epower conducts a power analysis on beyond BACI sampling designs through INLA based inference and Monte Carlo simulation. Currently epower considers only univariate responses. The analysis allows for cross random effects between Time and Site in the design. The epower toolbox has been designed for use by ecologists with a sound understanding of the statistical principles of BACI designs. To assist the non-R user, BACI parameters are defined in an excel workbook template. The template epower_interface_V1.3.xlsx should have been supplied with the package tar.gz file and is also available on github where this code is currently maintained. The 'epower' (V1.3) package is a BMT product which has been developed in collaboration with the Australian Institute of Marine Science (AIMS), Queensland University of Technology (QUT) and Pink Lake Analytics. The method/code supporting 'epower' (v1.3) is described in: Fisher R, Shiell GR, Sadler RJ, Inostroza K, Shedrawi G, Holmes TH, McGree JM (2019). 'epower' - an R package for power analysis of Before-After-Control-Impact (BACI) designs. Methods in Ecology and Evolution.
A detailed manual is available for 'epower'. This should have been supplied withe the package tar.gz file and is also available on github where this code is currently maintained. The manual is also available as Appendix S1 in Fisher R, Shiell GR, Sadler RJ, Inostroza K, Shedrawi G, Holmes TH, McGree JM (2019). 'epower' - an R package for power analysis of Before-After-Control-Impact (BACI) designs. Methods in Ecology and Evolution.
There are only two functions that are called by the user (if using the excel interface), including: fitData(), and assessPower(). While fitData must be called at the outset to run the ‘epower’ package, the function assessPower() is optional. The outputs generated by the ‘epower’ package and written to file include the model fit statistics and a PIT histogram generated by fitData(); and optionally a table of scenario details and associated proportions of detected impacts (across the number of simulation iterations) generated by assessPower().
For those proficient in R, the excel interface can be by-passed. This is achieved by calling the function supplyData instead of fitData, followed by a manual call to powerScenario. Once powerScenario has been successfully run, the necessary objects will be available to run assessPower. See ?supplyData for more details.
epower is freely available to users under licence. Licence for use explicitly precludes the incorporation of the
epower code into other programs which are subsequently sold or used for commercial purposes.
Users are asked to please include the following acknowledgment:
Analyses were carried out using the software 'epower' V1.3 (BMT 2019) as described in Fisher et al (2019),
(BMT 2019) and based on the statistical programming platform R (R-Core Team, 2019).
This software has been prepared jointly by BMT, the Australian Institute of Marine Science and Queensland University of Technology. BMT, the Australian Institute of Marine Science and Queensland University of Technology accept no liability or responsibility for in respect of any use of or reliance upon this software.