Print methods for hyper2 and hyper3 objects
# S3 method for hyper2
print(x, ...)
# S3 method for hyper3
print(x, ...)Returns the hyper2 or hyper3 object it was sent,
invisibly. Function pnv() converts a named vector to a
character string that is used in the hyper3 print method.
An object of class hyper2 or hyper3
Further arguments, currently ignored
Robin K. S. Hankin
Used mainly for their side-effect of printing the log-likelihood
function. In the print method, a natural logarithm is indicated with
“log()”---not “ln()”---consistent with R
builtin terminology base::log().
The hyper2 print method is sensitive to option
give_warning_on_nonzero_power_sum. If TRUE, a warning is
issued if the powers have nonzero sum. This is usually what you want
because observations are typically multinomial. If FALSE, the
warning is suppressed. One often needs hyper2 objects with
nonzero sum as the following snippet attests:
H1 <- hyper2(list('a'),1)
H2 <- hyper2(list('a',c('a','b')),c(1,-1))
Above, both objects are legitimate likelihood functions; both correspond
to the observation “player a won once”. But they differ
in that H1 corresponds to “player a came first in a
field of unspecified players” while H2 corresponds to
“player a came first in a field comprising a and
b”. Sometimes you want H1 but (in my experience) mostly
it is a result of coding error. Also note the following snippet:
icons["L"] %<>% inc
icons[c("L","NB")] %<>% dec
The above two lines correspond to “Take the icons
likelihood function and add to it the (independent) observation that a
correspondent found ‘London’ to be more concerning than
‘the Norfolk Broads’”. But after the first line and before the
second, icons has a nonzero power sum, pending addition of
another term. At this point, icons is arguably an incomplete
likelihood function as the players have not been specified; its nonzero
power sum is an indicator that it is a temporary object. That's OK as
long as we remember to add the denominator (as carried out in the next
line) which would mean multiplying by (L+NB)^{-1}, restoring the
zero power sum. If we forget to do this, the print method gives us a
warning which should prompt us to check the coding.