Converts reviewed candidate dimensions into a reusable scoring specification. The fitted object stores indicator orientation, centers, scales, missing-value treatment, dimension coefficients, and dimension-score standardization parameters. It can therefore be applied unchanged to later panel years by a scoring function.
Usage
fiscal_health_fit(
data,
dimensions,
dimension_names = NULL,
scoring = "standardized_mean",
standalone = NULL,
indicator_direction = NULL,
missing = c("median", "complete"),
min_indicators = 2L,
fm = "minres",
strict_direction = FALSE
)Arguments
- data
Reference data used to estimate the measurement model.
- dimensions
A
fiscal_dimensionsobject or a named list whose elements are character vectors of indicators.- dimension_names
Optional named character vector mapping discovered dimension names to substantive names.
- scoring
Dimension scoring method. Supply one value for every dimension or a named vector. Supported methods are
"standardized_mean","pca", and"factor".- standalone
Optional character vector of standalone indicators. The default uses standalones carried by a
fiscal_dimensionsobject.- indicator_direction
Optional named vector overriding indicator directions with
"higher","lower", or"context". Other directions are inferred fromfiscal_metrics()when possible.- missing
Training-data missing-value rule:
"median"or"complete". Median replacement here is statistical missing-value treatment, not e-file data normalization.- min_indicators
Minimum number of indicators required in a modeled dimension.
- fm
Factoring method passed to
psych::fa()for factor scores.- strict_direction
If
TRUE, error when an indicator is context dependent or absent from the metric registry without an explicit override.