. Data analysis
In order to assess the reliability of ISV, several parameters were computed. Unless stated, all parameters were computed separately for each participant on each task, session, and, where appropriate, condition.
Firstly, an accuracy measure (correct responses/total number of trials) was computed. After this, trials where responses were either incorrect or were made within 120 ms (and thus were presumably pre-emptive) were removed from the data The second group was made up of measures of variance: Standard deviation (SDRT) is probably the most familiar measure of variability. Due to its squaring of deviations in order to obtain uniformly positive values, it gives a somewhat disproportionately high weighting to extreme scores. Coefficient of variation (CVRT), defined as SDRT/MRT, is a measure of how much variability exists per unit of MRT. Mean absolute deviation (MAD) is the mean deviation from the MRT, rectified so that opposite signed deviations do not cancel. As MAD uses absolute rather than squared values it should be less influenced by extreme values than SDRT. Range is simply the difference between the fastest and slowest RT. Interquartile range (IQR) is the difference between the first and third quartile of the RT distribution. r is the estimated standard deviation of the Gaussian component of the ex-Gaussian distribution.
Finally the standard deviation of reciprocal RTs (RecipSDRT) and r
parameter of the LATER model were used. The third group consisted of measures of asymmetry and of the
size of the right-handed tail of the distribution: skewness is a measure of a distribution’s asymmetry. RT distributions are generally positively skewed, possibly because RTs are effectively capped at the fast end, but largely uncapped in terms of how slow a response can be. Thus a high positive skew is indicative of these very slow responses. s is the estimated mean and standard deviation of the exponential component of the ex-Gaussian distribution.