NR-669 · Week 3 of 8 · Analyzing post-implementation results

NR-669 Week 3 Analyzing Post-Implementation Results: How to Write It

The short answer

Stage three of NR-669 is the analysis: turning a closed data set into a stated result. The task is to describe what the outcome, process and balancing measures did across the whole project period, choose displays that show change over time rather than two averages, and write the finding in language proportionate to a single-site design. Analysis here means arranging evidence, not arguing about it. Your section may print this as NR 669 or NR669; 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-669 Week 3 grading scale at Chamberlain, the criterion levels this assessment is scored on, from Chamberlain Tutors
How Chamberlain grades NR-669 Week 3, visualized by Chamberlain Tutors.

What NR-669 Week 3 asks for

A recall process for children on stimulant medication needing quarterly follow-up produces a data set that looks encouraging until it is arranged properly. Baseline: thirteen of forty-seven eligible children seen within the target interval. Post-change: thirty-one of fifty-two. Arranged by month, though, the picture is more interesting. The first post-change month barely moved, the second jumped, the third held. That shape is worth more than the summary pair, because it tells the reader when the process actually took hold, and it usually corresponds to something real, such as the point at which the front desk stopped needing to be reminded.

The written work at this stage is normally a results-and-analysis document, and it has to cover all the measures rather than the flattering one. Outcome, process and balancing measures each get their counts. Where you have enough time points, the outcome gets a display across the whole period with the intervention marked. Where you do not, the honest move is to say so and present counts by period without pretending to a trend.

The line that students most often cross here is the one between describing and explaining. This stage produces the sentence a stakeholder audience will hear in the room later, and it must be defensible in that room. Documented follow-up within the target interval rose from thirteen of forty-seven to thirty-one of fifty-two is defensible. The recall process reduced missed follow-up by thirty percent is a causal claim a pre and post design at one site cannot carry. Where a section runs a discussion at this stage, expect it to ask for your headline result. Post it as final copy, since Canvas responses do not reopen.

Where the boundary sits in an analysis stage. The 168 clinical hours, the immersion, the data you personally gathered at the organization and every relationship there are yours and cannot be delegated. Hour logs, encounter counts, census entries, site paperwork, signatures and evaluations completed about you are your own record, never drafted, reconstructed or estimated with help. A manual addresses only the written and analytic layer: how to arrange a result, how to choose a display, how to phrase a finding proportionately. All data on the page is de-identified and aggregated so no individual could be recognized.

The NR-669 Week 3 method, step by step

Six moves for analyzing a closed improvement data set.

  1. Lay out the whole period before you compare anything

    Put every time point in sequence, baseline through post-change, with counts and denominators. Look at the shape before you calculate a difference, because the shape often contains the finding and a summary comparison erases it.

  2. Apply the analysis plan you wrote earlier

    Use the comparison and the display you specified before the data existed. If you now want a different cut, say in the document that you are departing from the plan and why. Undeclared changes of analysis after seeing results are the definition of a compromised finding.

  3. Read the run before you read the difference

    Look for a sustained shift, a trend, or points sitting outside the baseline range. The published rules for interpreting run charts give you a defensible way to say whether the process changed, and they suit small ambulatory data sets far better than tests built for randomized comparisons.

  4. Put delivery beside effect

    Report the process measure in the same paragraph as the outcome. A rise in outcome with high delivery is one finding; a rise with delivery on a third of eligible days is a different and more complicated one that you would rather discover now than in the room.

  5. Check the balancing measure honestly

    Look at whatever you chose to watch for unintended harm and report it whichever direction it moved. If visit length grew or another task slipped, that belongs in the result, and reporting it is what makes the rest believable.

  6. Write the finding in one sentence and stress-test it

    Draft the single sentence you would say to a director, then ask whether the data as arranged actually support it. If the sentence needs a word like reduced or improved to work, check whether it survives being rewritten as rose or was higher, and use the version the design supports.

A layout and word budget for a results and analysis document

Our frame for the analysis write-up, sized for roughly 1,200 to 1,500 words plus figures and tables. It is our own outline rather than anything the university issues, and your scoring guide outranks it wherever they disagree.

SectionWhat belongs in itWord target
Analysis approachThe comparison and display specified in advance, plus any declared departure from that plan with its reason.150 to 190
Sample across periodsEligible encounters by period, exclusions carried from the data account, and any change in volume over time.180 to 220
Outcome measureCounts by time point, the display, and a plain description of the pattern including when it appeared.280 to 330
Process measureDelivery reported beside the outcome, with any pattern in when the intervention did and did not run.200 to 240
Balancing measureWhat was watched and what it showed, with the confidence the data source supports.140 to 180
Statement of findingOne or two sentences of result in language a single-site design can carry, with no explanation attached.110 to 150

Evidence craft for analyzing improvement data

Show variation, do not average it away. Two summary numbers cannot distinguish a genuine shift from ordinary week-to-week swing. Displaying the points across the whole period is both more honest and more persuasive, because a reader can see for themselves that the post-change points sit outside the range the process produced before.

Say when the change appeared, not just that it did. Improvement rarely arrives on the day of implementation. Naming the lag, and the month at which performance stabilized, is one of the more useful things your analysis can offer the organization, because it tells them how long a similar change needs before it can be judged.

Match the statistic to the design if you use one. Where a test is required, choose one appropriate to consecutive observations from one site, state its assumptions, and report the effect in real units alongside it. A p value attached to sixteen encounters persuades nobody who reads the denominator carefully.

Report subgroups only where you planned to look. If one provider or one clinic day carried most of the movement, that is worth stating as an observation with its counts. What it is not is a separate analysis, unless your plan specified it, and repeated slicing until something appears is a recognized error rather than a discovery.

Keep cells large enough to protect people. In small pediatric panels, a monthly cell can contain very few children, and combined with a clinic day or a provider it can identify them. Aggregate until that is impossible and say in the document that you did.

Five mistakes that cost points in this week's territory

  • Two-point comparison. Before and after averages discard the information that makes improvement data interpretable and invite an overstated claim.
  • Analysis changed after seeing the data. Switching comparison or window without declaring it converts a result into an artifact of the choice.
  • Delivery omitted. An outcome reported without its process measure leaves the finding unexplainable in the room where it will be presented.
  • Causal verbs at analysis stage. Reduced, improved and drove all import an explanation the design has not earned.
  • Silence on the measure that moved badly. A balancing measure quietly dropped is the omission most likely to be caught by an audience member who works in the process.

Before you submit

  • Every proportion is accompanied by its counts and denominator
  • The outcome is displayed across the whole period with the change point marked
  • Any departure from the planned analysis is declared with a reason
  • The process measure appears alongside the outcome
  • The balancing measure is reported whichever direction it moved
  • The stated finding uses verbs the design can support

Analyzing NR-669 results this week?

Send the scoring guide and your aggregate data out of Canvas. A premium original draft comes back in 24 to 48 hours with the whole period displayed, delivery reported beside effect, and a finding phrased so it survives the room, and revisions run until the grade lands. Your hours, your site and your logs stay entirely yours.

Questions students ask about this stage

My result improved and then slipped back. What do I report?
All of it, in sequence, because the slip is one of the most valuable things a closing capstone can document. A rise followed by partial decay is the normal life cycle of an improvement that depended on attention rather than on structure, and knowing when the decay started tells the organization exactly what needs building into the workflow to hold the gain. Report the peak, report the decline, give both their counts, and note anything that changed at the site during the decline, such as a staffing change or a competing initiative. Then let the discussion and the sustainability section do the interpretive work. Presenting only the peak is the version most likely to be contradicted by whoever is still working in that process three months later, and in a stakeholder presentation that contradiction happens out loud.
Can I use statistical process control charts, or is that overreach?
If you have enough points and a stable enough process to calculate limits properly, control charts are the right tool for this kind of data and using one well is a strength. If you have six monthly points, calculating control limits from them produces limits nobody should trust, and a simple run chart with the median and the published run rules is both more honest and easier to defend. The decision should turn on the number of points and the stability of the baseline, not on which chart looks more sophisticated. Whichever you choose, name the method, cite the source you followed for the rules, and interpret according to those rules rather than by eye. An audience that includes quality staff will know the difference, and getting this right is one of the clearer signals of graduate-level analytic competence in the whole paper.
How do I analyze results when the intervention only ran on some days?
Report the overall result as planned, then present the delivery pattern as a second layer, and be careful about the temptation to compare days when the intervention ran against days when it did not. That comparison is intuitive and treacherous, because the days it ran are usually the days a particular person was working or the schedule was lighter, so any difference confounds the intervention with everything else about those days. If you present it at all, present it as an observation with the confounding named in the same paragraph. The more defensible framing is that the project achieved partial delivery and that the overall effect reflects that dose, which sets up an honest discussion about what full delivery would require. Reviewers respect that framing; they are unimpressed by a per-protocol comparison presented as if it were clean.

Keep going

Online now