[solved] Accuracy of Reaction times

paul

Dear Sebastiaan,

I am thinking of using Open Sesame to run a whole series of psycholingustic experiments. My colleagues, although impressed by the functionality of the program, are wondering about the accuracy of the reaction times. I did run the experiment entitled test_suite, but at first sight the results (excel output) haven't proved that much enlightening to me. Am I supposed to compare them for consistency in some way? In general what am I supposed to look for in the results that would show that my system is providng good reaction times?

Your help would be greatly appreciated!

Thanks,

Paul

sebastiaan

Hi Paul,

Thank you for your interest! The test_suite.opensesame tests the timing accuracy of display presentation for the legacy back-end. There are now two new scripts available, which also allow you to test the psycho back-end. You can find them here: http://osdoc.cogsci.nl/miscellaneous/timing

But these tests are only about accuracy of display timing, i.e. whether displays are shown when you want to be shown. Accuracy of responses timing is another matter. In order to test this for yourself, you would need to use some kind of external device, you cannot rely on the computer's self report. We have done this in a benchmark experiment that is described in the paper: http://www.springerlink.com/content/n264513n66704v33/. Basically, timing can be excellent, provided of course that you use a response device that is capable of giving accurate timestamps.

If you'd like to verify this on your own systems, which is quite sensible, you'd have to run a similar benchmark experiment. If you decide to do so, I can certainly give some (my) thoughts on how to do this best, and I'd love to hear your results!

Kindest regards,
Sebastiaan