Time-series diagnostics and sensitivity

Assess seasonality shocks and specifications

Check residual structure and sensitivity to trend, timing and comparison choices. A good-fitting model does not rule out concurrent changes or establish that the extrapolated trajectory is a valid counterfactual.

Flow diagram

Time-series diagnostics and sensitivityF1.S1Inspect residual structureF1.S2Review concurrent eventsF1.S3Vary justified specificationsF1.S4Assess control-series assumptionsF1.S5Report fragility and limitationsF1.N1Time-series diagnostics planConcernDiagnosticAlternativeRationaleSerial dependenceSeasonalityConcurrent shockDefinition changeImpact lagF1.N2Assumptions and rationaleF1.N3Method and protocol evidenceF1.S0 Start flowTime-series diagnostics andsensitivityF1.S999 End flowEnd flow

Steps

Inspect residual structure

Plan checks for autocorrelation, seasonality and model fit.

Review concurrent events

Assess alternative explanations and changes in outcome ascertainment.

Vary justified specifications

Compare lag, interruption dates, trend forms and interval widths.

Assess control-series assumptions

Evaluate contamination or different background changes where controls are used.

Report fragility and limitations

Retain all prespecified sensitivity comparisons and contextual evidence.

Time-series diagnostics and sensitivity

End flow

Tables and supporting information

Time-series diagnostics plan

Time-series diagnostics planConcernDiagnosticAlternativeRationaleSerial dependenceSeasonalityConcurrent shockDefinition changeImpact lag

Diagnostics inform assumptions; they do not prove intervention causality.

Assumptions and rationale

Short or unstable series can limit identification and precision even with many individuals per interval.

Method and protocol evidence

Attach the protocol, methodological sources and versioned software specification supporting these decisions. Record citations and provenance; do not enter participant identifiers.

Sources