-paced, no facilitator Keep dated weekly checkpoints with automated reminders; add an eight-minute goal, obstacle and if-then plan activity at enrolment; show each learner their plan against their progress weekly. Self-paced courses lost more completers [-M] (college PSI); implementation intentions [~S] (helped completion in individualist cultures only)
Learners come mostly from collectivist cultures, or schooling where choice reads as lack of guidance Do not rely on an if-then planning activity alone to raise completion; plan with a peer group or the facilitator, and present choices as options the teacher endorses. Implementation intentions [~S] (no effect for learners in collectivist cultures in the online experiments); the cultural note on autonomy support [~M] is the claim page's discussion, not a finding
Wide-open self-direction (open online course, networked learning, learners choose everything) Add the plan, checkpoints and a small cohort or buddy; give a default path learners may leave. Not all students in connectivist courses could direct their own learning [~M] (second-hand reports from a review)
Scarce facilitator time (volunteers, a busy shift, one tutor checking in briefly) Make the check-in five minutes against the written plan: what was done, what is next, one obstacle; batch asynchronous replies twice a week; let peers check each other's plans between. Untested proposal
One teacher or expert for many learners Run check-ins in small groups of 4–6 on a rota, each learner reporting against the plan in two minutes; the teacher reads only plans flagged as off track. Untested proposal
Mixed modes: some in the room, some remote; phone-only learners Keep the plan and tracker in one shared place that works on a phone; hold check-ins by short message or voice note for remote learners; put the menu in text, not a large file. Untested proposal
A single session only Do not run the staged release; run steps 1 and 2 only (probe, a written plan with one if-then plan, a date), and hand the learner the plan to keep. Untested proposal
A fixed short time budget per session (e.g. 45 minutes) Spend about 5 minutes on the plan, 30–35 on the chosen work, 5 on the self-check against the criterion; the plan revision moves to every second or third session. Untested proposal
Adults in work settings Tie the plan's goals to a task the learner does at work; let the employer's requirement fix the outcome and the learner choose route and pace; check-ins fit shift patterns. Untested; facilitated discussion felt empowering to one adult educator [+W] is one first-person reflection
The brief names its own routine for planning or review (its own learning contract, journal, check-in form, rubric) Keep it; change only what the evidence says matters: one if-then plan for the main obstacle, process goals for novices, dated checkpoints, and a self-rating compared with an outside check. Implementation intentions [+S]; process goals [+S]
The brief gives its own categories, criteria or competency framework Use them as the plan's goals and as the menu's labels; do not substitute this page's stages or criteria. Untested proposal
Private or sensitive material (personal goals, health, a reflective journal) The learner controls who sees the plan and journal; the facilitator sees only what the learner chooses to share and the checkpoint dates. Untested proposal
The product is for a real audience, or a whole-task performance the brief wants kept whole Let the learner plan the route to the whole product and set checkpoints on drafts of it; do not split it into separate exercises for evaluation. Untested proposal
Learners with low confidence or limited literacy Write the plan with the learner, in short lines or pictures; offer two choices at a time; read the self-check aloud. Untested proposal
Complex multimedia or video content that learners pace themselves Segment the material and add pause points rather than relying on learners to pause; pace control alone did little. Learner pacing helped on difficult questions in one experiment; learner control in general near zero [~S]
An AI assistant or hint budget is available during self-directed work Teach a deliberate rule for when to ask for help and record it in the plan; a budget alone rewarded learners who already had a strategy. Hint budget managed poorly [~W] (one programming course pilot)
High stakes: an exam or licence at the end Fix the outcome and the assessment; hand over pace and resources only, and run a practice check under exam conditions before each stage release. Untested proposal

Ten of 21 rows link a claim, and each carries it beyond the learners or setting it tested, as the Basis cell says (the rows for primary pupils and adults at work remain proposals with a claim beside them); the other eleven are untested proposals or the earlier page's guidance.

Choosing configurations from evidence

Do not rank this pattern against a teacher-directed course from these claims: none compares them, the outcomes differ (performance, strategy use and motivation combined; completion; calibration; self-determination scales), and the 0.69, 0.40 and .65 figures are not on one measure. A precise abstention names the missing comparison (a course that hands over goals, resources and pace in stages against the same course teacher-directed, on the course outcome and on completion, at the end of the course and later) and the local observations that stand in for it: completion by checkpoint, calibration, and the course outcome.

The evidence by question

Further evidence, not yet read against this model

Claims read for this rewrite but not used as core evidence above, or used only in one row; each says how far it bears on the page.

Illustrative design instance and observation record

Invented illustration, not a tested design. Pharmacy technicians, adults working shifts, take a ten-week online-self-paced course on sterile compounding before an employer competency check, with a tutor who has about ten minutes a week per learner. Because the outcome is fixed and stakes are high, only pace, resources and order are handed over. Week 1: a 15-item probe and a predicted score; one learner predicts 80% and scores 52%. Her written plan has process goals (the aseptic steps in order), two if-then plans ("if a night shift runs over, I watch the 10-minute video on my next break, not the reading"), and checkpoints every Friday. Weeks 1–3 run at Stage A with the tutor modelling a self-check against the competency checklist on a recorded demonstration. Week 4 her last two self-ratings were within 10 points of her scores and both checkpoints were met, so she moves to Stage B and chooses among a video, a simulation and a reading for each topic. Week 6 she misses two checkpoints after a rota change; she steps back to Stage A for one week and adds an if-then plan for rota changes. She says she wants to stop feeling slow on the ward; the employer's objective is passing the check. Both are recorded, and the check, completion by week 10 and her calibration are reported separately. The claims behind these choices tested other learners and settings; these are local choices to observe.

Record: what was handed over and what stayed fixed → probe score, prediction and plan → stage and its release criterion → checkpoints met or missed, with the stated obstacle → choices made from the menu → self-rating against outside check at each cycle → candidate interpretations (knowledge gap, poor calibration, time, avoidance, preference for guidance) and their basis → stage change and why → course outcome and completion at the stated horizon. Preserve the learner's stated purpose and any change in it. An agent may summarize this record but must not fabricate missing inputs or assign numerical probabilities that a learner is "ready for self-direction" without a calibrated model.

Elements and limits

Learning contracts, goal setting, learning goals, learner choice, choice boards, progress tracking, check-ins, self-monitoring, self-assessment, reflection, portfolio, fading scaffolding and mastery progression. Principles it draws on: Self-Directed Learning, Self-Regulated Learning, Autonomy and Goal Setting and Monitoring.

This pattern is scoped to a course or unit in which control is handed to learners in stages within a structured environment. Fully open networked learning, informal learning outside any course, adaptive systems that make the choices for the learner, and within-task prompting are other configurations. Whether a staged hand-over of goals, resources and pace produces better course outcomes or completion than the same course teacher-directed, for which learners and at what horizon, remains to be tested; the release criteria above are proposals to observe, not thresholds.

Claims

  • Self-Regulated Learning — the within-task plan–monitor–act prompting this pattern uses in its teaching step
  • Mastery Learning — mastery gating with group pacing; its self-paced variant (PSI) carries the completion cost this pattern plans for
  • Competency-Based Learning — fixed competencies with flexible pace and route, a common frame for Stage B

Examples

  • An adult upskilling course where the employer fixes the competency and each learner plans route and pace in a written contract with weekly checkpoints (see the illustration above)
  • A secondary independent-study unit run as a choice board of tasks mapped to the unit's criteria, with a weekly five-minute conference

Key Sources

  • Knowles, M. S. (1975). Self-directed learning. Follett.
Page metadata
Field Value
type pattern
id self-directed-learning
title Self-Directed Learning
description A staged policy for handing learners control over goals, resources, pace and evaluation within a structured course, with a written plan, taught self-direction moves, dated check-ins and calibrated self-evaluation; no claim in this wiki tests the hand-over as such, and learner control alone showed a near-zero effect while self-pacing cost completion.
status review
grain_size course
author self-directed learning tradition
canonical True
generated by claude/unspecified on 2026-10-15
trust unverified — no one has confirmed this page's content

Frontmatter is the page's machine-readable record — see the Schema & Guide.