Teacorps — Think tank · Worker cooperative

Türkçe

Sciences / Flow · Laminar

Method before conclusion

A result is worth exactly as much as the procedure that produced it.

Our view

We judge any claim, our own first, by the procedure behind it: whether the analysis was fixed before the data arrived, whether the result could have failed, and whether someone else can run it again. Agreement between independent methods outweighs any single striking result.

We build instruments to that standard. They state what they measure together with how far it can be trusted, and they are designed to be checked by the people who rely on them.

Our work

Two of our applications are measuring instruments.

  • Aurore — Inference that states its own probability of being right.
  • Jujiang — Data recovered from published figures.

Our selection

Reading

The whole library

An effective path

For a society

  1. Register analyses before the data arrives.
  2. Fund replication as research in its own right.
  3. Publish data and code with every paper.

For an individual

  1. Write your prediction down before you look.
  2. Read the primary source, not the summary.
  3. Keep a log of your predictions and how they turned out.

Next: Arts