What a D-score is
A D-score is a standardized difference between two pairing conditions in the seven-block IAT. In one condition two categories share a response key. In the other condition the pairing is reversed. People are usually faster when the pairing matches how they associate those categories. D is that difference, scaled so it is not just “milliseconds on a fast computer.”
IAT Design scores the improved algorithm published by Greenwald, Nosek, and Banaji in 2003. The Excel export is that number, plus the latencies and error counts it came from. If you publish, cite that paper for the scoring, not this page.
What you will see in the export
Each administration produces a D, an error count, and latencies by block. Survey answers sit on the same row as that D if you included a survey. The workbook is a record of the session. It is not a report writer and it will not tell you whether the score is “significant.”
What one administration is not
One person, one sitting, is a demonstration. It is not a diagnosis, not a stable trait score, and not a finding you can publish. D can move with practice, with how you worded the keys, with how tired the person was, and with how you built the blocks. Treat a single number as a check that the procedure ran, not as a conclusion about the person.
A class project or a study needs a question you wrote down before you looked at D, more than one participant, and the same test for each of them. This software will collect that. It will not design the study.
If you are relating D to something else
When you have many administrations and a survey item or a behavior you care about, the usual coefficient is Pearson’s r: the linear association between D and that other number. Positive r means higher D tends to travel with higher values of the other measure. Near zero means they do not move together in a straight line.
r does not prove that one thing caused the other. It does not replace looking at the scatter. It is not required to run or score an IAT. Compute it only if you have a second variable and a reason.
This page will not walk through the algebra. If you need the formula, any first-course statistics text is enough. If you need the scoring rules themselves, use Greenwald, Nosek, and Banaji (2003), Understanding and using the Implicit Association Test: I. An improved scoring algorithm.
This page does not put interpretive labels (“slight / moderate / strong”) on D. The export is the number.