Run the full battery of diagnostic checks

Executes every implemented check and returns a rcdiag_report() collecting the results. Checks whose required inputs are missing are skipped gracefully and listed in the report's $skipped_checks element.

Usage

run_diagnostics(
  responses,
  method = NULL,
  rdata = NULL,
  noise_matrix = NULL,
  baseimage = "base",
  expected_n = NULL,
  col_participant = "participant_id",
  col_stimulus = "stimulus",
  col_response = "response",
  col_rt = NULL,
  infoval_iter = NULL,
  ...
)

Arguments

responses

A data frame of trial-level responses.

method

"2ifc" or "briefrc", or NULL to auto-detect.

rdata

Optional. Path to an rcicr .RData file. Enables check_stimulus_alignment(), check_version_compat() (2IFC), compute_infoval_summary(), check_response_inversion(), and cross_validate_rt_infoval().

noise_matrix

Optional. Path to a Brief-RC noise-matrix text file, or an already-loaded numeric matrix. Enables validate_noise_matrix() and check_stimulus_alignment() (Brief-RC).

baseimage

Name of the base image used at stimulus-generation time (key in base_face_files in the rdata). Default "base". Only consulted when the infoval-dependent checks run.

expected_n

Optional. Passed to check_trial_counts().

col_participant, col_stimulus, col_response, col_rt

Column-name overrides.

infoval_iter

Number of iterations for rcicr's reference distribution; NULL disables infoVal checks even if rdata is supplied. Default NULL because the reference simulation is slow and unwanted by default; set e.g. 10000 to enable.

...

Unused.

Value

An object of class rcdiag_report.

Details

The orchestrator auto-detects the method when only rdata or only noise_matrix is supplied; otherwise pass method explicitly.

Examples

responses <- data.frame(
  participant_id = rep(1:3, each = 100),
  stimulus       = rep(1:100, 3),
  response       = sample(c(-1, 1), 300, replace = TRUE),
  rt             = round(exp(rnorm(300, 6.7, 0.4)) + 100, 1)
)
report <- run_diagnostics(responses, method = "2ifc", col_rt = "rt")
print(report)
#> == Data-quality report (2ifc) ==
#> 
#> [PASS] Response coding
#>   All 300 responses coded as {-1, 1}.
#> [PASS] Trial counts
#>   3 participants total.
#>   Trial counts observed: 100 trials (3 participants).
#> [PASS] Duplicates
#>   0 of 300 rows are fully duplicated.
#>   0 rows share a (participant, stimulus) pair with differing response or RT.
#> [PASS] Response bias
#>   0 of 3 participants gave the same response on every trial.
#>   0 participants have |mean response| > 0.6 (extreme bias).
#>   Group mean response: -0.033.
#> [PASS] Response times
#>   0 of 3 participants exceed 5% fast trials (< 200 ms).
#>   0 of 3 participants exceed 5% slow trials (> 5000 ms).
#>   0 of 3 participants have RT coefficient of variation below 0.1.
#> 
#> Summary: pass=5, warn=0, fail=0, skip=0
#> 
#> Skipped checks:
#>   - check_stimulus_alignment (no rdata / noise_matrix)
#>   - check_version_compat (no rdata)
#>   - compute_infoval_summary (need rdata + infoval_iter)
#>   - check_response_inversion (needs infoval)
#>   - cross_validate_rt_infoval (needs infoval)