User Manual Privacy Policy Disclaimer Contact us
  Advanced SearchBrowse




Journal Article

Re-examining the use of the LSI technique in zooarchaeology


Wolfhagen,  Jesse
Archaeology, Max Planck Institute for the Science of Human History, Max Planck Society;

External Ressource
No external resources are shared
Fulltext (public)
There are no public fulltexts stored in PuRe
Supplementary Material (public)
There is no public supplementary material available

Wolfhagen, J. (2020). Re-examining the use of the LSI technique in zooarchaeology. Journal of Archaeological Science, 123: 105254. doi:10.1016/j.jas.2020.105254.

Cite as: http://hdl.handle.net/21.11116/0000-0007-4BC2-A
Biometric analysis of faunal remains is crucial for estimating the age/sex composition of assemblages and exploring large-scale processes that affected animal biology in the past. The LSI technique is a premier method for examining biometry in different zooarchaeological scenarios, particularly domestication research and regional-scale surveys. Despite the technique's popularity, several early arguments describing limitations or concerns about the LSI technique still impact interpretations and applications today. More generally, though, the LSI technique is treated as a method of increasing sample sizes as a last resort when unmodified measurements are too scarce to use. This paper re-examines the theoretical foundations of the LSI technique to update best practices in LSI analyses in zooarchaeology. Redefining the LSI technique as a pseudo-centering process shows why LSI values are preferable to unmodified measurements for biometric analyses. This new definition also highlights the arbitrary nature of standard animal and logarithm base choice, though certain decisions (smaller standard animals and base e logarithms) can aid interpretation by closely linking changes in LSI values to proportional changes of the original measurements relative to the standard. Of more consequence on LSI analyses, however, is the way to aggregate LSI values from different measurement types; this paper shows how multilevel modeling uses partial pooling to balance the trade-offs of bias and variance caused by aggregation. To showcase the benefits of the Bayesian multilevel LSI model, the biometric variation of ten simulated sites using a reference set of Shetland sheep measurements (Popkin Peter et al., 2012). Modeling all ten sites within a single multilevel structure provides a clear way to evaluate biometric differences while accounting for potential allometries and variation in body part representation between different sites. These results clarify earlier arguments about the limitations of the LSI technique, summarized in a set of best practices for LSI applications.