Research
How HyreRoof research works
What we build our studies from, how we build them, and where they stop being reliable.
Last updated .
Every HyreRoof study is built from primary sources, the agency that produced the data, not a site quoting it. Each one states the window it covers, the date the data was retrieved, the formula behind any figure we derived, and the geography that figure applies to. Where coverage is incomplete we leave the gap visible rather than impute a value. This page is the long version of all of that, including the limitations.
The distinction that matters most. Source data is what an agency published. Analysis is what we did to it. Every study labels which is which, and a derived number presented as an official statistic is the single most common way a research page misleads.
Why we do research at all
A great deal of what circulates about roofing is asserted and almost never measured. Whether there is a roofer shortage. Whether licensing means anything. How much of the housing stock is due for replacement. Where storm chasers actually go. Those are answerable from public data, and nobody was answering them, so we did.
Primary sources
We work from the body that produced the data, not from a secondary site quoting it. Where a figure has passed through an intermediary, we say so.
| Source | What we use it for |
|---|---|
| State licensing boards and statutes | Whether roofing is licensed, what the licence requires, and what the authority actually checks before issuing. Read from the statute and the board, not from aggregator summaries. |
| Bureau of Labor Statistics | Roofer employment, wages and occupational fatality data. Note that BLS employment counts exclude most self-employed roofers, which is why published roofer counts run low. |
| U.S. Census Bureau and the American Housing Survey | Housing stock, age of housing and unit counts, the denominator for almost every per-home rate we publish. |
| NOAA National Centers for Environmental Information | The Storm Events Database, for hail and wind exposure by county and date. |
| FEMA | Disaster declarations and hazard exposure. |
| Municipal permit portals | Roofing permit volumes and requirements where a jurisdiction publishes them. Coverage is patchy and we say where it is. |
| State insurance departments | Filings, bulletins and consumer guidance on wind and hail deductibles and claims practice. |
| Model codes and trade standards | The IRC and NRCA material for technical claims about assemblies, ventilation and fastening. |
Not every study uses every source. Each study lists the ones it actually used, with the date the data was retrieved.
Source data is not our data. A figure an agency published is labelled with that agency and the retrieval date.
Something we derived, such as a rate, an index, a ranking or a per-home normalisation, is labelled a HyreRoof calculation, with the arithmetic stated.
A derived number presented as an official statistic is the single most common way a research page misleads. We treat that distinction as non-negotiable.
How we collect
Public datasets are downloaded from the publishing agency. Statutory and licensing material is read from the statute or the board’s own pages, state by state, rather than inferred from a national summary. Our licensing study exists precisely because the national summaries disagree with the statutes. Where a figure comes from a single jurisdiction’s portal, the study names the portal.
Calculations
Every derived figure states its formula, its numerator and its denominator. Where a normalisation choice changes the answer (per household against per housing unit, say, or employment against establishment counts) we state which we used and why, because the choice is usually doing more work than the data.
Geographic limitations
County-level storm data is not homeowner-level exposure. A county recording hail does not mean every roof in it was struck. State-level licensing does not describe municipal requirements, which can be stricter. City figures drawn from a metropolitan statistical area are not the city. We say which geography a figure is at, every time.
Time periods
Every study states the window it covers and the date the data was retrieved. Roofing data has long revision lags. BLS revises and NOAA’s storm database is updated retrospectively, so a figure retrieved today may differ from the same figure retrieved next year.
Missing data
Coverage is genuinely incomplete in places: permit portals are inconsistent, several states publish no usable licensing register, and self-employment makes roofer counts structurally uncertain. Where data is missing we leave the gap visible and say what is missing. We do not impute a value and present it alongside measured ones without saying which is which.
Updates and corrections
Studies carry the date they were last reviewed. When we find an error we correct it and say what changed, silent edits to a research page are how a research programme loses the right to be cited. The full corrections policy is in the editorial policy.
What this research is not
- Not a substitute for an inspection of your roof.
- Not insurance, legal or engineering advice.
- Not a prediction. A county with high historical hail frequency is not a forecast for your roof.
- Not a contractor rating. None of it ranks or evaluates a named roofing company.
The studies · Data and downloads · For journalists
If you came here to check a number
Checking a number against its method is the right instinct, and every study page carries the working. If instead you have a roof to make a decision about, the arithmetic is more use than the method: the roof cost calculator shows every rate it uses, and the repair vs replacement calculator shows the thresholds behind its answer. When you want to speak to someone, tell us about the project.