learning_objectives_writing v1.0.0

---

name: learning_objectives_writing description: "Write learning objectives that can actually be assessed: pick one observable verb at the right Bloom level, add condition and criterion, align activities and checks to the verb, and cut objectives that promise too much. Use when someone asks to write, rewrite, review, or fix learning objectives, learning outcomes, course or lesson objectives, training objectives, or success criteria for a lesson, module, course, or program."

---

Learning Objectives Writing

Purpose

Turn a vague aim like "cover fractions" or "understand leadership" into objectives a second person could grade without guessing. An objective that cannot be observed cannot be assessed fairly, and an objective that is never assessed is a wish, not a plan.

The method rests on four published frameworks, used for what each does best. The revised Bloom taxonomy (Anderson and Krathwohl, 2001) names the level of thinking. Mager's performance, condition, and criterion model, in its common ABCD form, gives the sentence its parts. Constructive alignment (Biggs) checks that activities and assessments require the same verb the objective states. Backward design (Wiggins and McTighe) sets the order: results first, evidence second, activities last.

This skill is source neutral. It works for a school lesson, a university module, workplace training, an online course, or a self study plan, in any tool or on paper.

Workflow

1. Start from the performance, not the topic

• Ask what the learner must do at the end that they could not do at the start. Write that as an action in a real setting, not as a topic heading.

• Separate three levels and keep them apart: a program outcome spans a whole program, a course objective spans a course, and a lesson objective spans one session. Do not promote a lesson objective to course level, and do not cram a course into one lesson objective.

• Count before you write. A common working range is 3 to 8 objectives for a course and 1 to 3 for a single lesson. If a list runs past that, group the small ones under a larger performance or split the unit.

• Name the audience once if it stays the same across the set. Objectives are written for the learner, in learner terms, not as a list of what the teacher will cover. "The teacher will introduce..." is a teaching plan, not an objective.

2. Pick one verb at one level

• Choose the cognitive process first, then the verb that names it. Full tables, with the knowledge dimension and borderline verbs, are in references/verb-tables.md.

• Remember: define, list, recall, identify, name
• Understand: explain, classify, summarize, compare, paraphrase
• Apply: solve, demonstrate, implement, execute, calculate
• Analyze: differentiate, organize, contrast, examine, attribute
• Evaluate: critique, justify, judge, check, defend
• Create: design, construct, formulate, produce, plan

• Use one main verb per objective. Two verbs usually means two performances, so split them. "Analyze and design a solution" becomes an Analyze objective and a Create objective, each with its own check.

• Ban these as the main verb: know, understand, learn, appreciate, be aware of, be familiar with, grasp, master. They name a private state. Replace each with the behavior that would prove it. To show understanding of photosynthesis, a learner might explain it, diagram it, or predict what changes when light drops. Pick the proof, then use its verb.

• Match the verb to the knowledge type. Naming parts is Remember with factual knowledge. Choosing which procedure fits a new case is Apply with procedural knowledge. An objective that says Apply but only asks for a definition is mislabeled, and the fix is to change the task, not to relabel it.

• Treat verb lists as aids, not law. A few verbs show up at more than one level depending on the task. Judge the task the learner actually performs, then pick the level that task sits at.

3. Build the sentence with ABCD

Write each objective with four parts. The order in the sentence can vary, and putting the condition first often reads best.

• Audience: who performs. "First year nursing students", "new support agents", "the learner". Leave it as a shared heading when the whole set shares one audience.

• Behavior: the one observable verb plus what it acts on. "Calculate dosage from a weight based chart", not "do calculations".

• Condition: what the learner is given, allowed, or denied when they perform. Given a case file, using a calculator, without notes, from memory, within the simulation, after the lab. Conditions are part of honesty: a closed book recall and an open book lookup are different objectives wearing the same verb.

• Degree: the criterion for good enough. How many, how accurate, how complete, how fast, or against which rubric or standard. "Four of the five steps in order", "within 5 percent of the reference value", "meeting all rows marked required on the rubric".

Worked builds:

• Weak: "Students will understand fractions."
• Strong: "Given a set of 10 pictured fraction problems (condition), grade 5 learners (audience) will write the equivalent fraction in lowest terms (behavior) for at least 8 of the 10 (degree)." Level: Apply, procedural knowledge.

• Weak: "Trainees will know the safety rules."
• Strong: "Without reference materials (condition), new warehouse staff (audience) will list the lockout steps in the posted order (behavior), naming all six steps with none out of sequence (degree)." Level: Remember, factual knowledge.

• Weak: "Learners will be able to critically evaluate sources."
• Strong: "Given two conflicting news reports on the same event (condition), learners (audience) will judge which claim is better supported (behavior), citing at least two specific pieces of evidence from each report and naming one source limitation (degree)." Level: Evaluate, conceptual knowledge.

When a degree would be fake precision, use a rubric reference or a product standard instead of an invented number. "Meeting the lab report rubric at proficient or above" is a real criterion. "With 90 percent accuracy" on an essay is not.

4. Check for overpromising

Run every drafted objective through these cuts before keeping it.

• Time cut: can a learner plausibly reach this level in the time and practice available? Create level work on a topic first met that morning is an overpromise. Drop a level or extend the practice.

• Scope cut: does the objective bundle a whole field into one sentence? "Design a marketing strategy" for a two hour workshop hides research, analysis, budgeting, and writing. Split it into the enabling performances and keep only the ones this unit teaches and checks.

• Evidence cut: could you collect proof of this in your setting? If the setting has no lab, no observation, and no product, an objective that needs one of those must change or the setting must.

• Prerequisite cut: does the performance assume knowledge or skill the unit never teaches? Name the prerequisite in the condition ("after completing the dosage basics module") or teach it first.

• Attitude cut: aims about valuing, appreciating, or enjoying something are real educational aims, but they are not graded performances. Keep them as stated course aims if they matter, label them as aims, and do not pretend a quiz measures them.

Rewrite pattern for an overpromise:

• Before: "By the end of the lesson, students will master academic writing."
• After: "Given a marked draft paragraph (condition), students (audience) will revise it to fix topic sentence and citation faults (behavior), correcting at least three of the four faults flagged in the marking guide (degree)." The word master is gone, the performance fits one lesson, and a marker can check it.

5. Align evidence and activities to the verb

Use constructive alignment as a three column check for each objective: objective verb, practice activity, assessment task. All three must require the same process at the same level.

• Objective says design, activity is a lecture, test is multiple choice recall: broken twice. The learner never practiced designing, and the test never asked for it. Fix the activity to a guided design task and the test to a design product judged with the rubric named in the degree.

• Objective says explain, test asks learners to list: the test sits below the objective. Either raise the test to an explanation task or rewrite the objective honestly at Remember level.

• Write the assessment idea at the same sitting as the objective, following backward design. If no assessment comes to mind, the objective is probably vague, unobservable, or out of scope. Fix the objective before planning activities.

• Give learners a practice version of the assessed performance before the graded one. The first time a learner meets a performance should not be the time it counts.

Record the check as a small alignment table, one row per objective, with columns for verb and level, practice activity, assessment, and degree or rubric. A row with an empty cell is unfinished work, not a formatting gap.

6. Run the linter, then read as a learner

• Run the bundled checker on a draft list saved as plain text, one objective per line: python3 scripts/check_objectives.py draft.txt. It flags banned main verbs, missing condition signals, missing degree signals, and multi verb lines. Read references/verb-tables.md for the signal word lists it uses.

• Treat the checker as a prompt for judgment, not a grader. A condition can be real without using one of its signal words. Fix genuine gaps, note deliberate exceptions, and never pad a sentence with filler just to pass a check.

• Read the finished set as a learner would. Could a learner use it to tell what to practice, what counts as done, and how they will be judged? If not, the set fails its main job even if every sentence is well formed.

Output Contract

Deliver, in this order:

1. The objective set: numbered objectives, each ABCD complete, with the Bloom level and knowledge type labeled after each one in brackets, for example [Apply, procedural].
2. The alignment table: one row per objective, with practice activity, assessment task, and degree or rubric.
3. Rewrites shown: for any objective the requester supplied, show before and after side by side, with one line on what changed and why.
4. Cuts and cautions: objectives you split, dropped, downgraded, or relabeled as aims, each with the reason from step 4.
5. Open assumptions: audience, time, setting, or prerequisites you had to assume because the requester did not state them.

Keep the learner facing version clean. If the requester needs a syllabus ready list, give the objectives once without the bracket labels as a separate short block, and keep the labeled version as the working copy.

Operating Rules

• Never ship an objective whose main verb you could not watch, hear, or collect as a product. If nobody could observe it, rewrite it.

• Never attach an invented percentage, time limit, or count as a degree. A criterion must come from the task, a standard, a rubric, or the requester. When none exists, say "criterion to be set" and name who should set it.

• Never raise a verb to a higher Bloom level to make a course sound stronger. The level is set by the task learners will actually do and be judged on.

• Keep frameworks in their lanes. Bloom names the thinking. ABCD shapes the sentence. Alignment checks the fit across objective, activity, and assessment. Backward design sets the order of work. Do not credit one framework with another one's parts.

• Present SMART only as an optional plain language check if the requester uses it. It comes from management writing, not from the education sources behind this skill, and it never overrides an ABCD or alignment failure.

• When the requester wants many objectives fast, produce fewer, finished ones and list the remainder as planned. A short aligned set beats a long set nobody can assess.

• For attitude, values, or appreciation goals, label them aims, keep them out of the graded objective list, and say plainly how, if at all, the course will notice growth in them.