flatlizards {BradleyTerry} | R Documentation |
Data collected at Augrabies Falls National Park (South Africa) in September-October 2002, on the contest performance and background attributes of 77 male flat lizards (Platysaurus broadleyi). The results of exactly 100 contests were recorded, along with various measurements made on each lizard. Full details of the study are in Whiting et al. (2006).
data(flatlizards)
This dataset is a list containing two data frames: flatlizards$contests
and flatlizards$predictors
.
The flatlizards$contests
data frame has 100 observations on the following 2 variables:
winner
lizard003
... lizard189
loser
lizard003
... lizard189
The flatlizards$predictors
data frame has 77 observations (one
for each of the 77 lizards) on the following 18 variables:
id
throat.PC1
throat.PC2
throat.PC3
frontleg.PC1
frontleg.PC2
frontleg.PC3
badge.PC1
badge.PC2
badge.PC3
badge.size
testosterone
SVL
head.length
head.width
head.height
condition
repro.tactic
resident
and floater
There were no duplicate contests (no pair of lizards was seen fighting more than once), and there were no tied contests (the result of each contest was clear).
The variables head.length
, head.width
,
head.height
and condition
were all computed as
residuals (of directly measured head length, head width, head height
and body mass index, respectively) from simple least-squares regressions
on SVL
.
Values of some predictors are missing (NA
) for some lizards,
`at random', because of instrument problems unconnected with the value of the
measurement being made.
The data were collected by Dr Martin Whiting, University of the Witwatersrand, http://web.wits.ac.za/Academic/Science/APES/Research/MWLab/People/, and they appear here with his kind permission.
Whiting, M.J., Stuart-Fox, D.M., O'Connor, D., Firth, D., Bennett, N.C. and Blomberg, S.P. (2006). Ultraviolet signals ultra-aggression in a lizard. Animal Behaviour 72, 353-363. (http://dx.doi.org/10.1016/j.anbehav.2005.10.018)
library(BradleyTerry) data(flatlizards) ## ## Fit the standard Bradley-Terry model, using the bias-reduced ## maximum likelihood method: ## BTmodel <- BTm(flatlizards$contests ~ .., br = TRUE) summary(BTmodel) ## ## That's fairly useless, though, because of the rather small ## amount of data on each lizard. And really the scientific ## interest is not in the abilities of these particular 77 ## lizards, but in the relationship between ability and the ## measured predictor variables. ## ## So next fit (by maximum likelihood) a "structured" B-T model in ## which abilities are determined by a linear predictor. This ## reproduces results reported in Table 1 of Whiting et al. (2006). ## BTmodel2 <- BTm(flatlizards$contests ~ throat.PC1 + throat.PC3 + head.length + SVL, data = flatlizards$predictors) summary(BTmodel2)