JMIR Publications

ABSTRACT

The development of mobile technology with substantial computing power (i.e. smart phones) has enabled the adaptation of performance-based cognitive assessments to remote administration and novel intensive longitudinal study designs (e.g. measurement burst designs). While an “ambulatory” cognitive assessment paradigm provides new opportunities to extend current psychological theories, the adaptation of conventional measures to a mobile format conducive to intensive repeated measurement involves balancing measurement precision, administration time, and procedural consistency. Across 3 studies, we adapted ‘complex span’ tests of working memory capacity (WMC) for ultra-brief, smart phone-based administration and examined their reliability, sufficiency, and associations with full-length, lab-based computerized administrations. In a lab-based setting, Study 1 examined associations between ultra-brief, smart phone adaptations of the Operation Span, Symmetry Span, and Rotation Span tasks and full-length computerized versions. In Study 2, we conducted a 4-day ecological momentary assessment (EMA) study (4 assessments/day) where we examined the reliability of ultra-brief, ambulatory administrations of each task. In Study 3, we conducted a 7-day EMA study (5 assessments/day) involving the ultra-brief Rotation Span Task where we examined reliability in the absence of extensive onboarding/training. Measurement models in Study 1 suggest that comparable estimates of latent WMC can be recovered from ultra-brief complex span task performance on smart phones. Significant correlations between the ultra-brief tasks and respective full-length versions were observed in Study 1 and 2, ranging between r = 0.4-0.57. Results of Study 2 and 3 suggest reliable between-person estimates of Operation Span, Symmetry Span, Rotation Span, and latent WMC can be obtained in 2-3 ultra-brief administrations (equivalent to <1 day of testing in an EMA study design). Results of Study 3 replicated our findings showing that reliable between-person estimates of Rotation Span may be obtained in as few as 2 ultra-brief administrations, in the absence of extensive onboarding/training. In addition, modification of task parameterization for Study 3 improved estimates of reliability of within-person change. Considerations for future mobile cognitive assessment design and parameterization are discussed.