The primary command for server-side experiment configuration in Hackystat is called "Register Experiment". It accomplishes several important tasks. First, it enables the administrator to specify a set of Hackystat users who are to be viewed as subjects in a specific experiment. Aggregating a group of users into an Experiment facilitates both experimental monitoring (by indicating the users whose data transmission should be monitored) and experimental analysis (by indicating the user data to export for further analysis.)
Second, the Register Experiment command automates several of the server-side configuration tasks for the users declared in the Experiment. This includes declaring their user keys, setting their workspace roots, and defining their individual Projects.
To begin the definition of our example Experiment, you must login as the Administrator to a Hackystat server configured with the hackyApp_Experiment module, and then go to the Administrator page. There you will see a command called "Register Experiment". Clicking on this button brings up the first page of the Register Experiment command as illustrated in Figure 11.1, “ Register Experiment: Initial Page ”.
As you can see, the Register Experiment initial page allows you to specify an Experiment name, instance, description, treatments, start and end day, and subject emails. Figure 11.1, “ Register Experiment: Initial Page ” illustrates how we might fill out this page in the case of our example experiment with the Experiment name, the instance (i.e. the pilot study), a short description, the treatments, the start and end days, and the email addresses of the subjects.
Once you've entered this top-level information about the experiment, you can go on to configure subject-specific data by pressing the "Configure Subjects" button.
Figure 11.2, “ Register Experiment: Configure Experiment Subjects (initial) ” shows the screen that appears after pressing the Configure Subjects button.
The purpose of this page is to enable the Experiment Administrator to type in the workspace roots and workspaces to be associated with each of the subjects. This is accomplished by selecting one or more subjects, then one or more treatments to be associated with the selected subject(s), and then finally the workspace roots and workspaces to be associated with the subjects. Press the "Apply Configurations" button to save the entry.
In our example experiment, both Smith and Jones will work on a windows machine in a directory called c:\exp\bowling, but Smith will be doing the TLD treatment and Jones will be doing the TFD treatment. Figure 11.3, “ Register Experiment: Configure Experiment Subjects (completed) ” shows the screen that appears after completing the configuration.
Once all of the subjects are configured, Press the "Finalize Experiment" button to move to the confirmation screen.
Figure 11.4, “ Register Experiment: Experiment Registration Confirmation ” shows the screen that appears after pressing the Finalize Experiment button.
This Confirmation page shows that the data associated with each subject as specified in the prior screens. It also shows the user key that the system automatically generates for each user. (In the case of a pre-existing user, it will use the existing user key.)
After experiment registration, each of these users will have an account created for them with the specified workspace root setting. In addition, a Project instance for this experiment will be created with the date interval and workspace setting specified for this Experiment.
Using the key provided on the confirmation page, you can login to the newly created user account and go to the Preferences page. Invoking the Project Management command brings up a screen illustrated in Figure 11.5, “ User Project Management page showing automatically created Project ”.
As the figure illustrates, the user smith@hawaii.edu now has a Project named "smith@hawaii.edu-TFDvTLD-Pilot1" with the start and end day and workspaces specified from the Register Experiment command. Similar Project instances will be defined for all of the other subjects in your study. Note that Project names share a global namespace and must be unique, which is why the implicitly defined project concatenates together the user email, experiment name, and instance name.