Double Labelling Experiment

A double labelling experiment is a statistical way of estimating a number that can only be sampled, like the population of a wild animal.

Imagine that you want to know how many rabbits there are in a given area. To perform a double-labelling experiment to find out you set a number of rabbit traps in that area then come back the next day. You tag every rabbit you have caught, release them, and reset your traps. Then you come back the day after that and count the number of your tagged rabbits that you have caught again.

Suppose that the first night you re-caught 100 rabbits, and the second night you re-caught 10 tagged (labelled) rabbits and 90 un-tagged ones (ignore the unlikely coincidence of catching exactly 100 on two nights running). You can now estimate that the total rabbit population of the area is about 1000, because you caught 100 rabbits the first night, and your second sample of tags tell you that that's about one tenth of the total population.

There are statistical and biological complications, of course. As described above the experiment assumes that the rabbits and traps are statistically independent from night to night, and in their geographical location. For instance, it assumes that once-caught rabbits are not made more wary than their fellows by their experience.

Double labelling experiments have very wide application. For example they can be used to estimate the number of undiscovered species (sample species, count the undiscovered ones, then sample again and count how many of your previous undiscovered species you see a second time). They can also be used in the social sciences on people, and in physics and chemistry.

[This page need to be improved and referenced. I added it quickly as a stopgap when I discovered Wikipedia had no page for this important technique.]