Date Published: October 11, 2018
Publisher: Public Library of Science
Author(s): Ratko Đokić, Maida Koso-Drljević, Nermin Đapo, Benjamin Peters.
The study examined the psychometric characteristics of three automated working memory span tasks: operational, reading, and symmetry span task, administered in groups of participants. For each task, the metric characteristics of six scoring procedures were evaluated: partial, absolute, partial non-weighted, absolute non-weighted, partial weighted, and absolute weighted scoring. Metric characteristics of all measures were compared across two parallel analyses: with and without application of a typical 85% accuracy criterion on the processing component of the tasks. The study demonstrates that the group administration of span tasks does not compromise their psychometric characteristics. All the tasks had an adequate internal consistency with Cronbach’s αs equal to or above .70; the exception being all types of the symmetry span task absolute scores with α values close to .60. Furthermore, all tasks have satisfactory convergent construct validity as well as criterion validity estimated in relation to measures of fluid intelligence. Omitting the 85% accuracy criterion on the processing component of the span tasks did not impair their psychometric properties. Thus, it is recommended that researchers discard this accuracy criterion as a criterion for filtering the results for further statistical analyses.
By developing further the concept of the central executive from Baddeley and Hitch’s working memory (WM) model , Engle and colleagues proposed a dual-component model of working memory capacity (for an overview, see: [2–6]). Within the dual-component model, working memory capacity (WMC) is defined by two controlled functions of the central executive that surface in the conditions of interference. The first function is executive attention, i.e. one’s capability to direct attention within the domain of primary memory in order to maintain relevant information in a highly activated state in the presence of interferences . The second function is a strategic search of secondary memory, which reflects the capability to retrieve relevant information from longer-term storage, in the presence of competitive, irrelevant information [5,6]. This twofold nature of the WMC is reflected in the structure of WM span tasks (span tasks). In span tasks, in between presentations of the individual to-be-remembered items (e.g. individual letters), which comprise the memory component of the tasks, participants are engaged in processing portion of the tasks (e.g. solving simple mathematical equations). In that way, WMC–operationalized as the number of to-be-remembered items recalled in correct serial order–is measured in conditions of interference caused by the second, processing component of the task. At the conceptual level, individual differences in WMC have two main sources. The first is an individual’s ability to maintain the relevant information in the focus of attention. If this is not possible (for example, because of interferences triggered by the processing component of the span task) this information needs to be efficiently recalled from the secondary memory in the presence of concurrent information (see: [5–7]). Thus, the ability to retrieve information from secondary memory comprises the second source of variability in WMC.
Of the initial 504 participants, two forgot to bring their spectacles to testing; testing for two participants of OSPAN and SSPAN was terminated because of distractions, whereas data for four participants of SSPAN, three of OSPAN, and two of RSPAN were not saved due to the experimenter’s error. Due to computer malfunction, individualised times for the processing component at RSPAN were not set for two participants. One participant started performing SSPAN before the other participants. The results of these participants on the named span tasks were excluded from further analyses.
The results of the study show that the group administration of automated span tasks does not compromise their psychometric characteristics. All the tested tasks demonstrate an adequate internal consistency with α values equal to or above the commonly used criterion of .70 (see [7,8]); the only exception being all types of SSPAN absolute scores, with αs in the range of .60. Furthermore, all tasks have satisfactory convergent construct validity as well as criterion validity estimated in relation to the gF measures. The obtained psychometric indicators correspond to the norms established by the individual administration of span tasks [7–9].