Constructors for the publication-bias priors used by robma() and
robma_sensitivity(). Each returns a robma_bias_prior object.
Usage
prior_bias(
type = c("weight_function", "pet", "peese", "copas", "jung", "none"),
parameters = list(),
prior_weight = 1,
...
)
prior_weight_function(
steps = c(0.025, 0.05),
alpha = NULL,
prior_weight = 1,
sided = "one"
)
prior_pet(distribution = "cauchy", location = 0, scale = 1, prior_weight = 1)
prior_peese(distribution = "cauchy", location = 0, scale = 5, prior_weight = 1)
prior_copas(prior_weight = 1)
prior_jung(prior_weight = 1)
prior_no_bias(prior_weight = 1)Arguments
- type
Bias-prior family. One of
"weight_function","pet","peese","copas","jung","none".- parameters
Named list of family-specific parameters.
- prior_weight
Numeric. Prior model weight in the bias-prior mixture.
- ...
Additional fields stored on the bias-prior object.
- steps
Numeric vector of p-value cutpoints for a step weight function.
- alpha
Numeric vector of Dirichlet concentration parameters; defaults to
rep(1, length(steps) + 1).- sided
One of
"one"or"two". Determines whether the weight function is one- or two-sided.- distribution
Distribution for the PET / PEESE slope prior. Currently
"cauchy".- location, scale
Location and scale of the PET / PEESE slope prior.