I do have "rt:any" as the first screen on every trial so this would explain the problem. I'll just use the button-press data recorded in the output file to determine response accuracy in a new column. Thanks for clearing this up - it's reassuring to understand why the "Correct" output did not match what I thought it should be.