Practical guides · Page 4
Search and filter all practical guidesRepresent interval-censored event times correctly
Keep last-negative and first-positive observations instead of inventing exact event dates.
State the population selected by a landmark analysis
Define a fixed landmark, available exposure information and the conditional question being answered.
Audit Language Restrictions in an Evidence Search
Make language decisions visible and test whether translation limitations are changing which studies can enter a review.
Preserve Leading Zeros in Research Identifiers
Import identifiers as text and detect collisions before a spreadsheet or parser converts them to numbers.
Handle delayed entry in a survival analysis
Align the time origin and entry time so participants join the risk set only when they become observable.
Distinguish Leverage, Residual Size, and Influence
Investigate observations that shape a regression without treating a diagnostic threshold as an automatic exclusion rule.
Keep detection limits distinct from observed values
Preserve censoring flags and assay limits before choosing how to analyse low measurements.
Use a Logarithmic Axis for Multiplicative Comparisons
Label original ratio values while preserving equal visual distances for reciprocal effects.
Back-Transform a Log-Scale Effect and Its Interval Together
Exponentiate ratio estimates consistently while keeping standard errors and interpretation on their proper scales.
Reshape Repeated Measurements Without Aggregating Them Away
Specify the observation key and expected cells before converting between long and wide research tables.
Publish a Results Table That Can Be Reused Without Guessing
Separate numeric estimates from display formatting and attach the metadata needed to interpret each row.
Preserve matched sets in case-control analysis
Audit matching identifiers and exposure variation before fitting a conditional logistic model.