Github pages site for A-maze.

View the Project on GitHub vboyce/Maze

Thought pages
What is A-maze?
An argument for 'redo' mode
New delay feature for 'redo' mode
Experiment design
Parameter considerations

Practical pages
Install instructions
Basic use
Advanced options
Using Ibex for Maze
Hosting an Ibex-Maze Server

Implementation of Maze for Ibex

To run Maze experiments online, there is a custom Ibex module that runs the Maze task. It is an adaptation of the SPR module, and uses the same timing mechanism as SPR. For general information on using Ibex, please see user documentation at https://github.com/addrummond/ibex/blob/master/docs/manual.md.

The Ibex implementation is now kept in a separate repository from the A-maze implementation, although they are meant to work together.


Note: If you want to add Maze functionality to an existing Ibex implementation (for instance, to run experiments on Ibex farm), these are the changed files:


If you’re building maze materials automatically using https://github.com/vboyce/Maze/tree/master/maze_automate, you can get those materials in ibex-format, and just copy those lines into the items list of a ibex experiment file.

If you’re not, these are the available arguments and specifications.

Options for Maze:

var defaults = [“Maze”, {redo: true}];

to the top of the data file. (Note that to check all experimental items, you may need to relabel the types of sentences to override Latin square behavior.) This mode can also be used when running the experiment; both the time until first press, and the total time to correct are recorded, so it you can tell how long participants are spending after making mistakes. (From some limited testing, this mode seems to work, and is useful if your items are long or you don’t proofread your distractors.) We recommend you consider whether your experiment would benefit from having participants continue after mistakes or not.

Passing results to next item:

When a participant makes an incorrect selection, maze will record the time it took them to do so, but then it will stop that sentence, and pass “failed” to the next element. It is recommended to use “followEachWith” for maze items to follow each with a separator saying whether it was correct or not, so participants know when a new sentence is starting.

Maze also displays a count of words gotten right so far. If you use the normal separator, the count will reset to 0 at the beginning of each item. If you want the count to be cumulative, you need to use MazeSeparator (.css and .js files included) to separate. All items that intervene between Maze tasks must contain the lines below for cumulative counting to work.

var x = this.utils.getValueFromPreviousElement(“counter”);

if (x) this.utils.setValueForNextElement(“counter”,x);

Note that this means that if you have blocks, with a message in between blocks, the count will reset at the start of each block, which may be desirable.

Result columns:

The first seven are same as for other Ibex modules; see Ibex documentation for an explanation.