oottest implements the out-of-treatment testing from Kuelpmann and Kuzmics (2020). Out-of treatment testing allows for a direct, pairwise likelihood comparison of theories, calibrated with pre-existing data.

You can install the development version of oottest from GitHub with:

```
# install.packages("devtools")
::install_github("PhilippKuelpmann/oottest") devtools
```

Input data should be structured in the following way:

columns represent different treatments

rows represent actions

cells record the number of subjects who chose each action on each treatment

Prediction data should be structured in the following way:

columns represent different treatments

rows represent the predicted probability of each action

the different tables represent the different theories

cells record the probability of choosing an action on each treatment depending on the theory

Here is a basic example on how you can use the vuong_statistic using predictions from two theories:

```
library(oottest)
<- c(1,2,3)
data_experiment <- c(1/3,1/3,1/3)
prediction_theory_1 <- c(1/4,1/4,1/2)
prediction_theory_2 vuong_statistic(data_experiment, pred_I = prediction_theory_1, pred_J = prediction_theory_2)
```

Here is a basic example how to compare three theories, using data from two treatments:

```
library(oottest)
<- c(1,2,3)
treatment_1 <- c(3,2,1)
treatment_2 <- data.frame(treatment_1, treatment_2)
data_experiment <- matrix(c(1/3,1/3,1/3, 1/3, 1/3, 1/3), nrow = 3, ncol=2)
theory_1 <- matrix(c(1/4,1/4,1/2,1/2,1/4,1/4), nrow = 3, ncol=2)
theory_2 <- matrix(c(1/3,1/3,1/3, 1/4,1/4,1/2), nrow = 3, ncol=2)
theory_3 <- array(c(theory_1,theory_2,theory_3), dim=c(3,2,3))
theories vuong_matrix(data_experiment, theories)
```