Longitudinal Research with Latent Variables

Bivariate Latent-Change-Score Analysis of Peer Relations From Early Childhood to Adolescence: Leading or Lagging Indicators of Psychopathology

“…We then fit five theoretically specified BLCSMs (baseline, interpersonal-risk, interpersonal-scar, transactional, and developmental) to determine whether there was also a leading or lagging relationship. On the basis of prior studies using BLCSMs (Ferrer et al, 2007; Kievit et al, 2018; McArdle & Grimm, 2010), the self-feedback parameters (β) for each construct were fixed to be equal across assessment waves. This guards against potential overfitting while also aiding in model convergence, and we did not have a theoretical reason to believe that this parameter would vary across development.…”

“…Across each of these models, coupling coefficients were fixed to be equivalent across time points. Similar procedures have been conducted to test time-dependent associations between reading and antisocial behaviors (McArdle & Grimm, 2010).…”

“…However, when the temporal association between peer relations and psychopathology is allowed to differ across development, we hypothesize that the developmental model will best fit the data, with psychopathology predicting change in peer relationships in pre-adolescence, but peer relationships predicting increases in psychopathology in adolescence. We tested these models using bivariate latent change score analysis (e.g., Ferrer et al, 2007;Ferrer & McArdle, 2010;McArdle & Grimm, 2010;Usami, Hayes, & McArdle, 2015. Additionally, while it is likely that both psychopathology and peer relations will show relationships with other factors such as gender, socioeconomic status, adverse childhood events, race, intelligence quotient, or family history of psychopathology, we predict that the hypothesized models of the links between peer relationships and psychopathology will hold even when accounting for these variables.…”

“…Our results supported longitudinal relationships between social withdrawal and depression and anxiety symptoms, whereby growth in social withdrawal was positively associated with depression and anxiety severity over development. (Ferrer et al, 2007) and reading and antisocial behaviors (McArdle & Grimm, 2010). Compared to auto-regressive cross-lagged models, bivariate latent change score models take into account growth (i.e.…”

Understanding Individual-level Change Through the Basis Functions of a Latent Curve Model

“…More generally, the use of latent characteristics to account for individual-level responses is common and lies at the heart of another major framework for longitudinal data analysis, namely, mixed-effects model. Several seminal books and articles concerning the more general approach are available, with variation in technical coverage and applications (Bollen and Curran 2006;Davidian and Giltinan 1995;Diggle, Liang, and Zeger 1994;Goldstein 2003;Hox 2002;Kreft and de Leeuw 1998;Laird and Ware 1982;Longford 1993;McArdle 1988;McCulloch and Searle 2001;Tisak 1984, 1990;Raudenbush and Bryk 2002;Singer and Willett 2003;Skrondal and Rabe-Hesketh 2004;Snijders and Bosker 1999;Verbeke and Molenberghs 2000). Beyond the quantitative literature, several articles have been published in substantive domains that have introduced latent curve models, emphasizing the potential benefits of the technique in a particular area of study (e.g., T. E. Duncan and Duncan 2004;Hedeker, Flay, and Petraitis 1996;McArdle 1986;McArdle and Epstein 1987).…”