Get contrasts between groups in a Bayesian model

This function wraps the machinery needed to compute the contrasts between the levels of a grouping variable in our data (e.g., Group, Cluster) for a given variable. It fits Bayesian models using the fit_stan_glm() function for the prior and posterior distributions, computes the contrasts using emmeans::emmeans() and pairs(), and then computes the Bayes Factor for the contrasts using bayestestR::bf_parameters(). The results are then formatted in a clean table.

Usage

get_contrast_bf(df, groups)

Arguments

df

A data frame in long format containing the variable to be analysed and the required columns. This is for example the output of get_longer(study_data).

groups

A grouping variable (e.g. Group or Cluster) without quotes.

Value

A data frame summarising the contrasts between the levels of the grouping variable, with the variable name in a Variable column, the comparison in a Comparison column, the difference in a Difference ($\Delta$) column, the 95% credible interval in a 95% CrI column, and the log Bayes Factor in a $log(BF_{10})$ column.

Examples

df_vviq <- study_data |> get_longer() |> dplyr::filter(Variable == "VVIQ")
vviq_contrast <- get_contrast_bf(df_vviq, Group)
print(vviq_contrast)
#> # A tibble: 1 × 4
#>   Comparison           `Difference ($\\Delta$)` `95% CrI`      `$log(BF_{10})$`
#>   <chr>                                   <dbl> <chr>                     <dbl>
#> 1 Control - Aphantasic                     39.7 [36.22, 43.17]             56.6