Skills

Epidemiological Study Design & Analysis

10 weeks · 0 milestones

Design and analyse an epidemiological study using real or publicly available health data. Proof requires: (a) a study design document (research question, study type, population, sampling strategy, outcome measures, bias considerations — minimum 1,000 words); and (b) a data analysis of a real dataset (publicly available datasets from WHO, CDC, ONS, or equivalent are acceptable) with documented methodology, results, and a critical discussion of limitations. Reviewed by a public health practitioner or researcher with epidemiology experience who can challenge the study design choices. Proof is the analysis and design reasoning, not the dataset itself.

Part of: Health & Medicine · Proves for: UG

Milestone map

3 milestones

Formulate a PICO research question and select study design

1–2 weeks

Choose a public health question relevant to your training area. Structure it as a PICO question (Population, Intervention/Exposure, Comparator, Outcome). Then select and justify an appropriate epidemiological study design (RCT, cohort study, case-control, cross-sectional, ecological study) for your question. Justify: why this design fits the question, what its main strengths and limitations are, and what alternative design you considered and rejected and why.

Proof required

Submit your PICO question, your study design selection with justification (1 page), and a brief analysis of one alternative design you considered and its limitations for your specific question.

What gets checked

PICO is specific and answerable — 'does diet affect health?' is not a PICO question; 'in adults 40–65 (P), does a Mediterranean diet vs. standard Western diet (I vs. C) reduce 10-year cardiovascular event rate (O)?' is.
Design justification addresses both strengths and limitations specific to your PICO — not generic strengths/limitations of the design type.
Alternative design comparison is genuine — explains specifically why the alternative is less suitable for this question, not just less common.

Common mistakes

PICO question that cannot be answered by any single study — too broad or too vague for epidemiological design.
Study design selected based on familiarity (e.g. always choosing RCT) without addressing feasibility for the specific question.
Alternative design not considered — this step requires genuine comparative reasoning.

Powstik Guide

A PICO question and a justified study design.

Steps

Write each PICO element.
List designs that could answer it.
Pick one and justify it.
Name its main bias risk.

Template

P:        I:        C:        O:
Design:
Why:
Main bias risk:

What gets sent back

PICO elements missing.
No justification.
No bias considered.

Resources

Foundationstart here

Depthgo deeper

What a verifier looks for

Ask the student to defend their design choice against a specific methodological challenge ('why not a randomised trial?').
Check that PICO is specific and answerable — ask 'what data would you need to answer this question?'
Ask what the greatest threat to internal validity in their chosen design is.

You'll sign in first, then come straight back here.

Design the study protocol with bias analysis

2–3 weeks

Produce a study protocol for your chosen design: sampling strategy (inclusion/exclusion criteria, sample size justification), data collection methods, exposure/outcome definitions, confounding variables and how you'll control for them, and ethical considerations (consent, data anonymisation). Then complete a bias analysis identifying potential selection, information, and confounding biases in your design and how each would be minimised.

Proof required

Submit your study protocol (3–5 pages) and bias analysis (1 page), each section clearly labelled. Include a sample size justification using a standard calculation (you may use online calculators such as EpiInfo or G*Power).

What gets checked

Sample size justification shows the inputs used (alpha, power, effect size, baseline prevalence) — not just the output number.
Bias analysis identifies specific biases for your design, not generic biases — explain why this study design is susceptible to these particular biases.
Confounding section identifies named confounders relevant to your specific exposure-outcome pair.

Common mistakes

Sample size justified only by 'convenience' or 'feasibility' without a power calculation — not acceptable for an epidemiological design.
Bias analysis that lists three types of bias without specificity to the design or question.
Ethical considerations that are generic ('data will be kept confidential') without addressing the specific ethical issues of your population or question.

Resources

Foundationstart here

Depthgo deeper

What a verifier looks for

Ask the student to explain the sample size calculation inputs — if they can't name them, they used the calculator without understanding it.
Ask what the most important confounder in their study is and how they'd control for it.
Check that the bias analysis is specific to the study design, not a textbook copy.

You'll sign in first, then come straight back here.

Defend the protocol in a simulated expert review

1–2 weeks (preparation + session + revision)

Present your study protocol in a 30-minute simulated ethics/expert review session with someone who has experience in epidemiological research, public health, or clinical research methods (an academic, a public health registrar, or a research methods tutor). The reviewer challenges your design choices — you must defend them, acknowledge genuine weaknesses, and propose modifications. Write a post-session protocol revision document capturing any changes made as a result of the review.

Proof required

Submit a written session record (reviewer name and credentials, date, duration, main challenges raised) countersigned by the reviewer, your protocol revision document identifying at least two changes made as a result of the session, and the rationale for each change.

What gets checked

Reviewer has epidemiological, public health, or clinical research methods experience — not a general clinical practitioner.
Revision document identifies at least two specific changes — not a general 'the protocol is now improved'.
Rationale for each change links the reviewer's challenge to the specific revision made.

Common mistakes

Reviewer without research methods experience — a clinician who has never designed a study is not a qualified protocol reviewer.
Revision document that lists changes without explaining why the review challenge necessitated them.
Protocol that was not meaningfully changed — if nothing needed revision, either the design was perfect (unlikely for a student) or the review wasn't substantive.

Powstik Guide

A countersigned record of an expert protocol review.

Steps

Book a qualified epidemiologist or researcher.
Send the protocol in advance.
Record each challenge raised.
Revise the protocol in response.

Template

Reviewer:        Credentials:        Date:        Duration:
| Challenge | My response | Change |
|-----------|-------------|--------|
Reviewer signature:

What gets sent back

Not countersigned.
Reviewer credentials missing.
No revisions.

Resources

Foundationstart here

Depthgo deeper

What a verifier looks for

Verify that the reviewer has epidemiological or research methods credentials — not just general clinical experience.
Check that at least two specific changes appear in the revision document with linked rationale.
Ask the student what the hardest challenge raised in the review was and whether they agreed with it.

You'll sign in first, then come straight back here.

Part of

Prove Your Public Health Chops →

Related outcomes

Milestone map

3 milestones

Formulate a PICO research question and select study design

1–2 weeks

Proof required

Submit your PICO question, your study design selection with justification (1 page), and a brief analysis of one alternative design you considered and its limitations for your specific question.

What gets checked

PICO is specific and answerable — 'does diet affect health?' is not a PICO question; 'in adults 40–65 (P), does a Mediterranean diet vs. standard Western diet (I vs. C) reduce 10-year cardiovascular event rate (O)?' is.
Design justification addresses both strengths and limitations specific to your PICO — not generic strengths/limitations of the design type.
Alternative design comparison is genuine — explains specifically why the alternative is less suitable for this question, not just less common.

Common mistakes

PICO question that cannot be answered by any single study — too broad or too vague for epidemiological design.
Study design selected based on familiarity (e.g. always choosing RCT) without addressing feasibility for the specific question.
Alternative design not considered — this step requires genuine comparative reasoning.

Powstik Guide

A PICO question and a justified study design.

Steps

Write each PICO element.
List designs that could answer it.
Pick one and justify it.
Name its main bias risk.

Template

P:        I:        C:        O:
Design:
Why:
Main bias risk:

What gets sent back

PICO elements missing.
No justification.
No bias considered.

Resources

Foundationstart here

Depthgo deeper

What a verifier looks for

Ask the student to defend their design choice against a specific methodological challenge ('why not a randomised trial?').
Check that PICO is specific and answerable — ask 'what data would you need to answer this question?'
Ask what the greatest threat to internal validity in their chosen design is.

You'll sign in first, then come straight back here.

Design the study protocol with bias analysis

2–3 weeks

Proof required

What gets checked

Sample size justification shows the inputs used (alpha, power, effect size, baseline prevalence) — not just the output number.
Bias analysis identifies specific biases for your design, not generic biases — explain why this study design is susceptible to these particular biases.
Confounding section identifies named confounders relevant to your specific exposure-outcome pair.

Common mistakes

Sample size justified only by 'convenience' or 'feasibility' without a power calculation — not acceptable for an epidemiological design.
Bias analysis that lists three types of bias without specificity to the design or question.
Ethical considerations that are generic ('data will be kept confidential') without addressing the specific ethical issues of your population or question.

Resources

Foundationstart here

Depthgo deeper

What a verifier looks for

Ask the student to explain the sample size calculation inputs — if they can't name them, they used the calculator without understanding it.
Ask what the most important confounder in their study is and how they'd control for it.
Check that the bias analysis is specific to the study design, not a textbook copy.

You'll sign in first, then come straight back here.

Defend the protocol in a simulated expert review

1–2 weeks (preparation + session + revision)

Proof required

What gets checked

Reviewer has epidemiological, public health, or clinical research methods experience — not a general clinical practitioner.
Revision document identifies at least two specific changes — not a general 'the protocol is now improved'.
Rationale for each change links the reviewer's challenge to the specific revision made.

Common mistakes

Reviewer without research methods experience — a clinician who has never designed a study is not a qualified protocol reviewer.
Revision document that lists changes without explaining why the review challenge necessitated them.
Protocol that was not meaningfully changed — if nothing needed revision, either the design was perfect (unlikely for a student) or the review wasn't substantive.

Powstik Guide

A countersigned record of an expert protocol review.

Steps

Book a qualified epidemiologist or researcher.
Send the protocol in advance.
Record each challenge raised.
Revise the protocol in response.

Template

Reviewer:        Credentials:        Date:        Duration:
| Challenge | My response | Change |
|-----------|-------------|--------|
Reviewer signature:

What gets sent back

Not countersigned.
Reviewer credentials missing.
No revisions.

Resources

Foundationstart here

Depthgo deeper

What a verifier looks for

Verify that the reviewer has epidemiological or research methods credentials — not just general clinical experience.
Check that at least two specific changes appear in the revision document with linked rationale.
Ask the student what the hardest challenge raised in the review was and whether they agreed with it.

You'll sign in first, then come straight back here.