The biggest challenge facing Windows experimentation software is providing reliable, accurate timing data. The problem is that Windows is not a real time operating system, meaning that it may introduce anomalies into stimulus presentation times and response latency measurement. For example, an experiment might present an very large number of large images on a single trial, the combination of which exceeds display memory capacity. Or, a high priority system thread may steal CPU cycles from Inquisit while it is measuring response time.
To be sure, the experimenter can minimize the chances of anomalies by making sure the system has enough display memory and by shutting down other applications while running Inquisit. Furthermore, Inquisit can minimize the chance that other programs, including Windows itself, will steal CPU cycles from it by setting its process and threads to the highest priority level.
The bottom line is that no software package can guarantee accurate timing every time. However, a well designed application can take steps that make timing anomalies very rare. Inquisit has been developing and tuning its design since 1995 when Microsoft introduced DirectX for Windows 95 in order to provide reliable, millisecond-accuracy timing on Windows.
De Clercq, Crombez, Buysse, & Roeyers (University of Ghent, Belgium) have conducted an independent audit of Inquisit using FASTLOG, a new program they developed for testing timing accuracy of PC experimentation software. Their results confirm that Inquisit provides millisecond accuracy stimulus presentation and response timing.
To read a summary of their results, click here . The full report was published in the following article:
De Clercq, Crombez, Buysse, and Roeyers (2003). A simple and sensitive method to measure timing accuracy. Behavior Research Methods, Instruments and Computers, 35, 109-115.