NR-586 · Week 4 of 8 · Study designs and reading association

NR-586 Week 4 Study Designs and Association: How to Write It

The short answer

Midway through NR-586 the writing shifts from describing a population to reasoning about why something happens in it. That means designs and measures of association: cohort studies that follow people forward and produce relative risk, case control studies that work backward from outcome and produce odds ratios, cross-sectional surveys that photograph a moment and cannot establish sequence, and ecological studies that compare places and cannot speak about persons. The written task is to read a number correctly and to say what the design does and does not permit you to claim. Your section may print this as NR 586 or NR586; 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-586 Week 4 grading scale at Chamberlain, the criterion levels this assessment is scored on, from Chamberlain Tutors
How Chamberlain grades NR-586 Week 4, visualized by Chamberlain Tutors.

What NR-586 Week 4 asks for

How does a pattern noticed on a clinic floor become a question a design can answer? A federally qualified health center opens a produce voucher program in the spring, and by autumn the care coordinator notices that patients who picked up vouchers seem to have better glycemic numbers than the ones who did not. That observation is real and it is also, as written, uninterpretable, because the people who collected vouchers had transport to the pickup site, a working phone to get the reminder text and enough slack in the week to go. Every one of those things independently predicts better control. The design question is not whether the association exists. It is what would have to be true for the association to mean the vouchers did something.

That is the territory. Each design buys a different amount of protection against being fooled and charges a different price. A prospective cohort follows exposed and unexposed people forward, establishes that exposure came first, and produces incidence in both groups, which is why relative risk is available. It costs time and it loses people. A case control study starts with cases and picks comparison subjects, which makes rare outcomes affordable and makes control selection the thing that can quietly destroy the study. A cross-sectional survey measures exposure and outcome at once, which is fast and cheap and cannot tell you which came first. An ecological comparison uses group-level data and is genuinely useful for hypothesis generation and genuinely dangerous as a basis for claims about individuals.

Deliverables at this depth usually run 1,100 to 1,500 words and often ask you to read a study, name its design from the methods, interpret its measure of association and evaluate its threats. Some sections instead ask you to propose a design for a question arising from your own population. Both versions are graded on the same thing: whether your verbs match what the design supports. Associated with, more likely to occur among and predicts belong to observational work. Reduced, caused and prevented belong only where an exposure was assigned and compared.

Confounding, bias and chance are the three rival explanations you are expected to work through before you accept a finding. A confounder is a third factor related to both the exposure and the outcome and not on the causal path, and naming a plausible one specific to the study in front of you is worth more than a paragraph reciting the definition. Selection bias and information bias each have a direction. Chance is what the interval speaks to. A paper that handles all three explicitly is doing the graded task.

The NR-586 Week 4 method, step by step

Six moves for writing about a design and the number it produced.

  1. Name the design from the methods section

    Ignore the title and the abstract. Ask whether there was a comparison group, whether exposure was measured before outcome, and whether anything was assigned. Those three answers name the design.

  2. Identify the measure the design can produce

    Cohort and trial designs yield risk and relative risk. Case control designs yield odds ratios. Cross-sectional work yields prevalence and prevalence ratios. Using the wrong term for the wrong design is a scored error.

  3. Read the point estimate in plain language before you touch the statistics

    Write the sentence a colleague would understand: the exposed group experienced the outcome roughly twice as often as the unexposed group over the follow-up period. Then check that the numbers actually say that.

  4. Read the interval around the estimate

    An interval crossing one is compatible with no association. A very wide interval is a statement about sample size. Both matter more to your interpretation than whether a threshold was crossed.

  5. Name one confounder that actually applies here

    Not a generic list. Say which variable is related to both exposure and outcome in this specific population, whether the authors adjusted for it, and which way the estimate would move if they had not.

  6. Close with what the design licenses you to say

    One paragraph translating the finding into the strongest defensible claim, and one sentence naming the study that would be needed to go further.

A layout and word budget for a design and association paper

Our frame for reading or proposing a design, sized for roughly 1,200 to 1,500 words. It is our own outline rather than anything the university issues, and your week's rubric outranks it wherever they disagree.

SectionWhat belongs in itWord target
Question and design namedThe exposure, the outcome, the population, and the design identified from the methods in your own words.140 to 180
How participants enteredRecruitment, comparison group selection, follow-up length and losses, with numbers at each stage.200 to 250
How exposure and outcome were measuredThe instruments or records used, who did the measuring, and whether the measurer knew the group assignment.180 to 220
The measure of associationThe estimate in plain language, the interval, and the absolute difference alongside the relative one.230 to 280
Confounding, bias and chanceOne named confounder specific to this study, one plausible bias with its direction, and what the interval says about precision.280 to 340
What can be claimedThe strongest defensible statement, with the verb matched to the design, and the next study that would strengthen it.150 to 190

Evidence craft for reading association

Report the absolute difference next to the relative one. A doubling of risk means something entirely different when the baseline is four in a hundred than when it is four in ten thousand. Give both, and the reader can judge whether the finding matters in a population your size.

Match the verb to the design every single time. Observational designs support associated with and occurred more frequently among. Reserve reduced and prevented for assigned comparisons. This is the most reliably scored sentence-level habit in the stage.

Say who was compared to whom. An odds ratio without its reference group is unreadable. Name the referent explicitly in the sentence, because the same number reversed tells the opposite story.

Do not treat significance as size. A statistically significant finding is one unlikely under the null, not a large or important one, and a non-significant result in a small study is not evidence of no effect. Precision language is directly graded here.

Give each bias a direction. Writing that recall bias may be present is a definition. Writing that mothers of affected children are likelier to remember an exposure, which would inflate the estimate, is an analysis.

Keep ecological findings at the ecological level. Counties with more of something having more of something else is a hypothesis about places. It licenses a study about persons; it does not license a claim about them.

Five mistakes that cost points in this week's territory

  • Causal verbs on observational data. The single fastest way to lose the interpretation row in this stage.
  • Calling every ratio a relative risk. A case control study cannot produce one, and using the term signals the design was never really identified.
  • Generic limitation paragraphs. Small sample size and further research is needed appear in every weak paper and demonstrate nothing about this one.
  • Reporting only the relative measure. Without the baseline risk the reader cannot tell whether the finding is clinically meaningful.
  • Skipping the comparison group. A paper that never says who the exposed were compared with has not described the design at all.

Before you submit

  • The design is named from the methods rather than from the abstract
  • The measure of association matches what the design can produce
  • The estimate appears in plain language before it appears as a statistic
  • An absolute difference sits beside every relative one
  • One named confounder specific to this study is discussed with its likely direction
  • Every causal-sounding verb has been checked against what was assigned

Reading a study for NR-586?

Send the article and the rubric out of Canvas. A premium original draft comes back in 24 to 48 hours with the design read from the methods, the measure interpreted in plain language, and confounding named specifically rather than generically, with revisions until the grade lands.

Questions students ask about this stage

When can an odds ratio be read as a relative risk?
When the outcome is uncommon in the population studied, the odds ratio approximates the relative risk closely enough that authors often discuss it in those terms. As the outcome becomes common, the odds ratio drifts away from the risk ratio and exaggerates the apparent effect, sometimes substantially. Practically, the safe habit for a graduate paper is to describe an odds ratio as an odds ratio, state the referent group, and add one clause noting that because the outcome was rare it approximates the risk ratio, if that is true in the study you are reading. Where the outcome is common, say so and avoid translating the number into risk language at all. Graders in this stage notice the distinction precisely because it is the point of teaching the designs separately, and a careful sentence about it is cheap to write and reliably rewarded.
How do I tell a confounder from a variable on the causal pathway?
Ask whether the variable is a step between the exposure and the outcome or a separate cause of both. If a neighborhood food environment affects diet, and diet affects glycemic control, then diet is on the pathway and adjusting for it would remove part of the very effect you are trying to measure. If household income affects both whether someone lives in that neighborhood and whether they can afford medication, income is a confounder and belongs in the adjustment. The distinction matters in your writing because over-adjustment is a real error that pushes estimates toward the null and can make a genuine effect disappear. When you are unsure, draw the arrows on paper before you write the paragraph, and say in the paper which role you think the variable plays and why. Explaining your reasoning is worth more than getting the classification perfectly right.
Can I propose a randomized trial for my population health question?
You can, and often the more useful answer is a cluster design or a staged rollout rather than an individually randomized trial. Many population health exposures cannot ethically or practically be assigned to individuals: you cannot randomize housing quality, and you would not withhold a service you believe helps. What you can often do is randomize which clinics, schools or neighborhoods receive an intervention first, which gives you a comparison group without denying anyone the program permanently. A staged rollout where every site eventually receives the intervention is frequently acceptable to the organizations involved for exactly that reason. If you propose one, address the practical objections in advance: how sites are allocated, how contamination between neighboring sites is handled, and what outcome you can measure in the time available.

Keep going

Online now