Capturing environmental dimensions of adversity and resources in the context of poverty: A moderated nonlinear factor model

• In Event: Poster Session 03
  In Poster Session: PS 03 Topic - Family Context & Processes

Abstract

Traditional measures of poverty—such as indices of cumulative risk and metrics of socioeconomic status—obscures heterogeneity in the mechanisms through which poverty influences individual differences in child developmental outcomes (Duncan & Magnuson, 2003). Similarly, common indicators of poverty may not have the same meaning across time or across racial-ethnic groups, leading to empirical imprecision and conclusions that do not adequately characterize the experience of any sociocultural group (Borsboom, 2006). Reducing such racial, social, and developmental bias in our measures is a major goal of developmental science. Towards this end, rigorous methods testing and adjusting for possible bias (i.e. non-invariance) should be considered alongside substantive developmental questions among diverse populations. In the current study, we draw from contemporary theoretical models of adversity (Hostinar & Miller, 2020; McLaughlin & Sheridan, 2016) and leverage recent advances in latent variable modeling to develop more nuanced representations of the constraints and supports experienced by families struggling with economic adversity.

Data comes from the Family Life Project (N=1,292)—a longitudinal sample of children living in predominately low-income rural U.S. communities (Vernon-Feagans & Cox, 2013). Home visits were conducted when children were 6 months, and 1, 2, 3, 5, 7, and 12 years of age. 49% children were identified by their parents as female, 42% as African American, and 58% reside in NC (42% in PA). The Economic Strain (ES; Conger et al., 1994), HOME (Caldwell & Bradley, 1984), and Conflict Tactics Scale (CTS; Straus & Gelles, 1990) questionnaires were chosen to represent environmental measures of material deprivation, sociocognitive resources, and psychosocial threat, respectively. After establishing configural invariance using confirmatory factor analysis, we leveraged moderated nonlinear factor analysis (MNLFA; Bauer, 2017) to establish group- and longitudinally-invariant factor scores as a function of age, racial and gender category, and research site. To investigate the utility of using MNLFA compared to raw mean scores, we used multi-level growth curve analysis to examine developmental change and group differences; criterion validity was examined via relations with family income and cognitive and behavioral child outcomes at age 12.

Broadly, results demonstrate the utility of applying MNLFA to improve substantive conclusions about the effects of poverty-related environmental exposures. The three derived factors were largely invariant across site, racial group, and development, with some non-invariance as a function of age that was adjusted for. Scores were also minimally to moderately related to each other and with income; in other words, these forms of poverty-related adversity do not always occur together (Table 1). Average racial group differences were found across all three constructs and presumably reflect larger systemic inequalities and longstanding racial income disparities. We also found differences in growth trajectories and individual variability between the scoring approaches (Figure 1), which resulted in substantively different conclusions that have far reaching implications for policy and theory.

Findings highlight an innovative measurement approach that revealed three distinct environmental exposures in the context of poverty—providing tangible targets for the creation of individualized interventions as our nation works to ameliorate poverty and structural racial and economic inequalities.

Authors

• Meriah L. DeJoseph, University of Minnesota - Twin Cities
  Presenting Author

• Robin Sifre, University of Minnesota
  Non-Presenting Author

• Cybele Raver, New York University
  Non-Presenting Author

• Clancy Blair, New York University
  Non-Presenting Author

• Daniel Berry, University of Minnesota - Twin Cities
  Non-Presenting Author