Method-Specific Variance in the Implicit Association Test.

Abstract

can be used to assess interindividual differences in the strength of associative links between representational structures such as attitude objects and evaluations. Four experiments are reported that explore the extent of method-specific variance in the IAT. The most important findings are that conventionally scored IAT effects contain reliable interindividual differences that are method specific but independent of the measures' content, and that IAT effects can be obtained in the absence of a preexisting association between the response categories. Several techniques to decrease the impact of method-specific variance are evaluated. The best results were obtained with the D measures recently proposed by A. G. Greenwald, B. A. Nosek, and M. R. Banaji (2003).

Validity of the Salience Asymmetry Interpretation of the IAT: Comment on Rothermund and Wentura (2004)

“…Second, R&W used millisecond-unit IAT measures rather than using either the log-transformed latency measure used in most of the existing published IAT literature or the improved D measure introduced by Greenwald et al (2003). Proneness of the millisecond-unit measure to cognitive skill artifact was demonstrated by Cai, Sriram, Greenwald, and McFarland (in press), Greenwald et al (2003), and Mierke and Klauer (2003). The two new experiments reported here avoided these problematic procedures.…”

“…If the finding of Mierke and Klauer's (2003) Experiment 1a and present Experiment 1 are taken together, salience asymmetries appear neither necessary (Mierke and Klauer's Experiment 1a) nor sufficient (present Experiment 1) to induce IAT effects. Nevertheless, it remains plausible that salience asymmetries might cause IAT effects in the absence of stronger cues to association (such as shared meaning or size-color contingency).…”

Personalizing the Implicit Association Test Increases Explicit Evaluation of Target Concepts

“…The IAT effect relies on response competition -the conflicting tendency to respond with the wrong key press when the response pairings are incompatible with one's associations (Greenwald & Nosek, 2001). Indeed, the mental costs associated with overriding the tendency to evaluate the target concepts when evaluation and categorization require different responses is thought to be a central component of the IAT effect itself (Mierke & Klauer, 2003). It is possible that the greater error-rate differential for the personalized task is a function of the increased response competition rather than evidence for task recoding.…”

“…That is, an error in the IAT could reflect either a misunderstanding of the rules, or a manifestation of the phenomenon of interest -e.g., strong associations between Bush and bad for Democrats lead them to respond more slowly and make more errors when Bush and bad have opposing response assignments. Indeed, response latency and error-rate differences are positively correlated in the original IAT showing that response competition both slows down responding and increases errors when the key assignments are "incompatible" (Greenwald et al, 1998;Greenwald & Nosek, 2001;Klauer & Mierke, 2005;Mierke & Klauer, 2003).…”

Consequential validity of the Implicit Association Test: Comment on Blanton and Jaccard (2006).

“…Although one might assume that scholarly criticism is unwelcome, we -as researchers identified with the IAT -find it extremely valuable. This value is evident in the several recent publications of method-focused IAT work (e.g., Cai et al, 2004;Greenwald et al, 2003;Greenwald, Nosek, Banaji, & Klauer, 2005;Greenwald et al, 2006;Mierke & Klauer, 2003;Pinter & Greenwald, 2005). Without the instigation provided by scholarly criticism of the IAT, there would have been less motivation to pursue questions of method.…”