Brian S. Cade. 2015. Model averaging and muddled multimodel inferences. Ecology 96:2370–2382. http://dx.doi.org/10.1890/14-1639.1
R code to compute model-averaged regression coefficients for Burnham and Anderson (2002) college gpa example and to simulate multi-part compositional predictors and model-averaged estimates similar to Rice et al (2013) zero-truncated Poisson regression count model for Greater Sage-Grouse.
Ecological Archives E096-210-S1.
File list (downloads)
Brian S. Cade
U. S. Geological Survey
Fort Collins Science Center
150 Centre Ave., Bldg. C.
Fort Collins, CO 80526 USA
E-mail: [email protected]
Supplement1.txt (MD5: faff226df1ad4adfa82f2ca1700d5010)
Supplement2.txt (MD5: 51eeb647483834e7ea0742d81ffc4372)
Supplement1.txt has R script to read in the college GPA example data from Burnham and Anderson 2002:226, estimates the 16 linear least squares regression models, computes AICc, AICc weights, variance inflation factors, partial standard deviations of predictors, standardizes estimates by partial standard deviations, computes model-averaged standardized estimates and their standard errors, and computes the model-averaged ratio of t statistics for unstandardized estimates (equivalent to model-averaged ratio of standardized estimates). The code is written to be transparent with respect to the mathematical operations rather than for efficiency.
Supplement2.txt has R script to generate the simulations in Appendix B for multi-part compositional predictors within a zero-truncated Poisson regression count model similar to the breeding sage-grouse count model of Rice et al. (2013).