Computes a Bayes factor quantifying evidence that a reported mean of integer-scale data is inconsistent with the reported sample size, accounting for rounding uncertainty.
Arguments
- mean_value
Numeric. The reported mean.
- n
Integer. The sample size.
- decimals
Integer. Number of decimal places in the reported mean. If NULL (default), inferred from the reported value.
- max_items
Integer. Maximum plausible value on the integer scale (default 10). Used to define the range of possible integer sums.
- n_tolerance
Integer. How many values around the reported N to consider as plausible (default 0, exact N only).
Value
A list with components:
- bf_inconsistent
Numeric. Bayes factor in favour of inconsistency (fabrication).
- posterior_prob_inconsistent
Numeric. Posterior probability of inconsistency assuming equal prior odds.
- consistent_at_n
Logical. Classical GRIM result at exact N.
- consistent_nearby
Logical. GRIM-consistent at any nearby N.
- interpretation
Character. Evidence strength label.
Details
Under H0 (genuine data), the mean must equal k/n for some integer k, rounded to the reported decimal places. Under H1 (fabricated data), the mean is drawn uniformly from the plausible range. BF_10 = P(data | H1) / P(data | H0).
Examples
# Consistent mean: evidence for genuine data
bayes_grim_test(2.60, n = 20)
#> $bf_inconsistent
#> [1] 1.005
#>
#> $posterior_prob_inconsistent
#> [1] 0.5012469
#>
#> $consistent_at_n
#> [1] TRUE
#>
#> $consistent_nearby
#> [1] TRUE
#>
#> $interpretation
#> [1] "weak_evidence"
#>
#> $mean_value
#> [1] 2.6
#>
#> $n
#> [1] 20
#>
#> $decimals
#> [1] 1
#>
# Inconsistent mean: evidence for fabrication
bayes_grim_test(2.47, n = 20)
#> $bf_inconsistent
#> [1] Inf
#>
#> $posterior_prob_inconsistent
#> [1] 1
#>
#> $consistent_at_n
#> [1] FALSE
#>
#> $consistent_nearby
#> [1] FALSE
#>
#> $interpretation
#> [1] "strong_inconsistency"
#>
#> $mean_value
#> [1] 2.47
#>
#> $n
#> [1] 20
#>
#> $decimals
#> [1] 2
#>