The impact of relevant versus irrelevant media multitasking on academic performance during online learning: a serial of mediating models

The paper addresses how media multitasking during online learning relates to academic performance when distinguishing academically relevant (on-task) from academically irrelevant (off-task) behaviors. With the rise of online education and frequent digital distractions, prior research has largely focused on off-task multitasking and found generally negative associations with academic outcomes, often attributed to attentional distraction. However, learners also perform course-relevant media activities (e.g., note-taking, accessing materials), whose effects may differ due to task relevance. Drawing on the attentional distraction hypothesis and cognitive load theory, the authors propose that academically relevant multitasking may increase germane cognitive load and support learning via self-regulation and flow, whereas academically irrelevant multitasking may impair attention and disrupt flow. The study tests three hypotheses: H1: Self-regulation strategies mediate the relationship between media multitasking type and academic performance. H2: Flow experience mediates the relationship between media multitasking type and academic performance. H3: Self-regulation strategies and flow experience serially mediate the relationship between media multitasking and academic performance.

Prior work consistently links media multitasking—especially off-task activities—to poorer academic outcomes, reduced attention control, and more attentional lapses (attentional distraction hypothesis). Yet task relevance can moderate multitasking effects; secondary tasks aligned with primary goals may not harm performance. Cognitive Load Theory suggests learning benefits from minimizing extraneous load and enhancing germane load; academically relevant multitasking (e.g., note-taking, targeted searches) could strategically increase germane load. Self-control includes both effortful inhibition and proactive self-regulation strategies; the latter may be especially crucial in autonomous online settings and relate positively to academic performance. Flow theory defines flow as a state of deep absorption with clear goals and feedback; focused attention is central. Off-task multitasking likely disrupts attentional focus and flow, whereas on-task multitasking may support germane load, skill–challenge balance, and thus flow. Limited prior studies have simultaneously examined both types of multitasking; mixed findings underscore the need to parse relevance and underlying mechanisms (self-regulation, flow).

Design: Cross-sectional survey. Participants: 608 recruited via WeChat and online questionnaire platform (June 2022), with snowball sampling; after excluding speeders and outliers (<180 s or >15 min completion), N=557 valid responses (effective rate 91.61%). Mean age 19.95 (SD=1.52); 126 males (22.62%), 431 females (77.38%). All students had eight weeks of online learning during COVID-19. Ethics approval obtained; informed consent collected; anonymized participation; single IP submission enforced. Measures: (1) Media Multitasking Questionnaires adapted from Baumgartner et al. measured frequency during online courses on 4-point Likert scales. Academically relevant items (4): e.g., class-related discussions via messaging apps; course-related web searches; reading related materials; note-taking. Academically irrelevant items (9): e.g., music, games, videos, social media, non-course messaging, unrelated browsing, shopping, other unrelated tasks. CFA acceptable though not ideal: AR-MMQ: χ²/df=9.78, CFI=0.93, TLI=0.80, RMSEA=0.13, SRMR=0.03; reliability α=0.65, AVE=0.34, CR=0.66. AIR-MMQ: χ²/df=7.02, CFI=0.91, TLI=0.87, RMSEA=0.10, SRMR=0.05; reliability α=0.90, AVE=0.51, CR=0.90. (2) Self-Regulation Strategies (SRS) from OL-MARS v.2 (Behavioral Strategies 6 items; Outcome Appraisal 3 items), 5-point scale; CFA fit χ²/df=4.26, CFI=0.93, TLI=0.90, RMSEA=0.08, SRMR=0.05; α=0.81, AVE=0.42, CR=0.87. (3) Flow Experience Scale (Chinese revised, 4 items), 5-point scale; CFA χ²/df=4.13, CFI=0.99, TLI=0.97, RMSEA=0.08, SRMR=0.02; α=0.83, AVE=0.56, CR=0.84. (4) Academic Performance Scale (5 items, self-evaluated within a named course; varied disciplines), 5-point scale; CFA χ²/df=2.51, CFI=0.99, TLI=0.99, RMSEA=0.05, SRMR=0.02; α=0.89, AVE=0.63, CR=0.89. Data Analysis: Descriptive statistics and Pearson correlations in SPSS 26. Harman’s single-factor test indicated no serious common method bias (first factor 24.81%, <40%). Structural equation modeling in Mplus 7.4 tested serial mediation, with centralized variables, age and gender as covariates. To reduce estimation bias, balanced item parceling: AR-MMQ, Flow, and AP parceled into two indicators each; AIR-MMQ and SRS into three each. Model reporting focused on structural paths and fit indices.

- Descriptives/correlations (Table 1): AR-MMQ correlated positively with SRS (r=0.322, p<0.01), FL (r=0.400, p<0.01), and AP (r=0.392, p<0.01). AIR-MMQ showed no significant correlations with SRS (r=-0.044, p=0.297), FL (r=0.063, p=0.136), or AP (r=-0.008, p=0.849). - Due to lack of associations, mediation analyses were not pursued for AIR-MMQ. - Serial mediation model (predictor: AR-MMQ; mediators: SRS then FL; outcome: AP; covariates: age, gender) fit well: χ²/df=2.91, CFI=0.97, TLI=0.96, RMSEA=0.06, SRMR=0.04. Path coefficients: AR-MMQ → SRS β=0.425, p<0.001; AR-MMQ → FL β=0.314, p<0.001; SRS → FL β=0.469, p<0.001; SRS → AP β=0.140, p<0.05; FL → AP β=0.644, p<0.001; direct AR-MMQ → AP β=0.101, p=0.068 (ns). - Bootstrapped indirect effects (Table 2): Total indirect=0.469 (SE=0.080, 95%CI [0.330, 0.650]). Indirect 1 (AR-MMQ→SRS→AP)=0.072 (SE=0.035, 95%CI [0.015, 0.159]), 12.2% of total effect. Indirect 2 (AR-MMQ→FL→AP)=0.243 (SE=0.070, 95%CI [0.128, 0.404]), 41.1%. Indirect 3 (AR-MMQ→SRS→FL→AP)=0.154 (SE=0.034, 95%CI [0.099, 0.237]), 26.1%. Direct effect non-significant (p=0.068). - Overall, academically relevant media multitasking improved AP indirectly via SRS and FL (both individually and serially). Academically irrelevant multitasking showed no significant associations with SRS, FL, or AP.

Findings indicate that academically relevant media multitasking supports learning through enhanced self-regulation and greater flow, aligning with cognitive load theory’s notion of increased germane load for strategic on-task behaviors. While prior literature often reports deficits in attention and self-control among frequent multitaskers, those effects likely reflect off-task multitasking and effortful inhibition. In contrast, on-task multitasking seems associated with the proactive use of self-regulation strategies (e.g., device management, focus maintenance), facilitating attentional control and flow, which in turn predicts better academic performance. The null associations for academically irrelevant multitasking suggest that its impact may be context-dependent, moderated by self-control or compensated by re-engagement strategies, and sensitive to measurement choices (e.g., outcome metrics). Overall, the results nuance the multitasking-performance relationship by emphasizing task relevance and identifying self-regulation and flow as key mechanisms.

The study differentiates on-task (academically relevant) from off-task (academically irrelevant) media multitasking in online learning and demonstrates that only academically relevant multitasking relates to improved academic performance, indirectly via self-regulation strategies and flow experience. This advances a mechanism-based account consistent with cognitive load theory, suggesting that strategically aligned multitasking can enhance germane cognitive load and learning outcomes. Practical implications include designing interventions that nudge media use toward course-relevant activities (e.g., content modification tools that foreground learning-related content and minimize distractors). Future research should refine the taxonomy of academically relevant multitasking (e.g., note-taking vs. targeted searches), integrate objective and multimodal measures (e.g., eye-tracking, cognitive load indices, standardized grades), and employ longitudinal/experimental designs to establish causal pathways and test moderators such as self-control.

- Measurement granularity: Academically relevant multitasking was captured via frequency of broad behaviors without distinguishing heterogeneous subtypes; some goal-relevant behaviors may still induce extraneous load. Future work should disaggregate subtypes and assess dynamic cognitive load (germane vs. extraneous) with objective tools (e.g., eye-tracking). - Self-report method: Sole reliance on questionnaires during COVID-19 may miss transient/contextual dynamics; multimethod approaches (e.g., diaries, intensive longitudinal designs) are needed. - Outcome assessment: Academic performance measured via self-evaluation limits comparability with objective metrics; future studies should include standardized/objective outcomes. - Design: Cross-sectional data limit causal inference; longitudinal and experimental studies are needed to confirm temporal ordering and test mediators/moderators.