ITA Reanalysis

This script allows you to reanalyze typically reported results from studies on unconscious priming. The typical study compares an direct vs. an indirect task to infer an indirect task advantage (ITA), that is, a higher sensitivity in the indirect task compared to the direct task. However, the reported results typically only contain a t value from the indirect task but no sensitivity. This script estimates the indirect task sensitivity and compares it to the direct task sensitivity.

The underlying methods are developed in our paper: The unconscious priming fallacy: When can scientists truely claim an ITA?.

Meyen, S., Zerweck, I. A., Amado, C., von Luxburg, U., & Franz, V. H. (2022). Advancing research on unconscious priming: When can scientists claim an indirect task advantage? Journal of Experimental Psychology: General, 151(1), 65–81. https://doi.org/10.1037/xge0001065

Reported results to be reanalyzed


The typical result reported in the direct task is an average d' value (alternatively, an average %-correct value) from a repeated measures design sampling N participants in K trials. The indirect task result is typically a mean congruency effect on reaction times captured by a t value (alternatively, an F value) from a paired-t-test (repeated measures ANOVA) with N participants in a total of K trials. Setting a value for K assumes balanced trials in each of the two conditions (e.g., congruent vs. incongruent condition). Alternatively, to reanalyze unbalanced designs, you can specify M1 and M2 as the number of trials in each of the two conditions. (Setting K implies M1 = M2 = K/2.)

One additional parameter, q, has to be estimated. This q is the standard deviation of true sensitivities across participants. That is, participants vary in their true sensitivity, for example, some may have d'=0.1 as a true value, others have d'=0.3, and the standard deviation of these true values is q. A conservative assumption for RT data (in the sense of rejecting evidence for an ITA) is q2=0.152=0.0225. A balanced assumption is q2=0.12=0.01 and a liberal assumption is q2=0.052=0.0025. Caution is required when using neuroimaging data (EEG, fMRI, MEG etc.) instead of RT data because we cannot recommend plausible assumptions on q for these measures yet.

Direct task

Indirect task

d' value
%-correct (alternative)
N
K
 
 
q2

Insert:

t value
F value (alternative)
N
K
M1 (alternative)
M2 (alternative)
q2

Insert:







Reanalysis results


Our reanalysis method estimates from the t or F value from the indirect task the sensitivity and compares this to the direct task results. In other words, we test directly for a difference in sensitivity, which is necessary to provide evidence for an ITA.