3.2 Módulo 2. Políticas de empleo inclusivas, desde la perspectiva de la
3.2.4 Recomendaciones globales para generar políticas de empleo
Before providing a final evaluation of the current state of knowledge on the relation- ship between media multitasking and cognition, it is necessary to briefly consider the implications of a number of methodological factors present in research in this domain.
4.3.2.1 Paradigmatic Implications
Research in this domain has produced varied outcomes. In particular, the general pattern of findings produced on the basis of self-reports of everyday executive functioning differs from that indicated by laboratory assessments. A number of researchers have commented on the distinction between these two measures. Lin(2009, p. 15521), for instance, notes that “what happens in lab experiments does not often represent a complete picture of what happens in real life [...] the distractions in experiments are not necessarily dis- tractions in real life”. Outside of artificial laboratory settings decisions about what is a primary task and what is a distraction, or when switching should occur, can all affect cognitive control and task-switching performance. Ralph et al.(2014) suggest that the
links between performance-based measures and ‘real-world’ performance are often ten- uous. This assertion is supported by a recent review indicating that self-reported and performance-based assays of cognitive functioning do not correlate (Toplak et al.,2013). Performance-based assessments are administered in controlled conditions where perfor- mance is measured for accuracy or response times. In contrast, self-report assessments are designed to provide an indication of performance on tasks requiring cognitive control in everyday life (Gioia et al.,2015). Toplak et al.(2013) contend that, while related, these measures consider cognitive functioning at different levels. Performance-based assess- ments focus on the underlying information processing mechanisms of cognitive control, at a functional level. Self-reported, reflective assessments relate to goals, beliefs and reflections on action in context. They argue that, only at the reflective level do “issues of optimal decision making come into play” (p. 137). Performance, therefore, is grounded in context and is related to prevailing goals. They argue that performance-based mea- sures, in their assessments of functional efficiency, ignore the role of goals in directing behaviour and cognitive control. Toplak et al.(2013) support this assertion in quoting Salthouse et al. (2003, p. 569): “the role of executive functioning may also be rather limited in many laboratory tasks because much of the organisation or structure of the tasks is provided by the experimenter and does not need to be discovered or created by the research participant”. Finally, the authors note that this distinction is characteristic of the distinction between typical performance situations and optimal performance situa- tions. As typical performance situations are unconstrained by requirements to maximise performance, evaluations of performance in such situations assess goal prioritisation and the extent to which behaviour requiring cognitive control typically corresponds to these goals. The latter, however, are constrained, in that participants are required to maximise performance within the bounds of a given task, irrespective of their personal goals. Given ongoing psychometric debates surrounding these measures, it is not within the scope of this dissertation to assess their validity as indications of cognitive functioning. What is important, however, are implications for research in this domain. Irrespective of their validity as measures of cognitive control at a functional level, self-report measures capture the extent to which behaviour in context typically corresponds to goals. Given the variety of situations in which media multitasking occurs, such measures provide a valuable indication of situated action and reflections on combined everyday executive functioning. More research seeking to understand the differential outcomes extending from these measurement paradigms is required. Of importance is understanding the role of goals in driving media multitasking and the extent to which assessments capture this.
4.3.2.2 Small Sample Sizes
AsWiradhany and Nieuwenstein (2017) note, a key factor characterising studies in this domain is the reliance on small sample sizes. InOphir et al.(2009)’s early study the out- comes of only 30 participants were compared. This trend continued with similarly small samples characterising subsequent studies: 36 in Minear et al. (2013), 23 in Alzahabi and Becker (2013), 36 in Uncapher et al. (2016), 28 in Murphy et al. (2017). Similarly, in studies where the sample size differed across the groups the trend proceeded. For in- stance, in their second studyWiradhany and Nieuwenstein (2017), considered 14 LMMs and six HMMs. As noted previously,Ralph and Smilek(2017) suggest that, in such stud- ies, effect sizes can be overestimated and spurious. The reliance on small samples, across measurement paradigms, jeopardises the determination of the true relationship between media multitasking and cognitive control. Further research, with larger samples more adequately powered to statistically determine the nature of any relationship is required.
4.3.2.3 The Measurement of Media Multitasking
Studies have used either the MMI, as described by Ophir et al. (2009), or modified versions of this index. As discussed in Section3.1.1there are a number of shortcomings to the MMI as a measure of media multitasking. Baumgartner et al.(2017a) note that, as a relative measure, MMI scores cannot be interpreted as representing an absolute amount of simultaneous media use. While the formula accounts for each primary medium’s hours of use, this is nullified by dividing the overall score by the total use hours. An MMI score is therefore the amount of simultaneous media use relative to overall media use. A shortcoming with this calculation, as noted by Wiradhany and Nieuwenstein (2017, p. 20), is that “a person who spends only 1 hour per day using his laptop while watching television can have the same MMI as a person who does this 16 hours per day”. Consider an individual who always multitasks when they use media, but only uses media infrequently, and an individual who extensively uses media, but rarely multitasks. The first individual will receive a higher MMI score than the second, despite a similar amount of multitasking. Arguing that time spent media multitasking may be more important, Moisala et al.(2016) produced an index for the absolute time spent media multitasking. The calculation of the MMI has also been discussed as a shortcoming. Wilmer et al. (2017) argue that the matrix-structure, where each primary medium is considered in relation to a number of secondary media, regards all media as equal. The MMI increases
by the same amount regardless of the type of media, the attentional demands of a task or combination of tasks. As an example, they note how ‘playing video games’ and ‘lis- tening to music’ are regarded as equivalent activities, despite their different cognitive demands. Additionally, they contend that ‘playing video games’ while ‘reading print media’ presents a different cognitive demand than ‘listening to music’ while ‘instant messaging’. Such combinations have an equivalent effect on an MMI score. Moreover, an individual who extensively participates in one combination, ‘instant messages’ while ‘browsing’ for instance, but not other categories, will receive a low MMI. The MMI does not reflect extensive task-switching resulting from a single, highly performed, activity. AsMoisala et al. (2016) note, such shortcomings are indicative of the manner in which the MUQ considers media multitasking as specific primary and secondary task combi- nations. While in some instances this may reflect how media are used, in others it may not. Everyday media multitasking consists of a combination of task-switching, divided attention, and serialised dual-tasking (Moisala et al.,2016).
These shortcomings present in the dominant measure of media multitasking may have affected findings produced on its basis. If the MMI is not a valid representation of media multitasking, in terms of time spent media multitasking or the frequency of task- switching, the validity of findings indicating potential relationships with cognitive control, in any direction, is brought into question. However, despite its flaws, the measure does produce an indication of media multitasking relative to overall media use which, nonethe- less is useful for comparative purposes. While alternative measures have been proposed, Baumgartner et al.(2017a)’s MMI-S or Moisala et al.(2016)’s MMT for example, they suffer from many of the same shortcomings as the MMI. Baumgartner et al. (2017a), for instance, designed the MMI-S to decrease the time taken to collect and classify an individual’s media multitasking, not necessarily to produce a more accurate assessment. Only presenting a subset of media combinations implies that the measure is, by definition, an inaccurate reflection of an individual’s media multitasking.
4.3.2.4 The Extreme Groups Approach
Extending from the use of the MMI is the adoption of an extreme-groups approach to considering possible relationships between media multitasking and cognitive control. The performance of HMMs’ has been compared with that of LMMs. Ralph(2017, p. 21) notes, however, that, across studies considering media multitasking, “researchers have consistently found that there is no bimodal distribution of heavy media multitaskers
and light media multitaskers” —MMI scores are relatively normally distributed. Ralph (2017, p. 22) contends that “there is no clear reason why one ought to discard data from individuals whose scores fall in the middle portion of the distribution when one could examine the entire distribution”. Only three studies,Cardoso-Leite et al.(2016),Murphy et al. (2017) and Hadlington and Murphy(2018), consider comparisons for those falling between the arbitrary classifications of light and heavy media multitaskers. Other studies have conducted comparisons between the entire sample and various outcome measures (e.g.Baumgartner et al.,2014;Cain et al.,2016). In the extreme-groups approach studies adopted different notions for what constitutes a heavy or a light media multitasker. Some researchers, for instanceOphir et al.(2009),Cardoso-Leite et al.(2016), orAlzahabi and Becker (2013), took HMMs to have an MMI score one SD or more above the mean and LMMs to be one SD or more below the mean. In contrast, other researchers, for instance Minear et al. (2013) or Yap and Lim(2013), divided the sample in half, taking the top 50% of scores to be HMMs and the bottom 50% to be LMMs. Murphy et al. (2017) divided their sample into tertiles, with HMMs being those in the top third of the sample. The extreme groups approach regards all within a particular group as equal in terms of media multitasking. However, as a result of the normal distribution of MMI scores, those within each group —LMM or HMM— display a diversity of scores. Commenting on this characteristicRalph(2017) notes that in many studies in this domain (e.g.Ophir et al., 2009; Minear et al., 2013; Cardoso-Leite et al., 2016), the range of MMI scores falling between the upper bound of LMM scores and the lower bound of HMM scores is smaller than the range of scores for a particular extreme group. This outcome, result- ing from the normal distribution of MMI scores, combined with the small sample sizes typical of research in this domain, implies that the scores for both extreme groups are skewed —higher for HMMs and lower for LMMs. While some studies account for this by eliminating outliers from their comparisons (e.g.Alzahabi and Becker,2013), a majority of those considered have not. This skewing of scores has implications for the determina- tion of potential relationships with media multitasking, in terms of both direction and magnitude, and thus presents as a possible explanation for the inconsistencies reported.