Statistical Approaches to Analyzing Electrodermal Activity in Children with Autism Spectrum Disorder

• In Event: Poster Session 11
  In Poster Session: PS 11 Topic - Methods, History, Theory

Abstract

Researchers are increasingly utilizing physiological data to understand how stress “gets under the skin” and impacts the socioemotional functioning of children and youth. The measurement of electrodermal activity (EDA), which refers to changes in the electrical properties of the skin, offers a window into children’s sympathetic nervous system responses and has been associated with anxiety, attention, affect and sensory input (Dawson et al., 2007). A similar interest in EDA measurement in children with autism spectrum disorder (ASD) can also be seen; however, results of EDA studies in children with ASD are mixed, with some finding hyperarousal in ASD, others suggesting hypo-arousal, and still others finding no difference compared to their neurotypical peers. Some of the variability in results likely stem from the different analytic techniques used to assess EDA. Therefore, the purpose of this study is to investigate and compare commonly used approaches to EDA analysis and their link to standard measures of child behavior.

This exploratory study will include EDA data collected from 28 children with ASD (5-11 years old) during a seven-minute play interaction with their primary caregiver. EDA was collected using Empatica E4 Wristbands (McCarthy et al., 2016). EDA was sampled at 4 Hz, with values ranging from 0.01 to 100 μSiemens. In addition, the primary caregiver completed measures of children’s ASD symptoms via the Social Responsiveness Scale (SRS-2; Constantino & Gruber, 2012) and their emotional and behavioral functioning via the Child Behavior Checklist (CBCL; Achenbach & Rescorla, 2001), which will be used as validation measures of the EDA analyses.

Data collection is complete, and the EDA data are currently being preprocessed and prepared for analysis. Data has been reduced to 2-second bins, resulting in 210 data points per person. The proposed poster will include the following analytic approaches: (1) standard deviations – measures EDA variability or magnitude of change in EDA across the interaction, (2) nonspecific fluctuations in skin conductance (NSCR) – measures EDA variability via the frequency of “spikes” (defined as an increase of at least 0.03 μSiemens over a 3-second period) in arousal, and (3) latent growth curve modeling – models change in EDA over time in an effort to capture the magnitude of reactivity and trajectory of arousal over the course of the interaction. We will also present correlations between each EDA measurement technique and scores on the SRS and CBCL. While we expect EDA, ASD symptoms, and emotional and behavioral functioning to be associated across measurement techniques, we hypothesize that latent growth curve analysis will reveal significant between person variability in EDA not present in other techniques. We also anticipate a heightened autonomic response specifically for children with more severe ASD symptoms and emotional and behavioral problems. We do not anticipate any problems completing the proposed data analysis by the time of the meeting.

Authors

• Jessica Greenlee, Kenyon College
  Presenting Author

• Emily Lorang, University of Wisconsin
  Non-Presenting Author

• Sigan Hartley, University of Wisconsin
  Non-Presenting Author