Millisecond Forums

IAT, errors, and response time

https://forums.millisecond.com/Topic12978.aspx

By Pelle - 4/13/2014

Hi

I've been using Inquisit 4 with a version of the Implicit Association Test script for Word Stimuli provided by you, followed by using your SPSS script for treating the data. My main question is - what kind of error penalty is used?

From what I can gather, the Inquisit scripts uses a "built-in" error penalty, but does this mean that any time is added when subjects make errors? Or simply that the feature where the "X" appears and subjects are forced to answer correctly itself serves as the "built-in error penalty"?

Part of the issue is that in the SPSS script, it is possible to alter the deletion of certain trials, but there seems to be nothing concerning added time as error penalties. So I guess this means that the Inquisit and SPSS scripts does not add any time, and thus produces a D1 / D2 measure (rather than, say, a D6 measure, according to Greenwald et al 2003)?

Thanks!
By Dave - 4/13/2014

> Or simply that the feature where the "X" appears and subjects are forced to answer correctly itself serves as the "built-in error penalty"?

Yes, that's exactly what built-in error penalty means (cf. Greenwald et al., 2003). Latency reflects submission of the corrected response. It's the recommended D-score variant (also see https://www.millisecond.com/forums/Topic3444.aspx).
By Pelle - 4/13/2014

Perfect! Thanks for the clarification!

As I found no reference to D1/D2, I took "built-in" as possibly referring to the fact that some error penalty was present already within the IAT script itself, rather than being added later in the processing of the data.