---
title: "Methodology — sources, cadence, coverage, and limits | Data Tactical Group"
route: "/docs/methodology"
updated: "2026-09-13"
---

# How the data is built

These pages publish the inputs, the sequence and the direction of every step, and withhold the estimators, parameters, thresholds and code. That line is drawn here once; every page beneath it follows it.

Updated 2026-09-13

## What is published

Every input, named: the Quarterly Census of Employment and Wages as the source of record, County Business Patterns, the Current Employment Statistics and State and Metro Area programs, the Employment Projections program, and the Occupational Employment and Wage Statistics program. The order in which they are used: published QCEW values first and never overwritten; withheld values estimated under the constraints the published record imposes; the series brought current with survey rates of change; then projected ten years forward and reconciled across industry and geography. The granularity: every NAICS level QCEW publishes and every SOC level, for the nation, the states and the District of Columbia, and every county and county-equivalent plus an unallocated area for every state but Utah. The span: quarterly from the first quarter of 2004. The cadence: fixed, named vintages, with observed release history rather than a published calendar.

Where a number would otherwise be misread, the page that reports it says so at that point. There is no general list of limitations, because a caveat stated away from the value it concerns is not read.

## What is withheld

The fitting method’s implementation, its estimation windows, the damping applied to interpolation, the caps and smoothing constants, convergence rules, tie-breaks, weights and code. These are the proprietary part of the work. What they produce is fully labeled, so their absence from these pages never leaves a reader unsure what a value is.

## The pages

Each page follows the same structure so that they can be read against one another, and each carries the date it was last reviewed.

- [Data sources](/docs/methodology/sources) — every dataset that contributes, what it provides, and what it is used for.
- [Industry data](/docs/methodology/industry) — what employment, establishments and earnings count, and how withheld values are completed.
- [Occupation data](/docs/methodology/occupation) — how occupation employment and earnings are constructed from the industry base, the national matrix and OEWS.
- [Projections](/docs/methodology/projections) — how the series is brought current, and how it is carried ten years forward.
- [Derived measures](/docs/methodology/derived-measures) — location quotient and what it can tell you, and the method shift share will use once it is available.
- [Geography](/docs/methodology/geography) — the areas reported, how area clusters are assembled, and what aggregation does to each measure.
- [Disclosure and provenance](/docs/methodology/disclosure) — the four disclosure codes, the reliability score, and the restatement flags.
- [Vintages](/docs/methodology/vintages) — fixed, named versions of the data, and why a cited figure can be recovered.

Release notes for each vintage are on the [releases](/releases) page.

Still stuck? A person can help.

[Get support](/support)