NR-707C · Week 6 of 8 · Equity and subgroup analysis

NR-707C Week 6 Equity and Subgroup Analysis: How to Write It

The short answer

A change that improves an average can widen a gap, and a 256-hour block has enough data and enough time to ask whether yours did. The written task at this stage is a disaggregated look at who the change actually reached and whom it worked for, handled with the care that small subgroups demand. Equity analysis is not a paragraph of intention; it is counts broken out by a stated characteristic, reported with their uncertainty. Your section may print this as NR 707C or NR707C; 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 707C Week 6 grading scale at Chamberlain, the criterion levels this assessment is scored on, from Chamberlain Tutors
How Chamberlain grades NR 707C Week 6, visualized by Chamberlain Tutors.

What NR-707C Week 6 asks for

A video visit program raised completion rates across a clinic panel and the leadership deck said so. Broken out, the picture was different: completion rose sharply among patients under sixty with a portal account and did not move at all among patients who needed a telephone visit because they had no reliable broadband. The program had improved access for people who already had it. Nothing about that finding is unusual, and nothing about it is discoverable without disaggregating. Technology-mediated care makes this pattern especially common, which is why a telehealth-adjacent project should expect to look for it rather than hope it is absent.

The boundary that governs this manual. Practicum hours, hour logs, encounter counts, site paperwork, and preceptor or mentor evaluations are your own record of work you personally performed and are never drafted, reconstructed, or estimated with anyone's help. What can be supported is the written layer: how a subgroup analysis is structured, how a small-cell result is reported responsibly, how an equity finding is stated without overclaiming. Everything on the page is de-identified, and small subgroups need extra care because a cell of three is close to identifying the people inside it.

There are two distinct questions and both belong in the section. The first is reach: did the change get to everybody it was meant to get to, or did enrollment, referral, or eligibility filter the population before your intervention ever appeared. The second is effect: among those it reached, did it work comparably. A project can pass one and fail the other, and the responses differ entirely. A reach problem is fixed upstream, in referral and enrollment. An effect problem is fixed in the intervention itself, usually because it assumed a resource that not everyone has.

Expect the deliverable to be a subgroup or equity analysis inside a late progress document, sometimes with a small table, and often a posted discussion about what disaggregation revealed. If your data does not permit disaggregation, the honest written analysis of why, and what would have to be collected to permit it, is a legitimate version of the same task and a genuinely useful contribution to your site.

The NR-707C Week 6 method, step by step

Six analytic moves for disaggregating responsibly.

  1. 1. Choose the strata before you look at the results

    Name the characteristics that plausibly interact with your mechanism and say why each was chosen. Selecting a stratum after noticing a gap in it converts an analysis into a search, and a reader can usually tell which happened.

  2. 2. Separate the reach question from the effect question

    First report who entered the eligible population and who did not, then report outcomes among those reached. Collapsing the two hides the most actionable finding, which is usually upstream of the intervention entirely.

  3. 3. Report every cell with its denominator

    Percentages computed on four cases are noise wearing a number. Write the counts, and where a cell is too small to report responsibly, say so rather than publishing a figure that identifies people or misleads readers.

  4. 4. Interrogate the mechanism for a hidden requirement

    Ask what your change silently assumes: a device, a data plan, literacy in a particular language, a daytime phone, a caregiver, transport. Named requirements are the usual explanation for a gap, and naming them is the doctoral part of the analysis.

  5. 5. State uncertainty in proportion to the numbers

    A difference between two small groups over a short window is a signal to investigate, not a conclusion. Write it as a pattern worth watching, with the counts visible, and refuse the temptation to describe it as a disparity your data has established.

  6. 6. Convert any finding into a specific upstream or design action

    A telephone pathway with equivalent content, an interpreter step built into the workflow, a referral audit at the clinic that under-enrolls. Equity findings earn their place when they produce an action somebody owns.

A layout for an equity and subgroup section

Our frame for this stage's written work, sized for roughly 1,100 to 1,400 words plus a disaggregated table. 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
Strata and rationaleThe characteristics examined, why each was selected in advance, and where each variable comes from.180 to 220
Reach by stratumWho entered the eligible population and who did not, with counts and the filter that operated.200 to 250
Delivery by stratumAmong those reached, whether the change was delivered comparably, with process counts.180 to 220
Outcome by stratumResults disaggregated, every cell carrying its denominator, with small cells handled explicitly.200 to 250
Hidden requirementsWhat the intervention assumes that not everyone has, and the evidence pointing at each assumption.190 to 230
Actions and ownersUpstream and design responses, each attached to a role, with what would show whether it worked.170 to 210

Evidence craft for equity writing

Say where each variable came from and how good it is. Demographic and social fields in operational systems are frequently incomplete, self-reported inconsistently, or entered by whoever registered the patient. Report the completeness rate for any field you stratify on, because a variable missing in a third of records cannot support a confident subgroup claim.

Describe differences without assigning cause. A gap between groups in your data reflects everything that shaped access before your project began. Write that a difference was observed, name the plausible upstream contributors, and avoid language that locates the cause in the patients rather than in the system that reached them.

Suppress cells that are too small to publish. Reporting an outcome for a subgroup of three is both statistically empty and a re-identification risk in a small setting. Write that the cell was too small to report and describe the direction only if it can be done without exposing anyone.

Ground the section in the published literature on access. Differential uptake of technology-mediated care, language access, and referral pattern variation are all documented phenomena with a scholarly literature. Citing it with years turns your local observation into an argument that a reader can situate.

Hold the improvement frame. This is a local, descriptive look at who your change reached. It does not establish a disparity in the epidemiological sense and it does not generalize. Where your organization requires a determination about human subjects oversight, describe the process you followed and its date rather than characterizing an outcome.

Five mistakes that cost points in this week's territory

  • Equity addressed as a statement of values. A paragraph affirming the importance of equity, with no disaggregated counts, does not perform the analysis.
  • Strata chosen after seeing the data. A gap found by searching many variables is a different and weaker claim than one found by testing a stated expectation.
  • Percentages on tiny cells. Sixty-seven percent of three is a number that should never appear in a doctoral document.
  • Ignoring data completeness. Stratifying on a field that is blank in a third of records produces subgroups that describe documentation habits.
  • A finding with no action. An observed gap with no upstream or design response attached leaves the most important part of the section unwritten.

Before you submit

  • Strata were named in advance with a stated rationale for each
  • Reach and effect are reported as separate questions
  • Every cell carries a denominator, and small cells are suppressed explicitly
  • Completeness of each stratifying variable is reported
  • Hidden requirements of the intervention are named with supporting evidence
  • Each finding is attached to an action with an owner by role

Writing the equity analysis for NR-707C?

Send the rubric and your disaggregated data out of Canvas. A premium original draft comes back in 24 to 48 hours with reach and effect separated, denominators visible, and small cells handled responsibly, and revisions run until the grade lands.

Questions students ask about this stage

My total population is forty people. Is subgroup analysis even possible?
Formal comparison is not, and pretending otherwise would be worse than declining. What remains genuinely valuable is descriptive reach reporting: who was eligible, who was enrolled, who completed, described by the one or two characteristics most relevant to your mechanism, with counts only and no percentages. Even at that scale, patterns such as every patient who needed a telephone rather than a video visit falling out before the second contact are visible and actionable, and reporting them as observations rather than as findings is entirely appropriate. Then write the methodological point explicitly: with this population size, differences of this magnitude cannot be distinguished from chance, and a longer observation period or a pooled analysis across a future block would be needed. That is a mature paragraph and it will be read as one.
Which characteristics should I stratify on?
Choose the ones that plausibly interact with how your change actually works, and justify each in a clause. If your mechanism depends on a patient using an application, then access to a device, connectivity, and comfort with the technology are directly relevant. If it depends on a phone conversation, then preferred language, hearing, daytime availability, and whether the recorded number reaches the patient all matter. Age, insurance type, and geography are frequently available and frequently informative. What you should avoid is stratifying on everything your data contains and reporting whichever split looks interesting, because that is a search rather than an analysis and it produces findings that will not replicate. Two or three well-argued strata, reported completely including the ones that showed nothing, is a stronger section than eight.
I found a gap. Should I fix it inside this block or report it?
Report it in full, and act only within what the remaining time and your authority honestly allow. With two or three weeks left, a well-chosen upstream adjustment is sometimes feasible, such as changing how eligible patients are identified so a referral pattern stops filtering the population, and if you do that, document it as a dated cycle with a prediction like any other. What you should not do is launch a second intervention in the closing weeks and then report it as though it were evaluated, because there will be no data to evaluate it with. The stronger move in most cases is a clear written recommendation with an owner, a specific first step, and a statement of what would need to be measured to know whether the gap closed. Handing your site a well-argued equity finding it can act on after you leave is a real contribution.

Keep going

Online now