Using Vector Autoregression Modeling to Reveal Bidirectional Relationships in Gender/Sex-Related Interactions in Mother–Infant Dyads

Abstract

Vector autoregression (VAR) modeling allows probing bidirectional relationships in gender/sex development and may support hypothesis testing following multi-modal data collection. We show VAR in three lights: supporting a hypothesis, rejecting a hypothesis, and opening up new questions. To illustrate these capacities of VAR, we reanalyzed longitudinal data that recorded dyadic mother-infant interactions for 15 boys and 15 girls aged 3 to 11 months of age. We examined monthly counts of 15 infant behaviors and 13 maternal behaviors (Seifer et al., 1994). VAR models demonstrated that infant crawling predicted a subsequently close feedback loop from mothers of boys but a subsequently open-ended, branched response from mothers of girls. A different finding showed that boys' standing independently predicted significant later increases of four maternal behaviors: rocking/jiggling, lifting, affectionate touching, and stimulation of infant gross-motor activity. In contrast, crawling by girls led mothers to later decrease the same maternal behaviors. Thus, VAR might allow us to identify how mothers respond differently during daily interactions depending on infant gender/sex. The present work intends to mainly showcase the VAR method in the specific context of the empirical study of gender/sex development.

Personalized risk predictor for acute cellular rejection in lung transplant using soluble CD31

“…1 ). Furthermore, sCD31 as a time series could be fully explained in terms of PaO 2 /FiO 2 as an absolute value ratio and respiratory SOFA (Sequential Organ Failure Assessment) using a multivariable function, which followed a vector autoregression model (VAR) 59 , 61 . Then, the predictors were only the lags of sCD31 and PaO 2 /FiO 2 series (see “ Methods ”).…”

A Dynamic Systems Framework for Gender/Sex Development: From Sensory Input in Infancy to Subjective Certainty in Toddlerhood

“…In a different approach, Eason and colleagues explored relationships in mother/infant interactions using vector autoregression analysis. This method promises an approach for testing gender/sex salience in multimodal, bidirectional effects in a dense, multimodal data set (Eason et al, 2020). Finally, both biologists and cognitive scientists are exploring the use of Bayesian statistics in the longitudinal study of development.…”

A tool to simulate and visualize dyadic interaction dynamics.

“…The VAR(1) model is a stationary model where multiple variables are regressed on themselves and the other variables at previous measurement occasions (Hamilton, 1994). It has often been applied in psychology with multivariate data (Haslbeck & Ryan, 2021), and also dyadic data (e.g., Bar-Kalifa & Atzil-Slonim, 2020; Bringmann et al, 2016; Eason et al, 2020; Kroemeke & Sobczyk-Kruszelnicka, 2019; Lunkenheimer et al, 2015; Okazaki et al, 2015). The term first-order (in the abbreviation (1)) refers to the fact that the model includes only lag 1 regressions.…”

“…We will discuss several related models in detail, focusing on vector autoregressive (VAR) models and extensions. The VAR model is a basic, fundamental, model that has been applied often to study the dynamics of various psychological processes (Bar-Kalifa & Atzil-Slonim, 2020; Bringmann et al, 2016; Eason et al, 2020; Lee et al, 2017; van der Krieke et al, 2017; Vrijen et al, 2018). However, given that the VAR model is relatively simple, the range of data patterns it can generate is limited and will not be able to mimic all empirical data well.…”