Marginal Relationships with Covariates

Evaluates and summarizes the marginal relationships between explanatory variables and recruitment, survival, or absence probability from a fitted DRM model.

Usage

marg(object, ...)

# S3 method for class 'adrm'
marg(
  object,
  process = c("rec", "surv", "pabs"),
  variable,
  newdata = NULL,
  n_pts = 100,
  summary = TRUE,
  prob = 0.95,
  ...
)

Arguments

object: An object of class adrm, typically the output of fit_drm().
...: Additional arguments passed to methods.
process: A character string indicating the process to evaluate: "rec" (recruitment), "surv" (survival), or "pabs" (probability of absence).
variable: A character vector with the name(s) of the focal variable(s) to examine.
newdata: An optional data.frame containing the values for the focal variable(s). If NULL, a grid is generated automatically based on the observed range in the model matrix.
n_pts: An integer specifying the number of points to generate for the sequence of each focal variable when newdata is NULL. Default is 100.
summary: Logical. If TRUE (the default), returns the quantiles of the posterior predictions. If FALSE, returns the raw posterior draws.
prob: A numeric scalar in \((0, 1)\) specifying the probability mass of the equal-tailed credible interval. Defaults to 0.9, which produces a 90\ together with the median.

Value

A data.frame with the posterior summaries (or draws) for the specified process. If summary = TRUE, it also receives the class marg_adrm to enable automated plotting.

Details

The marg function computes the predicted relationships across a sequence of values for a focal variable (or variables), holding all other non-focal variables in the model matrix at zero.

When summary = TRUE, the function calculates an equal-tailed credible interval and the median using quantile2, which is highly optimized for posterior draws.

Author

lcgodoy