This adapter fits unweighted ordinary least squares with an intercept and numeric predictors. It does not interpret a coefficient as a causal effect or certify population representativeness. Sampling and assignment declarations remain attached to the study but are not used by this estimator.
Usage
regression_request(
outcome,
predictors = character(),
cluster = NULL,
se_type = NULL,
missing = "fail",
target,
assumptions
)Arguments
- outcome
Stable numeric outcome identifier.
- predictors
Character vector of stable numeric predictor identifiers. Empty requests an intercept-only model. Create categorical indicators and interactions as documented studyframe derivations before requesting a model.
- cluster
Stable character or integer cluster identifier, or NULL.
- se_type
HC2 for unclustered inference; CR2 or stata for clustered inference. NULL selects HC2 without clusters and CR2 with clusters.
- missing
fail or omit. Omission applies jointly to all model variables and the cluster identifier and does not establish a missing-data assumption.
- target
Explicit description of the association and population of observations to which the model is intended to apply.
- assumptions
Nonempty character vector of the analyst's assumptions, including the justification for independence between observations or clusters and the interpretation of selection and missingness.
Value
A portable regression request for check_analysis and run_analysis. Scalar fields must be ordinary strings without attributes. Request arrays contain plain strings; classed values and matrices are not portable scalars.