6 Resampling Approaches
See Baxter (2001)
An example based on Raymond Dart’s right-handed australopith murderers - 7 of 9 baboon skulls damaged on the left side - thanks to Dave Daegling. (Dart 1949)
Really this is just an excuse to quote Dart’s famously lurid prose:
carnivorous creatures that seized living quarries by violence, battered them to death, tore apart their broken bodies, dismembered them limb from limb, slaking their ravenous thirst with the hot blood of victims and greedily devouring livid writhing flesh.
But even if we were to accept Dart’s inference that the damage is the result of bloodthirsty australopithecines (paleoanthropologists don’t, for the record), does 7 of 9 left-side injuries really constitute compelling evidence that austrolopithecines were right-handed? Is there a case to answer? We can address this through resampling by framing the question as, “If australopithecines were equally likely to be left- or right-handed, how likely is it that in a sample of nine only 2 would be left-handed?”
# make some data
hand <- c('R','L')
southpaw <- NULL
for (i in 1:1000) {
bash <- sample(hand,9,replace=TRUE)
southpaw[i] <- length(bash[bash=='L'])
}
#visualize
hist(southpaw)
## [1] 0.081