Habit 2.2.10 released

Habit 2.2.10 released

Quick Summary

  • Habit version 2.2.10 is an update to version 2.2.8.

Habit 2.2.10 released

A new version of Habit, version 2.2.10, has been released and is available on our downloads page. 

This version fixes a BUG that may have caused inaccurate results in some cases - see below for more. 

This version will run on macOS 10.13-13 (High Sierra, Mojave, Catalina, Big Sur, Monterey, Ventura) and Windows 10 (Windows 11 _should_ be supported, but I have not tested yet - anyone else?)

 Bug Fix

There was a bug, present in ALL versions of Habit from 2.2.3-2.2.8, that may affected experiments with specific settings and some perfectly-timed key presses. This bug is fixed in version 2.2.10. 

What experiments were potentially affected?

An experiment whose Trial Settings (for any phase) had a trial-ending criteria "Use Look Settings" chosen, and whose "Look Settings" had a maximum looking-away time value greater than zero is susceptible to the bug.  

Look Settings with max look-away time greater than zero.
Trial Settings using "Single Complete Look"
What must happen for the bug to occur?

   If your experiment had that combination of settings for a phase, the "Use Look Settings" & "Single Complete Look" checkboxes mean that when one of these trials is running, and Habit detects a "complete look", the trial ends. 

   The bug requires that the trial end with _very_ precise timing.

   In the example settings I've shown, the Look Settings have a Maximum looking-away time of 200ms. If you have a looking-away time of at least that time, the trial ends. But, if the looking-away time is exactly the same length as the maximum looking-away time (in this case, 200ms), then the trial ends and the look which ended the trial is reported twice in the results. The "extra look" has the same start and end time, and the same total looking time. The correct treatment of the data is to remove one of the two looks from your results, or to apply the fix I've included with version 2.2.10.

   I should emphasize - the chances of hitting this bug are remote but NOT ZERO.  In the example above, you'd hold down the "5" key to indicate looking for some time - at least 2000ms. The subject looks away, but only briefly. You release the "5" key, then quickly press it again. You must have done that exactly  200ms after releasing the key! If you press 201ms later, the trial will end correctly, and the look is reported correctly. If you press 199ms later, the trial will continue (199ms is less than the maximum looking-away time)

How can I determine if any of my experiments were affected?

   There are a couple of ways to ensure that you are not affected by this bug. 

  1. Check for duplicate looks. This is straightforward - depending on how you are handling the results of your experiments, you can simply check that no looks are duplicated. By "duplicated" here, you must make sure that you check for looks in the same trial, with identical start, end, and total looking times. Looks cannot be duplicated across trials or phases. 
  2. Habit 2.2.10 has a tool that will scan an entire folder of experimental results, and report if any of the results files contain duplicated looks. See below for more.
Can I check my existing results files using Habit?

   Yes! Habit 2.2.10 has a tool that will check every result for a given experiment in your workspace. To use the tool, first install Habit 2.2.10, then follow the steps below.

Open the Results Viewer.
Open Results Viewer
Select an experiment to check, and click "Check Results"
Select an experiment and click "Check Results"
Scroll through the results. Green is OK - no duplicates; files with duplicate looks are highlighted in yellow.
Scroll through output.