binomial_to_bernoulli_data: Convert binomial data to bernoulli data

In hojsgaard/doBy: Groupwise Statistics, LSmeans, Linear Estimates, Utilities

Convert binomial data to bernoulli data

Description

Convert binomial data to bernoulli data by expanding dataset.

Usage

binomial_to_bernoulli_data(
  data.,
  y,
  size,
  type = c("rest", "total"),
  response_name = "response",
  rest_name = NULL
)

Arguments

data.	A dataframe
y	Column with 'successes' in binomial distribution y~bin(size, p)
size	Column with 'failures', i.e. size-y or 'total', i.e. size.
type	Whether size is rest (i.e. 'failures') or 'total'
response_name	Name of response variable in output dataset.
rest_name	Name of 'failures' in column response_name.

Examples

dat <- budworm
dat <- dat[dat$dose %in% c(1,2), ]
dat$ntotal <- 5
dat
dat.a <- dat |>
  binomial_to_bernoulli_data(ndead, ntotal, type="total")
dat.b <- dat |>
  dplyr::mutate(nalive=ntotal-ndead) |> dplyr::select(-ntotal) |>
  binomial_to_bernoulli_data(ndead, nalive, type="rest")

m0 <- glm(cbind(ndead, ntotal-ndead) ~ dose + sex, data=dat, family=binomial())
m1 <- glm(ndead / ntotal ~ dose + sex, data=dat, weight=ntotal, family=binomial())
ma <- glm(response ~ dose + sex, data=dat.a, family=binomial())
mb <- glm(response ~ dose + sex, data=dat.b, family=binomial())

dat.a$response
dat.b$response ## Not same and therefore the following do not match

all.equal(coef(m0), coef(ma))
all.equal(coef(m0), coef(mb))
all.equal(coef(m1), coef(ma))
all.equal(coef(m1), coef(mb))

hojsgaard/doBy documentation built on May 4, 2024, 5:20 a.m.