Early Parenting Precursors of Child Developmental Outcomes: What Counts Beyond Parental Sensitivity?

SYNOPSIS

Objective. Parental sensitivity is a precursor of child developmental outcomes, but other parental behaviors such as warmth, stimulation of cognitive development, intrusiveness, and hostility are also associated with those outcomes. The aim of the current study was to untangle unique linkages among parental sensitivity, other parental behaviors, and outcome domains deemed relevant for child welfare in an ethnically homogenous sample. Design. Data from European American dyads of the NICHD Study of Early Child Care and Youth Development (N = 1,097) across 13 waves from 6 months to 15 years of age were split into two datasets for cross validation. Twelve of 284 outcome measures were selected a priori as domain indicators. Only one measure was eligible for the domain Coping, therefore this measure and domain were omitted. Results. Confirmatory factor analysis revealed acceptable model fit for robust outcome domains of Emotional Regulation Problems, Self-Concept, and Problem Behavior assessed during childhood and adolescence. Using structural equation modeling, parental sensitivity statistically predicted Emotional Regulation Problems during primary school and Problem Behavior throughout primary school and adolescence. However, prediction in adolescence was no longer statistically significant after adding demographic confounders. Stimulation of cognitive development predicted Problem Behavior during primary school over and above parental sensitivity, yet, this effect was no longer statistically significant after adding demographic confounders. Parental intrusiveness and detachment, positive and negative regard, and flatness of affect showed no added predictive value. Conclusions. These findings may guide parsimonious assessment of the quality of the caregiving environment of young children.

INTRODUCTION

In the study of parenting, sensitivity has emerged as a consistent predictor of developmental outcomes of children; however, other parental behaviors, like warmth, stimulation of cognitive development, intrusiveness, and hostility, have been found predictive as well (e.g., Cooke et al., 2022; Davidov & Grusec, 2006; Gulseven et al., 2021; Pinquart, 2017). Few studies have systematically integrated these findings to test the unique predictive value of parental behaviors beyond parental sensitivity. This knowledge gap hinders the development and justification of efficient assessment strategies for research and daily practice (Goldberg et al., 1999; Von Suchodoletz et al., 2011). Without a parsimonious set of parenting behaviors to focus attention on, it may be difficult to close the gap between research and practice in supporting families and informed decision-making for child welfare and child protection (Cyr & Alink, 2017; Forslund et al., 2022; Van IJzendoorn et al., 2018).

The aim of the current study was to untangle the unique predictive value of parental sensitivity and added predictive value of other parental behaviors over and above parental sensitivity for child developmental outcomes based on data of 13 waves of the National Institute of Child Health and Human Development Study of Early Child Care and Youth Development (NICHD SECCYD). To heighten relevance for child welfare practice, we aimed to optimize prediction across broad domains of child development that are known to be associated with parenting and well-being of children: emotional regulation problems (Morris et al., 2017), coping (Skinner & Zimmer-Gembeck, 2016), self-concept (Pinquart & Gerke, 2019), and problem behavior (Pinquart, 2017).

METHOD

Participants

One thousand ninety-seven mothers with European American children were followed in the longitudinal study NICHD SECCYD (1991–2007) from 1 month to 15 years of age (for further information see NICHD Early Child Care Research Network, 2001). Demographic characteristics of the full and—for analyses in this study—split samples are provided in Supplementary Table S1.

Procedures

Data in NICHD SECCYD were collected in four phases (see https://www.icpsr.umich.edu/web/ICPSR/series/00233/studies), with 1,364 mother-child dyads followed in Phase I, 1,220 in Phase II, 1,100 in Phase III, and 1,056 in Phase IV. Mother-child interaction data collected with videotaped semi-structured 15-min tasks at the 6-, 15-, 24-, and 36-month measurements of Phase I were used as predictors (see Dallaire & Weinraub, 2005, for information on the collection of these data). Child developmental outcome data collected through questionnaires filled out by the child, mother, father, teacher, after school caregiver or care provider, and other adults at 54 months, kindergarten, and Grade 1 of Phase II, Grades 2, 3, 4, 5, and 6 of Phase III, and 15 years of age Phase IV were used as outcomes.

Measures

Parental Behavior

NICHD SECCYD rating scales (Owen, 1992) were applied to observed mother–child interaction. Coders used age-appropriate coding manuals. The scales Sensitivity to Distress and Sensitivity to Non-Distress/Supportive Presence were aggregated into a Parental Sensitivity scale. The Parental Intrusiveness/Respect for Autonomy, Parental Detachment, Stimulation of Cognitive Development, Positive Regard, Negative Regard/Hostility, and Flatness of Affect scales were also used. Table S2 describes operationalization of the scales, measurements, and coder reliability. For further information on this measure see NICHD Early Child Care Research Network (1999).

Child Developmental Outcomes

For four outcome domains—selected by expert psychologists working in child and family welfare—12 measures were included through a predefined protocol. Four measures were selected as indicators of the Emotional Regulation Problems domain, three as indicators of the Self-Concept domain, and six as indicators of the Problem Behavior domain. Two measures were included for two outcome domains by selecting different scales of the same measure. Because only one measure was eligible—relevant for the domain, standardized, performing favorably on reliability (conform criteria in Cicchetti, 1994) and validity, and having norm scores available—for the domain Coping, this measure and domain were dropped from the study. Table S3 summarizes the measures used for the three domains left, including scales, informant, measurement occasions, and internal consistency. A full description of the selection of outcome measures with a flow diagram is provided in Supplemental Materials and Figure S1.

Control Variables

With previous effects found for parental behavior and child gender (e.g., Zvara et al., 2018), education (e.g., Tamis LeMonda et al., 2009), and income (e.g., Cooke et al., 2022), the following demographics were included as covariates: child assigned gender (0 = boy, 1 = girl), years of education of mother, and income-to-needs ratio (total family income divided by the poverty threshold) at child age 1 month.

Analytic Plan

Data files were prepared in IBM SPSS Statistics version 26. Predictors were mean aggregated across time points into one score per scale. Outcome variables were mean aggregated across time points and informants for 54 months until Grade 6 to reflect primary school age and at age 15 to reflect adolescence. Analyses were conducted in R version 3.6.2 and RStudio version 1.2.5033 with Lavaan package version 0.6–7. The dataset was randomly divided into two samples for performing primary analyses in Sample 1 and cross validation of the results in Sample 2. Domains of child developmental outcomes were represented by latent variables using confirmatory factor analyses (CFA) in Sample 1. Model fit was considered acceptable when Comparative Fit Index and Tucker-Lewis Index values exceeded .90 and Root Mean Square Error of Approximation values were smaller than .08 for at least two out of these three model fit indices (Hu & Bentler, 1998; Kline, 2015). If needed, models were adjusted according to modification indices and content.

Predictive associations were assessed using structural equation modeling (SEM). Figure 1 portrays the model that was tested. First, SEM models with only Parental Sensitivity as a predictor were tested in Sample 1 and cross validated in Sample 2. If the explained variance by Parental Sensitivity was statistically significant for an outcome domain, other parental behavior scales were added one by one to test whether the predictor added statistically significant variance in the outcome in Sample 1 and cross validated in Sample 2. If the predictor explained additional variance in the outcome, the procedure was repeated.

Figure 1. Model of predictive associations of parental behaviors on child developmental outcomes.

To assess robustness, demographics were added to the SEM models. First, we tested which demographics showed statistically significant associations with the outcome domain per age period in Sample 1. Results were cross validated in Sample 2. Next, we added demographics associated with the outcome domain per age period in the SEM models with statistical effects in Sample 1 and cross validated in Sample 2.

An alpha level of .05 was used for statistical significance for all SEM models. Effect sizes were interpreted with a field-specific benchmark (Schuengel et al., 2021).

RESULTS

Preliminary analyses including CFA are reported in Supplemental Materials. Means and standard deviations of parental behaviors are reported in Table S4; means, standard deviations and correlations between child developmental outcomes in primary school in Table S5 and in adolescence in Table S6; correlations between parental behaviors in Table S7; correlations between parental behaviors and child developmental outcomes in primary school in Table S8 and in adolescence in Table S9; and model fit statistics for CFA in Table S10. Results for SEMs with predictive associations of parental behaviors in the primary school age are shown in Table S11 and for adolescence in Table S12.

Predictive Value of Parental Behavior for Child Developmental Outcomes

Figure 2a shows predictive associations of Parental Sensitivity for Emotional Regulation Problems in primary school and Figure 2b for Problem Behavior in primary school. Parental Sensitivity shows medium-sized effects for Emotional Regulation Problems and medium to large-sized effects for Problem Behavior. Figure 2c shows the predictive associations of Parental Sensitivity for Problem Behavior in adolescence. Parental Sensitivity showed small to medium-sized effects for Problem Behavior. Further, Parental Sensitivity showed statistical effects for Emotional Regulation Problems in adolescence and for Self-Concept in primary school, yet in Sample 2 no statistical effects were found. Alternatively, Parental Sensitivity showed no statistical effect for Self-Concept in adolescence in Sample 1, yet in Sample 2 a statistical effect was found. Thus, Parental Sensitivity in early childhood predicted Emotional Regulation Problems throughout the primary school age period, and Problem Behavior in both primary school and adolescence. Self-concept in primary school and adolescence was not consistently predicted by Parental Sensitivity.

Figure 2. Predictive associations of Parental Sensitivity in early childhood for (a) Emotional Regulation Problems, (b) Problem Behavior in primary school, and (c) Problem Behavior in adolescence.

Standardized regression estimates and variances before the slash represent Sample 1 and after the slash Sample 2.

*p < .05. **p < .01. ***p < .001.

For primary school, there were no statistical effects for other parental behaviors over and above Parental Sensitivity for Emotional Regulation Problems. Stimulation of Cognitive Development showed small to medium-sized effects for Problem Behavior over and above small to medium-sized effects for Parental Sensitivity in primary school (Figure 3). For adolescence, there were no statistical effects for other parental behaviors over and above Parental Sensitivity for Problem Behavior.

Figure 3. Predictive associations of Parental Sensitivity and Stimulation of Cognitive Development in early childhood for Problem Behavior in primary school.

Standardized regression estimates and variances before the slash represent Sample 1 and after the slash Sample 2.

*p < .05. **p < .01. ***p < .001.

Influence of Demographics

To test the robustness of the SEMs with statistically significant effects, we added demographics that were statistically associated with Emotional Regulation Problems and Problem Behavior in Sample 1 and cross validated in Sample 2. Results for primary school are shown in Table S13 and for adolescence in Table S14. Predictive associations of Parental Sensitivity for Emotional Regulation Problems and Problem Behavior in primary school remained statistically significant in Samples 1 and 2 after controlling, respectively, for education of mother at child age 1 month and for child gender, education of mother at 1 month, and income-to-needs ratio at 1 month. Other previously statistically significant SEMs proved not to be robust after adding demographics.

DISCUSSION

A parsimonious set of parental behaviors with broad predictive value across primary school to adolescence contained parental sensitivity in mothers with European American children. Only stimulation of cognitive development predicted problem behavior in primary school over and above parental sensitivity. In contrast to parental sensitivity in primary school, the predictive values of stimulation of cognitive development and of parental sensitivity in adolescence were not robust controlling for demographic characteristics. Parental intrusiveness and detachment, positive and negative regard, and flatness of affect individually may show associations with outcomes, but including them did not contribute to the predictive value of parental sensitivity. Further, in accordance with previous findings on effects of parental sensitivity on child developmental outcomes (Cooke et al., 2022), parental sensitivity in early childhood predicted outcomes across the domains of emotional regulation problems and problem behavior during primary school. Predictive effects for self-concept did not cross validate.

The effect sizes found were small to large relative to field-specific benchmarks (Schuengel et al., 2021). Changes that happen within families and their context and measurement unreliability need to be considered. Our findings indicate that parental sensitivity and to a lesser extent stimulation of cognitive development are meaningful, but links with outcomes are probabilistic, requiring careful explanation when findings are disseminated with practitioners.

Limitations

First, the aggregation of a wide range of outcome measures across different ages means that cascading effects may not be detected. Second, the coping domain lacked sufficient measures in NICHD SECCYD and was excluded. The self-concept domain did not replicate across the two samples. Further research is needed to untangle the predictive values of different parental behaviors on these domains of child developmental outcomes. Third, sampling procedures in NICHD SECCYD precluded examination of effects within different ethnic groups (NICHD Early Child Care Research Network, 2001). The current study, therefore, only included European American families. Further research with other ethnic groups is needed. Fourth, because NICHD SECCYD does not include an at-risk sample, replications of the current study in at-risk populations and populations under the attention of child welfare are needed to heighten the relevance of the findings for daily practice. Fifth, shared coder effects reduce unique variance across parenting behaviors.

Conclusions

Beyond parental sensitivity, no other parenting behaviors were robustly associated with child developmental outcomes. Stimulation of cognitive development showed unique predictive value for problem behavior during primary school, but was not robust across demographic confounders; parental intrusiveness and detachment, positive and negative regard, and flatness of affect did not show any unique effects. Early parental sensitivity predicted emotional regulation problems during primary school and problem behaviors throughout primary school and adolescence—although in adolescence not robust. Predictions of self-concept were equivocal.

IMPLICATIONS FOR PRACTICE AND THEORY

With limited resources and high constraints in research and daily practice, a focus on a parsimonious set of parenting behaviors, including parental sensitivity and perhaps stimulation of cognitive development, could be a powerful strategy to make observational assessment of parental behavior feasible, putting transparent and shared decision-making, intervention plans, and evaluation on firmer footing.

AFFILIATIONS AND ADDRESSES

Mirte L. Forrer, Vrije Universiteit Amsterdam, Van der Boechorststraat 7, 1081 BT Amsterdam, The Netherlands. Email: [email protected]. Marleen H. M. de Moor is at Erasmus University Rotterdam, The Netherlands, and Carlo Schuengel and Mirjam Oosterman are at Vrije Universiteit Amsterdam, The Netherlands.

ARTICLE INFORMATION

Conflict of Interest Disclosures

Each author signed a form for disclosure of potential conflicts of interest. No authors reported any financial or other conflicts of interest in relation to the work described.

Ethical Statement

The authors affirm having followed professional ethical guidelines in preparing this work. These guidelines include obtaining informed consent from human participants, maintaining ethical treatment and respect for the rights of human or animal participants, and ensuring the privacy of participants and their data, such as ensuring that individual participants cannot be identified in reported results or from publicly available original or archival data.

Role of the Funders/Sponsors

None of the funders or sponsors of this research had any role in the design and conduct of the study; collection, management, analysis, and interpretation of data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication. The ideas and opinions expressed in this manuscript are those of the authors alone, and endorsement by the authors’ institutions or the funding agency is not intended and should not be inferred.

Data Availability Statement

The data that support the findings in this study are available upon reasonable request through https://www.icpsr.umich.edu/web/ICPSR/series/00233/studies.

Supplementary Material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15295192.2024.2310853

Additional information

Funding

This work was supported by a grant from the Netwerk Effectief Jeugdstelsel Amsterdam (NEJA) and a grant from the foundation Pro Juventute Amsterdam.