Abstract
In the present study an assumption of retrieval theories of negative priming was tested. In particular, retrieval theories assume that the same underlying process causes benefits in target-to-target repetition trials and cost effects in distractor-to-target repetition trials in selection tasks in which targets are accompanied by distractor stimuli. More specifically, retrieval theories predict a negative correlation: The higher the benefit in target-to-target repetition trials is, the higher the cost effect in distractor-to-target repetition trials should be. In fact, in a study with considerable power (N = 110) and a design that allowed a confound-free detection of correlations, I observed a significant negative correlation between benefit and cost effects. This result is interpreted as evidence for a common retrieval mechanism.
Acknowledgments
The research reported in this article was supported by a grant of the Deutsche Forschungsgemeinschaft to Christian Frings (FR 2133/1–1). The author thanks Dirk Wentura for helpful discussions concerning statistics and stylistic suggestions. Further I would like to thank Axel Buchner, Jason Leboe, and an anonymous reviewer for their helpful comments on an earlier version of this article.
Notes
1 Strictly speaking, this assumption holds only if we additionally assume that interindividual differences in retrieval (or in the latent variable causing NP) do exist. The contrary view would be to assume that all participants have the same latent NP effect, and any empirical difference in NP is due to measurement error. However, there are several studies in which meaningful correlations between NP and other measures were reported (e.g., Conway, Tuholski, Shisler, & Engle, Citation1999; Frings, Wentura, & Holtz, Citation2007; Green & Williams, Citation1999; Long & Prat, Citation2002; Stavridou & Furnham, Citation1996; Tipper & Baylis, Citation1987), which make it very plausible to interpret empirical differences in the NP effect—at least in part—as differences in the latent variable that causes NP.
2 All power analyses reported in this article were calculated with the G-Power software Version 3.03 (Faul, Erdfelder, Lang, & Buchner, Citation2007).