## Summary

If a certain feature (e.g. distribution, significance test, descriptive statistic or multiplicity adjustment procedure) is not implemented in the Mediana package, the user can defined a custom function and use it within the package. The user must follow the guidelines presented below to create valid custom functions.

## User-defined functions for a data model

If a certain distribution is not implemented in the Mediana package, the user can defined a custom distribution function. The user must follow the guidelines presented below to create valid custom functions.

### Custom functions for defining distributions

The following template must be used by the user to define a custom distribution function that can be used to specify an outcome, enrollment or dropout distribution in a data model. The distribution-specific components that should be modified if a different distribution needs to be implemented are identified in the comments.

As an example, the following function is used to generate observations following the Template distribution that has two parameters, parameter1 and parameter2.

## User-defined functions for an analysis model

If a significance test, descriptive statistic or multiplicity adjustment procedure is not included in the Mediana package, the user can easily define a custom function that implements a test, computes a descriptive statistic or performs a multiplicity adjustment. The user must follow the guidelines presented below to create valid custom functions.

### Custom functions for implementing a significance test

The following template must be used by the user to define a function that implements a significance test in an analysis model. The test-specific components that need to be modified if the user wishes to implement a different test are identified in the comments.

As an example, the function defined below carries out a test named TemplateTest with a single required parameter labeled parameter1.

### Custom functions for implementing a descriptive statistic

The following template must be used by the user to define a function that implements a descriptive statistic in an analysis model. The statistic-specific components that need to be modified if the user wishes to implement a different statistic are identified in the comments.

As an example, the function defined below carries out a descriptive statistic named TemplateStatistic with a single required parameter labeled parameter1 and two samples.

As an illustration, the function defined below implements a criterion named TemplateCriterion based on two significance tests and one descriptive statistic with two required parameters. The first parameter, labeled parameter1, is applied to the significance tests and the second parameter, labeled parameter2, is applied to the statistic. Success is achieved if at least one of the two tests is less than parameter1 and the statistic is greater than parameter2.