The Statistics of Tiny Samples: The Utility of ACTUS, an Alternative Method of Contingency Table Analysis Using Simulation, in Human Skeletal Biology

Document Type

Presentation

Presentation Date

4-21-2017

Abstract or Description

Contingency table analyses of count data can be incredibly useful for the inference of relationships between variables and proportional differences between skeletal populations. Of the most frequently used methods for making these inferences, the chi-share test cannot be applied to small samples and Fisher's Exact test cannot be applied to tables larger than 2x2 . Consequently, many bioarchaeological and paleopathological studies involving few instances of the variable of interest rely on sample proportions to draw conclusions about population proportions. An alternative approach to these methods of analysis is ACTUS (Analysis of Contingency Tables Using Simulation), a computationally intense simulation program developed by George Estabrook and used mainly in biology. Compared to traditional methods, ACTUS offers four advantages: it enables rigorous statistical inferences despite low counts; it generates probability values for each individual cell, aiding interpretation of probability values generated for the entire table; even the least quantitatively oriented anthropologist can understand how the program works; and the free software operates rapidly, producing results based on thousands of simulated tables in seconds. We explain the methodology of how ACTUS works and use several examples related to incidence of traumatic injuries and other pathologies to demonstrate its utility for comparative analyse

Sponsorship/Conference/Institution

American Association of Physical Anthropologists Meeting

Location

New Orleans, LA

Source

https://meeting.physanth.org/program/2017/session40/estabrook-2017-the-statistics-of-tiny-samples-the-utility-of-actus-an-alternative-method-of-contingency-table-analysis-using-simulation-in-human-skeletal-biology.html

Share

COinS