MPH-504 · Week 2 of 8 · Measuring health across populations

MPH-504 Week 2 Measuring Health Across Populations: How to Write It

The short answer

The second stage of MPH-504 is the measurement stage, and it is where most of the session's marks are quietly won or lost. You are being taught to hold two numbers from two populations and say whether they may honestly be set beside each other: same case definition, same denominator, same reference period, comparable completeness of reporting, and age structures either matched or standardized. The written product is usually a small comparative table with prose that defends every figure in it. Your section may print this as MPH 504 or MPH504; 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.

MPH-504 Week 2 grading scale at Chamberlain, the criterion levels this assessment is scored on, from Chamberlain Tutors
How Chamberlain grades MPH-504 Week 2, visualized by Chamberlain Tutors.

What MPH-504 Week 2 asks for

Two under-five mortality figures sit in a draft on a screen: 5 deaths per 1,000 live births in one country and 58 per 1,000 in another. The paragraph beneath them says the second country's rate is more than eleven times higher. Everything about that sentence is arithmetically true and analytically premature, because one of those numbers came from a civil registration system that captures nearly every birth and death, and the other came from a household survey that asked mothers to recall birth histories over the preceding five years. They are both usable. They are not the same kind of object, and this week exists to make you write the sentence that says so.

The territory is measurement literacy applied comparatively. Expect the deliverable to involve selecting indicators for a population pair, retrieving them from named sources, presenting them in a form a reader can scan, and then writing the defence. Where the guide asks for a table, the table is not the assignment; the assignment is the paragraph that tells a reader why the table can be trusted and where it cannot. In our experience the papers that score in the top band at this stage are the ones that volunteer a limitation the grader had not thought of yet.

The core distinctions to have straight before you draft are few and they repeat all session. Prevalence counts who has the condition now and is inflated by longer survival; incidence counts new cases in a defined period and is the measure of risk. Crude rates describe the population as it actually is; age-standardized rates describe what the population's risk would look like if its age structure matched a reference. A composite measure that combines years of life lost with years lived in reduced health is answering a different question again, and it carries value judgments about disability inside its construction that a careful writer names rather than hides.

One habit belongs here rather than later. When you retrieve a figure, record the retrieval in the same keystroke: the database or report, the indicator name exactly as that source writes it, the geography, the reference year and the date you pulled it. Global databases revise back-series, and a paper defended in week seven with a number pulled in week two and never re-checked is the commonest source of an internal contradiction in a final submission.

The MPH-504 Week 2 method, step by step

Six moves for building a comparison a reviewer cannot dismantle.

  1. Choose indicators that answer your question, not the ones that are easy to find

    If your problem is access, an outcome rate alone will not show it; pair the outcome with a coverage or utilization measure. Two well-chosen indicators beat six retrieved because they were on the same page.

  2. Write down the case definition each source used

    Low birth weight, stunting, a notified case, a maternal death: each has a formal definition, and countries and systems do not always apply the same one. Where the definitions differ, that is a finding to report rather than a nuisance to smooth over.

  3. Match the reference period or explain the gap

    Comparing a 2019 figure with a 2023 figure across a period when service delivery was disrupted is not a comparison of two places, it is a comparison of two places and two eras. Say which years you used and why.

  4. Decide on standardization and state the decision

    If your populations differ in age structure and your indicator is age-sensitive, use age-standardized figures where the source provides them. Where only crude rates exist, name that as a limit in the sentence that reports them.

  5. Interrogate completeness before you interpret a difference

    A lower notification rate can mean less disease or less detection. Ask what proportion of events the system captures, whether the denominator comes from a census or a projection, and how old that census is.

  6. Report uncertainty as part of the number

    Where a source publishes an interval, carry it. A rate of 58 with a published range of 46 to 72 tells a different story from a bare point estimate, and dropping the range is a decision the reader should not have to reverse-engineer.

Build the indicator table and budget the prose around it

Our frame for a comparative measurement piece, sized for roughly 1,000 to 1,300 words plus a table. It is our own outline rather than anything the university issues, and your week's scoring guide outranks it wherever the two disagree.

SectionWhat belongs in itWord target
Why these indicatorsThe question each measure is chosen to answer, and the one you rejected with your reason for rejecting it.140 to 180
The table itselfIndicator, both populations, value, unit and denominator, reference year, source system. One row per measure, no orphan figures.Table, plus 60 of lead-in
Provenance paragraphWhat produced each number: registration, survey, surveillance notification or model, and what that implies about its error.200 to 250
Comparability defenceCase definitions, periods, denominators, completeness and standardization, each addressed rather than listed.250 to 320
The pattern, describedDirection, magnitude and internal distribution, using the units in the table and no causal verbs at all.180 to 230
What the measures cannot showThe question your indicators are silent on, and what data would be needed to answer it.120 to 160

Evidence craft for numbers that crossed a border

Cite the producing system, not the aggregator. A global portal that republishes a national survey estimate is a route to the figure, not its origin. Name the survey, the round and the year, then the portal if your style requires it. Reviewers check this and the correction is cheap before submission and expensive after.

Keep the unit attached to the number every time it appears. Per 1,000 live births and per 100,000 population are not interchangeable, and a figure that migrates from a table into prose without its unit is where most arithmetic errors in these papers begin. Rewrite any sentence in which a bare number appears more than once.

Do not subtract across incompatible measures. Ratios of ratios, differences between a crude and a standardized rate, and gaps computed from two different reference years all look like analysis and are not. Where you want a single summary of the difference, choose one comparable pair and state what you computed.

Let disaggregation carry the argument. A national figure hides who inside the population is bearing the burden. Where a source publishes by wealth quintile, residence, maternal education or region, use it: the within-country spread is often larger than the between-country gap, and pointing that out is the single most reliable way to lift an analysis row at this stage.

Five mistakes that cost points in this week's territory

  • Percentages with no base. Twenty-three percent of what, out of how many, in which year. A proportion with its numerator and denominator visible is evidence; without them it is decoration.
  • Treating a modeled estimate as a count. Model outputs are legitimate and frequently unavoidable. Presenting one in the same voice as a registration count overstates what your source knows.
  • Comparing years of convenience. Using whichever year each source happened to publish, without saying so, quietly turns a place comparison into a time comparison.
  • Causal verbs in a descriptive section. Reduced, improved and drove belong to later weeks and to designs that support them. This stage describes.
  • A table nobody reads. An unreferenced table sitting between two paragraphs earns nothing. Point at specific rows in the prose and make the table do work.

Before you submit

  • Every figure in the table carries a unit, a reference year and a named source system
  • Case definitions are compared explicitly rather than assumed to match
  • Standardization is either applied or its absence is named as a limit
  • Published uncertainty ranges travel with their point estimates
  • At least one indicator is disaggregated within a population
  • No causal verb appears anywhere in the descriptive section

Building the comparison table this week?

Send the scoring guide and the two populations you chose. A premium original draft comes back in 24 to 48 hours with every figure sourced to its producing system and the comparability defence written out, and revisions run until the grade lands.

Questions students ask about this stage

Two sources give different values for the same indicator and year. Which do I use?
Report both and explain the disagreement rather than choosing the convenient figure. Differences almost always trace to one of four things: a different case definition, a different estimation method, a different denominator source such as a census versus a projection, or different assumptions about reporting completeness. Work out which one is operating, say so in a clause, then choose one figure for the body of your analysis and carry the other as the stated range. Graders read that sequence as measurement literacy, and it is worth more than a paper in which every number is tidy because the inconvenient source was never opened.
Do I have to use age-standardized rates?
Only when age structure could plausibly explain the difference you are describing, which is most of the time for mortality and chronic disease and much less often for measures already restricted to a narrow age band. An under-five indicator is already age-bounded, so standardization matters less; an all-cause mortality comparison between a country with a median age in the twenties and one in the forties is meaningless without it. The rule that keeps you safe is to say which you used. A sentence stating that the rates are age-standardized to a named reference population, or that they are crude and that the populations differ in age structure, protects the evidence row either way.
My indicator is not available for one of my populations at all. Do I change topic?
Not necessarily, and the absence is itself information worth a paragraph. Missing data usually means the system that would produce it does not exist, is not funded, or does not reach the group in question, and any of those is a finding about the health system you were going to have to discuss anyway. The practical fix is to substitute the nearest available proxy, state clearly that it is a proxy, and say in one sentence what it over- or under-states relative to the measure you wanted. What loses points is silently swapping in a related indicator and letting the reader assume it is the one your question named.

Keep going

Online now