TY - JOUR

T1 - Proposing network analysis for early life adversity

T2 - An application on life event data

AU - de Vries, Tjeerd Rudmer

AU - Arends, Iris

AU - Rod, Naja Hulvej

AU - Oldehinkel, Albertine J.

AU - Bultmann, Ute

PY - 2022

Y1 - 2022

N2 - Commonly used methods for modelling early life adversity (e.g., sum-scores, latent class or trajectory ap-proaches, single-adversity approaches, and factor-analytical approaches) have not been able to capture the complex nature of early life adversity. We propose network analysis as an alternative way of modelling early life adversity (ELA). Our aim was to construct a network of fourteen adverse events (AEs) that occurred before the age of 16 in the TRacking Adolescents Individual Lives Survey (TRAILS, N =1029). To show how network analysis can provide insight into why AEs are associated, we compared findings from the resulting network model to findings from tetrachoric correlation analyses. The resulting network of ELA comprised direct re-lationships between AEs and more complex, indirect relationships. A total of fifteen edges emerged in the network of AEs (out of 91 possible edges). The correlation coefficients suggested that many AEs were associated. The network model of ELA indicated, however, that several associations were attributable to interactions with other AEs. For example, the zero-order correlation between parental addiction and familial conflicts (0.24) could be explained by interactions with parental divorce. Our application of network analysis shows that using network analysis for modelling the ELA construct allows capturing the constructs' complex nature. Future studies should focus on gaining more insight into the most optimal model estimation and selection procedures, as well as sample size requirements. Network analysis provides researchers with a valuable tool that allows them as well as policy- makers and professionals to gain insight into potential mechanisms through which adversities are associated with each other, and conjunctively, with life course outcomes of interest

AB - Commonly used methods for modelling early life adversity (e.g., sum-scores, latent class or trajectory ap-proaches, single-adversity approaches, and factor-analytical approaches) have not been able to capture the complex nature of early life adversity. We propose network analysis as an alternative way of modelling early life adversity (ELA). Our aim was to construct a network of fourteen adverse events (AEs) that occurred before the age of 16 in the TRacking Adolescents Individual Lives Survey (TRAILS, N =1029). To show how network analysis can provide insight into why AEs are associated, we compared findings from the resulting network model to findings from tetrachoric correlation analyses. The resulting network of ELA comprised direct re-lationships between AEs and more complex, indirect relationships. A total of fifteen edges emerged in the network of AEs (out of 91 possible edges). The correlation coefficients suggested that many AEs were associated. The network model of ELA indicated, however, that several associations were attributable to interactions with other AEs. For example, the zero-order correlation between parental addiction and familial conflicts (0.24) could be explained by interactions with parental divorce. Our application of network analysis shows that using network analysis for modelling the ELA construct allows capturing the constructs' complex nature. Future studies should focus on gaining more insight into the most optimal model estimation and selection procedures, as well as sample size requirements. Network analysis provides researchers with a valuable tool that allows them as well as policy- makers and professionals to gain insight into potential mechanisms through which adversities are associated with each other, and conjunctively, with life course outcomes of interest

KW - Early life adversity&nbsp

KW - Adverse experiences&nbsp

KW - Network analysis&nbsp

KW - Negative life events&nbsp

KW - Cross-sectional&nbsp

KW - Childhood adversity

KW - INDIVIDUAL-LIVES SURVEY

KW - CHILDHOOD EXPERIENCES

KW - CUMULATIVE RISK

KW - COHORT PROFILE

KW - MORTALITY

KW - TRAUMA

KW - HEALTH

KW - POLICY

U2 - 10.1016/j.socscimed.2022.114784

DO - 10.1016/j.socscimed.2022.114784

M3 - Journal article

C2 - 35152049

VL - 296

JO - Social Science & Medicine

JF - Social Science & Medicine

SN - 0277-9536

M1 - 114784

ER -