Methodology

How the salary data is built

Written so every number on the salary pages can be traced to a source and a rule. Questions the data can't answer honestly, we don't answer.

Sources

Posted salaries — every job listing Alaska Jobs syncs (currently 1096 postings with parseable pay, open and archived) from Workplace Alaska, the municipal and borough job boards listed on the sources page, the Foraker Group nonprofit board, and USAJOBS. The board never deletes a listing, so the dataset accumulates daily.
The state salary schedule — the ASEA General Government (GG/GP), Class 1, 37.5-hour week, parsed directly from the Department of Administration's published schedule, effective 2026-07-01. We cross-validate the parse against live postings that state their range (e.g., Range 11 Step A = $23.01/hr matches postings exactly).
Geographic differentials — the duty-station table from the GGU 2025–2028 collective bargaining agreement, which implements AS 39.27.020.

Annualization

Postings quote pay hourly, biweekly, monthly, or annually. For comparison we annualize: hourly × 2,080 · biweekly × 26 · monthly × 12. When a posting gives a range, we use its midpoint for medians and keep the endpoints for the observed span. Figures outside $12,000–$500,000/yr after annualization are treated as parse failures and excluded.

Title normalization

"Office Assistant II (PCN 06-1234) — Anchorage" and "Office Assistant 2" are the same role. We strip PCNs, locations, level numerals, and posting boilerplate, then group what remains. A role page publishes only at three or more parseable observations — below that, a median is noise and we'd rather show nothing.

Known limitations

Posted ≠ paid. These are the ranges employers advertise, not settled offers or payroll data.
Hiring-range bias. State postings usually quote the early steps; long-tenured employees earn above the posted figure.
Mixed employers. Municipal, nonprofit, and federal postings don't follow the state schedule; role pages note when postings cite a state range.
Small samples. Alaska is small; a three-posting median is a starting point, not gospel. Every page shows its N.

Update cadence

Observations update with the morning sync. Per-role medians snapshot weekly, building the trend series shown on role pages. The state schedule is re-parsed when the state publishes a new one (annually, ~July 1).

Corrections welcome — this data is published to be checked. Contact.