NR-723 Week 4 Direct and Indirect Measures: How to Write It

What NR-723 Week 4 asks for

A critical care education team once presented a year of residency data to its practice council. The slides showed that 94 percent of residents rated themselves confident managing a deteriorating patient at the end of the program, up from 41 percent at entry. It was a genuinely large change and it was completely uninformative, because nothing in the data set showed a single resident recognizing deterioration. Confidence is a perception. Recognition is a behavior. The program had spent a year measuring how residents felt about an ability it never once observed, and the gap between those two things is the whole subject of this stage.

Direct measures require a performance and a judgment. A scored simulation of an unstable postoperative patient, a written care escalation plan evaluated with a rubric, a medication calculation examination, a structured observation of a bedside handoff against defined criteria, a portfolio artifact judged by two raters. What makes them direct is that a person did something and someone applied criteria to it. Indirect measures include learner satisfaction surveys, self-rated confidence, exit interviews, employer perception surveys and preceptor impressions gathered without a defined instrument. They are cheap, fast, and easy to over-read.

The written work at this stage usually asks you to select or design measures for outcomes and to defend the selection. That defense is the graded part. For each measure you should be able to say what it produces, who scores it, against what criteria, when it is collected, where the data is stored and what its known weaknesses are. Measures chosen because they already exist are perfectly legitimate, provided you say so and treat the convenience as a limitation rather than pretending the choice was ideal.

The last thing this stage asks is a proportion argument. An evidence set that is entirely direct is expensive and blind to experience; one that is entirely indirect proves nothing. Doctoral writing here should state deliberately how the balance was struck for each outcome and why, without inventing a rule about how many of each a program must have. There is no published quota to cite, and claiming one would be a fabricated requirement in a document whose value depends on being checkable.

The NR-723 Week 4 method, step by step

Seven moves for building an assessment set that can defend itself.

1. Split the rubric into selection rows and justification rows

   Choosing measures is quick. Defending them against alternatives is the work that carries the weight. Find out which rows expect a comparison of options before you commit your word budget to describing a single instrument in detail.

2. Start from the outcome's verb and ask what it would look like

   If the outcome says prioritize, the measure must present competing demands and record what the learner chose. If it says lead, the measure must involve other people. Working from the verb prevents the most common failure, which is attaching a knowledge test to a performance outcome.

3. Classify every candidate measure before you evaluate it

   Label each one direct or indirect, and formative or summative, in a table. The classification does analytic work: it shows instantly whether your evidence for an outcome rests on performance or on perception, and it is the fastest way to find the outcomes with no direct evidence at all.

4. Name the scoring instrument, not just the activity

   A simulation is not a measure. A simulation scored by two trained raters using a defined performance checklist is. Say which instrument, whether it is published or locally built, and what evidence exists for its consistency.

5. Decide who scores and how disagreement is handled

   An evaluator who also taught the content sees what they hoped to see. Say who rates, whether any blinding is possible, whether two raters are used on a sample, and what the plan is when raters diverge. This single paragraph separates serious assessment writing from wishful assessment writing.

6. Fix collection points to the sequence, not to intentions

   State the term, module or week where each measure is gathered, tied to the map you built earlier. Ongoing and as needed are not collection schedules, and an assessment plan that cannot say when data arrives will not produce any.

7. Write the weakness of each measure in a clause

   Every instrument has one: small numbers, rater variability, self-selection into an elective, a survey answered by the satisfied, a simulation that rewards familiarity with the manikin. Naming it is what makes the plan defensible; leaving it out is what invites the grader to name it for you.

A layout and word budget for an assessment measure plan

Our frame for a measure selection document with defense, sized for roughly 1,000 to 1,300 words of prose beside the table. It is our own outline rather than anything the university issues, and your week's rubric outranks it wherever the two disagree.

Evidence craft for assessment selection

Prefer an instrument with published evidence, and say what that evidence is. When a validated tool exists for the ability you are measuring, name it with its authors and year and state briefly what its reliability evidence looks like. A locally built checklist is acceptable and often necessary, but it should be labelled as local rather than presented as though it arrived with psychometric backing.

Report counts with their denominators. Twenty-two of 34 residents met the escalation criteria on the summative scenario is evidence. Sixty-five percent is a figure a reader cannot weigh, because it hides whether the cohort was 34 people or 340.

Keep satisfaction out of the outcome column. Learners can be delighted by a program that taught them nothing measurable and frustrated by one that changed their practice. Report perception data as perception data, in its own row, with that limitation stated in the sentence rather than in a footnote.

Do not promise psychometric properties you have not established. Writing that a locally developed rubric is valid and reliable is an unsupported claim. Writing that the rubric will be piloted with two raters on a sample of ten artifacts, with agreement examined before full use, is a plan, and plans are what this document is for.

Attribute your assessment framework. The direct and indirect distinction, the formative and summative distinction and the criteria for quality assessment all come from published assessment literature. Cite the source you are working from with its year so a reader can check your categories against a standard rather than against your judgment.

Five mistakes that cost points in this week's territory

• Confidence used as competence. Self-rated confidence is the most collected and least probative data in nursing education, and treating it as outcome evidence is the defining error of this stage.
• An activity named where an instrument is needed. A capstone project, a simulation, a preceptor conversation. None of them is a measure until criteria and a scorer are attached.
• No collection schedule. Measures without a stated term, module or week produce no data, and the plan will not survive a single cycle.
• Silence about who scores. Rater identity and rater training decide how much a direct measure is worth, and skipping them leaves the strongest part of the plan undefended.
• Invented validity claims. Calling a locally built tool validated is a fabricated fact, and it undermines every other claim in the document.

Before you submit

• Every outcome has at least one direct measure attached
• Each measure is labelled direct or indirect and formative or summative
• The scoring instrument is named for every direct measure, with local tools identified as local
• Scorers, rater training and a consistency check are described
• Each collection point is tied to a specific term, module or week
• Every measure carries a stated weakness and the plan carries an honest burden estimate

Questions students ask about this stage

Keep going