Add an exposure offset to a count-model formula — add_exposure

Appends offset(log(<exposure>)) to a Poisson or negative binomial formula (zero-inflated or not), so the coefficients describe the rate per unit of exposure (session length, number of intervals, opportunities) rather than the raw count. The exposure column must be numeric, present on every row, and strictly positive.

Usage

add_exposure_offset(formula_str, data, exposure, family)

Arguments

formula_str: Model formula as a string.
data: Data holding the exposure column.
exposure: Name of the exposure column, or NULL to return formula_str unchanged.
family: brms family name; "poisson", "negbinomial", "zero_inflated_poisson", or "zero_inflated_negbinomial".

Value

The formula string with the offset term appended (unchanged when it already carries that offset); an error when the exposure column or family is not usable, or when the formula already holds a different offset.

Examples

d <- data.frame(outcome = c(3, 5, 2), level_AB = c(0, 1, 1), minutes = c(10, 20, 15))
add_exposure_offset("outcome ~ level_AB", d, "minutes", "poisson")
#> [1] "outcome ~ level_AB + offset(log(minutes))"