Computer-paced versus experimenter-paced working memory span tasks: Are they equally reliable and valid?

The Predictive Utility of a Working Memory Span Task Depends on Processing Demand and the Cognitive Task

“…From a methodological viewpoint, it is worth noting that our study supports the earlier finding that computer-paced WM span tasks are good predictors of higher order cognition (Bailey, 2012;Lépine et al, 2005;McCabe, 2010;Redick et al, 2012), and generalizes this finding to the more traditional, widely used Reading Span task as well as the new Word Problem Span task. Furthermore, our results indicate that the main difference between the two WM tasks used in this study is indeed the processing demands: the participants recalled fewer items in the memory task and made more errors in the processing task in the relatively more demanding Word Problem Span task than in the less demanding Reading Span task.…”

“…The first approach varies the existence of the processing task's time-limits (Lépine et al, 2005;McCabe, 2010). Studies using this approach have consistently demonstrated that a time-limited computer-paced WM span task is a more powerful method for predicting higher order cognition than a task that is not time-limited (Bailey, 2012;Lépine et al, 2005;McCabe, 2010). Lépine et al (2005) showed that regardless of the simplicity of the processing task, the computer-paced task predicted more of the variance in children's academic skills (mathematics and literacy) than the traditional self-paced WM task.…”

Does Controlling for Temporal Parameters Change the Levels-of-Processing Effect in Working Memory?

“…Other studies have indicated that a model with two different underlying latent factors either better accounts for the variability common to fixed-pace and unfixed-pace tasks ( Bailey, 2012 ) or is not significantly different than a model with a single latent factor ( Lucidi, Loaiza, Camos, & Barrouillet, 2014 ). Thus, in three different regards, the literature has shown that fixed and unfixed tasks are not identical; in fact, they often differentiate in terms of overall WM capacity (e.g., Barrouillet et al, 2004 ), predictive utility (e.g., McCabe, 2010 ), and in latent structure (e.g., Bailey, 2012 ). Such results highlight the possibility that complex span tasks with unfixed temporal parameters do not measure WM either as accurately as or similarly to tasks with fixed temporal parameters.…”

“…Indeed, in their recent studies, Rose and colleagues ( 2014 , 2015 ) showed that instructions to rehearse the word during an interval between the LoP judgment and recall of the word yielded a null LoP effect. Moreover, such strategy use could even change the validity of the WM measure (e.g., Bailey, 2012 ; Engle et al, 1992 ; Friedman & Miyake, 2004 ) to the extent that it may no longer properly reflect the construct. Thus, from a broader theoretical perspective, the validity of the WM measure must be ensured in order to clarify the effect of LoP in WM.…”

“…Accordingly, our study aimed to replicate the design of Rose and colleagues’ ( 2010 ) study except for one crucial methodological, but theoretically meaningful, difference: a fixed presentation of the processing component during which participants make LoP judgments. We sought to clarify that the results reported thus far persist even when accounting for theoretically meaningful methodological parameters that have been shown to, at the very least, reduce WM capacity (e.g., Barrouillet et al, 2004 ) and, at most, yield two different latent variables suggesting overlapping but still independent measures of WM capacity ( Bailey, 2012 ). This would allow researchers to better consider their theoretical implications, especially for as important a topic as the distinction between WM and EM.…”