Skip to contents

Runs the built-in checks configured in _rules.yml on every data point: internal consistency of each statistic type (e.g. Q1 <= median <= Q3, SD >= 0, n/N matching the reported percentage), missing dispersion, sample sizes larger than the group, duplicates, unknown variable names, unit mismatches and implausible values. Nothing is changed; the flags guide human review in review_values().

Usage

check_values(path = ".", study_ids = NULL, rules = read_rules(path))

Arguments

path

Project root.

study_ids

Studies to check. Defaults to all.

rules

Rules to use. Defaults to read_rules().

Value

A tibble with study_id, measurement_id, check, severity and message (one row per flag).