1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
|
% Generated by roxygen2: do not edit by hand
% Please edit documentation in R/proportionBF.R
\name{proportionBF}
\alias{proportionBF}
\title{Function for Bayesian analysis of proportions}
\usage{
proportionBF(
y,
N,
p,
rscale = "medium",
nullInterval = NULL,
posterior = FALSE,
callback = function(...) as.integer(0),
...
)
}
\arguments{
\item{y}{a vector of successes}
\item{N}{a vector of total number of observations}
\item{p}{the null value for the probability of a success to be tested against}
\item{rscale}{prior scale. A number of preset values can be given as
strings; see Details.}
\item{nullInterval}{optional vector of length 2 containing
lower and upper bounds of an interval hypothesis to test, in probability units}
\item{posterior}{if \code{TRUE}, return samples from the posterior instead
of Bayes factor}
\item{callback}{callback function for third-party interfaces}
\item{...}{further arguments to be passed to or from methods.}
}
\value{
If \code{posterior} is \code{FALSE}, an object of class
\code{BFBayesFactor} containing the computed model comparisons is
returned. If \code{nullInterval} is defined, then two Bayes factors will
be computed: The Bayes factor for the interval against the null hypothesis
that the probability is \eqn{p_0}{p0}, and the corresponding Bayes factor for
the compliment of the interval.
If \code{posterior} is \code{TRUE}, an object of class \code{BFmcmc},
containing MCMC samples from the posterior is returned.
}
\description{
Bayes factors or posterior samples for binomial, geometric, or neg. binomial data.
}
\details{
Given count data modeled as a binomial, geometric, or negative binomial random variable,
the Bayes factor provided by \code{proportionBF} tests the null hypothesis that
the probability of a success is \eqn{p_0}{p_0} (argument \code{p}). Specifically,
the Bayes factor compares two hypotheses: that the probability is \eqn{p_0}{p_0}, or
probability is not \eqn{p_0}{p_0}. Currently, the default alternative is that
\deqn{\lambda~logistic(\lambda_0,r)} where
\eqn{\lambda_0=logit(p_0)}{lambda_0=logit(p_0)} and
\eqn{\lambda=logit(p)}{lambda=logit(p)}. \eqn{r}{r} serves as a prior scale parameter.
For the \code{rscale} argument, several named values are recognized:
"medium", "wide", and "ultrawide". These correspond
to \eqn{r} scale values of \eqn{1/2}{1/2}, \eqn{\sqrt{2}/2}{sqrt(2)/2}, and 1,
respectively.
The Bayes factor is computed via Gaussian quadrature, and posterior
samples are drawn via independence Metropolis-Hastings.
}
\examples{
bf = proportionBF(y = 15, N = 25, p = .5)
bf
## Sample from the corresponding posterior distribution
samples =proportionBF(y = 15, N = 25, p = .5, posterior = TRUE, iterations = 10000)
plot(samples[,"p"])
}
\seealso{
\code{\link{prop.test}}
}
\author{
Richard D. Morey (\email{richarddmorey@gmail.com})
}
\keyword{htest}
|
