DatabaseDefinitionsSeverity Mix

Severity mix (5-Day share)

Do these operators report crashes the same way?

Reporting obligations behind it
1
The filings a company is required to make that this figure is built from
Known biases ledgered
2
Each carrying the direction it pushes the figure, rather than a warning that data may be imperfect
Of those, unsigned
2
Entries where the record does not say which way the error runs

Formula

Unit: % of filings

count(filed under the five-day trigger) ÷ count(filed under either the five-day or the monthly trigger)

What it is built on

  • Federal crash reporting for automated driving systems, under Standing General Order 2021-01

Bias ledger

Each known problem with this number, and which way it pushes it. A signed direction is more useful than a warning that data may be imperfect.

Effect
Two opposite causesdirection unknownA high share can mean genuinely more severe crashes, or it can mean the operator files only what the tow-away trigger obliges. The number cannot distinguish them and must never be read alone.
Small samplesdirection unknownBelow 20 filings this share is noise. Those operators are shown as raw fractions rather than percentages.
No rows match this filter.
Direction is the sign of the error, not its size: an overstating source pushes the figure up, an understating one pushes it down, and a direction unknown means the record does not say which.

How it is computed

Every step that changes the answer, in the order it happens. Nothing here is a summary of a longer method: this is the method.

  1. Take every crash report on the driverless file, latest version of each.
  2. Split each operator's reports into the ones filed under the five-day trigger and the ones filed on the monthly schedule.
  3. The share is the five-day count over the two of them together. Reports filed under neither are outside the question and are not counted either way.

Not for

Concluding that one operator is more dangerous than another. It measures filing behaviour, which is exactly why raw crash counts between operators are not comparable.

← All metrics and the nine marks