NR-702A Week 7 The Evaluation Plan: How to Write It

What NR-702A Week 7 asks for

What is the difference between a measure and a metric people already collect? A mobile health unit adding a standing referral step could measure the number of referrals generated, which is easy and available, or the proportion of eligible clients who reached a first appointment, which is harder and is what anyone actually cares about. Evaluation writing at this stage is largely the discipline of choosing the measure that answers the question over the measure that is convenient, then being honest in writing about what the convenient one would and would not have told you. That trade-off, argued explicitly, is a doctoral move.

The second demand is the three-measure structure that improvement work uses. An outcome measure says whether the thing you care about changed. A process measure says whether the change was actually delivered, which is what separates a failed intervention from an intervention that was never really implemented. A balancing measure watches for harm somewhere else: longer visits, a task displaced onto another role, a delay in something that was working. Students routinely write outcome measures alone, and then cannot explain a flat result. Process and balancing measures are what make a null finding interpretable rather than embarrassing.

Third, the evaluation has to fit 128 hours and the sequence that follows. A plan that requires twelve months of monthly data is not wrong, but it belongs to a project timeline that extends beyond this course, and the writing should say which portion falls inside your practicum courses. Equally, an evaluation resting on a single before-and-after comparison of two weeks is thin. Where the volume allows, plotting the measure over consecutive weeks tells a far richer story than two aggregate numbers, and it protects you against the ordinary swings a small denominator produces.

The boundary that governs every page in this manual. Practicum hours, hour logs, encounter counts, site documentation, preceptor evaluations and signatures are the student's own record and are never drafted, reconstructed or estimated with help from anyone. No tutor collects, extracts, or documents clinical data. Support belongs to the written layer: defining a measure in prose, structuring an evaluation plan, and making the analysis approach legible to a practice audience. Every patient detail in your writing must be de-identified, and small-denominator results should be reported at a level that cannot expose an individual patient or staff member.

The NR-702A Week 7 method, step by step

Seven moves for writing an evaluation plan a committee will accept and a site could run.

1. Restate the primary outcome in the baseline's exact words

   Copy the operational definition from your baseline section rather than rewriting it. Any difference in wording between the two, however small, will produce a comparison someone can challenge.

2. Add a process measure that proves delivery

   The proportion of eligible encounters where the step actually happened. Without it, a flat outcome is uninterpretable, because you cannot distinguish an ineffective change from one that was never consistently delivered.

3. Choose one balancing measure and say what harm it watches

   Visit length, a competing task's completion rate, wait time, or staff-reported burden. Name the specific unintended consequence you consider most likely, then measure that rather than a generic satisfaction score.

4. Write a data collection line for every measure

   Source, who pulls it, how often, where it is stored, and in what form. A measure without a named collection route is an intention, and intentions do not produce data during a busy clinic week.

5. Set the observation rhythm before the analysis

   Weekly or biweekly points across the implementation window if volume allows, because the pattern over time is what improvement work reads. Say how many eligible encounters each point is expected to contain.

6. State the analysis in terms suited to improvement data

   Proportions with denominators, plotted over time, with a stated rule for what would count as a signal rather than noise. If you intend a statistical comparison, say what test and why, and keep the claim proportionate to the sample.

7. Write in advance what different results would mean

   What you would conclude if the outcome moved, if it did not move but the process measure was high, and if the process measure was low. Committing to interpretations before you have data is the strongest evidence of methodological seriousness in the whole plan.

A layout and word budget for an evaluation plan

Our frame for an evaluation deliverable, sized for roughly 1,200 to 1,500 words. It is our own outline rather than anything the university publishes, and your week's rubric outranks it wherever they disagree. If a measures table is required, the prose paragraphs shorten and the definitions live in the rows.

Evidence craft for evaluation writing

Cite an improvement methodology rather than a trial methodology. Quality improvement has its own literature on measurement, run charts and reporting standards, and drawing on it signals that you know which kind of study you are running. Borrowing the vocabulary of clinical trials for a site-level project is the most common category error at this stage.

Report proportions with both numbers, every time. Nineteen of 63 eligible encounters, not thirty percent. In small denominators the difference between one and two additional cases can look like a large percentage swing, and the base is what allows a reader to see that.

Keep the language of causation proportionate. A before-and-after comparison at one site supports language about association and about change coinciding with implementation. It does not support caused or proved. This restraint is scored, and overreach here undoes an otherwise careful plan.

Say how data will be handled and protected. Where it lives, who has access, how identifiers are removed, and what happens to the working file afterward. Even where a project sits outside formal research oversight, the expectation of careful data stewardship does not relax.

Describe the review pathway without predicting its outcome. State what determination process your program and site use and what you will submit. Do not assert in advance that a project will be classified in a particular way or that approval will be granted, because those decisions belong to the bodies that make them.

Five mistakes that cost points in this week's territory

• An outcome measure with no process measure beside it. A flat result becomes uninterpretable, and the plan cannot distinguish a weak intervention from one never delivered.
• Measures that drift from the baseline definition. A single changed word between sections invalidates the comparison the whole project rests on.
• Statistical machinery on a tiny denominator. Significance testing applied to a handful of encounters produces numbers with more confidence than the data can carry.
• No balancing measure at all. Improvement work that never asks what got worse elsewhere reads as advocacy rather than evaluation.
• Collection with no owner. A plan that says data will be collected weekly, without naming who does it and from where, will not survive the second week of a clinic schedule.

Before you submit

• The outcome definition matches the baseline definition word for word
• A process measure of delivery is defined with a fidelity threshold
• One balancing measure names the specific harm it watches for
• Each measure has a source, an owner, a frequency and a storage location
• The display method and the signal rule are stated before any analysis is described
• Causal language is proportionate to a single-site before-and-after design
• Data handling and de-identification are described explicitly
• Interpretation of at least three possible result patterns is written in advance

Questions students ask about this stage

Keep going