Editorial standards

Every published figure is computed from stored postings, not estimated. We show what the data says and, where it matters, how large the sample is. When evidence is thin, we say so rather than imply confidence we do not have.

A statistic is only reported over the postings that actually state the underlying field. Where an attribute (for example work arrangement, compensation, or application deadlines) is stated on too few postings to be meaningful, we withhold the distribution and note that we are doing so, rather than publish a misleading number.

We never invent a statistic, a company, or a job. Where AI assists in reading a posting, it only extracts fields that the posting states; it never fills in a missing value. Aggregate numbers are computed deterministically from those stored fields.

When we get something wrong, we fix it and, where relevant, note it. See corrections.