3.1. Psychometric Test
For the 39 participants with ERP data, we merged the psychometric
results from all blocks and across Asian and Caucasian faces. In the
difficult task, accuracy was M = 74.23% correct (SD =
7.55; range 60.00 – 92.50%) and mean inverted RTs (1/s) were M= 0.90 (SD = 0.18; range: 0.56 – 1.18). In the easy task,
accuracy was M = 95.17% correct (SD = 4.49; range 80 –
100%) and mean inverted RTs (1/s) were M = 1.14 (SD =
0.17; range: 0.68 – 1.46). Paired t -tests revealed that in the
easy task accuracy was significantly higher than in the hard task
(\(t_{\left(61.89\right)}=14.88,p<0.001\)); similarly, inverted
RTs showed higher speed in the easy task
(\(t_{\left(38\right)}=11.17,p<0.001\)).These results demonstrate
that the two tasks differ in accuracy and response speed as intended,
yielding easy versus difficult versions indeed.
In order to assess the reliability of the psychometric tasks, we
calculated the correlations of the performance between Run 1 and Run 2
(after averaging across Asian and Caucasian faces) for each test
versions. For the easy task, inverted RTs in Run 1 and Run 2 were
correlated r = 0.820, corresponding to a split-half reliability
of 0.901 after Spearman-Brown correction. For the difficult tasks,
accuracy in Run 1 and Run 2 was correlated r = 0.541,
corresponding to a split-half reliability of 0.702 after Spearman-Brown
correction. These results are similar as those of Cepulic et al. (2018;
online supplement), where the inverted RTs in the two runs in easy
Caucasian and Asian faces correlated r = 0.71 and r =
0.56, respectively, and for the accuracy in the hard versions the
correlations were r = 0.59 and r = 0.58, respectively.
Moreover, the present results indicate excellent reliability for the
easy task and sufficient reliability for psychometric purposes for the
difficult task.