BradleyTerry2 (version 1.1-2)

chameleons: Male Cape Dwarf Chameleons: Measured Traits and Contest Outcomes

Description

Data as used in the study by Stuart-Fox et al. (2006). Physical measurements made on 35 male Cape dwarf chameleons, and the results of 106 inter-male contests.

Usage

chameleons

Format

A list containing three data frames: chameleons$winner, chameleons$loser and chameleons$predictors.

The chameleons$winner and chameleons$loser data frames each have 106 observations (one per contest) on the following 4 variables:

ID

a factor with 35 levels C01, C02, ... , C43, the identity of the winning (or losing) male in each contest

prev.wins.1

integer (values 0 or 1), did the winner/loser of this contest win in an immediately previous contest?

prev.wins.2

integer (values 0, 1 or 2), how many of his (maximum) previous 2 contests did each male win?

prev.wins.all

integer, how many previous contests has each male won?

The chameleons$predictors data frame has 35 observations, one for each male involved in the contests, on the following 7 variables:

ch.res

numeric, residuals of casque height regression on SVL, i.e. relative height of the bony part on the top of the chameleons' heads

jl.res

numeric, residuals of jaw length regression on SVL

tl.res

numeric, residuals of tail length regression on SVL

mass.res

numeric, residuals of body mass regression on SVL (body condition)

SVL

numeric, snout-vent length (body size)

prop.main

numeric, proportion (arcsin transformed) of area of the flank occupied by the main pink patch on the flank

prop.patch

numeric, proportion (arcsin transformed) of area of the flank occupied by the entire flank patch

Details

The published paper mentions 107 contests, but only 106 contests are included here. Contest number 16 was deleted from the data used to fit the models, because it involved a male whose predictor-variables were incomplete (and it was the only contest involving that lizard, so it is uninformative).

Examples

# NOT RUN {
##
## Reproduce Table 3 from page 1268 of the above paper:
##
summary(chameleon.model <- BTm(player1 = winner, player2 = loser,
  formula = ~ prev.wins.2 + ch.res[ID] + prop.main[ID] + (1|ID), id = "ID",
  data = chameleons))
head(BTabilities(chameleon.model))
##
## Note that, although a per-chameleon random effect is specified as in the
## above [the term "+ (1|ID)"], the estimated variance for that random
## effect turns out to be zero in this case.  The "prior experience"
## effect ["+ prev.wins.2"] in this analysis has explained most of the
## variation, leaving little for the ID-specific predictors to do.
## Despite that, two of the ID-specific predictors do emerge as
## significant.
##
## Test whether any of the other ID-specific predictors has an effect:
##
add1(chameleon.model, ~ . + jl.res[ID] + tl.res[ID] + mass.res[ID] +
  SVL[ID] + prop.patch[ID]) 

# }