Skip to contents

Run the INSPECT-SR Trustworthiness Assessment

Source: R/inspect_sr.R
inspect_sr.Rd

Takes a data frame with one row per study and runs the automated Domain 4 checks (Carlisle's test, participant-number consistency, GRIM, p-value verification). Manual items (D1, D2, D3, and the non-automated D4 items) are read straight from the input. Domain-level and overall judgements are derived per INSPECT-SR guidance (overall = most severe domain).

Usage

inspect_sr(
  data,
  studyvar = study,
  bayes = FALSE,
  prior_prob_trustworthy = 0.9,
  pvalue_tolerance = 0.01,
  carlisle_method = "fisher",
  verbose = TRUE
)

Arguments

data

A data frame or tibble with one row per study. See Expected columns in the package vignette, or the bundled inspect_sr_example dataset for the exact layout.

studyvar

Unquoted column name identifying the study (tidyeval). Defaults to study.

bayes

Logical. If FALSE (default) produces frequentist pass/fail judgements. If TRUE produces Bayes factors and a posterior probability of trustworthiness.

prior_prob_trustworthy

Numeric in (0, 1). Prior probability that each study is trustworthy, used only when bayes = TRUE (default 0.90).

pvalue_tolerance

Numeric. Tolerance for the frequentist p-value check (default 0.01).

carlisle_method

"fisher" (default) or "ks" — see carlisle_test().

verbose

Logical. Print a summary to the console (default TRUE).

Value

If bayes = FALSE: an object of class inspect_sr (a data frame with columns Study, D1, D2, D3, D4, Overall), with per-study details in attr(x, "details").

If bayes = TRUE: an object of class bayes_inspect_sr (a data frame with columns Study, Prior, Posterior, Combined_BF, Interpretation), with individual Bayes factors in attr(x, "details").

See also

inspect_sr_table() for a per-check gt table; inspect_plot() for the traffic-light visualisation; filter_trustworthy() for filtering a meta-analysis dataset.

Examples

if (FALSE) { # \dontrun{
data(inspect_sr_example)

# Frequentist
res <- inspect_sr(inspect_sr_example, studyvar = study)

# Bayesian
res_bayes <- inspect_sr(inspect_sr_example, studyvar = study, bayes = TRUE)
} # }