Testing measures of animal social association by computer simulation
Techniques used to measure patterns of affiliation among social animals have rarely been tested for accuracy. One reason for this lack of validation is that it is often impossible to compare sample data to the true distribution of social assortment of a group of animals. Here we test some methods of...
| Published in: | Behaviour |
|---|---|
| Main Authors: | , |
| Format: | Article in Journal/Newspaper |
| Language: | English |
| Published: |
2007
|
| Subjects: | |
| Online Access: | https://risweb.st-andrews.ac.uk/portal/en/researchoutput/testing-measures-of-animal-social-association-by-computer-simulation(2019e9d6-d7b2-4ce0-8e85-9e8d9921233e).html https://doi.org/10.1163/156853907782418259 http://www.scopus.com/inward/record.url?scp=35748966971&partnerID=8YFLogxK http://www.ingentaconnect.com/content/brill/beh/2007/00000144/00000011/art00008 |
| Summary: | Techniques used to measure patterns of affiliation among social animals have rarely been tested for accuracy. One reason for this lack of validation is that it is often impossible to compare sample data to the true distribution of social assortment of a group of animals. Here we test some methods of assessing social assortment by using a computer simulation of organisms whose assortment patterns were under our control. We created male and female organisms that moved in a direction that was based on a social bias parameter. As the weight of this parameter increased, organisms were more likely to move in the direction of others of their sex. We then created virtual observers to sample assortment of the organisms under different social bias conditions. Observers used three different techniques of measuring assortment. These were (1) group membership: noting all organisms that were associated in the same 'group', (2) nearest neighbour: noting the nearest organism to a randomly selected individual and (3) neighbourhood: noting all organisms near a selected individual. Neighbourhood was taken either by all-occurrence sampling or by focal sampling the associations of randomly selected individuals. Some techniques emerged as more sensitive than others under different conditions and biases were revealed in some measures. For example, the group membership method was biased toward finding significant assortment differences between the sexes when no difference actually existed. Nearest neighbour was insensitive to finding a difference in assortment between sexes when one existed. Focal sampling was less sensitive to finding effects than all-occurrence sampling. The computer simulation revealed properties of each technique that would have been impossible to detect in the field. |
|---|