Why English? | Reporting STROBE-MR Mendelian Randomisation Studies Through Genetic Instruments, Assumptions, Data Sources, Sensitivity Analyses and Causal Limits

ORIENTATION

What this guide owns

Current fact check

STROBE-MR was finalised in July 2021 and published in JAMA in October 2021. Its six-section checklist contains 20 main items and 30 subitems and covers one-sample, two-sample and multivariable Mendelian-randomisation settings.

Canonical boundary

This article owns transparent reporting of Mendelian-randomisation studies. The general STROBE article owns cohort, case-control and cross-sectional reports, while GWAS reporting and laboratory genetics require their own guidance.

Worked example

Every chapter returns to a fictional two-sample Mendelian-randomisation study asking whether genetically proxied longer habitual sleep influences examination-related anxiety in adolescents. It is fictional, low-stakes and educational; it does not report a real school, clinical decision or guaranteed outcome.

Warm up with eduKate’s guides to explaining scientific evidence and asking better questions in class.

CHAPTER 1 OF 17

1. Identify the Study as Mendelian Randomisation

Back to contents

Why this move matters

Use the title and abstract to name MR, the exposure, outcome, design, data sources, principal method and main estimate.

Draw the causal question before the analysis

Use the title and abstract to name MR, the exposure, outcome, design, data sources, principal method and main estimate. Name exposure, outcome, population, time frame, estimand and why genetic variation may help answer this question.

Audit one genetic instrument

Worked example: The title identifies a two-sample MR study of habitual sleep and anxiety. Record the variant source, association strength, allele orientation, population, harmonisation, exclusions and evidence relevant to every MR assumption.

Triangulate rather than certify

Compare estimates across methods and sensitivity analyses, then ask which violations each method can and cannot detect. Agreement strengthens an interpretation only when shared biases are considered.

Use causal language proportionately

The MR label does not establish that the instrument assumptions hold. Mendelian randomisation can inform causal inference under assumptions; it does not make those assumptions automatically true or replace trials, biology and other observational evidence.

Student, parent and career transfer

Make one evidence card for “Identify the Study as Mendelian Randomisation”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports genetic epidemiology, biostatistics, public health, bioinformatics, causal inference, peer review and science communication without promising a grade, admission, funding decision or job.

CHAPTER 2 OF 17

2. Explain Why MR Helps This Particular Question

Back to contents

Why this move matters

Describe conventional evidence, possible confounding or reverse causation and why genetic proxies may add information.

Draw the causal question before the analysis

Describe conventional evidence, possible confounding or reverse causation and why genetic proxies may add information. Name exposure, outcome, population, time frame, estimand and why genetic variation may help answer this question.

Audit one genetic instrument

Worked example: Self-reported sleep and anxiety may influence each other, motivating an additional causal-inference approach. Record the variant source, association strength, allele orientation, population, harmonisation, exclusions and evidence relevant to every MR assumption.

Triangulate rather than certify

Compare estimates across methods and sensitivity analyses, then ask which violations each method can and cannot detect. Agreement strengthens an interpretation only when shared biases are considered.

Use causal language proportionately

MR does not automatically solve measurement error, selection or population structure. Mendelian randomisation can inform causal inference under assumptions; it does not make those assumptions automatically true or replace trials, biology and other observational evidence.

Student, parent and career transfer

Make one evidence card for “Explain Why MR Helps This Particular Question”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports genetic epidemiology, biostatistics, public health, bioinformatics, causal inference, peer review and science communication without promising a grade, admission, funding decision or job.

CHAPTER 3 OF 17

3. State Prespecified Causal Hypotheses and Estimands

Back to contents

Why this move matters

Name direction, exposure contrast, outcome scale, target population and whether the estimate concerns lifelong liability or a modifiable intervention.

Draw the causal question before the analysis

Name direction, exposure contrast, outcome scale, target population and whether the estimate concerns lifelong liability or a modifiable intervention. Name exposure, outcome, population, time frame, estimand and why genetic variation may help answer this question.

Audit one genetic instrument

Worked example: The hypothesis concerns the effect of genetically proxied longer habitual sleep on anxiety score. Record the variant source, association strength, allele orientation, population, harmonisation, exclusions and evidence relevant to every MR assumption.

Triangulate rather than certify

Compare estimates across methods and sensitivity analyses, then ask which violations each method can and cannot detect. Agreement strengthens an interpretation only when shared biases are considered.

Use causal language proportionately

A genetic-liability estimate may not equal the effect of changing sleep for six weeks. Mendelian randomisation can inform causal inference under assumptions; it does not make those assumptions automatically true or replace trials, biology and other observational evidence.

Student, parent and career transfer

Make one evidence card for “State Prespecified Causal Hypotheses and Estimands”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports genetic epidemiology, biostatistics, public health, bioinformatics, causal inference, peer review and science communication without promising a grade, admission, funding decision or job.

CHAPTER 4 OF 17

4. Write the Three Core Instrument Assumptions in Context

Back to contents

Why this move matters

Explain relevance, independence from confounders and exclusion restriction using the actual variants, exposure, outcome and population.

Draw the causal question before the analysis

Explain relevance, independence from confounders and exclusion restriction using the actual variants, exposure, outcome and population. Name exposure, outcome, population, time frame, estimand and why genetic variation may help answer this question.

Audit one genetic instrument

Worked example: Variants must predict sleep, be independent of common causes of sleep and anxiety, and affect anxiety only through sleep. Record the variant source, association strength, allele orientation, population, harmonisation, exclusions and evidence relevant to every MR assumption.

Triangulate rather than certify

Compare estimates across methods and sensitivity analyses, then ask which violations each method can and cannot detect. Agreement strengthens an interpretation only when shared biases are considered.

Use causal language proportionately

The assumptions are substantive claims, not boxes proved by one statistical test. Mendelian randomisation can inform causal inference under assumptions; it does not make those assumptions automatically true or replace trials, biology and other observational evidence.

Student, parent and career transfer

Make one evidence card for “Write the Three Core Instrument Assumptions in Context”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports genetic epidemiology, biostatistics, public health, bioinformatics, causal inference, peer review and science communication without promising a grade, admission, funding decision or job.

CHAPTER 5 OF 17

5. Describe Every Data Source and Sample

Back to contents

Why this move matters

Report study names, recruitment, eligibility, dates, ancestry, geography, genotyping, phenotype measurement, approvals and sample sizes.

Draw the causal question before the analysis

Report study names, recruitment, eligibility, dates, ancestry, geography, genotyping, phenotype measurement, approvals and sample sizes. Name exposure, outcome, population, time frame, estimand and why genetic variation may help answer this question.

Audit one genetic instrument

Worked example: Exposure associations come from one consortium and outcome associations from a separate adolescent cohort. Record the variant source, association strength, allele orientation, population, harmonisation, exclusions and evidence relevant to every MR assumption.

Triangulate rather than certify

Compare estimates across methods and sensitivity analyses, then ask which violations each method can and cannot detect. Agreement strengthens an interpretation only when shared biases are considered.

Use causal language proportionately

Large samples can still carry selection, measurement and ancestry limitations. Mendelian randomisation can inform causal inference under assumptions; it does not make those assumptions automatically true or replace trials, biology and other observational evidence.

Student, parent and career transfer

Make one evidence card for “Describe Every Data Source and Sample”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports genetic epidemiology, biostatistics, public health, bioinformatics, causal inference, peer review and science communication without promising a grade, admission, funding decision or job.

CHAPTER 6 OF 17

6. Report Sample Overlap and Participant Flow

Back to contents

Why this move matters

Quantify or discuss overlap between exposure and outcome samples and show exclusions, missingness and final analytic numbers.

Draw the causal question before the analysis

Quantify or discuss overlap between exposure and outcome samples and show exclusions, missingness and final analytic numbers. Name exposure, outcome, population, time frame, estimand and why genetic variation may help answer this question.

Audit one genetic instrument

Worked example: The consortia are checked for shared cohorts, and likely overlap is estimated. Record the variant source, association strength, allele orientation, population, harmonisation, exclusions and evidence relevant to every MR assumption.

Triangulate rather than certify

Compare estimates across methods and sensitivity analyses, then ask which violations each method can and cannot detect. Agreement strengthens an interpretation only when shared biases are considered.

Use causal language proportionately

Unrecognised overlap can change weak-instrument bias. Mendelian randomisation can inform causal inference under assumptions; it does not make those assumptions automatically true or replace trials, biology and other observational evidence.

Student, parent and career transfer

Make one evidence card for “Report Sample Overlap and Participant Flow”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports genetic epidemiology, biostatistics, public health, bioinformatics, causal inference, peer review and science communication without promising a grade, admission, funding decision or job.

CHAPTER 7 OF 17

7. Define Exposure, Outcome and Covariate Measurement

Back to contents

Why this move matters

State instruments, units, transformations, case definitions, time frames and whether measurements are comparable across contributing studies.

Draw the causal question before the analysis

State instruments, units, transformations, case definitions, time frames and whether measurements are comparable across contributing studies. Name exposure, outcome, population, time frame, estimand and why genetic variation may help answer this question.

Audit one genetic instrument

Worked example: Sleep duration is reported in hours with study-specific harmonisation, while anxiety uses the named scale. Record the variant source, association strength, allele orientation, population, harmonisation, exclusions and evidence relevant to every MR assumption.

Triangulate rather than certify

Compare estimates across methods and sensitivity analyses, then ask which violations each method can and cannot detect. Agreement strengthens an interpretation only when shared biases are considered.

Use causal language proportionately

Similar labels can hide different questionnaires or thresholds. Mendelian randomisation can inform causal inference under assumptions; it does not make those assumptions automatically true or replace trials, biology and other observational evidence.

Student, parent and career transfer

Make one evidence card for “Define Exposure, Outcome and Covariate Measurement”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports genetic epidemiology, biostatistics, public health, bioinformatics, causal inference, peer review and science communication without promising a grade, admission, funding decision or job.

CHAPTER 8 OF 17

8. Explain Variant Selection and Genome-Wide Association Sources

Back to contents

Why this move matters

Give thresholds, clumping or pruning, linkage disequilibrium reference, region exclusions, replication and whether variants were discovered in the same data.

Draw the causal question before the analysis

Give thresholds, clumping or pruning, linkage disequilibrium reference, region exclusions, replication and whether variants were discovered in the same data. Name exposure, outcome, population, time frame, estimand and why genetic variation may help answer this question.

Audit one genetic instrument

Worked example: Independent variants meeting the prespecified threshold are selected using a matched reference panel. Record the variant source, association strength, allele orientation, population, harmonisation, exclusions and evidence relevant to every MR assumption.

Triangulate rather than certify

Compare estimates across methods and sensitivity analyses, then ask which violations each method can and cannot detect. Agreement strengthens an interpretation only when shared biases are considered.

Use causal language proportionately

Selection based on the analysed outcome can introduce bias. Mendelian randomisation can inform causal inference under assumptions; it does not make those assumptions automatically true or replace trials, biology and other observational evidence.

Student, parent and career transfer

Make one evidence card for “Explain Variant Selection and Genome-Wide Association Sources”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports genetic epidemiology, biostatistics, public health, bioinformatics, causal inference, peer review and science communication without promising a grade, admission, funding decision or job.

CHAPTER 9 OF 17

9. Report Instrument Strength and Direction

Back to contents

Why this move matters

Provide associations with exposure, uncertainty, units, F statistics or other strength metrics and checks that variants predict exposure rather than outcome direction.

Draw the causal question before the analysis

Provide associations with exposure, uncertainty, units, F statistics or other strength metrics and checks that variants predict exposure rather than outcome direction. Name exposure, outcome, population, time frame, estimand and why genetic variation may help answer this question.

Audit one genetic instrument

Worked example: Per-allele sleep associations and conditional strength statistics are tabulated. Record the variant source, association strength, allele orientation, population, harmonisation, exclusions and evidence relevant to every MR assumption.

Triangulate rather than certify

Compare estimates across methods and sensitivity analyses, then ask which violations each method can and cannot detect. Agreement strengthens an interpretation only when shared biases are considered.

Use causal language proportionately

A high average strength measure can hide a weak component in multivariable MR. Mendelian randomisation can inform causal inference under assumptions; it does not make those assumptions automatically true or replace trials, biology and other observational evidence.

Student, parent and career transfer

Make one evidence card for “Report Instrument Strength and Direction”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports genetic epidemiology, biostatistics, public health, bioinformatics, causal inference, peer review and science communication without promising a grade, admission, funding decision or job.

CHAPTER 10 OF 17

10. Document Allele Harmonisation and Ambiguity Rules

Back to contents

Why this move matters

Explain effect alleles, strand alignment, palindromic variants, proxies, allele frequencies, incompatible variants and final exclusions.

Draw the causal question before the analysis

Explain effect alleles, strand alignment, palindromic variants, proxies, allele frequencies, incompatible variants and final exclusions. Name exposure, outcome, population, time frame, estimand and why genetic variation may help answer this question.

Audit one genetic instrument

Worked example: Alleles are aligned to sleep-increasing orientation, with ambiguous high-frequency variants removed by rule. Record the variant source, association strength, allele orientation, population, harmonisation, exclusions and evidence relevant to every MR assumption.

Triangulate rather than certify

Compare estimates across methods and sensitivity analyses, then ask which violations each method can and cannot detect. Agreement strengthens an interpretation only when shared biases are considered.

Use causal language proportionately

A silent allele flip can reverse an estimate. Mendelian randomisation can inform causal inference under assumptions; it does not make those assumptions automatically true or replace trials, biology and other observational evidence.

Student, parent and career transfer

Make one evidence card for “Document Allele Harmonisation and Ambiguity Rules”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports genetic epidemiology, biostatistics, public health, bioinformatics, causal inference, peer review and science communication without promising a grade, admission, funding decision or job.

CHAPTER 11 OF 17

11. Describe the Primary Estimator and Its Assumptions

Back to contents

Why this move matters

Name the estimator, weighting, correlation handling, model choices, software and conditions under which it estimates the target effect.

Draw the causal question before the analysis

Name the estimator, weighting, correlation handling, model choices, software and conditions under which it estimates the target effect. Name exposure, outcome, population, time frame, estimand and why genetic variation may help answer this question.

Audit one genetic instrument

Worked example: Inverse-variance weighting is primary, with correlated variants handled by a stated matrix. Record the variant source, association strength, allele orientation, population, harmonisation, exclusions and evidence relevant to every MR assumption.

Triangulate rather than certify

Compare estimates across methods and sensitivity analyses, then ask which violations each method can and cannot detect. Agreement strengthens an interpretation only when shared biases are considered.

Use causal language proportionately

A familiar estimator still depends on valid instruments and modelling choices. Mendelian randomisation can inform causal inference under assumptions; it does not make those assumptions automatically true or replace trials, biology and other observational evidence.

Student, parent and career transfer

Make one evidence card for “Describe the Primary Estimator and Its Assumptions”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports genetic epidemiology, biostatistics, public health, bioinformatics, causal inference, peer review and science communication without promising a grade, admission, funding decision or job.

CHAPTER 12 OF 17

12. Report Pleiotropy and Heterogeneity Diagnostics

Back to contents

Why this move matters

State which diagnostics assess directional pleiotropy, heterogeneity, outliers or influential variants and what each can detect.

Draw the causal question before the analysis

State which diagnostics assess directional pleiotropy, heterogeneity, outliers or influential variants and what each can detect. Name exposure, outcome, population, time frame, estimand and why genetic variation may help answer this question.

Audit one genetic instrument

Worked example: Intercept, heterogeneity and leave-one-out results are reported with limitations. Record the variant source, association strength, allele orientation, population, harmonisation, exclusions and evidence relevant to every MR assumption.

Triangulate rather than certify

Compare estimates across methods and sensitivity analyses, then ask which violations each method can and cannot detect. Agreement strengthens an interpretation only when shared biases are considered.

Use causal language proportionately

A non-significant diagnostic is not proof that pleiotropy is absent. Mendelian randomisation can inform causal inference under assumptions; it does not make those assumptions automatically true or replace trials, biology and other observational evidence.

Student, parent and career transfer

Make one evidence card for “Report Pleiotropy and Heterogeneity Diagnostics”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports genetic epidemiology, biostatistics, public health, bioinformatics, causal inference, peer review and science communication without promising a grade, admission, funding decision or job.

CHAPTER 13 OF 17

13. Use Sensitivity Analyses With Different Failure Modes

Back to contents

Why this move matters

Choose methods such as weighted median, mode, MR-Egger, robust approaches, colocalisation or negative controls because they answer different concerns.

Draw the causal question before the analysis

Choose methods such as weighted median, mode, MR-Egger, robust approaches, colocalisation or negative controls because they answer different concerns. Name exposure, outcome, population, time frame, estimand and why genetic variation may help answer this question.

Audit one genetic instrument

Worked example: Several estimators are compared, and disagreement is investigated rather than averaged away. Record the variant source, association strength, allele orientation, population, harmonisation, exclusions and evidence relevant to every MR assumption.

Triangulate rather than certify

Compare estimates across methods and sensitivity analyses, then ask which violations each method can and cannot detect. Agreement strengthens an interpretation only when shared biases are considered.

Use causal language proportionately

Many analyses can create selective emphasis if their roles were not prespecified. Mendelian randomisation can inform causal inference under assumptions; it does not make those assumptions automatically true or replace trials, biology and other observational evidence.

Student, parent and career transfer

Make one evidence card for “Use Sensitivity Analyses With Different Failure Modes”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports genetic epidemiology, biostatistics, public health, bioinformatics, causal inference, peer review and science communication without promising a grade, admission, funding decision or job.

CHAPTER 14 OF 17

14. Present Estimates With Units, Precision and Direction

Back to contents

Why this move matters

Report variant-level and combined estimates, confidence intervals, scaling and transformations so readers can interpret magnitude.

Draw the causal question before the analysis

Report variant-level and combined estimates, confidence intervals, scaling and transformations so readers can interpret magnitude. Name exposure, outcome, population, time frame, estimand and why genetic variation may help answer this question.

Audit one genetic instrument

Worked example: The estimate is expressed per genetically proxied one-hour longer sleep with a confidence interval on anxiety score. Record the variant source, association strength, allele orientation, population, harmonisation, exclusions and evidence relevant to every MR assumption.

Triangulate rather than certify

Compare estimates across methods and sensitivity analyses, then ask which violations each method can and cannot detect. Agreement strengthens an interpretation only when shared biases are considered.

Use causal language proportionately

A precise estimate may still be biased by violated assumptions. Mendelian randomisation can inform causal inference under assumptions; it does not make those assumptions automatically true or replace trials, biology and other observational evidence.

Student, parent and career transfer

Make one evidence card for “Present Estimates With Units, Precision and Direction”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports genetic epidemiology, biostatistics, public health, bioinformatics, causal inference, peer review and science communication without promising a grade, admission, funding decision or job.

CHAPTER 15 OF 17

15. Show Results of Assumption Checks and Data Quality Work

Back to contents

Why this move matters

Report instrument strength, harmonisation losses, heterogeneity, pleiotropy tests, influential variants, sample overlap and missingness alongside the main result.

Draw the causal question before the analysis

Report instrument strength, harmonisation losses, heterogeneity, pleiotropy tests, influential variants, sample overlap and missingness alongside the main result. Name exposure, outcome, population, time frame, estimand and why genetic variation may help answer this question.

Audit one genetic instrument

Worked example: The main table connects each diagnostic to the concern it addresses. Record the variant source, association strength, allele orientation, population, harmonisation, exclusions and evidence relevant to every MR assumption.

Triangulate rather than certify

Compare estimates across methods and sensitivity analyses, then ask which violations each method can and cannot detect. Agreement strengthens an interpretation only when shared biases are considered.

Use causal language proportionately

Passing reported checks does not validate untestable assumptions completely. Mendelian randomisation can inform causal inference under assumptions; it does not make those assumptions automatically true or replace trials, biology and other observational evidence.

Student, parent and career transfer

Make one evidence card for “Show Results of Assumption Checks and Data Quality Work”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports genetic epidemiology, biostatistics, public health, bioinformatics, causal inference, peer review and science communication without promising a grade, admission, funding decision or job.

CHAPTER 16 OF 17

16. Interpret Causality With Biological and Population Limits

Back to contents

Why this move matters

Discuss alternative explanations, lifelong genetic liability, developmental timing, ancestry, selection, canalisation and how estimates compare with other evidence.

Draw the causal question before the analysis

Discuss alternative explanations, lifelong genetic liability, developmental timing, ancestry, selection, canalisation and how estimates compare with other evidence. Name exposure, outcome, population, time frame, estimand and why genetic variation may help answer this question.

Audit one genetic instrument

Worked example: The discussion avoids claiming that a short sleep intervention would reproduce the genetic estimate. Record the variant source, association strength, allele orientation, population, harmonisation, exclusions and evidence relevant to every MR assumption.

Triangulate rather than certify

Compare estimates across methods and sensitivity analyses, then ask which violations each method can and cannot detect. Agreement strengthens an interpretation only when shared biases are considered.

Use causal language proportionately

Transport across ancestry or age groups requires evidence, not optimism. Mendelian randomisation can inform causal inference under assumptions; it does not make those assumptions automatically true or replace trials, biology and other observational evidence.

Student, parent and career transfer

Make one evidence card for “Interpret Causality With Biological and Population Limits”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports genetic epidemiology, biostatistics, public health, bioinformatics, causal inference, peer review and science communication without promising a grade, admission, funding decision or job.

CHAPTER 17 OF 17

17. Report Ethics, Funding, Conflicts, Data and Code Access

Back to contents

Why this move matters

State approvals or source-study governance, funder role, interests and where harmonised data, variant lists, code and analytical decisions can be accessed.

Draw the causal question before the analysis

State approvals or source-study governance, funder role, interests and where harmonised data, variant lists, code and analytical decisions can be accessed. Name exposure, outcome, population, time frame, estimand and why genetic variation may help answer this question.

Audit one genetic instrument

Worked example: The repository includes variant tables, code, software versions and a data-access explanation. Record the variant source, association strength, allele orientation, population, harmonisation, exclusions and evidence relevant to every MR assumption.

Triangulate rather than certify

Compare estimates across methods and sensitivity analyses, then ask which violations each method can and cannot detect. Agreement strengthens an interpretation only when shared biases are considered.

Use causal language proportionately

Open code cannot make restricted participant data public or repair poor reporting. Mendelian randomisation can inform causal inference under assumptions; it does not make those assumptions automatically true or replace trials, biology and other observational evidence.

Student, parent and career transfer

Make one evidence card for “Report Ethics, Funding, Conflicts, Data and Code Access”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports genetic epidemiology, biostatistics, public health, bioinformatics, causal inference, peer review and science communication without promising a grade, admission, funding decision or job.

PRACTICE LAB

Build a instrument-to-inference decision trail

Student route

1. Choose one original report inside this guide’s scope.
2. Copy its objective and underline actor, action, context, evidence and intended decision.
3. Draw the specific reasoning chain used in this article.
4. Mark each step reported, inferred, contested or unknown.
5. Rewrite the conclusion so it preserves the narrowest important limitation.

Parent or teacher route

Ask the learner to explain STROBE-MR’s central reporting method twice: once to a classmate and once to a method specialist. Praise a precise correction and a well-stated unknown; both show disciplined attention to evidence.

Early-career route

Create a one-page STROBE-MR audit with source link, date checked, scope, method logic, decisive wording, contrary evidence, limit and next action. A colleague should be able to continue without guessing.

Continue with eduKate’s guides to reading course and career information and giving useful feedback in group projects.

USEFUL NEXT READING

Connect this narrow owner to its neighbour

Read the general STROBE observational-study reporting guide for a related but different reader job. Return here when the question is specifically how to make every mendelian-randomisation assumption and data transformation visible before using causal language. This boundary keeps one canonical owner per intent.

For English practice across research, school and career communication, visit the English at work and in public life guide.

FREQUENTLY ASKED QUESTIONS

Questions students and parents often ask

What is STROBE-MR?

A stand-alone extension for reporting Mendelian-randomisation studies.

How many checklist items are there?

Twenty main items and 30 subitems across six sections.

Does MR prove causation?

It can inform causal inference when assumptions are sufficiently credible; it does not make them automatically true.

What is horizontal pleiotropy?

A genetic variant affects the outcome through a pathway other than the exposure of interest.

Why report ancestry and samples?

Variant associations, linkage patterns, selection and transfer can differ across populations.

Are sensitivity analyses a guarantee?

No. They test different vulnerabilities and must be interpreted with their own assumptions and power limits.