Visual primitive

Error bars / confidence intervals `has_error_bars`

Visual representation of uncertainty around each point.

Required when reporting estimates with uncertainty (clinical trials, polling, A/B tests). Without error bars, point estimates can mislead. Standard in research papers; rare in consulting decks (often to consulting decks' cost — they present point estimates as if they were certain).

When critical

Forest plots, A/B test results, polling, any aggregate statistic with sample variance.

When optional

Census data, exact accounting figures, deterministic calculations.

Failures when missing

point estimates presented as if certain
reader cannot judge significance

Failures when present (badly)

error bars without specifying confidence level (95%? 99%?)
asymmetric error bars labelled as symmetric

Critical for families

distribution
correlation

Aspose property bindings

ErrorBars.type
ErrorBars.amount
ErrorBars.direction