10. Create a custom experiment ruleset

createruleset command

To create a rules file for custom experiment, you can use the following command: ./scripts/manage createruleset

This will prompt you to provide an experiment name for your new experiment's ruleset. If no rules file by that name exists yet, the command will do the following:

• create a file {ruleset_name}.py in experiment/rules
• create a file {ruleset_name}_test.py in experiment/rules/tests - this implements a rudimentary unit test
• add references to your new rules file to experiment/rules/__init__.py

Set up the experiment

Go to the admin interface at localhost:8000/admin. If you click on Add next to Experiments, you can verify that your rules now appear in the Rules dropdown. Give the experiment a name and assign a slug to it. Also, tie a playlist to it. Now you can see the experiment in action if you navigate to localhost:3000/{your_slug}. The experiment plays audio files and presents a BooleanQuestion "Do you like this song?" as many times as there are rounds in the experiment (you can adjust this in the admin interface), and then shows a Final action.

The rules file

Open your new rules file experiment/rules/{experiment_name}.py and edit away! You can see some methods implemented for you.

__init__

This method initializes questions to collect information on demographics or musical training from your participant. Note that you can easily change these questions through the admin interface.

first_round

When the frontend first retrieves information about the experiment from the backend, this method will provide the actions to show during the setup of the experiment, i.e., before the experiment proper starts. Most experiments display some introductory information here (as an Explainer), as well as a consent form (Consent). Optionally, experimenters may choose to let participants choose between multiple playlists before starting the experiment (Playlist). All this information is optional.

next_round

This is the centrepiece of your experiment's logic. You can check here in which round you are (session.get_next_round()), or whether all rounds defined in the experiment have been completed (session.rounds_complete()). The number of rounds passed is measured by the number of Result objects saved during the cause of the experiment session.

Most experiments present a Final action when the rounds are complete, and before that, Trial actions, with optionally some Score actions in between to show intermediate results to the participant.