Perform Bayesian Repeated-Measures ANOVA Using generalTestBF

Performs a Bayesian repeated-measures ANOVA using the generalTestBF(..., whichModels = "top") function from the BayesFactor package, assuming one subject column, multiple within-subject factor columns, and one dependent variable column.

Usage

rmANOVA_bf(
  x,
  rscaleFixed = "medium",
  rscaleRandom = "nuisance",
  rscaleCont = "medium",
  rscaleEffects = NULL,
  method = "auto",
  progress = TRUE,
  multicore = FALSE,
  summarize = TRUE,
  inc_ranef = FALSE,
  seed = NULL
)

Arguments

x: A data frame in long format. The first column should be the subject identifier. The last column should be the dependent variable, and the columns in between should be within-subject factors.
rscaleFixed: Prior scale for fixed effects. Default is "medium".
rscaleRandom: Prior scale for random effects. Default is "nuisance".
rscaleCont: Prior scale for continuous covariates. Default is "medium".
rscaleEffects: Optional vector of prior scales for individual effects.
method: Method for computing Bayes factors. See BayesFactor::generalTestBF(). Default is "auto".
progress: Logical. Whether to display progress bar. Default is TRUE.
multicore: Logical. Whether to use multicore processing. Default is FALSE.
summarize: Logical. Whether to output summarized results or not. Default is 'TRUE'.
inc_ranef: Logical. Whether to output results of random effects.
seed: Optional. A numeric seed to fix the random number generator state for reproducibility.

Value

A tibble summarizing Bayes factors for each model compared to the full model. Columns include:

effect: The effect excluded from the full model
BF: Bayes factor for the null (exclusion) over the alternative hypothesis
error: Estimated numerical error
log10_BF: Base-10 logarithm of the Bayes factor
favor: Indicates whether data favor the null or alternative hypothesis
evidence: Strength of evidence ("anecdotal", "moderate", "strong", "very strong", "extreme")

Details

The function currently supports designs with 2 or 3 within-subject factors only. The subject column is automatically renamed to "s", factor columns to "fw1", "fw2", etc., and the outcome variable to "y" internally for formula construction.

Examples

library(BayesFactor)
# Simulated data with subject, 2 factors, and outcome
set.seed(123)
dat <- data.frame(
  id = rep(1:30, each = 4),
  A = rep(c("low", "high"), times = 60),
  B = rep(c("left", "right"), each = 2, times = 30),
  y = rnorm(120)
)
res <- rmANOVA_bf(dat)
#> 
  |                                                                            
  |                                                                      |   0%
  |                                                                            
  |==========                                                            |  14%
  |                                                                            
  |====================                                                  |  29%
  |                                                                            
  |==============================                                        |  43%
  |                                                                            
  |========================================                              |  57%
  |                                                                            
  |==================================================                    |  71%
  |                                                                            
  |============================================================          |  86%
  |                                                                            
  |======================================================================| 100%
#> Error in str_detect(full_model_vars, "s"): could not find function "str_detect"
print(res)
#> Error in h(simpleError(msg, call)): error in evaluating the argument 'x' in selecting a method for function 'print': object 'res' not found