Measured, derived, absent

A data-honesty taxonomy that labels every reading as exactly one of three states. Measured is a published series shown as-is — the highest confidence. Derived is a labeled estimate built from public inputs when no direct series exists — honest that it's an estimate, never passed off as a measurement. Absent is no usable public data, so the product says so plainly instead of inventing a number.

Why it matters

Any data product can adopt it, and the whole point is that you can always tell which of the three any reading is — a measured fact, a labeled estimate, or an honest gap — before you bet a decision on it. The Cost Index tags every ingredient as exactly one of the three: measured shows a solid line and a source date, derived shows a dashed line and an estimate badge, absent shows a greyed card that names the structural reason rather than a fabricated price.

Watch the 90-second explainer

Frequently asked

What do measured, derived, and absent mean?

They are the three honesty states a data product can put on any number. Measured means a published series shown exactly as the agency published it — the highest confidence. Derived means a labeled estimate built from public inputs because no direct series exists; it is honest that it is an estimate and is never dressed up as a measurement. Absent means there is no usable public data, so the product says so plainly instead of inventing a figure. The point is that a reader can always tell which of the three any reading is.

Why label a gap as absent instead of just leaving the number blank?

A blank invites the reader to guess, or to assume the product missed it. Naming the gap as absent — and saying why no public series exists — turns a silence into information. It tells you the reason the number is missing, so you know not to wait for one and can go find it another way. A named gap is more trustworthy than a quietly invented one.

How does the Muntin Cost Index use the measured/derived/absent spine?

Every ingredient in the Cost Index is tagged as exactly one of the three states, and an untagged ingredient fails the build. A measured reading shows a solid line with a source link and report date; a derived reading shows a dashed line, a confidence band, and an estimate badge; an absent reading shows a greyed card that names the structural reason — for example, a confidentiality suppression — rather than a fabricated price. So you can always see which state any reading sits in before you trust it.

Is derived data less reliable than measured?

Yes, and the Cost Index says so. A derived reading is an estimate built from public inputs when no direct series exists — honest, but a step below a measured series shown as-is. That is why it ships with a dashed line, a confidence band, and an estimate badge: useful, clearly labeled, and never mistaken for a measurement.