Test Task PLS Design Subspaces — test_design

Computes observed covariance energy in term-specific design subspaces and, when requested, compares those statistics to a global Task PLS permutation null. The permutation null asks whether term-aligned covariance is larger than expected under the package's ordinary global Task PLS row permutation.

Usage

test_design_subspaces(
  x,
  fit = NULL,
  spec = NULL,
  design = NULL,
  formula = NULL,
  condition_key = NULL,
  terms = "all",
  statistic = c("trace", "largest_root"),
  weights = c("cell_equal", "subject_count", "row_weights"),
  nperm = 0L,
  permutation = c("none", "global_task_pls"),
  correction = c("none", "maxT"),
  permsamp = NULL,
  progress = FALSE
)

Arguments

x: A Task PLS pls_result, or a pls_spec when fit is supplied.
fit: Optional fitted pls_result when x is a pls_spec.
spec: Optional pls_spec used to recompute permuted cross-block matrices. Required for nperm > 0 when x is a pls_result.
design: Optional pls_design object.
formula: Optional one-sided formula overriding design$formula.
condition_key: Optional condition metadata.
terms: "all" or a character vector of term labels to test.
statistic: "trace" for total subspace covariance energy, or "largest_root" for the dominant singular-root statistic.
weights: "cell_equal", "subject_count", or "row_weights" to use stored result$task_pls$row_weights.
nperm: Number of permutations.
permutation: "none" or "global_task_pls".
correction: "none" or "maxT" adjustment across tested terms.
permsamp: Optional permutation order matrix.
progress: Logical; show permutation progress messages.

Value

A data frame with term, rank, statistic, p-value, and null labels.