Using observation-level random effects to model overdispersion in count data in ecology and evolution.
Harrison XA. · 2014
Read Using observation-level random effects to model overdispersion in count data in ecology and evolution. by Harrison XA. (2014) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc4194460.txt.
Description
Xavier A. Harrison and Chiyuan Miao wrote a methods study on count data in PeerJ in 2014. Ecologists often count animals or events and model the counts, but real counts swing more widely than the standard Poisson model allows. That extra swing, called overdispersion, can distort results if ignored.
Using computer simulations, the study tested a fix that adds a random effect for each single observation. When the extra swing came from random noise or clumped data, the fix gave truer estimates than ignoring the problem. It failed on data with too many zeros, and sometimes made bias worse there. Ignoring overdispersion also inflated how much of the pattern a model seemed to explain. The paper advises using the fix with care, matched to the cause.
Catalog record
| CREATOR | Harrison XA. |
|---|---|
| YEAR | 2014 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 49 KB |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |