The correlations between the reaction time measures are
shown in Table 4. With regard to the Simple Reaction Time
(SRT) tasks, there was a large, significant positive correlation
between the mean response times of the Numbers task and
the Lights task (r = .68). There was a significant positive
correlation between the standard deviations (SD) of response
times of the Numbers task and the Lights task (r = .26). The
correlations of the means and SD within both reaction time
tasks were also all significant.
With regard to the Choice Reaction Time (CRT) tasks, there
was a very large, significant positive correlation between the
mean response times of the Numbers task and the Lights task
(r = .81). There was a large significant positive correlation
between the standard deviations (SD) of response times for the
Numbers task and the Lights task (r = .57). The correlations of
the means and SD within each task were also large and
significant: Numbers task (r = .78) and Lights task (r = .84).
Faster participants were less variable. There was a significant
positive correlation between the number of errors made in the
Numbers task and the Lights task (r = .19). There were few
errors overall. The number of errors and mean response times
within each task was slightly negatively correlated: Numbers
task (r = −.25) and Lights task (r = −.23).
Not all reaction time parameters are perfectly normally
distributed, especially errors forwhich even transformationswill
not achieve a normal distribution. Therefore, the above correlations
were all re-run using the non-parametric Spearman's ρ
coefficient. The results are shown in parentheses below the
Pearson correlations in Table 4. These differ very little from the
Pearson's r coefficients, and show that the Pearson coefficients
have not led to any over-estimation of effect sizes.