Why a Standard IAT Effect Cannot Provide Evidence for Associative Learning: The Role of Similarity Construction. Bading, K.; Stahl, C.; and Rothermund, K.
abstract   bibtex   
Moran and Bar-Anan (2013) demonstrated that evaluation on a direct measure reflected information on both US valence and CS-US relations, while evaluation on an indirect measure reflected only information on US valence: alien creatures that ended a horrifying scream were rated more favourably than alien creatures that ended a pleasant melody; yet, in the IAT, the creatures that ended the melody were preferred over the creatures that ended the scream. The IAT effect was interpreted in terms of associative learning, and hence supported dual process models of EC. In the present study, we present an alternative explanation that is based on the similarity account of the IAT. This account claims that IAT performance is driven by flexible constructions of target-attribute similarity that rely on information which is salient in a given IAT and context. Across two experiments, we consistently found that IAT effects did not reflect rigid associations between creatures and valence, but instead drew on whichever information was highlighted by a given IAT. The findings are discussed with regard to theoretical models of EC as well as in reference to prominent accounts of IAT performance.
@article{bading_why_nodate,
  title = {Why a Standard {{IAT}} Effect Cannot Provide Evidence for Associative Learning: {{The}} Role of Similarity Construction},
  abstract = {Moran and Bar-Anan (2013) demonstrated that evaluation on a direct measure reflected information on both US valence and CS-US relations, while evaluation on an indirect measure reflected only information on US valence: alien creatures that ended a horrifying scream were rated more favourably than alien creatures that ended a pleasant melody; yet, in the IAT, the creatures that ended the melody were preferred over the creatures that ended the scream. The IAT effect was interpreted in terms of associative learning, and hence supported dual process models of EC. In the present study, we present an alternative explanation that is based on the similarity account of the IAT. This account claims that IAT performance is driven by flexible constructions of target-attribute similarity that rely on information which is salient in a given IAT and context. Across two experiments, we consistently found that IAT effects did not reflect rigid associations between creatures and valence, but instead drew on whichever information was highlighted by a given IAT. The findings are discussed with regard to theoretical models of EC as well as in reference to prominent accounts of IAT performance.},
  language = {en},
  author = {Bading, Karoline and Stahl, Christoph and Rothermund, Klaus},
  pages = {18},
  file = {G\:\\_lokal\\Zotero\\storage\\AK9V3RVA\\Bading et al. - Why a standard IAT effect cannot provide evidence .pdf}
}
Downloads: 0