Data limitations
The specific reasons these figures should not be treated as population statistics.
Selection bias, stated plainly
Owners who hire an eviction service differ from owners who don't. They tend to hold more units, be further from the property, and file more consistently. Every figure here inherits that selection.
Geographic coverage
EvictIQ operational measures come from cases EvictIQ worked, concentrated in metro Atlanta and surrounding Georgia counties. That book is not national field coverage. Geography pages exist for every U.S. county and do not imply EvictIQ completed those cases. Ranking volumes are cited from Princeton Eviction Lab or omitted.
We are part of what we measure
Judgment-to-writ intervals reflect our own application timing. Setout delay coding reflects our own operational practice. These are honest measures of our book, not neutral measurements of a county.
Court dockets are not our source
Our filing date is when we filed, not the docket's timestamp. These usually agree and occasionally don't.
Coding involves judgement
Delay reasons are coded by people applying published rules. Another analyst applying the same rules would produce similar but not identical results.
Cost figures are one provider's prices
Cost benchmarks reflect what we charged and what we paid. They are not a market rate survey and should not be cited as one.
Lost rent is estimated
We observe rent and dates; we compute lost rent by proration. The estimation rule is published and reasonable people could choose a different one.
Time-boundedness
Court calendars, county staffing and executing-office practice change. A figure from one period does not predict another, and we date every publication.
What would fix this
Docket-level data across all U.S. counties, which we do not have and are not claiming. If a research partner has it, we're interested.
Questions about the method?
We'd rather answer them before you cite us than after.