Person-Based vs. Event-Based Rates

The source material includes measures that count people and others that count events. This page clarifies the difference, shows how each rate is calculated, and outlines how the choice of metric shapes the trend you see.

What We Track

Person-based rate: Each individual is counted once in the period or follow-up window. Example: “Share of youth with at least one new adjudication within 12 months.”
Event-based rate: Every qualifying incident is counted, even if multiple events come from the same person. Example: “Number of petitions per 100 youth in the cohort.”
Denominators: Person rates typically divide by the number of people in the cohort; event rates may divide by people (yielding events-per-person) or by population (yielding incidence).
Time base: Many person rates use fixed follow-up windows (e.g., 6/12/24 months), while event rates are often calendar-period counts (e.g., CY2022).

Formulas at a Glance

Person-based: rate = (# people with ≥1 qualifying outcome) ÷ (# people in cohort)
Event-based (per 100): rate = (total qualifying events ÷ # people in cohort) × 100
Population incidence (per 1,000): rate = (events ÷ mid-year population) × 1,000

Typical Flow

Identify the unit: Decide whether the analysis is person- or event-focused.
Build the denominator: Cohort count for follow-up studies; population or cohort for event series.
Aggregate outcomes: For person rates, collapse multiple outcomes per person to one flag.
Compute and label: Include the unit (“people” vs. “events”), the time base, and any normalization (per 100, per 1,000).
Document exceptions: Note if technical violations are included, if multiple events per day are bundled, or if partial-year exposure exists.

Comparability & Caveats

Inflation from repeat events: Event-based series will exceed person rates when the same individuals generate multiple incidents. This can exaggerate peaks in periods with short cycles of re-contact.
Denominator drift: Switching between population and cohort denominators changes the rate’s meaning. Keep denominators consistent within a chart.
Window vs. calendar: Mixing fixed follow-up windows with calendar-year event counts can distort trend timing. Align periods when possible.
Suppression and pooling: Small cohorts may require year pooling or suppression; see Small-N & Suppression.

Data & Methods

This summary is distilled from the research text file. When both person- and event-based fields are available, consider building parallel series and labeling each clearly. If exposure time varies (e.g., partial-year supervision), note the convention used (fixed-day window vs. calendar year) and any adjustments for multiple events occurring on the same day or within the same proceeding.

Transparency note: When cohorts are small or exposure is uneven, parallel displays (person vs. event) and clear labels help prevent misinterpretation.