Request a supported analysis with explicit design choices

Usage

analysis_request(
  kind,
  outcome,
  weight = NULL,
  assignment = NULL,
  domain = FALSE,
  missing = "fail"
)

Arguments

kind: mean or effect. These target a declared population mean or randomized-roster effect respectively.
outcome: Stable numeric outcome identifier.
weight: Named weight declaration for a mean.
assignment: Named assignment declaration for an effect.
domain: TRUE explicitly declares a restricted mean analysis to be a population domain.
missing: fail or omit for means. Effects require complete numeric outcomes.

Value

A plain analysis request. check_analysis validates it against a study. Scalar fields must be ordinary strings or logical values without attributes; classed values and one-cell matrices are not portable JSON scalars.