Modeling Poverty in the Development of Maltreated Infants Using Latent Curve Models

Poverty has a negative influence on infants' development and is often included in analyses of child development. There are multiple conceptualizations and operationalizations of poverty in quantitative analysis and these may influence. Common operationalizations of poverty include a dichotomous measure using US Census Bureau thresholds based on household size and income, poverty as a continuous income-to-needs ratio, and poverty as low socioeconomic status. Three questions were asked; (a) Are maltreated infants developmental trajectories linear or non-linear? (b) How do differing operationalizations of poverty influence model fit and estimates of independent variables? (c) Do the previous models hold true in smaller, somewhat more homogenous subgroups? Using the infants participating in the National Survey of Child and Adolescent Well-Being (NSCAW), these longitudinal analyses used Latent Curve Models to model infant development and multiple operationalizations of poverty. The sample consisted of 1,196 infants enrolled in NSCAW and drew data from all five waves of data collection; baseline and 12, 18, 36, and 66 month follow-up interviews. Development domains modeled were adaptive behavior development, cognitive development, and language development and indicators were scaled to a common metric to facilitate comparison. Poverty was defined by data gathered at baseline. Multiple Imputation by Markov Chain Monte Carlo was used to estimate missing data as well as change the time interval to chronological years. Latent Curve Models were fitted to 10 augmented data sets consisting of developmental data, operationalizations of poverty, and case characteristic control variables. Models were estimated in Mplus using the 10 imputed data sets and NSCAW weights and complex survey sample design information. Results indicated that between years of life two and six, most infants experienced non-linear development in the domains of adaptive behavior and cognitive development while experiencing linear development in the domain of communication development. The income-to-needs ratio operationalization of poverty was the preferred model in all three domains based on the criteria of substantively meaningful parameter estimates, acceptable fit indices, and absence of improper solutions. While the more simple dichotomous operationalization of poverty had similar results, the income-to-needs operationalization clearly demonstrated that ratios near, but above, the poverty threshold used in the dichotomous operationalization also produced a negative effect on developmental trajectories. Indicators based on the social capital conceptualization of socioeconomic status did not produce effects on the developmental trajectories. Findings about the shape of the developmental trajectories held in smaller groups of maltreated infants defined by gender, minority status, and case characteristics Predictors of the developmental trajectories, particularly foster care placement and minority status, did not hold in all sub-group populations. There are several key implications. First, the effects exerted by poverty on maltreated infants' development vary by how poverty is conceptualized and operationalized. Second, maltreated infants are not a homogenous group, therefore effects observed in large heterogeneous samples may be misleading. Substantively, poverty was found to have a consistent, negative effect on development. No support was found for the social capital conceptualization of socioeconomic status.