Single-Category IAT (SC-IAT)

The SC-IAT is a variant of the Implicit association test (IAT). This chapter describes in which aspects the SC-IAT differs from the IAT.

In the “normal” IAT, the association of a dimension (spanned between two opposing terms/categories) and a second dimension (usually an evaluation/assessment) is determined. The SC-IAT identifies the association between a single term (category) and a dimension.

The test procedure and scoring in SoSci Survey follow Karpinski and Steinman (2006).

As with the IAT, the strength of association is calculated as a D-score for the SC-IAT. The procedure is described in Karpinski and Steinman (2006).

The cleanup is performed according to the following criteria:

• In how many trials (of all blocks including practice blocks) did the participant incorrectly match the stimulus? If more than 20% of the answers are incorrect, the test is invalid.
• Trials with a response time below 350 ms are excluded from the analysis.
• Trials in which the participant has not answered in the given time will be excluded.
• If a participant has incorrectly assigned the stimulus in a trial, the mean value of all non-excluded trials (in the block) plus 400 ms is used as the response time for this trial.

From the response times adjusted and corrected in this way, the result is calculated as follows:

1. The difference between the average response times of block 4 and block 2 is calculated (the shorter exercise blocks 1 and 3 are not included in the evaluation).
2. This difference is divided by the standard deviation of all response times from those trials in which the participant answered correctly.

A positive D-score means that the target concept (“object”) is more strongly associated with the concept entered for “Positive” in the evaluative dimension (or with the stimuli entered for “Positive”) than with the concept entered for “Negative”.

In the SC-IAT, only two blocks are recognizable for the respondents. Each of the two blocks consists of 24 practice trials and 3 times 24 = 72 measurement trials. The transition from practice to measurement is not apparent to the respondents.

• Instructions
• 1 = Practice: good + self ↔ bad
• 2 = Test: good + self ↔ bad
• Instructions
• 3 = Practice: good ↔ bad + self
• 4 = Test: good ↔ bad + self

The variables on the number of trials include all trials (i.e., also the practice), the D-score only uses the 2 times 72 trials from the actual measurement.

To estimate the reliability, SoSci Survey calculates not only the D-score for all (measurement) trials, but also three further D-scores for the first, second, and third 24 trials of the (measurement) trials. The reliability can be estimated from their correlation.

Karpinski, A., Steinman, R. B. (2006). The Single Category Implicit Association Test as a Measure of Implicit Social Cognition. Journal of Personality and Social Psychology, 91(1), 16 –32. doi: 10.1037/0022-3514.91.1.16