Skip to content

Constructive Alignment

Pattern · All patterns
Evidence · 19 claims (11 for, 7 mixed, 1 against) · 20 studies (6 quant-synthesis, 4 causal, 4 theoretical, 3 associational, 2 review, 1 qualitative), q1–q4 · 9 of 20 report an effect size · 16 claims rest on one study

Description and scope

A reusable course- or programme-level policy for checking that what learners are told to achieve (intended outcomes), what they spend their time doing (activities) and what they are scored on (assessment tasks) call for the same performance, and for repairing the system where they do not. It instantiates the converted Assessment for Learning principle, which models how assessment evidence is used during instruction (shared criteria, stakes that let partial understanding show, a revision step), and it depends on the judgement modelled by the converted Competency-Based Assessment principle: whether a task represents the capability, whether its threshold is defensible, and whether assessors agree. This page owns the more general relationship around both: whether the outcome, the practice and the assessment are about the same capability at the same level. It does not repeat either sibling's model; use them for the reading of a single response and for the quality of a single judgement.

The relationship the policy assumes is this: for learners working towards a capability they have not yet reached, when the assessed task demands the capability named in the outcome, and the activities give repeated practice of that capability with feedback before the assessed task, performance on the assessment (general-achievement, or skill-performance for a complex-skill) is expected to be higher, and learners' study effort is expected to go to the intended capability rather than to whatever the assessment happens to reward. No claim in the wiki tests alignment as such: none compares an aligned course with a misaligned one, or measures whether realigning a course shifts learners from surface to deep study. The claims below test parts of it (practice of the assessed content, active work against lecture, assessment used during learning, stated objectives) and one qualifies its central "same verb" rule. The study configurations are evidence; the policy, the audit and the table of responses are untested design proposals. Biggs (1996), the source of the pattern, is in the frontmatter; it has no claim page here, so nothing below rests on it.

Inputs: establish the design brief

Field Record or elicit
Intended outcomes as written The outcomes the course actually states, verbatim, with their verbs and conditions. Topic lists ("know about", "understand") are recorded as such; they are an observation, not yet an outcome.
What is assessed, and who controls it Every graded task, its weight, its format, its criteria, and whether the designer can change it (an external examination, an accreditation rubric or a shared departmental exam usually cannot be changed).
What learners actually do The activities and the time spent on each, from the timetable or a week of observation, not from the syllabus; which of them produce a response the teacher sees.
Learners and starting observation Population band (pre-k-12, postsecondary, adult-workplace), expertise on the course's tasks (novice, intermediate, advanced), and a short sample of independent work on a representative task. Ask how learners have prepared for past assessments in this subject.
Goal kind and level verbal-association, concept, procedure, principle, complex-skill or attitude-motivation, and the level at which it must be performed (recall, application, analysis, design).
Learner-valued goal What learners want from the course (a pass, a credential, a job skill, interest), kept apart from the designer's objective.
Grain and time Programme, course or unit; the number of weeks; how many outcomes the course carries.
Outcome and horizon The instrument that will show the capability, a locally chosen criterion, and whether a later use (the next course, the workplace) is a target.

A request to "align the course" does not specify these inputs. Ask for them before rewriting outcomes or tasks. If the assessment cannot be changed, record that first: it decides which half of the policy below is available.

Sequence and conditional policy

  1. Audit before redesigning. Build a three-column grid with one row per intended outcome: the outcome's verb and conditions, the activities that require that performance (with the class time each takes), and the assessed items that require it (with their weight). Mark every outcome with no activity, every outcome with no assessment, and every assessed item that matches no outcome. Untested proposal; from the earlier page's requirement that outcomes, activities and assessment "point in the same direction".
  2. Fix the outcomes first, and keep them few. Rewrite topic statements as performances with an action verb and conditions. Aim for a short list per course (the earlier page warned that "long checklists of granular outcomes fragment learning"; it gave no number, so a figure such as five to eight per course is an untested proposal). Write them at the level the course is meant to reach: alignment amplifies whatever level the outcomes specify, so an outcome written as "list" aligns the whole course to listing. Untested; from the earlier page.
  3. Give each outcome practice with feedback before it is assessed. At least one activity per outcome in which learners perform it and get feedback before the graded task; more for a complex-skill. Evidence: including practice with feedback raised posttest scores in a randomised study of a computer-based lesson (practice presence [+M]); active work against lecture in undergraduate STEM (active learning [+M]). The number of practice episodes per outcome is untested.
  4. Practise at or above the assessed level, not below it. Where the assessment asks for recall or comprehension, practice at a higher level can still serve it: in one course, higher-order clicker questions produced better lower-order exam performance than lower-order ones (HOCS clickers [~M]). Read "same verb" as "the assessed performance must be practised", not "practice must be restricted to the assessed level".
  5. Make the assessed task demand the outcome, and grade against criteria. Replace items that can be answered by recall where the outcome asks for analysis or design; publish the criteria; grade against them rather than by ranking learners. Untested here; the criteria and threshold questions are modelled on Competency-Based Assessment, and the argument against ranking is one essay (competitive grading [+W]).
  6. Use some assessment during learning, with low stakes and a revision step. One checkpoint per outcome before the graded task, read against the same criteria. Evidence: assessment for learning [+M]; the policy for reading the responses is on the formative assessment pattern.
  7. Reobserve and compare with the audit. After one run, compare assessed performance per outcome, and ask learners how they prepared. If preparation still targets something other than the outcome, the assessment still rewards it: revise the task, not the learners. Untested proposal.

When a stated condition is not met: if the assessment cannot be changed (an external examination), do steps 1–4 and 6 against the outcomes you hold, add explicit practice on the examination's format as a separate, labelled activity, and record the mismatch as a known limit rather than rewriting outcomes to match the examination. If only one unit can be redesigned, do the audit for the whole course anyway, so the redesigned unit's outcomes are not contradicted by the final assessment; the earlier page held that a single aligned lesson inside a misaligned course "changes little" (untested).

Observation Discriminating probe Proposed action, with its uncertainty
Learners do well on the graded task but cannot perform the outcome on a new task Give an ungraded task that demands the outcome in a changed context; compare item by item with the graded task. If the graded task can be passed by recall, rewrite it (step 5). If both tasks are passed in isolation but not together, the gap may be integration, not alignment. The rewrite is untested.
Learners report studying past papers or memorising, where the outcome asks for analysis Ask what they expect to be asked and how they decided; inspect what the past papers actually reward. If the assessment rewards memorising, change the assessment; if it does not, the expectation may come from other courses or from stakes, so make the criteria and an exemplar visible and observe again. Neither reading is validated.
An outcome has no activity in the grid, or one activity only Check the timetable for where the performance is practised; ask learners where they first tried it. Add practice with feedback before the graded task (step 3). The evidence supports including practice, not a dose.
Activities are "active" but none demands the outcome's performance Compare the verb the activity actually calls for (discuss, list, watch) with the outcome's verb. Redesign one activity per outcome so the response it produces is the outcome's performance; keep the others if they serve engagement. Untested proposal.
Assessed performance improves but learners disengage or complain of grading pressure Ask what changed for them; check whether the formative checkpoints now carry marks. Remove marks from checkpoints and keep the revision step; see the stakes branch on Assessment for Learning, which is itself a hypothesis there.

These rows are proposals for interpreting a course's state, untested with learners or teachers. One probe can shift confidence between explanations; it does not establish a cause.

Fitting the design to a situation

The three facts that most change the decision:

  1. Whether the designer controls the assessment. If the graded task is fixed by someone else (an external examination, an accreditation rubric), alignment can only be worked from the activity side, and the mismatch must be recorded. If unstated, ask: who sets and marks the final assessment, and can its format change this year?
  2. The kind and level of the goal. A recall goal is easy to align and easy to over-align downward; a complex-skill needs whole-task practice and a performance assessment, which cost time and assessor agreement. If unstated, ask: what should a learner be able to do at the end that they cannot do now, and in what situation?
  3. The learners' expertise and how they prepare now. novice learners need more practice with feedback per outcome and visible criteria; experienced learners may already study strategically to the assessment. If unstated, ask: have they done tasks like the assessed one before, and how did they prepare for the last assessment in this subject?
If the brief says… Then change… Basis
Learners are novice on the assessed tasks Add a worked or modelled example and at least one practised attempt with feedback per outcome before the graded task; show an exemplar against the criteria. Practice presence [+M] for including practice (undergraduates, one computer-based lesson); the per-outcome dose and exemplar are untested proposals
Learners are advanced or professionals Fewer, more integrated outcomes; assess on a realistic task in their own setting; cut practice of components they already perform. Untested proposal
Learners are children or adolescents (pre-k-12) Write outcomes as short "I can…" statements the teacher shares and checks with learners; keep frequent low-stakes checkpoints with a revision step. Assessment for learning [+M] for formative checkpoints (K–12 syntheses); the "I can" wording is untested; the earlier page described the pattern as developed in higher education
Goal is facts or recall (verbal-association) and the assessment is recall Do not restrict practice to recall: include some questions at a higher level (explain, apply) in class practice. HOCS clickers [~M] (one nonmajors biology course, controlled-nonrandom)
Goal is a complex-skill Assess a whole task with criteria and calibrated assessors; schedule whole-task practice across the course, not only at the end. Untested; from the earlier page. Assessor agreement is qualified on Competency-Based Assessment
Goal is an attitude or disposition (attitude-motivation) Do not grade it on a scale; assess the behaviour it should produce, or report it ungraded; give learners bounded choices within the aligned tasks. Untested proposal; the motivation claim is under Further evidence
Setting is online-self-paced with no teacher present Build practice with immediate feedback into each module ahead of its assessed task; make criteria and exemplars part of the module; audit automatically scored items for recall-only questions. Practice presence [+M] (computer-based lesson, undergraduates); the audit is untested
Large lecture class, one session at a time Replace part of lecture with questions or problems answered in class that demand the outcome; use some at a higher level than the exam. Active learning [+M] (undergraduate STEM; held across class sizes); HOCS clickers [~M]
Only one session or unit can be changed Audit the whole course, then align the one unit's outcome with an assessed item that exists; do not promise a course-level effect. Untested; from the earlier page ("a single aligned lesson inside a misaligned course changes little")
High-stakes final assessment fixed externally Keep the outcomes you hold; add labelled practice on the examination's format; check its items' levels and record where they fall below the outcomes. Exam questions at low levels [~W] shows one university's papers sat mostly at comprehension and knowledge (associational); the response is an untested proposal

Six rows rest at least partly on a claim (novice, children, recall goal, self-paced, large class, fixed examination), and each is carried no further than its claim page reports; none of those claims tests alignment itself. The other four (advanced learners, complex skill, attitude goals, one unit only) are untested proposals or come from the earlier page.

Choosing configurations from evidence

  • In one experiment with 256 undergraduates, computer-based lesson versions that included practice produced significantly higher posttest scores than versions without practice [+M]: 256 undergraduate computer-literacy students were randomly assigned within pretest blocks to six versions of one lesson; the four versions that included practice scored above 17 on the posttest and the two without practice below 15 (randomized, immediate posttest, no effect size reported). It supports giving each outcome practice before it is assessed. The claim page does not report how closely the practice items matched the posttest, so it does not test alignment of practice to assessment as such. Not yet checked against its source.
  • HOCS clicker questions during learning improve performance on LOCS exam questions more than LOCS clicker questions do [~M]: in a counterbalanced quasi-experiment over two semesters of a nonmajors Human Biology course, units taught with higher-order clicker questions produced better lower-order exam performance than units taught with lower-order ones (76% against 72%, d = 0.47). This qualifies a strict reading of the pattern: practice above the assessed level served the lower-level assessment better than practice matched to it. A companion claim from the same study reports that a semester difference, not clicker type, separated performance on higher-order exam questions (semester cohort [~M]), so the study does not show that higher-order practice improved higher-order exam performance. One course, sample size not reported on the claim page, not yet checked.
  • Active learning improves exam performance [+M]: a meta-analysis of undergraduate STEM courses compared with traditional lecturing reports exam and concept-inventory scores 0.47 SD higher (158 studies), largest in classes of 50 or fewer and on concept inventories; a second synthesis found exam gaps for underrepresented and low-income students narrowed only under high-intensity active learning. It supports activities in which learners perform rather than listen. The claim page lists alignment of active elements with assessment as an expected moderator still to be established, so it does not show that aligned active learning beats unaligned active learning. Partly checked against its sources.
  • Assessment for learning improves achievement [+M]: syntheses of K–12 formative assessment studies with control conditions report small positive effects on achievement (d = .29; about .20), with larger feature estimates for supported self-assessment (d = .61) and a medium cycle within or between units (d = .52); the claim page says these feature estimates are not randomised comparisons, not additive and not a forecast. It supports step 6, assessment during learning; it does not test whether the formative tasks were aligned with the summative ones. Partly checked against its sources.
  • Individually removing objectives, examples, or review from a well-designed computer-based lesson did not significantly reduce achievement [~M]: in the same six-version experiment as the practice claim, the three versions that each removed one event, one of them the stated objectives, scored between 17.16 and 17.36 against 17.61 for the full lesson, a nonsignificant difference; equivalence was not tested. It limits what step 2 may promise: writing outcomes well is a design discipline for the designer, and showing them to learners did not by itself raise achievement in this lesson. Not yet checked.
  • Instructional design models are rarely tested against outcomes; their credibility comes from practitioners finding them useful [~W]: an expert account in a textbook chapter, from a survey of 34 design models, that models are rarely evaluated against predetermined criteria and gain credibility from adoption. It applies to this pattern: its standing in higher education is not evidence that it raises learning, and the evidence for it lives in the claims above, about its parts.

Do not combine these into one strength or rank them: two are single lessons or courses, two are syntheses of different populations, one is a nonsignificant difference and one is an argument. None measured learners' approach to study (surface or deep), which is the mechanism the earlier page put at the centre of the pattern; that link is a hypothesis here.

The evidence by question

Each question below links a general claim that gathers the studies behind it into one table, one row per distinct study, with each study's design and effect as recorded and no averaged verdict.

Further evidence, not yet read against this model

Claims this page cited before it was rewritten as a conditional model. They are evidence about the relationship, but none has yet been re-read against the model above; each says how far its own sources have been checked.

Illustrative design instance and observation record

Invented for illustration; not a tested design. A logistics company runs a self-paced online induction for new shift supervisors (adult-workplace, online-self-paced, two weeks, no trainer present). The stated outcomes are a topic list ("understand health and safety", "know the escalation policy"), and the only assessment is a 20-item multiple-choice quiz the compliance team owns and will not change this year.

How the situation table changed the design. The fixed-assessment row applies: the quiz stays, its items are audited, and most turn out to ask for recall of policy wording. The outcomes are rewritten as performances the company actually wants ("decide whether an incident on the floor must be escalated, and to whom, and say why"). The self-paced row adds, in each module, three short incident scenarios answered with immediate feedback before the quiz (the number is an untested choice). The recall-goal row adds scenario questions above the quiz's level rather than drilling quiz wording. Because the quiz cannot show the outcome, an ungraded scenario task at the end of week two is added and its result is reported to the shift manager, labelled as not part of the compliance record. A supervisor's valued goal ("not to look out of my depth on my first night shift") is recorded beside the company's objective (a pass on the quiz and correct escalations in the first month); the scenarios serve the first, the quiz only the second.

Record: outcome as written → rewritten outcome → activity and practice attempts with feedback → assessed item(s) and who controls them → learner response on the graded task and on an ungraded representative task → how the learner prepared → candidate interpretations → next change to outcome, activity or assessment → next observation at the stated horizon. An agent may summarise this record but must not invent missing inputs, and must not report a course as "aligned" from the grid alone without the learner observations.

Elements and limits

Learning objectives, practice, application, class discussion, case studies, feedback, formative assessment, assess performance, summative assessment, rubrics and exemplars. For the design method that starts from outcomes and works back to evidence, see backward design and the Understanding by Design process.

This pattern is about the coherence of a course or programme, not about any one activity. It does not say which activities teach an outcome best, how to set a threshold, or how to read a single learner's response; those belong to the pages it links. The theoretical claim at its centre, that misalignment drives learners to surface study, has no claim page here and is untested in this wiki. Outcomes written low align the course low; alignment cannot repair a poorly chosen outcome, and it can make a narrow examination more powerful rather than less. Redesign is time-intensive and may meet resistance where predictable assessment formats are removed (from the earlier page, untested).

Claims

Key Sources

  • Martin, F., Klein, J., & Sullivan, H. (2003). Effects of Instructional Events in Computer-Based Instruction. https://eric.ed.gov/?id=ED484984
  • Deborah K. Anderson, Michelle Schoenleber, and Stuart Korshavn. (2023). Higher-Order Clicker Questions Engage Students and Prepare Them for Higher-Order Thinking Activities. Journal of Microbiology and Biology Education. https://doi.org/10.1128/jmbe.00151-22
  • Freeman, S., Eddy, S. L., McDonough, M., Smith, M. K., Okoroafor, N., Jordt, H., & Wenderoth, M. P. (2014). Active learning increases student performance in science, engineering, and mathematics. Proceedings of the National Academy of Sciences, 111(23), 8410–8415. doi:10.1073/pnas.1319030111
  • Theobald, E. J., Hill, M. J., Tran, E., Agrawal, S., Arroyo, E. N., Behling, S., Chambwe, N., Cintrón, D. L., Cooper, J. D., Dunster, G., Grummer, J. A., Hennessey, K., Hsiao, J., Iranon, N., Jones, L., Jordt, H., Keller, M., Lacey, M. E., Littlefield, C. E., … Freeman, S. (2020). Active learning narrows achievement gaps for underrepresented students in undergraduate science, technology, engineering, and math. Proceedings of the National Academy of Sciences, 117(12), 6476–6483. doi:10.1073/pnas.1916903117
  • Lee, H., Chung, H. Q., Zhang, Y., Abedi, J., & Warschauer, M. (2020). The effectiveness and features of formative assessment in US K-12 education: A systematic review. Applied Measurement in Education, 33(2), 124–140. doi:10.1080/08957347.2020.1732383
  • Kingston, N., & Nash, B. (2011). Formative assessment: A meta-analysis and a call for research. Educational Measurement: Issues and Practice, 30(4), 28–37. doi:10.1111/j.1745-3992.2011.00220.x
  • Black, P., & Wiliam, D. (1998). Assessment and classroom learning. Assessment in Education: Principles, Policy & Practice, 5(1), 7–74. doi:10.1080/0969595980050102
  • Anees, S. (2017). Analysis of Assessment Levels of Students’ Learning according to Cognitive Domain of Bloom’s Taxonomy. https://eric.ed.gov/?id=ED586762
  • Dousay, T. A. (2018). Instructional design models. In Foundations of Learning and Instructional Design Technology (1st ed., pp. 277–295). EdTech Books. https://edtechbooks.org/lidtfoundations/instructional_design_models
Page metadata
Field Value
type pattern
id constructive-alignment
title Constructive Alignment
description Constructive alignment is expected to raise performance on the intended capability when the stated outcome, the practised activities and the assessed task demand the same performance at the outcome's level, with practice and feedback before assessment; no claim here tests alignment itself, and one study found practice above the assessed level served a lower-level exam better.
status review
grain_size course, program
author Biggs (1996)
moved_from principle/bind-teaching-testing-task-based-washback, principle/include-aligned-practice-in-cbi, principle/learning-activity-selection-principles
canonical True
generated by claude/unspecified on 2026-10-02
trust unverified — no one has confirmed this page's content
sources 4 — biggs-1996, martin-2003, ali-panahi-2012, the-iowa-guide-for-curriculum-improvement-in-industrial-arts-1975

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