NR-575 · Week 6 of 8 · Measuring whether the capstone worked

NR-575 Week 6 The Evaluation Plan: How to Write It

The short answer

The evaluation section is the row most capstones lose, and they lose it the same way: several hundred words about the importance of evaluation and one vague sentence about monitoring outcomes. What earns the band is arithmetic. Name the measure, define it operationally down to what counts and what does not, say where the data physically lives, say who pulls it and how often, state the baseline and the threshold that would count as improvement, and name what else could produce the same change. Your section may print this as NR 575 or NR575; it is the same course. Chamberlain publishes no syllabi outside Canvas. The placement here is our teaching judgment from the course's catalog arc; your section's rubric decides what your week actually asks. Clinical hours, logs and preceptor evaluations remain your own record and are never drafted, reconstructed or estimated with help.

NR-575 Week 6 grading scale at Chamberlain, the criterion levels this assessment is scored on, from Chamberlain Tutors
How Chamberlain grades NR-575 Week 6, visualized by Chamberlain Tutors.

What an evaluation plan has to make measurable

Sit in on a quality meeting where somebody presents a rate and watch the first three questions. Out of what? Since when? Who pulled it? A measure that cannot answer those on demand is not a measure, it is an impression, and every well-run audit dies at that table. Your evaluation section is written to survive the same interrogation from a grader who cannot see your unit and will judge entirely on whether the definitions hold.

Three families of measure belong in a capstone plan and each does a different job. Process measures say whether the thing was done: the proportion of eligible encounters where the tool was completed. Outcome measures say whether it mattered: time to reassessment, unplanned transfers, escalations that met criteria. Balancing measures say whether you broke something else while fixing this: documentation time, alert volume, delayed care elsewhere. A plan with only outcome measures cannot tell a failed intervention from an intervention nobody used, which is the single most useful thing an evaluation can distinguish.

Operational definition is where the marks concentrate. Time to reassessment is not a measure until you say what starts the clock, what stops it, which encounters are eligible, which are excluded and why, and what happens to the records where the field is blank. Two competent people applying your definition to the same twenty charts should produce the same number. If they would not, the definition is not finished, and a grader with a scoring guide will see the ambiguity before you do.

The boundary holds here too, and this stage is where it matters most in practice. Writing a plan for how data would be collected is coursework. Actually extracting patient data, auditing real records for a project, or collecting anything from staff requires your site's own approvals and your program's guidance, and those are yours to obtain and yours to conduct. Any illustrative figure that comes from your own precepted observation must be de-identified and labelled as your own local observation rather than presented as institutional data.

Deliverables at this stage are typically the written evaluation section of the capstone, sometimes with a measure table or a data collection instrument in an appendix. Where a discussion runs alongside, write it as final copy; posts do not reopen after submission in Canvas.

The NR-575 Week 6 method, step by step

Six moves that turn intended outcomes into a plan somebody could execute.

  1. Write the measure sentence before anything else

    The proportion of eligible transfers with a documented reassessment within the defined interval, drawn from the electronic record, reported monthly. If you cannot write that sentence, the problem is upstream in your product, not in this section.

  2. Define numerator, denominator, exclusions and missing data

    Say exactly what qualifies for each, then write the exclusions with reasons. Missing data needs its own rule, because in real audits blanks are the largest single category and how you count them changes the result more than the intervention does.

  3. Choose at least one process and one balancing measure

    The process measure tells you whether the tool was used. The balancing measure tells you what it cost. Together they let you interpret a null outcome result, which is otherwise uninterpretable.

  4. Name the data source and the person who touches it

    Which system, which report, which role runs it, at what interval, and what happens to the file afterwards. Data that would require a build request nobody has approved is a feasibility problem worth stating rather than assuming away.

  5. State a baseline and a threshold in advance

    What the measure reads now, or how you would establish it, and what value would count as meaningful improvement. Setting the bar before the data arrives is what separates evaluation from justification after the fact.

  6. Write the rival explanation paragraph

    Seasonal variation, a concurrent initiative, a staffing change, the attention the project itself created. Name the two most likely and say how you would tell them apart from a real effect.

A layout and word budget for the evaluation section

Our frame for the evaluation section of a capstone, sized for roughly 1,000 to 1,300 words with the measure table in an appendix. It is our own outline rather than anything the university issues, and your week's rubric outranks it wherever the two disagree.

SectionWhat belongs in itWord target
What success would look likeThe change you are claiming the product would produce, stated as a direction and a magnitude before any method appears.90 to 120
Primary outcome measureOperational definition with numerator, denominator, eligibility, exclusions and the rule for missing data.250 to 300
Process and balancing measuresAt least one of each, defined to the same standard, with what each would tell you that the outcome cannot.210 to 260
Data source and collectionSystem, report, role responsible, interval, storage and how identifiers are handled.170 to 210
Baseline, threshold and windowCurrent value or how it would be established, the threshold for meaningful change, and how long you would look.170 to 210
Rival explanationsThe two most plausible alternative causes and how you would distinguish them from an effect of the product.160 to 200

Evidence craft for an evaluation plan

Borrow measures that already have definitions. Where a nationally reported or professionally endorsed measure exists for your outcome, use its published specification and cite it. Inheriting an established definition is both less work and more defensible than inventing one, and it lets your result be compared with something.

Report every proportion as a fraction with a window. Nine of twenty over one quarter, not forty-five percent. In an evaluation section the denominator is not decoration; it is the thing that tells a reader whether a change of two cases means anything at all in your volume.

Say plainly what your design can and cannot support. A before-and-after comparison on one unit is the most common capstone design and it cannot exclude concurrent change. Write that sentence yourself rather than leaving it for the grader, then say what you would do to strengthen inference: more baseline points, a comparison unit, a balancing measure that would move if the cause were something else.

Match the statistical claim to the volume you actually have. With small numbers, run charts and simple proportions communicate more honestly than significance testing, and saying so demonstrates judgment. Claiming statistical power you could not have is a marked error in graduate work.

Handle identifiers in the plan itself. State that collection would be aggregate, that any record-level extract would be de-identified, and where such data would be stored. Evaluation sections that describe pulling patient-level information with no word about protection lose credibility with graders who work in the same systems.

Five mistakes that cost points at this stage

  • Satisfaction as the primary outcome. Staff liking a tool is a process signal at best. If your only measure is a survey, the plan cannot detect whether anything changed for patients.
  • Definitions that two auditors would apply differently. Improved documentation and timely reassessment are phrases, not measures, until eligibility and timing rules are written.
  • No baseline and no threshold. Without both, any result can be narrated as success, and a grader scoring rigour will read it exactly that way.
  • Silence on missing data. Blanks are the largest category in most real audits and the rule for counting them changes the answer more than the intervention does.
  • Rival explanations unmentioned. An evaluation that treats its own intervention as the only possible cause of change has not evaluated anything.

Before you submit

  • The primary measure has a numerator, a denominator, eligibility rules and exclusions
  • There is at least one process measure and one balancing measure
  • The rule for missing or blank data is written down
  • A data source and a responsible role are named for each measure
  • Baseline, improvement threshold and observation window all appear
  • Two rival explanations are named with a way to distinguish them
  • Identifier handling and storage are addressed in one explicit sentence

Writing the NR-575 evaluation plan?

Send the rubric and your product and implementation drafts out of Canvas. A premium original draft comes back in 24 to 48 hours with operational definitions two auditors could apply identically and rival explanations handled, and revisions run until the grade lands.

Questions students ask about this stage

I have no baseline data and no way to get any. Can I still write this section?
Yes, and the way you handle it is itself gradeable. Write a baseline establishment step into the plan: what would be measured, over what period before launch, using which source, and by whom. Where a published benchmark exists for comparable settings, cite it and say explicitly that it is a proxy rather than a local value. What loses marks is inventing a starting figure or writing the section as if a baseline already existed. Naming the absence and designing around it shows the grader you understand what a comparison requires, and it also protects you from the harder question at the intensive, which is where a confident-sounding number came from.
How long should the evaluation window be?
Long enough for the outcome to occur often enough to be countable, and long enough for adoption to stabilize past the initial attention effect. Those two requirements usually push past a single eight-week session, which is fine, because most capstone evaluations are plans rather than completed studies. Say what you would look at in the first month, which is almost always the process measure, and what you would look at at three and six months, which is the outcome. Naming the different questions each window answers is stronger than picking a single duration, and it gives the sustainment argument in your implementation section something concrete to attach to.
Does an evaluation plan for a capstone need ethics approval language?
It needs an honest paragraph about the distinction, and the distinction is one your faculty and site determine rather than you. Quality improvement conducted for internal purposes and research intended to produce generalizable knowledge are treated differently by organizations, and the same project can be classified either way depending on intent and dissemination. Write which you believe yours is, say what determination process would apply at your organization, and note that the determination is not yours to make unilaterally. Add the practical protections regardless: aggregate reporting, de-identified extracts, restricted storage. That paragraph is short, it is frequently scored, and it is one of the easiest to omit entirely.
What if the honest answer is that the product probably would not change the outcome much?
Then write the plan that would detect a small effect and say so. Capstones are not graded on optimism; they are graded on rigour, and a plan that names a modest expected effect, chooses a measure sensitive enough to see it, and explains why a small improvement still matters clinically is more sophisticated than one promising a dramatic change. It also sets up the strongest version of the process measure argument: if the outcome moves little but the tool is used consistently and the balancing measure stays flat, you have learned something real about where the leverage actually sits. Write that reasoning explicitly; it reads as judgment rather than as hedging.

Keep going

Online now