Analysis of affective instability in ecological momentary assessment: Indices using successive difference and group comparison via multilevel modeling.

Abstract

Temporal instability of affect is a defining characteristic of psychological disorders such as borderline personality disorder (BPD) and mood cycling disorders. Ecological momentary assessment (EMA) enables researchers to directly assess such frequent and extreme fluctuations over time. The authors examined 4 operationalizations of such temporal instability: the within-person variance (WPV), the first-order autocorrelation, the mean square successive difference (MSSD), and the probability of acute change (PAC). It is argued that the MSSD and PAC measures are preferred indices of affective instability because they capture both variability and temporal dependency in a time series. Additionally, the performance of these 2 measures in capturing within- and between-day instability is discussed. To illustrate, the authors present EMA data from a study of negative mood in BPD and major depressive disorder patients. In this study, MSSD and PAC captured affective instability better than did WPV. Given that MSSD and PAC are individual difference measures, the authors propose that group differences on these indices be explored using generalized multilevel models. Versions of MSSD and PAC that adjust for randomly elapsed time interval between assessments are also presented.

Sleep/Wake Patterns and Parental Perceptions of Sleep in Children Born Preterm

“…20 Variability of nightly scheduling of bedtime and wake time, and of sleep duration was calculated according to methods described by Sanchez-Ortuno et al 21 For each participant, differences between each consecutive measurement (e.g., bedtime night 2 minus bedtime night 1) were calculated and squared, capturing both variability and temporal dependency across time. 22 These values were used to determine differences in mean variability between groups. The square roots of the weighted means are reported.…”

Predicting Depression From Language-Based Emotion Dynamics: Longitudinal Analysis of Facebook and Twitter Status Updates (Preprint)

“…Negative affect instability, as measured by the MSSD, has been linked to more severe depression symptoms across several studies and has been identified as a concomitant and early indicator of depression [22,29,[31][32][33][34]. It has been shown to be a significant risk factor for more frequent and severe suicidal ideation [33] and may be a unique underlying emotion pattern in depression.…”

“…This time-structured measure of variability uses the temporal ordering of measurements to quantify the magnitude of incremental changes in emotion [28][29][30]. Crucially, unlike iSD where the same result would be obtained if the same set of emotion expression observations are shuffled through time, the MSSD is sensitive to the time-ordering of observations.…”

“…Thus, in addition to considering time-ordering, the time elapsed between successive observations also needs to be considered. Emotion instability, when operationalized as MSSD, assumes even sampling of observations to be computed meaningfully [28,29]. Where this is not possible, adjustments can be applied to the data to provide a weighted estimate of time-structured variability [29].…”

“…Where this is not possible, adjustments can be applied to the data to provide a weighted estimate of time-structured variability [29]. As in Jahng et al [29] the following time-adjusted MSSD, which accounts for an uneven sampling of observations through time was applied:…”