| Course | HS840 Advanced Health Research and Analysis Methods |
|---|---|
| Unit | Unit 4 |
| Paper type | Design justification |
| Length | About 1,100 words, 6 pages double-spaced |
| Format | APA 7 student paper |
| School | Purdue University Global |
| Program | Doctor of Health Science (DHSc) |
| Updated | October 2026 |
Free sample paper for HS840 Unit 4
Four Designs, One Rollout: A Unit 4 Design Justification for a Stepped-Wedge Trial of Same-Day Contraceptive Insertion
[Student Name]
Purdue University Global
HS840: Advanced Health Research and Analysis Methods
Unit 4 Design Justification
[Instructor Name]
[Date]
The network, its constraints and its schedule are composite; the methods literature cited is real.
The Question and the Conditions Any Design Must Meet
The question settled in the critique unit asks whether staggered access to same-day insertion raises the share of requesting patients who have their device placed within 30 days, measured against today's two-visit routine. Four conditions bind any design.
1. Every clinic will receive same-day service during the year; network leadership will not withhold it from any clinic beyond the rollout period.
2. Training capacity limits starts to two clinics every two months, because one clinician trainer must observe supervised insertions at each site.
3. The outcome comes from the electronic health record, where requests and insertions are recorded continuously for all patients.
4. A new contraceptive formulary takes effect network-wide in month five, which could change insertion rates on its own.
Each candidate design is judged on four criteria: protection against confounding, fit with the rollout, contamination risk and analytic burden.
Candidate One: Individual Randomization Within Clinics
Patients requesting a device could be randomized to same-day insertion or the usual pathway. This offers the strongest protection against confounding at the patient level. It fails on contamination and on fit. Same-day service depends on stocked devices, trained staff and a scheduling template that changes how the whole clinic runs; once a clinic can insert the same day, clinicians will find it hard to send a randomized patient home to return later, and patients talk to each other. It also asks clinicians to refuse a service they can deliver, which would erode trust in the study. Rejected for contamination.
Candidate Two: A Parallel Cluster Randomized Trial
Six clinics could be randomized to start same-day service and six to continue the usual pathway for the year. This protects against confounding by clinic and avoids contamination between arms. It fails the first condition: leadership will not hold half the clinics back for a full year, and training capacity would not allow six clinics to start at once anyway. A parallel trial over a shorter period would leave too few months of follow-up. Rejected for incompatibility with the rollout, although it would be the stronger design in a network free to hold clinics back.
Candidate Three: An Interrupted Time Series
Each clinic's monthly insertion proportion could be tracked before and after it starts same-day service, without randomizing the order. This fits the rollout and uses the continuous outcome data well. Its weakness is the formulary change in month five. Without randomized timing, a time series cannot separate the effect of same-day service from that of the formulary change, or from any other change that happens to coincide with a clinic's start. If leadership chose which clinics go first, the earliest clinics would likely be those most eager, which would bias the comparison further. Rejected for vulnerability to concurrent change and selection of order.
A design that cannot separate the intervention from a formulary change scheduled for the middle of the study would leave the network unable to say what worked.
Candidate Four: The Stepped-Wedge Cluster Randomized Design
In a stepped-wedge design, all clusters begin in the control condition and cross over to the intervention at randomized times until all are exposed (Hussey & Hughes, 2007). Here, after a two-month baseline in which no clinic offers same-day service, two clinics cross over at the start of each subsequent two-month period, so that by the end of the seventh period all twelve clinics have done so. The order of crossover is randomized.
Protection against confounding. Because the order is random, clinic characteristics such as eagerness, size and patient mix are balanced across steps on average. Comparisons are made both within clinics, before and after crossover, and between clinics at the same time point, one exposed and one not.
Fit with the rollout. Exact. The design uses the schedule leadership already intends, constrained by training capacity, and only randomizes its order.
Contamination. Low, since entire clinics change at once.
Analytic burden. Higher than the alternatives. Because the proportion of exposed clinic-periods rises over time, calendar time is correlated with exposure, and any secular trend, such as the formulary change, must be modeled with period effects to avoid confounding (Hemming et al., 2015). This is the design's main cost, and it is a cost the analysis can bear.
The stepped-wedge design is therefore chosen.
Randomizing the Order
The crossover order will be generated by a statistician outside the clinics, using a computer-generated sequence, and revealed to each pair of clinics two months before their step so that training can be scheduled. Simple randomization of twelve clinics could, by chance, send the four largest clinics into the last two steps, which would leave the early intervention periods dominated by small clinics with few requests. The sequence will therefore be constrained: clinics are grouped into large and small by their prior-year request volume, and each step pairs one large and one small clinic. Constrained randomization of this kind keeps the design balanced without giving anyone a choice over which clinics go first. The sequence and the constraint will be recorded before the baseline period begins, so that the analysis cannot be accused of fitting the schedule to the results.
The Costs Named
Time confounding. The analysis must include a fixed effect for each two-month period. The formulary change in month five will be absorbed by period effects as long as it affects all clinics similarly.
Duration. The study runs fourteen months, longer than a parallel trial of the same size would need.
Analysis. A mixed model with period effects and random effects for clinic is required, and possibly a random effect allowing the clinic effect to vary over time (Li et al., 2021), which will be specified in the analysis plan.
Ethics. Some clinics wait up to twelve months for a service that leadership believes beneficial. That wait is set by training capacity, not by the study, so the design adds no delay that would not have happened anyway.
What the Design Cannot Answer
The design answers whether access to same-day insertion increases completed insertions at the clinic level. It cannot tell which component, devices on site, trained staff or a changed template, drives the effect, and it cannot measure effects on unintended pregnancy, which would need longer follow-up. Chapter 3 will name both limits plainly.
References
Hemming, K., Haines, T. P., Chilton, P. J., Girling, A. J., & Lilford, R. J. (2015). The stepped wedge cluster randomised trial: Rationale, design, analysis, and reporting. BMJ, 350, h391. https://doi.org/10.1136/bmj.h391
Hussey, M. A., & Hughes, J. P. (2007). Design and analysis of stepped wedge cluster randomized trials. Contemporary Clinical Trials, 28(2), 182-191. https://doi.org/10.1016/j.cct.2006.05.007
Li, F., Hughes, J. P., Hemming, K., Taljaard, M., Melnick, E. R., & Heagerty, P. J. (2021). Mixed-effects models for the design and analysis of stepped wedge cluster randomized trials: An overview. Statistical Methods in Medical Research, 30(2), 612-639. https://doi.org/10.1177/0962280220932962
What the HS840 Unit 4 assignment asks for
This unit typically asks learners to justify the research design for their planned study by comparing it with plausible alternatives. Five to six pages is usual. The paper should restate the research question, state the practical and ethical conditions the design must meet and then evaluate each candidate against the same criteria. Instructors look for alternatives a competent researcher might actually propose, not straw options. Each rejection needs a reason tied to the setting. The chosen design's weaknesses should be named along with how the study will manage them. Methods literature, such as foundational papers on cluster or stepped-wedge trials, should support claims about each design's strengths and limits.
How this HS840 Unit 4 example is built
A numbered list sets out four binding conditions, then four criteria for judging designs. Individual randomization is rejected because same-day service runs through shared staff, devices and templates. A parallel cluster trial is rejected because leadership will not withhold service for a year, though the paper notes it would be stronger elsewhere. An interrupted time series is rejected because the month-five formulary change and self-selected order would confound it. The stepped-wedge design is then assessed on all four criteria. A section on randomizing the order pairs one larger and one smaller clinic at each step. The paper closes by naming time confounding, duration, analytic burden and waiting clinics, and what the design cannot answer.
HS840 Unit 4 rubric: what earns full points
Rubrics for design justification papers usually look at how clearly the question and constraints are set out, the fairness and depth of the comparison, the soundness of the final choice, recognition of limitations and use of methods sources. The comparison often carries the most points, because the unit is about reasoning between designs; rubrics reward applying identical criteria to every option. Soundness is judged on whether the chosen design actually fits the stated constraints. Limitations earn credit when the paper names the design's real costs, such as secular trends in stepped-wedge studies, and explains how the analysis will handle them. Citing methods papers rather than general textbooks strengthens the evidence score. Organization and APA style make up the rest.
HS840 Unit 4 help: mistakes that cost points
Weak justifications usually describe the chosen design in detail and dismiss alternatives in a sentence each. Another is rejecting designs for reasons that do not apply to the setting, such as cost, without explanation. Some learners choose a stepped-wedge design because it sounds sophisticated, without showing the rollout justifies it. Others forget to address concurrent changes during the study period. Leaving out how the crossover order is randomized weakens the paper. If your study has different constraints, perhaps a single hospital unit or a community program, send your question and setting details, and a justification can be built that tests realistic designs against them.
Get an HS840 Unit 4 example written to your instructions
Setting constraints decide design choices, so describe yours, including anything leadership has ruled out, alongside the question and the Unit 4 prompt and rubric. Conditions come first in the justification, each candidate design is tested against them, and the chosen design's costs are admitted openly. It arrives within 24-48h, and your first custom sample carries no fee. The paper above is an original model document written by our desk, not a submitted student paper and not an official Purdue University Global document.
More HS840 papers and DHSc samples
- HS840 Unit 1: Research Question Discussion Post
- HS840 Unit 2: Research Question Critique Paper
- HS840 Unit 3: Cluster Trial Methods Appraisal
- HS840 Unit 5: Sampling Plan
- HS840 Unit 6: Instrument Review
- HS840 Unit 7: Seminar Reflection
- HS840 Unit 8: Data Analysis Plan
- HS840 Unit 9: Ethics and Consent Memo
- HS840 Unit 10: Methods Chapter Draft
- HS830 Unit 4: Adult Obesity Trend Data Analysis
- HS820 Unit 4: France Health Financing Analysis
- HS810 Unit 4: Legislative History Brief
HS840 Unit 4 questions, answered
When is a stepped-wedge design appropriate?
When an intervention will reach every cluster anyway and the rollout can be staggered, or when withholding it would be unacceptable. It suits service changes more than drugs. It is less efficient than a parallel trial in many situations and harder to analyze, so it should be chosen for a practical reason the paper states, not for novelty.
How many alternatives should a justification consider?
Three or four realistic ones, meaning designs a competent researcher might actually propose for this question. Rejecting obviously unsuitable designs adds length without argument. The sample considers individual randomization, a parallel cluster trial and a time series, each a design someone on the team might plausibly have favored. A rival nobody would propose proves nothing when it loses.
Do I need to cite methods literature?
Yes, at doctoral level. Claims about a design's strengths and weaknesses should rest on methods sources, such as the Hussey and Hughes paper for stepped-wedge analysis or the CONSORT extensions for cluster trials. Citing a textbook for basic definitions is fine; contested points need primary methods papers. The sample cites both where its argument depends on them.
Where can I find a free HS840 Unit 4 sample paper?
This page presents the full Unit 4 design justification, comparing four designs for a same-day contraception study and choosing a stepped-wedge cluster trial.
Why not use an interrupted time series instead?
Without randomized timing, it cannot separate the effect of same-day service from a formulary change scheduled for month five or from bias in which clinics go first.