Edit me

Imitate a run from Mechanical Turk

You should always test your study before posting it anywhere. Testing that your study runs via a simple link is easy: just generate the link, start the study and run once through it. Testing studies posted in MTurk is especially cumbersome, because you should make sure that the confirmation codes are correctly displayed when the study is over. The standard way to test this is to create a study in MTurk’s Sandbox. JATOS offers a way to emulate MTurk, without having to set up anything in the sandbox. Here’s how.

If you think about it, MTurk simply calls a JATOS URL. The URL to start a study is normally http://your-jatos-server/publix/study-id/start (where study-id is a placeholder for the ID of the study you want to run). Two additional variables in the URL’s query string tell JATOS that this request comes from MTurk (and that it should display the confirmation code when the study is over): workerId and assignmentId. Both pieces of information are normally generated by MTurk; but they can be any arbitrary string. The only constraint is that the workerId does not already exist within JATOS. (Think of it this way: Because a MTurk worker can run a study only once, the same workerId can be used only once in JATOS.)

Here are some concrete examples:

To run the study with the ID 4 on a local JATOS use http://localhost:9000/publix/4/start?workerId=123456&assignmentId=abcdef.

To imitate a run from MTurk’s Sandbox, use any arbitrary values in the query strings workerId and assignmentId (In this example, workerId = 12345 and assignmentId = abcdef). Also, set turkSubmitTo to the value ‘sandbox’.

http://localhost:9000/publix/4/start?workerId=123456&assignmentId=abcdef&turkSubmitTo=sandbox

Lock your studies before running them

Each Study bar has a button that toggles between the ‘Unlocked’ and ‘Locked’ states. Locking a study prevents changes to its (or any of its components’) properties, change the order of components, etc.

GUI screenshot

Do a General Single Run more than once in the same browser

The problem here is that a General Single Run is intended to work only once in the same browser. Although this is a feature to limit participants doing the same study twice, it can be a hassle for you as a study developer who just want to try out the General Single Run a second time. Luckily there is an easy way around: Since for a General Single Run all studies that the worker already participated in are stored in a browser cookie, it can be easily removed. Just remove the cookie with the name JATOS_GENERALSINGLE_UUIDS in your browser. You can find this cookie in every webpage hosted by a JATOS server. If it doesn’t exist you probably never did a General Single run yet.