Hi there,
I had the same problem. Jason suggested that this might be due to the graphics card. Since I could not easily change the graphics card of our lab notebooks my colleague and I tried to find a workaround.
In the attachment you will find a sample IAT demonstrating what we thought might be a managable solution (also for data analysis). Please note that we aimed at an intertrial interval of 150ms for correct and incorrect responses. Therefore, in the style file we used 70ms for the intertrial interval. This means for the correct response: 70ms for the first trial, 70ms for the jump to the blank screen and 10ms for the duration of the blank screen. For the incorrect response, we add the remaining time (80ms) as a blank screen to complete 150ms.
For data analysis one will need the log-file in which both rt for the incorrect responses are stored. These rt need to be aggregated in order to derive at the rt until correct response is made. This should be easy since both rt are stored with the same trial number.
We hope this makes sense and helps.
Jenny & Kevin