The `R`

package **BFpack** contains a set of functions for exploratory hypothesis testing (e.g., equal vs negative vs postive) and confirmatory hypothesis testing (with equality and/or order constraints) using Bayes factors and posterior probabilities under commonly used statistical models, including (but not limited to) Bayesian t testing, (M)AN(C)OVA, multivariate/univariate linear regression, correlation analysis, multilevel analysis, or generalized linear models (e.g., logistic regression). The main function `BF`

needs a fitted model (e.g., an object of class `lm`

for a linear regression model) and (optionally) the argument `hypothesis`

, a string which specifies a set of equality/order constraints on the parameters. By applying the function `get_estimates`

on a fitted model, the names of the parameters are returned on which constrained hypotheses can be formulated. Bayes factors and posterior probabilities are computed for the hypotheses of interest.

Install the latest release version of `BFpack`

from CRAN:

The current developmental version can be installed with

```
if (!requireNamespace("remotes")) {
install.packages("remotes")
}
remotes::install_github("jomulder/BFpack")
```

Below several example analyses are provided using **BFpack**.

First a classical one sample t test is executed for the test value (= 5) on the therapeutic data

The `t_test`

function is part of the ** bain** package. The function is equivalent to the standard

`t.test`

function with the addition that the returned object contains additional output than the standard `t.test`

function.To perform a Bayesian t test plug the fitted object into the `BF`

function.

This executes an exhaustive test around the null value: `H1: mu = 5`

versus `H2: mu < 5`

versus `H3: mu > 5`

assuming equal prior probabilities for `H1`

, `H2`

, and `H3`

of 1/3. The output presents the posterior probabilities for the three hypotheses.

The same test would be executed when the same hypotheses are explicitly specified using the `hypothesis`

argument.

When testing hypotheses via the `hypothesis`

argument, the output also presents an `Evidence matrix`

containing the Bayes factors between the hypotheses.

First an analysis of variance (ANOVA) model is fitted using the `aov`

fuction in `R`

.

Next a Bayesian test can be performed on the fitted object.

By default posterior probabilities are computed of whether main effects and interaction effects are present. Alternative constrained hypotheses can be tested on the model parameters `get_estimates(aov1)`

.

An example hypothesis test is consdered under a logistic regression model. First a logistic regression model is fitted using the `glm`

function

```
fit_glm <- glm(sent ~ ztrust + zfWHR + zAfro + glasses + attract + maturity +
tattoos, family = binomial(), data = wilson)
```

The names of the regression coefficients on which constrained hypotheses can be formualted can be extracted using the `get_estimates`

function.

Two different hypotheses are formulated with competing equality and/or order constraints on the parameters of interest. These hypotheses are motivated in Mulder et al. (2019)

```
BF_glm <- BF(fit_glm, hypothesis = "ztrust > (zfWHR, zAfro) > 0;
ztrust > zfWHR = zAfro = 0")
summary(BF_glm)
```

By calling the `summary`

function on the output object of class `BF`

, the results of the exploratory tests are presented of whether each separate parameter is zero, negative, or positive, and the results of the confirmatory test of the hypotheses under the `hypothesis`

argument are presented. When the hypotheses do not cover the complete parameter space, by default the complement hypothesis is added which covers the remaining parameter space that is not covered by the constraints under the hypotheses of interest. In the above example, the complement hypothesis covers the parameter space where neither `"ztrust > (zfWHR, zAfro) > 0"`

holds, nor where `"ztrust > zfWHR = zAfro = 0"`

holds.

By default `BF`

performs exhaustice tests of whether the separate correlations are zero, negative, or positive. The name of the correlations is constructed using the names of the variables separated by `_with_`

.

Constraints can also be tested between correlations, e.g., whether all correlations are equal and positive versus an unconstrained complement.

For a univariate regression model, by default an exhaustive test is executed of whether an effect is zero, negative, or postive.

Hypotheses can be tested with equality and/or order constraints on the effects of interest. If prefered the complement hypothesis can be omitted using the `complement`

argument

```
BF2 <- BF(lm1, hypothesis = "Vehicle > 0 & Face < 0; Vehicle = Face = 0",
complement = FALSE)
print(BF2)
```

In a multivariate regression model hypotheses can be tested on the effects on the same dependent variable, and on effects across different dependent variables. The name of an effect is constructed as the name of the predictor variable and the dependent variable separated by `_on_`

. Testing hypotheses with both constraints within a dependent variable and across dependent variables makes use of a Monte Carlo estimate which may take a few seconds.

```
lm2 <- lm(cbind(Superficial, Middle, Deep) ~ Face + Vehicle,
data = fmri)
constraint2 <- "Face_on_Deep = Face_on_Superficial = Face_on_Middle < 0 <
Vehicle_on_Deep = Vehicle_on_Superficial = Vehicle_on_Middle;
Face_on_Deep < Face_on_Superficial = Face_on_Middle < 0 < Vehicle_on_Deep =
Vehicle_on_Superficial = Vehicle_on_Middle"
set.seed(123)
BF3 <- BF(lm2, hypothesis = constraint2)
summary(BF3)
```

`BF`

on a named vectorThe input for the `BF`

function can also be a named vector containing the estimates of the parameters of interest. In this case the error covariance matrix of the estimates is also needed via the `Sigma`

argument, as well as the sample size that was used for obtaining the estimates via the `n`

argument. Bayes factors are then computed using Gaussian approximations of the likelihood (and posterior), similar as in classical Wald test.

We illustrate this for a Poisson regression model

The estimates, the error covariance matrix, and the sample size are extracted from the fitted model

Constrained hypotheses on the parameters `names(estimates)`

can then be tested as follows

```
BF1 <- BF(estimates, Sigma = covmatrix, n = samplesize, hypothesis =
"woolB > tensionM > tensionH; woolB = tensionM = tensionH")
```

Note that the same hypothesis test would be executed when calling

because the same Bayes factor is used when running `BF`

on an object of class `glm`

(see `Method: Bayes factor using Gaussian approximations`

when calling `print(BF11)`

and `print(BF2)`

).

You can cite the package and the paper using the following reference

Mulder, J., van Lissa, C., Gu, X., Olsson-Collentine, A., Boeing-Messing, F., Williams, D. R., Fox, J.-P., Menke, J., et al. (2019). BFpack: Flexible Bayes Factor Testing of Scientific Expectations. (Version 0.2.1) [R package]. https://CRAN.R-project.org/package=BFpack

Mulder, J., Williams, D. R., Gu, X., Olsson-Collentine, A., Tomarken, A., Böing-Messing, F., Hoijtink, H., . . . van Lissa, C. (2019). BFpack: Flexible Bayes factor testing of scientific theories in R. Retrieved from https://arxiv.org/abs/1911.07728

If you have ideas, please get involved. You can contribute by opening an issue on GitHub, or sending a pull request with proposed features.

By participating in this project, you agree to abide by the Contributor Code of Conduct v2.0.