How we compare assessed values

What we do not do

We do not estimate what your home would sell for. There is no automated valuation model here and no appraisal — those are different things, done by licensed professionals under different rules.

What we do is narrower and, for a property tax appeal, more useful: we compare the assessor's number for your home against the assessor's numbers for comparable homes. Both sides of that comparison are public record. In Utah, sale prices are not public at all, which is precisely why comparing assessments is the practical route.

Why dollars per square foot is the wrong measure

The obvious approach is to divide assessed value by building area and compare that against the neighbourhood median. It is also wrong, and measurably so.

Across roughly 288,000 owner-occupied homes in Salt Lake County, the median assessed value per square foot falls steadily as homes get larger:

That is a 35% spread, and it is not an error in the roll — land and fixed improvements are spread over less area in a smaller home, so its value per foot is genuinely higher. The consequence is that comparing a 900 sq ft house against a neighbourhood whose median home is 2,000 sq ft on raw dollars per foot would flag it as over-assessed purely for being small. A county board would take that argument apart immediately, and rightly.

How we correct for it

Instead of assuming value rises in proportion to area, we measure how it actually rises. Fitting assessed value against building area on a county's own roll gives an exponent — for Salt Lake County it is 0.771, meaning value scales roughly as area raised to that power rather than to the power one.

So the figure we compare is assessed value divided by area^0.771. That quantity does not drift with house size, which makes homes of different sizes genuinely comparable. The exponent is fitted from the cohort itself where there are enough homes, otherwise from the county as a whole.

Measured on a fixed sample of 1,196 Salt Lake homes, this reduces the spread in average deviation across size groups from about 35 percentage points to 5.5.

How comparable homes are chosen

Comparables come from the subject's own subdivision as named on the county roll, restricted to homes within 40% of its building area and 20 years of its construction date, and matched on exemption status. Salt Lake County has roughly 9,400 subdivisions with enough homes to compute a median, and the typical one contains 21 comparable homes.

Where a subdivision is missing or too small, we fall back to nearby homes of similar size and age within about 0.4 miles. That is a weaker basis and a report says so on its face.

The median is computed across the whole cohort, not just the handful of comparables printed in a packet. Those are chosen for being closest to the subject in size and age — never for being the lowest assessed. Cherry-picking cheap neighbours would inflate the apparent gap and make the packet's own median contradict its argument the moment an assessor pulled the full list.

Why some results carry a warning

A deviation means nothing without knowing how tightly comparable homes cluster. In a tract of identical new townhouses, being 6% above the median is a real outlier. In a 1920s neighbourhood where assessed value per foot ranges from $596 to $1,005 among homes of similar size — a real example from Salt Lake City — 6% is noise.

So every result carries a measure of how much its cohort scatters, rated tight, moderate or wide. Where the scatter is wide, or where the gap does not clearly exceed the cohort's own noise, we say the finding is not strong enough to act on. We would rather decline than sell a weak case.

What this method still gets wrong

• A residual size effect remains. Most subdivisions are too small to fit their own exponent, so they inherit the county-wide one, which slightly under-corrects at the extremes. On our test sample the smallest homes still average about 3 percentage points high. That is small against a 5% threshold, but it is not zero, and we would rather state it than let someone discover it.
• Condition and quality are invisible. The roll records area, age and construction material, not whether a kitchen was gutted or a roof is failing. Two homes identical on paper can be genuinely worth different amounts, and an assessor may know something the roll does not say.
• Subdivisions are an imperfect cohort. They are a recording convention, not a market boundary. A large subdivision can span a busy road or a school catchment line that genuinely moves value.
• Roll vintages differ. Counties refresh at different times, and every report states the vintage it used. An older roll may not reflect recent construction or a recent revaluation.
• Being above the median is not proof of error. About half of any group sits above its own median by definition. What matters is the size of the gap relative to how tightly the group clusters, which is what the statutory threshold tests.