Methodology & Data

1. What we measure

The grades rest on a direct, time-matched sales-ratio study — the gold-standard measure of assessment fairness used by the IAAO and state tax departments. There is no model and no prediction. We take the assessor’s own number and divide it by what the open market actually paid:

For every arms-length residential sale, we pull the appraised value the assessor actually had on the books in that same year — not today’s value, not a forecast — and compare it to the revealed market value, the sale price. Matching the value to the sale year is what makes this fair: it asks whether the assessor was right at the moment the market spoke, not whether a later reappraisal eventually caught up.

A ratio near 1.0 means the office had the property valued at roughly what it sold for. Across West Virginia the ratio typically runs below 1.0 — values lag the market — because the statewide reappraisal cycle and a rising market keep assessed values behind sale prices. Why that is not counted against the assessor is explained in section 3.

2. The standards we grade

From the ratios we compute the standard IAAO statistics and grade each against its published threshold. Three things are assessor-attributable and carry a verdict: how evenly parcels are assessed relative to one another (uniformity), whether high- and low-value homes are treated alike (vertical equity), and whether the roll is being kept current (cadence).

• Uniformity (COD). The coefficient of dispersion measures how much individual ratios scatter around the median. A low COD means similar homes are assessed similarly. IAAO targets ≤ 15 for improved residential property.
• Vertical equity (PRD & PRB). These ask whether expensive and inexpensive homes are assessed at the same rate. PRD 0.98–1.03 and PRB within ± 0.05 indicate no systematic bias; values above the range are regressive (high-value homes under-assessed relative to low-value homes).
• Reappraisal cadence. The share of the residential roll with ≥20% uncaptured market movement — i.e. how much of the roll has gone stale. Under 15% stale means the roll is current.

Stated targets, verbatim from the scoring layer: COD ≤ 15 (IAAO, improved residential); PRD 0.98–1.03 (IAAO); PRB ±0.05 (IAAO); cadence < 15% of roll stale.

3. What we don’t grade

We do not grade the level of assessment — the median ratio itself, i.e. whether values sit at 80% or 95% of market. The level is context only — confounded by reappraisal cycle + market, not graded. West Virginia’s mandated reappraisal cycle and the housing market — both largely outside any one assessor’s control — push the level around far more than assessment practice does. A county can be perfectly uniform and perfectly equitable while still sitting below 100% of market simply because the cycle hasn’t come around. Penalizing that would punish assessors for the calendar.

So the median ratio is reported as context only, never as PASS/WARN/FAIL.

Real elected officials. Level (median ratio) is confounded by the statewide reappraisal cycle and the housing market; COD/uniformity is the more assessor-attributable signal. Not a judgment of any individual.

4. Grading gates — when an era is graded at all

Because these are real, named, elected officials, we hold the bar high before publishing a verdict. An assessor’s era is graded only when it clears both gates:

Why these gates exist. A county’s value roll in any single year reflects the last reappraisal — often set by the prior assessor — measured against that year’s sale prices. An assessor observed in only one partial year, especially the current unsettled roll year, would be scored on a roll they did not build against prices at a cyclical peak. That manufactures an inflated COD and a steep negative PRB that are artifacts of the cycle and market, not signals of that office’s practice. Requiring multiple distinct sale years forces an era to be measured across the assessor’s own roll, and the n ≥ 100 floor keeps COD and PRB statistically meaningful.

Eras that fail a gate are not hidden: they are listed with the assessor’s name, years, and sale count, but carry no published COD/PRB grade — only a note saying why. The current roll year (2025) is re-included once it becomes a full, settled prior year.

A single partial/current roll year reflects the prior assessor's reappraisal measured against peak sale prices, producing artifact COD/PRB; multi-year, n>=100 eras are required before a grade is published.

5. Confidence tiers

Sample size matters. A graded verdict from a large county carries more weight than one scraped from a few hundred sales, so every metric is labeled with a confidence tier based on the number of attributed sales. Small samples are flagged, not hidden — a tiny county should never read as equivalent to a large one.

6. Data sources & limits

• Source. County assessor records — appraised values from parcel_tax_history and sales from parcel_sales_history.
• Arms-length only. We include only validity-code 0 sales (open-market, arms-length). Family transfers, foreclosures, and other non-market deeds are excluded so the denominator reflects true market value.
• Residential. Property class R only. Commercial, agricultural, and other classes follow different valuation rules and are out of scope.
• No tax dollars. This dataset contains no billed tax amounts. Every figure here is about appraisal accuracy and uniformity, not anyone’s tax bill.
• Time-matched. Each sale is paired to the appraised value the office carried in the sale year, attributed to the assessor serving that county that year.