NR-539 · Week 6 of 8 · Designing the evaluation plan

NR-539 Week 6 Designing the Evaluation Plan: How to Write It

The short answer

Evaluation is the half of this course's title that students reach with the least word budget left, and it is the half that separates a plan from a proposal. The stage asks for two evaluations, not one: process measures that tell you whether the program was delivered as designed, and outcome measures that tell you whether anything changed. A plan carrying only the second cannot explain a null result, and saying that in your own paper earns marks. Your section may print this as NR 539 or NR539; 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.

NR-539 Week 6 grading scale at Chamberlain, the criterion levels this assessment is scored on, from Chamberlain Tutors
How Chamberlain grades NR-539 Week 6, visualized by Chamberlain Tutors.

What NR-539 Week 6 asks for

Developmental screening at well-child visits shows why the pairing matters. Suppose a pediatric practice adopts a standardized screening tool at the nine, eighteen and thirty-month visits and, six months later, the referral rate to early intervention has not moved. Three explanations sit on the table and only process data can separate them. The tool may never have been administered at most eligible visits. It may have been administered and scored inconsistently. Or it may have worked exactly as designed and found fewer children than expected. Without a delivery measure, the program's leaders will argue about which of the three happened and will guess. Design the delivery measure now and the argument is settled by a number.

Rubric rows in an evaluation stage tend to name indicators, data sources, timing and analysis. What they are really testing is whether each measure is collectable. A beautifully specified indicator that nobody can extract from any existing system is a fantasy with a definition attached. Every measure you write should carry the same four things: what exactly is counted, from what source, by whom, and how often. If any of those four is missing, the measure will not survive contact with a real program.

Deliverables here typically include an evaluation matrix or measurement table, a narrative describing design and analysis, and sometimes a data collection procedure. Some sections run a discussion about evaluation challenges. If yours does, write about one measure whose data source is genuinely awkward and how you would work around it, and treat it as final copy, since posts do not reopen after submission in Canvas.

The NR-539 Week 6 method, step by step

Six moves for building an evaluation someone could actually run.

  1. Return to your logic model and measure one link at a time

    Each column in the model earns at least one indicator: inputs consumed, activities delivered, outputs produced, short-term change, outcome. Measuring only the last column is how programs lose the ability to explain themselves.

  2. Define every indicator as a fraction with both halves stated

    Numerator, denominator, inclusion window. Children aged nine to thirty months with a completed screening documented, divided by children in that age band with a well-child visit in the same period. Ambiguous denominators are the most common defect in these tables.

  3. Attach a real data source to each indicator before keeping it

    Existing report, chart field, pharmacy feed, sign-in sheet, brief survey. If the source is a new data collection, price the collection burden and say who does it, because unbudgeted data collection is the first thing a busy site abandons.

  4. Set the baseline and the collection schedule in the same sentence

    What the measure reads before launch, when it will be read again, and how many readings you get inside the program window. Programs with a single post-launch reading cannot distinguish a trend from a fluctuation.

  5. Choose a comparison and name what it cannot rule out

    Pre and post at one site, a matched clinic, a staggered rollout across offices, a seasonal comparison. State which threats your comparison leaves open, since a design honestly limited outscores a design overclaimed.

  6. Write the decision rule before you have any data

    Say in advance what result would lead to spread, to modification, or to stopping. A decision rule written before results is evaluation; a conclusion written after them is commentary.

A layout and word budget for an evaluation plan

Our frame for the evaluation layer, sized for roughly 1,200 to 1,500 words with an indicator matrix carrying the definitions. It is our own outline rather than anything the university issues, and your week's scoring guide outranks it wherever they disagree.

SectionWhat it must nail downWord target
Evaluation questionsThe two or three questions the evaluation exists to answer, phrased so that a number could answer them.120 to 150
Process measuresDelivery, fidelity and reach indicators with numerators, denominators and sources.230 to 270
Outcome measuresShort-term and, where honest, intermediate indicators, each with a baseline value and a target.230 to 270
Design and comparisonThe comparison being used, why it is feasible here, and the alternative explanations it leaves standing.200 to 240
Data handling and analysisWho extracts, who checks, how the data are stored and protected, and what analysis is planned.190 to 230
Decision rules and reportingWhat each range of results would trigger, and who receives the findings on what schedule.170 to 200

Evidence craft for evaluation design

Borrow indicator definitions rather than inventing them. National measure sets, quality programs and published evaluations define numerators and denominators precisely, and adopting an established definition makes your results comparable to something. Cite the source of the definition in the table, not only in the narrative, so a reader can check the specification you used.

Say where the baseline number came from and how old it is. A baseline pulled from a twelve-month look-back on an existing report is credible. A baseline stated with no origin invites the reader to assume it was chosen to make the target reachable. Give the window and the extraction method in one clause.

Keep statistical language honest and modest. Small program evaluations rarely support strong inference. Describe direction and magnitude, use run charts or simple comparisons where they are the appropriate tool, and avoid implying causation from a single-site pre and post design. Precision about what the design cannot show is graded favourably in this course.

Address privacy in operational terms. Say which fields are identifiable, who has access, where the working file lives, and when identifiers are removed. Human subjects considerations for a program evaluation differ from those for research, and stating which one your project is, and why, is a mark of graduate command.

Five mistakes that cost points in this week's territory

  • Outcome measures only. Without delivery data, a flat result is uninterpretable and the program cannot be defended or fixed.
  • Indicators without denominators. Number of children screened rises simply because more visits happened, and tells you nothing about performance.
  • Data sources that do not exist. An indicator requiring a field nobody records is a measure the program will silently drop in week two.
  • Satisfaction standing in for effect. A well-liked program that changed nothing is a real outcome, and satisfaction alone cannot detect it.
  • No decision rule. Results without a prior rule for interpreting them turn evaluation into after-the-fact justification.

Before you submit

  • At least one indicator exists for each column of the logic model
  • Every indicator states a numerator, a denominator and a time window
  • Each measure names an existing data source or a costed new collection
  • Baselines carry their look-back period and extraction method
  • The comparison design is named along with what it cannot rule out
  • Decision rules are written before any result is discussed

Designing the evaluation for NR-539?

Send the prompt, the measurement template and the rubric out of Canvas. A premium original draft comes back in 24 to 48 hours with process and outcome indicators fully specified and decision rules written in advance, and revisions run until the grade lands.

Questions students ask about this stage

How many indicators is the right number?
Fewer than most drafts contain. Four to six well-specified measures beat a dozen half-defined ones, because every indicator carries a collection cost that someone has to pay every week. A workable set usually looks like two process measures covering delivery and reach, one or two short-term outcome measures, one balancing measure that would reveal harm or unintended burden, and at most one longer-run outcome you acknowledge will not move inside the window. If your table runs past eight rows, ask which of them anybody would actually act on, and cut the ones with no decision attached.
Is my project research or quality improvement, and does it matter for the paper?
It matters, and one clear paragraph handles it. Program evaluation intended to improve a local service and not to produce generalizable knowledge is usually handled through an organization's improvement pathway rather than as human subjects research, while a project designed to publish generalizable findings normally requires formal review. Say which yours is, give the reason in a clause, and note that the determination belongs to the reviewing body rather than to the writer. That sentence shows you know the distinction exists without overstating your authority to settle it, which is exactly the register graduate readers expect.
What do I do about an outcome that cannot possibly move in the program window?
Keep it, label it, and put a proxy in front of it. Write the long-run outcome as the program's purpose, state honestly that it will not be detectable inside the evaluation period, and then name the short-term measure that stands in for it along with the published basis for treating it as a reasonable proxy. That structure is more persuasive than either dropping the outcome or pretending it will move. It also protects your final synthesis, because the closing stage of the course will ask what your results would mean, and a plan that promised an impossible movement has nothing sensible to say there.

Keep going

Online now