| Course | PHLT 8064 Dissertation Seminar III |
|---|---|
| Week | Week 5 |
| Paper type | Sampling plan |
| Length | About 850 words, 6 pages |
| Format | APA 7 student paper |
| School | Walden University |
| Program | PhD in Public Health |
| Updated | October 2026 |
Free sample paper for PHLT 8064 Week 5
Sampling Plan for a Registry Census: Eligibility Rules, an Empty Flow Diagram and Power Calculated in Reverse for Every Infant Referred From Newborn Hearing Screening Across Six Birth Years
Student Name
Walden University
PHLT 8064: Dissertation Seminar III
Instructor Name
Month Day, Year
Sampling Plan for a Registry Census: Eligibility Rules, an Empty Flow Diagram and Power Calculated in Reverse for Every Infant Referred From Newborn Hearing Screening Across Six Birth Years
Who the Study Is About and Who It Can Reach
The target population is every infant born in the composite state whose newborn hearing screen ended in a referral for diagnostic audiologic evaluation. The accessible population is those infants as the state registry records them for births from 2017 through 2022, roughly 2,140 referred infants. The gap between the two is mostly births outside hospitals. Home births and some birthing center births are screened later or not at all, and infants who never received a screen cannot enter the registry as referred. They are likely to differ from hospital births in ways related to follow-up, so the plan names them as outside the study's reach rather than assuming they resemble the cohort.
Why No Sample Is Drawn
When a register holds the whole population of interest, drawing a sample from it throws away information without saving meaningful cost. Thygesen and Ersbøll (2014) describe this as a defining strength of register-based studies: selection at entry is avoided because no one has to agree to take part. The work of this plan therefore lies in deciding who counts as eligible and in showing what a cohort of fixed size can detect.
Eligibility Rules
Inclusion
1. Born in the composite state from January 1, 2017, through December 31, 2022, because those birth years fall under the current registry and the most recent year allows full three-month follow-up before data extraction.
2. Recorded as not passing the final inpatient or outpatient screen in one or both ears, because only those infants are referred for diagnostic evaluation.
3. Matched to a birth certificate, because the residential address and covariates come from that record.
Exclusion
1. Died before three months of age, because the outcome window could not be completed. Counted and reported.
2. Adopted or placed in care outside the state before three months, because the registry stops following them. Counted and reported.
3. Residential address that could not be geocoded to a street location, because drive time cannot be computed. Counted, and their characteristics compared with included infants.
4. Diagnosed with a condition requiring prolonged neonatal intensive care that the program follows on a separate pathway, because their evaluation timing follows different guidance. Counted and reported.
Each exclusion could be related to the outcome. Infants with failed geocoding, for example, may disproportionately live on rural routes, the very families the study is about. For that reason no exclusion is silent: every removal is counted, and the third is compared with the analytic cohort on payer, maternal education and county.
A Flow Diagram Drawn Before the Data Arrive
Table 1
Flow From Registry Records to the Analytic Cohort
| Step | Records remaining | Records removed and why | |
|---|---|---|
| All registry records, births 2017-2022 | n to be entered | Not applicable | |
| Referred after a non-pass screen | n to be entered | Passed screening, n to be entered | |
| Matched to a birth certificate | n to be entered | No match, n to be entered | |
| Alive and in state at three months | n to be entered | Died or moved out of state, n to be entered | |
| Address geocoded | n to be entered | Not geocoded, n to be entered | |
| Analytic cohort | n to be entered | Separate pathway, n to be entered | |
Note. Counts are left blank deliberately; the diagram is fixed before extraction so the path cannot be adjusted to the results, consistent with RECORD reporting guidance.
Power for a Fixed Cohort
A census still needs a power analysis, but it runs backward. The question is not how many infants to recruit; it is what size of association the fixed cohort could reliably detect. Using G*Power's sensitivity mode for logistic regression (Faul et al., 2009), the inputs are a two-sided alpha of .05, power of .80, an analytic cohort of about 2,000 after exclusions, a proportion with documented evaluation of 0.75 among infants at average drive time, drive time entered as a standardized continuous predictor, and an R squared of 0.10 between drive time and the other covariates. Those inputs return a smallest detectable odds ratio of about 0.86 per standard deviation of drive time. If the true association is weaker than that, the study could miss it, and the discussion will say so. This calculation is done before the data are seen; Hoenig and Heisey (2001) warn that power computed after a null result from the observed effect adds nothing.
Events per Covariate
With about a quarter of infants lacking a documented evaluation, the cohort should contain roughly 500 outcome events. The main model is planned with about twelve parameters, giving more than 40 events per parameter. Peduzzi et al. (1996) recommended at least ten events per variable for stable logistic regression estimates, and Vittinghoff and McCulloch (2007) showed that the rule can be relaxed in some settings. The planned model sits well above either standard, which leaves room for the payer interaction in the second research question without strain.
References
Faul, F., Erdfelder, E., Buchner, A., & Lang, A.-G. (2009). Statistical power analyses using G*Power 3.1: Tests for correlation and regression analyses. Behavior Research Methods, 41(4), 1149-1160. https://doi.org/10.3758/BRM.41.4.1149
Hoenig, J. M., & Heisey, D. M. (2001). The abuse of power: The pervasive fallacy of power calculations for data analysis. The American Statistician, 55(1), 19-24. https://doi.org/10.1198/000313001300339897
Peduzzi, P., Concato, J., Kemper, E., Holford, T. R., & Feinstein, A. R. (1996). A simulation study of the number of events per variable in logistic regression analysis. Journal of Clinical Epidemiology, 49(12), 1373-1379. https://doi.org/10.1016/S0895-4356(96)00236-3
Thygesen, L. C., & Ersbøll, A. K. (2014). When the entire population is the sample: Strengths and limitations in register-based epidemiology. European Journal of Epidemiology, 29(8), 551-558. https://doi.org/10.1007/s10654-013-9873-0
Vittinghoff, E., & McCulloch, C. E. (2007). Relaxing the rule of ten events per variable in logistic and Cox regression. American Journal of Epidemiology, 165(6), 710-718. https://doi.org/10.1093/aje/kwk052
PHLT 8064 Week 5 instructions, in plain terms
The fifth week's assignment asks for a sampling plan that defines the target and accessible populations, states eligibility criteria and justifies the sample size. For survey or trial designs that means recruitment and a priori power. For registry and administrative data studies, the plan instead explains why the full population is used and what the fixed number of records can detect. Faculty expect every exclusion to carry a reason and a plan to report how many records it removes, since exclusions can bias a census as surely as low response biases a survey. A participant flow diagram, usually following STROBE or RECORD conventions, is often required. Plans typically run three to five pages, with power software output described in enough detail to be repeated and sources on sample size and register-based research.
How this PHLT 8064 Week 5 example is built
The first section defines the target population as every referred infant born in the state and the accessible population as those the registry records, then names home and birthing center births as the gap. A short section argues, with Thygesen and Ersbøll, that sampling from a complete register discards information. Eligibility follows as numbered inclusion and exclusion rules, each with its reason, and a paragraph notes that infants with failed geocoding may live on rural routes and will be compared with the analytic cohort. Table 1 is a flow diagram with every count left blank so the path is fixed before data arrive. The power section reports each G*Power input and the detectable odds ratio and warns against post hoc power, and an events-per-variable section confirms about 500 outcome events for twelve parameters.
Where the marks sit in the PHLT 8064 Week 5 rubric
Grading for this plan centers on whether the population definitions, eligibility rules and size justification fit the design. A census plan that offers an a priori recruitment target misreads its own design and loses substance points; the sample avoids that by reframing power as sensitivity. Criteria for exclusions reward reasons tied to the outcome window or the exposure measurement and penalize silent removals. The blank flow diagram earns credit for transparency because it cannot be adjusted to results. Accuracy is checked in the power inputs, which must be stated fully enough to rerun, and in the events-per-variable reasoning. Rubrics also look for acknowledgment of who the accessible population leaves out. Citations to sample size and register research, along with APA tables and headings, round out the scoring.
Common PHLT 8064 Week 5 mistakes, and how to avoid them
Sampling plans for registry studies often copy the template written for surveys and end up describing recruitment that will never happen. If your data already contain everyone eligible, say so and spend your words on eligibility instead. Another frequent error is listing exclusions without reasons, or worse, without counts, so the reader cannot tell how much the cohort shrank or why. Power is the third trouble spot: some candidates skip it because the sample is fixed, while others calculate it after seeing results, which Hoenig and Heisey showed to be uninformative. Run a sensitivity calculation before data access and report the smallest effect you could detect. Draw the flow diagram early, even empty. If your registry, birth years or outcome differ, a custom plan written to your data dictionary is available through this page at no charge.
Get a PHLT 8064 Week 5 example written to your instructions
Describe the population you intend to study and the data source that holds it, then add the rubric and wording of the Week 5 sampling assignment. The plan separates target from accessible population, gives every exclusion a reason, draws the flow in advance and sets power to match your design. The first plan is free within 24 to 48 hours; its registry counts nothing real.
More PHLT 8064 papers and PhD in Public Health samples
- PHLT 8064 Week 1: Research Question Reading Post
- PHLT 8064 Week 2: Purpose Statement Rebuild
- PHLT 8064 Week 3: Design Comparison Paper
- PHLT 8064 Week 4: Method Defense Paper
- PHLT 8064 Week 6: Measurement Memo
- PHLT 8064 Week 7: Alignment Table Check
- PHLT 8064 Week 8: Analysis Plan
- PHLT 8064 Week 9: Peer Design Critiques
- PHLT 8064 Week 10: Feasibility Memo
- PHLT 8064 Week 11: Chapter 3 Method Draft
- PHLT 8062 Week 5: Matrix Evidence Table
- PHLT 8060 Week 5: Boundary-Testing Peer Critique
- PHLT 8048 Week 5: TPB Versus COM-B Comparison
- PHLT 8046 Week 5: Falsifiable Aims Peer Replies
PHLT 8064 Week 5 questions, answered
Why does a census need a power analysis?
Because a fixed population can still be too small to detect an association that matters. When every eligible record is already included, a power analysis cannot tell you how many to recruit, but it can tell you how small an association the cohort could reliably detect. The example runs G*Power in sensitivity mode for that reason. If the detectable association is larger than anything plausible, your committee needs to know before data are requested.
What does G*Power need for a logistic regression?
Inputs include the significance level, the desired power, the expected outcome proportion in the reference group, the distribution of the predictor and how strongly it relates to the other covariates. The example states where each input will come from, such as the registry's published annual summaries, rather than inventing values. Name your sources for these inputs; a committee will ask, and figures without a source are hard to defend.
Should infants with missing addresses be excluded or imputed?
The example excludes infants whose addresses cannot be geocoded, because drive time cannot be estimated without a location, and it reports how many are removed. It then compares their recorded characteristics with the analytic cohort to show whether exclusion might bias results. Where only a covariate is missing, imputation is usually better than exclusion, which the analysis plan addresses separately.
Where can I find a free PHLT 8064 Week 5 sample paper?
Read the full Week 5 sampling plan here: it covers a registry census of infants referred after newborn hearing screening, with eligibility rules, a blank flow diagram and a reverse power calculation. Every input to the power analysis is stated, and each reference can be checked.
How do target and accessible populations differ in a registry study?
The target population is everyone the study's conclusions are meant to describe; the accessible population is the part the data can actually reach. In the sample, home births screened late or never are in the target but outside the registry, so the plan names them rather than assuming they resemble the cohort.