Original citation. Payne, B. K. Burkley, M. A. Stokes, M. B. (2008). Why Do Implicit and Explicit Attitude Tests Diverge? The Role of Structural Fit. Journal of Personality and Social Psychology, 94, 16–31.

Target of replication. The finding we reproduced is related to the structural-fit hypothesis, which predicts that implicit and explicit measures with the most similar procedural qualities will show the strongest correlations.

A priori replication criteria. A successful replication would find a moderately intense correlation between the two measures (e.g., r>.25) and a moderation effect of the social pressure manipulation on the relationship between direct and indirect measures

Materials, Data, and Report Study materials can be found in the materials component of this project. Raw data and the analysis script can be found in the dataset node. The full report and other materials appear in the files section of this node.

Conclusions. The present replication of Study 4 from Payne et al. (2008) successfully replicated the original results. The relationship between the direct and indirect procedures employed by the authors is high and, most importantly, it is moderated by the perceived pressure to respond in a socially desirable manner. The structural-fit hypothesis seems to hold in a sample of different citizenship (Italy), although the effect we have observed in our replication (ΔR2=.016) is substantially smaller than the original effect (ΔR2=.09). We have no interpretation of this difference, although it might be due to the specific sample involved in this replication.

For the purposes of the Reproducibility Project: Psychology, to compare effect tests, the original article's t value ( t(67) = 3.08 ) was converted to an F value: F(1, 67) = 9.49. And, the F value from the replication ( F(1, 176) = 4.063 ) was converted to a t value: t(176) = 2.016.

Download the full report

Figure 1. Effect sizes compared. Error bars represent 90% confidence intervals.

Note: Confidence intervals and R-squared values have been estimated using Smithson's (2005) procedure (see Dataset node)