Research methodology
Every dataset declares its minimum sample size before collection opens, and no result is published below it. That is the whole point of publishing the method first.
The publication rule
Every dataset declares a minimum sample size before collection opens. Below that number, no result is published — not a partial figure, not a directional hint, not an “early signal”. The gate is enforced in code: a single function decides whether a dataset may show results, and the visitor-facing statement about its status comes from the same call that gates the rendering, so the two cannot disagree.
Structured data follows the same rule. Dataset markup is emitted only for a set that has genuinely reached publication, because describing an empty collection to a search engine as a dataset is a claim about data that does not exist.
What a dataset has to declare
- Provenance. Who collected it, how, on what evidence, when collection opened, when it was last reviewed, and how often it is revisited.
- Methodology. What it measures, the ordered collection procedure, what counts as a valid observation, what is deliberately excluded, the minimum sample size, and the unit of observation.
- Limitations. Never empty. A dataset that cannot say what it fails to capture has not been thought about hard enough to publish.
- Sample size. Shown on every publication, beside the result rather than in a footnote.
Prohibited
- Synthetic, illustrative, or seeded observations, in any dataset, for any reason
- Publishing a result below the declared minimum
- Changing a minimum sample size after seeing the data
- Aggregating across markets whose per-market samples are inadequate
- Any observation that would identify a specific property or person
- Naming an individual contractor in a dataset that measures the market
Banding and privacy
Continuous values are banded at collection time rather than at analysis time. Acreage, largest stem size, and location are all recorded as bands or as a market, never as exact figures, so no record can describe an identifiable property. That costs precision, and it is the right trade.
What happens when a published result turns out to be wrong
It is withdrawn, the withdrawal reason is published in place of the result, and the change is recorded in the revision log. A withdrawn dataset keeps its methodology page — removing the evidence of a mistake is worse than the mistake.
Work out the scope before you ask for prices
Answer the five questions every quote depends on — what is growing, where the material goes, how deep removal has to reach, whether equipment can get in, and what the land is for afterwards. Then send one properly described project to the right kind of contractor.
Nothing on this site is a recommendation of a specific company. The guidance describes categories of work and the questions worth asking. If you decide to request quotes, the introduction is free to you, contractors may pay to receive it, and you are under no obligation to hire anyone.