Theory & Argument Building (demog-theory-building)
At Demography a result is not a contribution until it is attached to a claim population science can
use — a mechanism that explains a demographic process, a sharper estimate that revises the record,
or a formal result that unifies or clarifies. This skill turns findings into argument: explicit
mechanisms, scope conditions, and observable implications, in the idiom appropriate to your kind of
work.
When to trigger
- The empirics are strong but the "so what / why" is thin
- A reviewer said the paper is "descriptive," "atheoretical," or "just a correlation"
- You need to state mechanisms, identifying assumptions, or scope conditions explicitly
- Formal demography: deciding what to model and what the model buys you
Build the argument (by mode of work)
Explanatory population study
- Concept & measure — define the demographic construct (e.g., parity progression, lifespan
inequality, net migration) precisely; distinguish it from neighbors and from its measure.
- Mechanism — the population story: which behaviors, exposures, or compositional shifts move the
rate, for whom, and why (incentives, constraints, selection, cohort experience).
- Observable implications — what we should see if the mechanism operates (age pattern, cohort
signature, subgroup contrast) and what we should not see. These become the tests in
demog-research-design.
- Scope conditions — which populations, periods, and regimes the argument covers.
Formal / mathematical demography
- State the substantive population puzzle the model addresses before the setup.
- Keep assumptions (stability, stationarity, Markov transitions, independence) transparent and
motivated; flag which results depend on which assumptions.
- Translate results into interpretable demographic quantities (e.g., contributions to life
expectancy, sensitivities/elasticities, equilibrium structure) a reader can recognize.
- Say what the model buys: a non-obvious decomposition, a unifying identity, a corrected intuition.
Measurement / decomposition contribution
- Make explicit what the new measure or decomposition separates that prior work conflated (e.g.,
tempo vs. quantum, composition vs. rate, age vs. cohort).
- Show the substantive payoff: the trend now attributes to a different component than was assumed.
The "portability" test (Demography-specific)
Ask: Could a demographer studying a different component or population import this mechanism, measure,
or decomposition? If yes, you have a population-science contribution. If it only works for your exact
case, generalize the logic or reframe (back to demog-topic-selection).
Anti-patterns
- "Hypothesizing after results are known" — state the argument before the tests
- A formal model with opaque assumptions chosen to produce the desired identity
- Mechanisms named but never made observable in age/cohort/subgroup patterns
- Treating a regression coefficient as a mechanism with no demographic story
- Universal claims with no scope conditions on population, period, or regime
Worked micro-example: from finding to population-science claim (illustrative)
A hypothetical study observes that completed cohort fertility fell across successive birth cohorts. The
argument is built in the idiom Demography — the Population Association of America flagship at Duke
University Press — rewards (numbers invented to illustrate):
- Bare finding: "Cohort TFR fell from ~2.1 to ~1.7 across the 1955-1975 birth cohorts."
- Mechanism: Postponement of first births raised the mean age at first birth, and recuperation at
older ages was incomplete — a quantum decline operating through tempo, not a uniform shift.
- Observable implication: Parity-progression ratios from parity 0 to 1 should fall most at younger
ages and only partly rebound later; a pure quantum story would show uniform decline across ages.
- Scope condition: The claim covers low-fertility settings with delayed childbearing.
- Portability: A mortality scholar can import the tempo-vs-quantum logic to lifespan compression;
that import makes it a population-science contribution, not a single-country fact.
Referee-pushback patterns and the theory-side fix
- "This is descriptive — where is the mechanism?" -> Name the behavior/exposure/compositional shift
that moves the rate, for whom, and translate it into an age or cohort signature the design can test.
- "You assert a mechanism but a compositional shift would produce the same trend." -> State the rival
(composition) explicitly and the observable that separates it from your account before the results.
- "The formal model's assumptions are chosen to deliver the identity." -> Flag which results depend on
stability/Markov/independence assumptions and motivate each substantively.
- "This only works for your one case." -> Generalize the mechanism so another demographer studying a
different component could reuse it; otherwise reframe via
demog-topic-selection.
Output format
【Core claim】one sentence
【Mechanism / identity】the population story or formal result
【Assumptions】(formal) the load-bearing ones
【Observable implications】testable demographic signatures -> research-design
【Scope conditions】which populations / periods it covers
【Portability】who else in population science can use this
【Next】demog-research-design
Supplementary resources
1---2name: demog-theory-building3description: Use when building the argument of a Demography (PAA / Duke University Press) manuscript into a population-science contribution — whether the work is formal/mathematical demography, an explanatory account of fertility/mortality/migration, or a measurement/decomposition advance. Demography rewards a clear mechanism or a sharpened estimate over a bare correlation. Structures the argument; it does not run analyses.4metadata:5 source: https://skillmd.com/skills/brycewang-stanford/demog-theory-building6---78# Theory & Argument Building (demog-theory-building)910At Demography a result is not a contribution until it is attached to a **claim population science can11use** — a mechanism that explains a demographic process, a sharper estimate that revises the record,12or a formal result that unifies or clarifies. This skill turns findings into argument: explicit13mechanisms, scope conditions, and observable implications, in the idiom appropriate to your kind of14work.1516## When to trigger1718- The empirics are strong but the "so what / why" is thin19- A reviewer said the paper is "descriptive," "atheoretical," or "just a correlation"20- You need to state mechanisms, identifying assumptions, or scope conditions explicitly21- Formal demography: deciding what to model and what the model buys you2223## Build the argument (by mode of work)2425### Explanatory population study261. **Concept & measure** — define the demographic construct (e.g., parity progression, lifespan27 inequality, net migration) precisely; distinguish it from neighbors and from its measure.282. **Mechanism** — the population story: which behaviors, exposures, or compositional shifts move the29 rate, for whom, and why (incentives, constraints, selection, cohort experience).303. **Observable implications** — what we should see if the mechanism operates (age pattern, cohort31 signature, subgroup contrast) and what we should *not* see. These become the tests in32 `demog-research-design`.334. **Scope conditions** — which populations, periods, and regimes the argument covers.3435### Formal / mathematical demography36- State the **substantive population puzzle** the model addresses before the setup.37- Keep assumptions (stability, stationarity, Markov transitions, independence) **transparent and38 motivated**; flag which results depend on which assumptions.39- Translate results into **interpretable demographic quantities** (e.g., contributions to life40 expectancy, sensitivities/elasticities, equilibrium structure) a reader can recognize.41- Say what the model **buys**: a non-obvious decomposition, a unifying identity, a corrected intuition.4243### Measurement / decomposition contribution44- Make explicit **what the new measure or decomposition separates** that prior work conflated (e.g.,45 tempo vs. quantum, composition vs. rate, age vs. cohort).46- Show the substantive payoff: the trend now attributes to a different component than was assumed.4748## The "portability" test (Demography-specific)4950Ask: *Could a demographer studying a different component or population import this mechanism, measure,51or decomposition?* If yes, you have a population-science contribution. If it only works for your exact52case, generalize the logic or reframe (back to `demog-topic-selection`).5354## Anti-patterns5556- "Hypothesizing after results are known" — state the argument before the tests57- A formal model with opaque assumptions chosen to produce the desired identity58- Mechanisms named but never made observable in age/cohort/subgroup patterns59- Treating a regression coefficient as a mechanism with no demographic story60- Universal claims with no scope conditions on population, period, or regime6162## Worked micro-example: from finding to population-science claim (illustrative)6364A hypothetical study observes that completed cohort fertility fell across successive birth cohorts. The65argument is built in the idiom Demography — the Population Association of America flagship at Duke66University Press — rewards (numbers invented to illustrate):6768- **Bare finding:** "Cohort TFR fell from ~2.1 to ~1.7 across the 1955-1975 birth cohorts."69- **Mechanism:** Postponement of first births raised the mean age at first birth, and recuperation at70 older ages was incomplete — a quantum decline operating *through* tempo, not a uniform shift.71- **Observable implication:** Parity-progression ratios from parity 0 to 1 should fall most at younger72 ages and only partly rebound later; a pure quantum story would show uniform decline across ages.73- **Scope condition:** The claim covers low-fertility settings with delayed childbearing.74- **Portability:** A mortality scholar can import the tempo-vs-quantum logic to lifespan compression;75 that import makes it a population-science contribution, not a single-country fact.7677## Referee-pushback patterns and the theory-side fix7879- *"This is descriptive — where is the mechanism?"* -> Name the behavior/exposure/compositional shift80 that moves the rate, for whom, and translate it into an age or cohort signature the design can test.81- *"You assert a mechanism but a compositional shift would produce the same trend."* -> State the rival82 (composition) explicitly and the observable that separates it from your account before the results.83- *"The formal model's assumptions are chosen to deliver the identity."* -> Flag which results depend on84 stability/Markov/independence assumptions and motivate each substantively.85- *"This only works for your one case."* -> Generalize the mechanism so another demographer studying a86 different component could reuse it; otherwise reframe via `demog-topic-selection`.8788## Output format8990```91【Core claim】one sentence92【Mechanism / identity】the population story or formal result93【Assumptions】(formal) the load-bearing ones94【Observable implications】testable demographic signatures -> research-design95【Scope conditions】which populations / periods it covers96【Portability】who else in population science can use this97【Next】demog-research-design98```99100## Supplementary resources101102- [`../../resources/external_tools.md`](../../resources/external_tools.md) — formal-demography and decomposition tooling103- [`../../resources/official-source-map.md`](../../resources/official-source-map.md) — Demography scope and contribution expectations