Sources and the dataset ledger

Four classes of source, and nothing else

A state regulator’s premium survey is a consumer publication in which insurers file a sample premium for hypothetical drivers the regulator defines. It is not a market survey and not a price list. Such a survey exists in 13 jurisdictions, in formats ranging from a public data endpoint to a several-hundred-page PDF, and each defines its drivers differently.

A state statute is the text that sets a jurisdiction’s minimum liability limits, its tort system and its rating rules. All 51 jurisdictions are covered from statute text, with a department or motor-vehicle agency page as the second source.

The federal price index is the Bureau of Labor Statistics motor-vehicle-insurance series CUUR0000SETE, with the all-items series CUUR0000SA0 used for real-terms comparisons. Both are federal government work and public domain.

The census population file decides only which geographies ship first. Where a regulator publishes more places than one batch can carry, the most populous are built first; the file settles that ordering and contributes no figure to any page.

Retrieval dates and effective dates are different things

Two dates appear against each premium source and they are never interchanged. The rates-effective date is the document’s own statement of when the filed rates were in force. The retrieval date is when we fetched the artefact. A retrieval date says nothing about how current the figures are, and never appears where a rates date belongs.

Among the premium sources, 3 publish no rates-effective date at all: Florida, Nevada and Utah. Their pages say so in those words and show only the retrieval date. Among the dated sources the oldest is Arizona at March 2023 and the newest is Maryland at August 2026, a spread of years inside one supposedly comparable class of publication, and the clearest single argument against a national average.

Licence position, by class

State government works are generally not subject to federal copyright, and none of the regulators publishing these surveys asserts a state-law claim over them. Most attach only a disclaimer that the figures are illustrative and that no insurer is guaranteed to sell at the price shown, which this site repeats rather than works around. One regulator permits reproduction of its guide only in its entirety; we reproduce none of it and publish only new statistics computed from the underlying premiums, which are uncopyrightable facts. Federal price-index data and statutes are public domain.

The two sources that are cited-only or excluded

The National Association of Insurance Commissioners publishes the report behind most national average figures in circulation. Its front matter prohibits reproduction without written permission, its data reaches publication roughly two years after the period it covers, and the report itself prints a caution against direct state-by-state comparison. It is cited in words on one explainer page and is a data source nowhere on this site.

A widely quoted industry institute permits personal and other non-commercial use only. This site carries advertising, so that body is excluded outright: not a source, not a citation, not a cross-check. A build gate fails if any reference to it reaches a published page.

State regulator premium surveys

State government works carry no reuse restriction in these jurisdictions. Where a publisher attaches a condition we follow it: one guide permits reproduction only in its entirety, so no part of it is reproduced here and only computed statistics are published.

Federal series

Sources we do not use

Two well-known sources are absent by decision rather than oversight. One trade association’s terms permit non-commercial use only, and this site carries advertising, so it is not a source, not a citation and not a cross-check. The other, a national association of state regulators, prohibits reproduction of its report without written permission, publishes its data more than two years after the year it describes, and prints its own caution against the state-by-state comparison that most published tables build on it. It is discussed on one page and used as a data source nowhere.