Dyadic Interaction Patterns During Infancy and Early Childhood

Restricted (Penn State Only)
Bray, Brandon A
Graduate Program:
Master of Science
Document Type:
Master Thesis
Date of Defense:
November 07, 2017
Committee Members:
  • Ginger A Moore, Thesis Advisor
  • Pamela Marie Cole, Committee Member
  • Jenae Marie Neiderhiser, Committee Member
  • dyadic interaction
  • parenting
  • dyadic regulation
  • Hidden Markov Model
Dyadic interaction patterns, the dynamic social interplay between caregiver and infant characterized by each partner’s response to the behavior of the other, are considered one of the foundational factors of infants’ emergent self-regulation (Beebe et al., 1992; Kopp, 1982; Schore, 1996). Theoretically, then, dyadic interaction patterns should change over time as infants develop regulatory autonomy and capabilities for mobility, social engagement, and independence (Kopp, 1982), and vary depending on caregivers’ interactive styles (Feldman, 2007). Although research has examined links between early dyadic interaction patterns, measured as dyadic synchrony between parents’ and infants’ behaviors, and later child outcomes, relatively little is known about the specific types of parents’ and infants’ behaviors that typically co-occur at different ages. To address this gap, the current study provided detailed descriptive data about dyadic interaction patterns across infancy and early toddlerhood for mother-child dyads. Following advances in time-series data analytic methods for modeling dyadic data (e.g., Stifter & Rovine, 2015), the current study used Hidden Markov Modeling (HMM; Visser & Speekenbrink, 2010) to identify patterns of moment-to-moment behaviors co-occurring between mothers and their children (latent dyadic states) and to compute probabilities of transitions among those states at three ages: 9-, 18-, and 27-months. The current study used microcoded observations of adoptive mothers’ and infants’ behaviors collected as part of the Early Growth and Development Study (Leve et al., 2013) during an observational Teaching Task at child ages 9-, 18-, and 27-months (N = 551). The Teaching Task elicits maternal support for infant autonomy and exploration, therefore relevant behaviors that were coded included, for example, maternal scaffolding, praise, and social bids, and child attention to task, compliance, and toy exploration. Separately at each age, HMM was used to compute the probabilities of all possible latent dyadic states (i.e., all possible co-occurring mother-child behaviors) and to determine the number of dyadic states that resulted in best model fit. In other words, HMM quantified specific patterns of dyadic interaction that were most likely to occur at 9-, 18-, and 27-months. Then, separately at each age, HMM was used to compute a set of transition probabilities (i.e., the probability of dyads’ moving from one latent state to any other latent state). Thus, HMM provided a rich description of the content of dyadic interactions (i.e., the most likely types of co-occurring behaviors) and the process of dyadic interactions (most likely patterns of movement among dyadic states) at each age. The dyadic interaction patterns at each age were discussed in terms of similarities and differences at the different ages. HMM can be used in future research to examine individual differences in dyadic interaction patterns, explore genetic and environmental contributions to the development of dyadic interaction patterns, and predict child outcomes in relation to early dyadic interaction patterns.