Learning Outcomes

What an outcome has to be before anything can be built on it, and the two directions it is checked in. The design-course skill writes outcomes against this file; make-exam checks them against it before building a blueprint. This file is reference only — it carries no steps of its own.

The host project governs. A programme, an institution, or a country imposes its own outcome framework and its own forms; where one exists it governs, and this file fills the gaps it leaves. What follows is not a standard — it is what the standards have in common.

That framework is found, not assumed: before checking, look for the project's own outcome standard — a verb list, a rule on how many verbs an outcome may carry, a rule relating verb level to the level the curriculum map assigns. Its knowledge index and the rule file of the area holding the course are where it usually sits. Name the files that governed in the heading of the check table, or state that the project has none. A check run against this file alone, in a project with stricter rules, reads exactly like one that ran against both.

Constructive alignment

Three things are designed together, and the middle one is what makes the other two true:

Learning outcome  ──►  Teaching and learning  ──►  Assessment
 (what it claims)      (the chance to do it)      (whether they can)

They carry the same verb at the same level. An outcome saying design that is taught by lecture and assessed by recalling design principles is not a strict outcome imperfectly served — it is a different outcome, and the marks say so. Alignment is the property being checked, in both directions:

• Vertically — each course outcome serves at least one programme outcome, and each programme outcome the course is assigned reaches at least one course outcome. An outcome serving nothing upward is the course's own invention; a programme outcome nothing reaches is a hole in the curriculum map, and it is the map that is wrong, not the course.
• Horizontally — each outcome has teaching hours, an assessment that elicits its verb, and a share of the marks. Any of the three missing makes the outcome decorative.

What an outcome must contain

The mandatory part is a behaviour: an observable verb and the content it acts on. Two parts are conditional:

Mager's original demand is the one that still does the work: use doing words — verbs whose performance can be seen — and never know, learn, appreciate, be aware of. The audience ("students will be able to…") is conventional and carries no information.

The four ways a verb fails

Each is a defect with a fix, and each is cheap now and expensive after the paper is written.

• Compound. More than one assessed action in one outcome — "create X and evaluate Y", "analyse A and identify B", "converse and present" — whether the actions sit at two cognitive levels or at one. No single assessment serves it, so the assessment becomes two assessments sharing one outcome's marks, and it reads that way to anyone sitting it. The deletion test decides what counts: delete the phrase from the outcome and look at the assessment. If a rubric criterion or a block of marks has to go with it, the phrase was an assessed action. If only the task setup changes — what is supplied, the channel, the constraint — it was a condition. The test is for phrases that name something the learner does; a phrase that describes the quality of what the main verb produces — "effectively", "drawing on verifiable sources" — is degree, and its rubric row belongs to the main verb. "By researching the topic" names something the learner does rather than something supplied, so it stays a condition only while nothing scores it. One verb with several objects — "demonstrates responsibility, punctuality, and honesty in citing" — is one action, but the outcome must settle whether meeting some of the objects passes; left open, no marker can apply it. Fix: split into two outcomes, or fold the second action's marks into a quality criterion of the product the first one makes. The marks and the hours then divide by outcome instead of being argued about inside one.
• Unobservable. Understand, know, realise, appreciate name a state, not a performance. There is no item that confirms them, so in practice something else gets assessed and the outcome is decorative. Fix: ask what the learner would do to show it, and make that the verb.
• Ambiguous. A verb that sits at two levels depending on reading — explain (restating a cause is Understand; working out why is Analyze), compare (against criteria given is Understand; choosing the criteria is Analyze). Fix: the condition disambiguates it. "Compare against the criteria supplied" and "compare, choosing your own criteria" are different outcomes.
• Inflated. The verb claims a level the course never teaches to and the paper never reaches. Fix: raise the teaching and the assessment, or lower the verb. Reading the verb loosely so the existing paper qualifies is the one move that is not available — that is how a course states an outcome it does not deliver, in writing, for as long as the syllabus stands.

What can evidence which level

An outcome is only as assessable as the formats available to it. Choosing the format at writing time rather than at design time is what quietly caps a course at its lowest level.

No selected-response item evidences Create, whatever its stem says. An outcome at that level with no constructed-response, product, or performance behind it is unassessed however the marks total.

The level the programme assigns

A curriculum map gives each course a level of responsibility for each programme outcome — introduce, reinforce, master; a weight from one to three; or the programme's own scheme. An outcome's verb level is read against it. A verb at Evaluate or Create under a programme outcome the course only introduces claims more than the programme asked this course to reach, and a programme review reads it that way. Where the host project sets how map levels translate to verb levels, apply its translation; where it sets none, report each outcome's verb level beside its map level and ask.

An outcome kept above its map level on purpose is an intentional exception, and it passes only with a record beside it: evidence that the teaching and the assessment both reach the verb's level, the cost of keeping it, the lower verb it would fall back to, and who confirmed it and when. Without the record it is an inflated outcome that happens to have been noticed.

When the learners cannot reach the outcome's level

Alignment says what the assessment must reach; it does not say the learners can reach it. A recommendation resting on alignment alone — "the outcome says analyze, so the items are at Analyze" — answers whether the design fits the course, never whether this cohort can do it, and only the person teaching them can say that. Where the two conflict, offer the ways out in this order, each with its cost:

1. Keep the level, cut the difficulty that is not the outcome. Level and difficulty are different properties: an Analyze item on a short stem and a familiar artifact can be within reach where the same level buried in a long narrative is not. Reading load, unfamiliar context, and option length make an item hard without making it measure more. Cost: it may still be out of reach.
2. Ladder the levels. Most items at the levels beneath, building towards the outcome, and a minimum share at the outcome's own level, chosen by the user. Cost: the evidence that the outcome was reached is thinner.
3. Assess below the outcome, and record the gap. Cost: the paper no longer evidences the outcome as written, which is the finding the next review of the outcomes needs.

Whichever is chosen, an assessment that ends up below its outcome's level is written down as a decision record, in the shape the grill-point skill gives.

The check table

The check is a table with one row per outcome, written into the design wherever its outcomes are recorded. One outcome, one row. A check that leaves only a verdict cannot be told apart from one that never ran, and a prose paragraph of ticks is a verdict.

Every cell other than the result quotes the outcome or cites a line of the design. A result is only ever one of those three words.

How many

For a full-weight course, roughly four to six. This is practice rather than a standard — no standards body names a number — and the reasoning is arithmetic: a term divided by five outcomes leaves enough weeks per outcome to teach it, practise it, and assess it. Fewer than three and the outcomes are too broad to attach marks to; more than seven and some of them are certainly not being assessed.

Three rules for the alignment table

The table that pairs each outcome with its teaching, its assessment, and its marks is where misalignment becomes visible. Three things in it are always defects:

1. The verb changes along the row. An outcome saying design taught as design but assessed by a written test about design principles is misaligned by definition, not by degree.
2. A block of marks with no outcome. Marks awarded for something no outcome claims — participation, neatness, effort — are either an outcome that has not been written down, or marks that should not exist.
3. A high-level outcome carrying trivial weight. An outcome at design or evaluate worth 5% beside a remember outcome worth 60% describes a course that says one thing and rewards another. Learners follow the marks.

Checking a set of outcomes

Run this over the whole set, not one outcome at a time — most of these are properties of the set:

• Every outcome has its row in the check table, and passes the four verb checks or is raised with the user.
• Every outcome maps upward to at least one programme outcome, and every programme outcome the course is assigned is reached.
• Every outcome's verb level is within its map level, or it is an intentional exception with its record.
• Every outcome has teaching hours behind it, and every scheduled session serves an outcome or is marked as not assessed.
• Every outcome has an assessment whose format can reach its level, and a share of the marks.
• The marks sum to the course's total, and their distribution across levels matches what the outcomes claim.
• Where the programme has more than one revision in force, the outcomes name which revision they map to. A course taught to two cohorts under two revisions maps to both, separately.