| Course | PHLT 8520 Advanced Epidemiological Methods |
|---|---|
| Week | Week 5 |
| Paper type | Selection analysis paper |
| Length | About 1,070 words, 6 pages |
| Format | APA 7 student paper |
| School | Walden University |
| Program | PhD in Public Health |
| Updated | October 2026 |
Free sample paper for PHLT 8520 Week 5
Three Groups of Pregnancies the Cohort Never Sees: A Selection Analysis Sorting Each Absence by Whether It Bends the Night-Shift Estimate or Only Narrows It
Student Name
Walden University
PHLT 8520: Advanced Epidemiological Methods
Instructor Name
Month Day, Year
Three Groups of Pregnancies the Cohort Never Sees: A Selection Analysis Sorting Each Absence by Whether It Bends the Night-Shift Estimate or Only Narrows It
The Entry Rule Under Examination
The record-linked cohort admits a pregnancy only if it is confirmed in the health system's own obstetric clinics while the woman holds a rostered job. That rule is convenient, since it guarantees linked records, but every entry rule is also a filter. This paper identifies three groups of pregnancies the rule keeps out and asks, for each, whether the condition that excludes them depends on night work, on preterm risk or on both. Hernán et al. (2004) show that selection biases an estimate when inclusion is a common effect of the exposure and the outcome, or of their causes; when inclusion depends on only one side, it changes whom the result describes without distorting it.
Group One: Losses Before Confirmation
Who They Are
Pregnancies that end before the first clinic visit, usually around weeks 6 to 8, never generate an obstetric record. In the composite system, the pregnancy registry shows confirmation at a median of 7.4 weeks, and early losses before that point are invisible.
Does Inclusion Depend on Exposure?
Plausibly. Danish payroll data linked two or more night shifts in the previous week with a higher hazard of miscarriage after week 8, a hazard ratio of 1.32 (Begtrup et al., 2019). If nights raise the risk of very early loss as well, night workers' surviving pregnancies are a more selected group than day workers'.
Does Inclusion Depend on Outcome Risk?
Also plausibly. Some causes of early loss, such as placental and immune factors, are thought to raise preterm risk in pregnancies that continue. A pregnancy that survives to confirmation despite night work may therefore carry less of that underlying risk.
Verdict
Both arrows point into inclusion, so inclusion is a collider. Conditioning on survival to confirmation can create a negative association between night work and underlying risk among included pregnancies, which would pull the estimate toward the null or below it. Liew et al. (2015) illustrate the same structure for live birth in pregnancy cohorts and find the bias is usually modest unless the exposure strongly affects loss. This group is classed as a source of bias, probably toward the null, and probably small.
Group Two: Employees Insured Through a Partner
Who They Are
Employees who decline the system's health plan, usually because a partner's plan covers the household, may receive prenatal care from outside practices. Their pregnancies never reach the system's obstetric records. In the composite benefits file, 17% of rostered women aged 20 to 44 waived system coverage in a typical year.
Does Inclusion Depend on Exposure?
Only indirectly. Waiving coverage depends on a partner's job, which is tied to household income, and income is tied to the likelihood of choosing nights. Night work does not cause a woman to waive coverage.
Does Inclusion Depend on Outcome Risk?
Only through the same household resources. Within levels of salary band and job grade, there is no clear path from preterm risk to the coverage decision, which is made before pregnancy.
Verdict
Inclusion here depends on causes the analysis already adjusts for, at least in part, and not on the exposure or outcome themselves. The absence narrows the study's reach, to women insured through their employer, without opening a new path. Residual bias would arise only if salary band poorly captures household income, which is noted for the Week 9 sensitivity argument.
Group Three: Women Who Left Before Conceiving
Who They Are
Women hired into rostered jobs who left the system before ever conceiving cannot appear in an employee pregnancy cohort. Turnover among night-rotating staff is higher than among day staff.
Does Inclusion Depend on Exposure?
Yes. Women who tolerate night work poorly are more likely to leave night jobs, or the employer, within their first years.
Does Inclusion Depend on Outcome Risk?
Plausibly. Women with conditions that make night work hard to sustain, such as migraine, hypertension or poor sleep, may also carry higher obstetric risk. If those conditions drive departure, the night workers who remain to conceive are a healthier group than the night workers who were hired. Arrighi and Hertz-Picciotto (1994) describe this ongoing filtering of workers as the healthy worker survivor effect.
Verdict
This is the strongest of the three. Inclusion depends on both night work and on health that predicts preterm birth, so the selection would make night work look less harmful than it is. Part of the bias can be blocked by adjusting for pre-pregnancy chronic disease, already in the Week 3 set, but tolerance for nights is not recorded.
Sorting the Three
Table 1
Selection Structures and Expected Consequences
| Absent group | Inclusion depends on exposure? | Inclusion depends on outcome risk? | Consequence | Likely direction | |
|---|---|---|---|---|
| Losses before confirmation | Plausibly | Plausibly | Bias | Toward the null, probably small | |
| Insured through a partner | Through income only | Through income only | Narrower reach | None if income is captured | |
| Left before conceiving | Yes | Plausibly | Bias | Toward the null | |
Note. Judgments for the composite cohort.
What the Study Can Do
Three responses follow. First, the registry's records of pregnancy tests and early ultrasounds can identify some losses after a positive test but before confirmation, allowing a sensitivity analysis that counts losses as a competing outcome. Second, a length-of-service restriction, limiting the cohort to women hired within two years before conception, would shorten the time over which departures can filter the night group. Third, the results section will report the waiver rate and describe the population the estimate applies to.
None of these responses removes selection entirely. The registry captures only losses that reached a pregnancy test in the system, the hire-date restriction shrinks the cohort by roughly a third and costs precision, and a careful description of who is missing is a statement of limits rather than a correction. The point of the exercise is to know which way the remaining selection leans, so that a reader can judge whether it could have produced, hidden or reversed the result the cohort reports.
Summary
Of three absent groups, two are likely to bias the night-shift estimate toward the null, and one mainly limits generalizability. Any positive association the cohort finds is therefore more likely to be understated than inflated by selection, a direction that will matter when the Week 9 argument weighs an observed estimate against unmeasured confounding.
References
Arrighi, H. M., & Hertz-Picciotto, I. (1994). The evolving concept of the healthy worker survivor effect. Epidemiology, 5(2), 189-196. https://doi.org/10.1097/00001648-199403000-00009
Begtrup, L. M., Specht, I. O., Hammer, P. E. C., Flachs, E. M., Garde, A. H., Hansen, J., Hansen, Å. M., Kolstad, H. A., Larsen, A. D., & Bonde, J. P. (2019). Night work and miscarriage: A Danish nationwide register-based cohort study. Occupational and Environmental Medicine, 76(5), 302-308. https://doi.org/10.1136/oemed-2018-105592
Hernán, M. A., Hernández-Díaz, S., & Robins, J. M. (2004). A structural approach to selection bias. Epidemiology, 15(5), 615-625. https://doi.org/10.1097/01.ede.0000135174.63482.43
Liew, Z., Olsen, J., Cui, X., Ritz, B., & Arah, O. A. (2015). Bias from conditioning on live birth in pregnancy cohorts: An illustration based on neurodevelopment in children after prenatal exposure to organic pollutants. International Journal of Epidemiology, 44(1), 345-354. https://doi.org/10.1093/ije/dyu249
What the PHLT 8520 Week 5 instructions ask for
Selection weeks generally ask students to identify who is missing from their study and to judge whether the omission biases the estimate. Prompts may cover loss to follow-up, nonparticipation, eligibility rules and survival bias. Doctoral instructors increasingly expect the structural approach: selection biases a result when inclusion depends on both exposure and outcome, or their causes. Students are asked to state a likely direction and, where possible, a rough size, not just a possibility. Three to five pages is typical, often with a summary table. Many prompts ask for responses the study could make, such as sensitivity analyses or restrictions, together with their costs. Citations to methods papers on selection structure, and to substantive work on the exposure, are usually required.
Inside the PHLT 8520 Week 5 example
An opening section states the cohort's entry rule and the structural test, with a citation. Each of the three absent groups then gets four short subsections: who they are, whether inclusion depends on exposure, whether it depends on outcome risk, and a verdict. The early-loss group draws on Danish payroll evidence about night work and miscarriage and on a paper about conditioning on live birth. The insurance group uses a composite benefits figure of 17% waiving coverage. The departure group cites the survivor concept. Table 1 summarizes all three by dependence, consequence and direction. A response section proposes registry-detected losses, a hire-date restriction and plain reporting, then admits what none of them fixes. A summary ties the direction of selection to later weeks.
PHLT 8520 Week 5 rubric: what earns full marks
Selection papers earn their top marks when each absence is judged by structure, not by intuition. The sample applies the same test to three groups and reaches different verdicts, which shows the test was used rather than recited. Distinguishing bias from reduced generalizability is a distinct criterion, and the insurance group demonstrates it. Directions are stated with cautious language and a reason, which graders prize over vague warnings. Grounding the early-loss argument in published miscarriage findings earns evidence credit. The response section is weighed for realism: costs such as lost sample size are named. The summary table, clear headings and correct references secure the mechanics row. Papers that declare every exclusion a bias usually lose the analytic marks.
PHLT 8520 Week 5 help from the desk
A common mistake here is treating every missing person as a source of bias. Ask, for each group, whether inclusion depends on the exposure, on the outcome or on both. If it depends on only one, you have a generalizability issue, not necessarily bias. Give each verdict a direction and say why. Use published evidence where it exists, such as studies linking your exposure to early outcomes that remove people before entry. Propose responses, but name what each costs and what it cannot fix. Avoid claiming that adjustment for baseline variables will repair selection that happens after exposure starts. A summary table helps graders see your verdicts at once. We can build a first selection analysis for your cohort without charge.
Get a PHLT 8520 Week 5 example written to your instructions
Forward the Week 5 prompt and rubric, and name the study or design under examination. The analysis identifies who could never have entered it, draws what decided each absence, and says which absences bend the estimate and which only narrow its reach. First analysis free, back in 24-48 hours.
More PHLT 8520 papers and PhD in Public Health samples
- PHLT 8520 Week 1: Error Taxonomy Brief
- PHLT 8520 Week 2: Design Selection Paper
- PHLT 8520 Week 3: Confounding Brief
- PHLT 8520 Week 4: Design Proposal
- PHLT 8520 Week 6: Peer Challenge Reply
- PHLT 8520 Week 7: Measurement Error Brief
- PHLT 8520 Week 8: Control Strategy Comparison
- PHLT 8520 Week 9: Sensitivity Argument
- PHLT 8520 Week 10: Validation Check
- PHLT 8520 Week 11: Final Methods Paper
- PHLT 8500 Week 5: Outcome Type Brief
- PHLT 8450 Week 5: Asset Inventory
- PHLT 8331 Week 5: Two-Group Mean Comparison
- PHLT 8270 Week 5: Data Quality Appraisal
PHLT 8520 Week 5 questions, answered
Where can I find a free PHLT 8520 Week 5 sample paper?
The complete Week 5 selection analysis is shown above, sorting three absent groups of pregnancies by structure, with a verdict table and verified references.
When does missing data cause selection bias?
When inclusion in the analysis depends on both the exposure and the outcome, or on their causes. If inclusion depends only on one side, the result describes a narrower population but is not necessarily biased.
What is bias from conditioning on live birth?
In pregnancy studies, only pregnancies that survive long enough enter. If the exposure causes early loss and loss shares causes with the outcome, survivors in the exposed group are less susceptible, which can bias the estimate.
Which way does the healthy worker survivor effect push an estimate?
Usually toward the null, because workers who tolerate the exposure poorly leave it, leaving a hardier exposed group that makes the exposure look less harmful.
Can selection bias be fixed with regression adjustment?
Only partly. Adjusting for causes of inclusion can help, but selection that happens after exposure begins, or depends on unmeasured factors, needs other methods such as weighting or sensitivity analysis.