Practical guides · Page 10
Search and filter all practical guidesAudit Survey Strata and Primary Sampling Unit Identifiers
Check nesting, singleton strata and design metadata before estimating survey uncertainty.
Review Weight Trimming as a Bias–Precision Tradeoff
Investigate influential survey weights and document a justified trimming sensitivity analysis.
Align Eligibility, Treatment Assignment and Follow-Up at Time Zero
Write the hypothetical trial before translating an observational cohort into analysis rows.
Distinguish technical repeats from independent replication
Identify the experimental unit before counting observations or calculating uncertainty.
Validate Predictions on Later Data Without Leaking Future Information
Freeze the prediction pipeline and align evaluation dates with the intended deployment sequence.
Choose a Classification Threshold Using the Intended Consequences
Separate probability estimation from the decision rule and validate any tuned cutoff.
Code time-varying exposure without using future information
Split follow-up into valid intervals and test exposure values at every event-time boundary.
Search Trial Registries Beyond Published Articles
Preserve registry coverage, recruitment status, and linked reports when looking for unpublished or ongoing studies.
Keep Unavailable Full Texts Separate from Ineligible Studies
Document retrieval attempts and unresolved eligibility without treating access failure as evidence of exclusion.
Audit Unicode Differences Before Matching Text Identifiers
Separate visually similar text from exact identifier equality and document any normalization that changes matches.
Test Derived Research Variables with Small Known Examples
Use boundary cases and independently calculated expected values to check derivations before running them on study data.
Plan an E-Value Analysis Without Treating It as Proof of Causality
Quantify a specified confounding sensitivity while keeping other biases and effect scales separate.