---
tag: XCWKYS1DFA
kind: concept
name: Blocking
msc: [62K10]
sources:
  - url: https://en.wikipedia.org/wiki/Blocking_(statistics)
same_as:
  wikidata: Q4927286
  wikipedia: Blocking (statistics)
---

## Summary

Blocking is the grouping of experimental units into blocks of similar units, with treatments compared within each block.
Variation between blocks, such as differences between fields, days or batches of material, is then removed from the comparison of treatments.
This makes the comparison more precise.
It is the local control principle of [[design of experiments|2TE42ZWJ3E]], and is usually combined with [[randomization|734Q6Q51X7]] inside each block.
The randomized complete block design, Latin squares and balanced incomplete block designs are the standard block designs.

## Explanations

### L1 human

To compare three wheat varieties, a farmer has four fields of different quality.
If each variety got its own fields, a good field could make a poor variety look good.
Instead, each field is split into three plots, and every variety is grown once in every field, in plots chosen at random.
Differences between fields then affect all varieties equally and cancel out of the comparison.