None of the reaction time parameters used in the present
study has perfect reliability—and, of course, neither does the
cognitive tests—and so it is useful to be reminded of the relative
reliabilities of the parameters in assessing the correlations
found here. For the Numbers task, in a previous study, we
found almost-period-free test–retest reliability (Spearman's ρ)
as follows: SRT mean = .67; SRT SD = .20; CRT mean = .92;
and CRT SD = .73 (Deary & Der, 2005c). Therefore the
correlations, especially thosewith SRT SD, could not be expected
to have reached beyond modest effect sizes.
The correlations we found between reaction time and higher
cognitive ability test scores were somewhat lower than those
reported by Deary et al. (2001). This is partly due to two factors.
The first is that our present sampleswere somewhat restricted in
ability range by comparison with the sample in Deary et al.,
which was fully representative of its background population.
Disattenuation of the coefficients in the present study led to at
least the older sample's coefficients approximating those of
Deary et al. However, that study used the Alice Heim 4 test,
which is almost certainly a more reliable and broad test of
general cognitive ability than Matrix Reasoning and therefore
likely to be more strongly correlated with reaction time