Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2020 Dec 1.
Published in final edited form as: Dev Psychol. 2019 Sep 19;55(12):2522–2533. doi: 10.1037/dev0000824

Additive and Synergistic Relations of Early Mother- and Caregiver-Child Interactions for Predicting Later Achievement

Robert J Duncan a, Sara A Schmitt a, Deborah Lowe Vandell b
PMCID: PMC6861641  NIHMSID: NIHMS1047342  PMID: 31535893

Abstract

This study examines associations between stimulating-responsive social interactions with mothers and nonparental childcare providers during the first three years of life and children’s vocabulary and mathematics skills through age 15 (N = 1364). Additive relations were found in which more stimulating-responsive interactions with mothers and with caregivers were linked to higher mathematics achievement in childhood and adolescence. More stimulating-responsive early interactions with mothers were also associated with larger child vocabularies through age 15. Synergistic relations, consistent with the dual risk hypothesis, also were found. Children whose early interactions with both mothers and caregivers were low in stimulation and responsivity had substantially lower mathematics skills. Implications for early childhood interventions and policies are discussed.

Keywords: mother-child interaction, caregiver-child interaction, early childcare, vocabulary, mathematics

Developmental theory and empirical research have emphasized the importance of early life experiences for children’s later development (Fraley, Roisman, & Haltigan, 2013; Golinkoff & Hirsh-Pasek, 2000; Phillips & Shonkoff, 2000; Raby, Roisman, Fraley, & Simpson, 2015; Shonkoff et al., 2012). The current study extends this research by examining relations between stimulating-responsive social interactions with mothers and nonparental childcare providers (i.e., caregivers) during the first three years of life and children’s vocabulary and mathematics achievement from age 4 ½ through 15 years. Specifically, our primary focus is to examine the extent to which interactions with mothers and caregivers are linked to children’s later skills from age 4 ½ to 15 years in ways that appear to be additive (i.e., no moderation) or synergtistic (i.e., moderation). We also examine to what extent the effects associated with the first three years persist or fadeout from age 4 ½ to 15 years and whether they are attenuated when stimulating-responsive interactions at age 4 ½ years are added to the model.

Are There Additive or Moderated Relations between Early Mother-Child and Caregiver-Child Interactions with Later Achievement?

A robust literature points to both short- and long-term associations between stimulating-responsive mother-child interactions during the first three years and children’s vocabulary and mathematics skills (Ayoub, Vallotton, & Mastergeorge, 2011; Burchinal, Vernon-Feagans, Cox, & Key Family Life Project Investigators, 2008; NICHD Early Child Care Research Network [ECCRN], 2000a; Rafferty, Griffin, & Lodise, 2011). For instance, Fraley et al. (2013) reported associations between maternal sensitivity during the first three years and children’s social competence and academic skills through age 15, associations that were not explained by maternal sensitivity in middle childhood or adolescence. Early caregiver-child interactions have also been linked to early language and cognitive abilities (Burchinal, Roberts, Nabors, and Bryant 1996; Li, Farkas, Duncan, Burchinal, & Vandell, 2013; NICHD ECCRN, 2000a). In research utilizing the SECCYD dataset, the behavioral counts of caregivers’ language and cognitive activities with specific children predicted their language comprehension, mathematics skills, and expressive vocabulary at 4 ½ years (NICHD ECCRN, 2003).

Relatively few studies have considered how early mother-child and caregiver-child interactions, together, are linked to later child developmental outcomes. However, the available research underscores the value of considering joint effects of children’s experiences with mothers and caregivers. There is evidence of additive or cumulative effects in which both mother-child and caregiver-child interactions during early childhood are significant positive predictors of children’s cognitive and language outcomes at age 3 years (NICHD ECCRN, 2000a), 4 ½ years (NICHD ECCRN, 2003), and first grade (Hirsh-Pasek & Burchinal, 2006). If the relations are additive (i.e., no moderation), then high quality interactions with caregivers would be associated with higher cognitive and language outcomes regardless of the quality of interactions with mothers (and vice versa). In other research that considers both mother-child and caregiver-child interactions, there is evidence of moderated relations in which effects associated with early caregiver-child interactions on early childhood outcomes (Vernon-Feagans, Bratsch-Hines, & The Family Life Project Key Investigators, 2013) and adolescent outcomes (Burchinal, Vandell, & Belsky, 2014) are moderated by mother-child interactions. If relations are synergistic (i.e., moderation), then relations between high quality interactions with caregivers and children’s cognitive and language outcomes would be conditional on the quality of the interactions with mothers (and vice versa).

Two moderated models derived from the work of Rutter (1987) are considered in the current study. The compensatory model hypothesizes that stimulating-responsive interactions with one adult can compensate for less stimulating-responsive interactions with another adult. Thus, positive caregiving from childcare providers may off-set poor quality interactions with mothers, or stimulating-responsive interactions with mothers may off-set or compensate for lower quality interactions with caregivers. The dual risk hypothesis posits that receiving little stimulation and responsivity from both mother and caregiver may be particularly detrimental for young children because of synergistic effects.

There is some evidence consistent with both the compensatory and the dual risk hypotheses. For example, using data from the Family Life Project, Vernon-Feagans and colleagues (2013) found higher quality caregiver-child language interactions buffered the effects of lower quality mother-child language interactions on children’s language outcomes. These relations were found at both the 36- and 48-month waves of data. Similarly, Miller and colleagues (2014) found evidence of compensatory effects of Head Start participation for children who received lower levels of academic stimulation in their homes. Also, consistent with the compensatory hypothesis were findings reported by Dearing, McCartney, and Taylor (2009) in which low-income children who attended higher quality child care demonstrated higher academic achievement in middle childhood than other low-income students. Evidence consistent with the dual risk hypothesis was reported by Nelson and colleagues (2007) who found experiences of extreme deprivation were exacerbated when children continued to be exposed to depriving living conditions but could be partially ameliorated if living conditions improved. The current study extends this prior research by examining the potential additive and synergistic effects of mother-child and caregiver-child interactions during the first three years and children’s vocabulary and mathematics achievement through age 15 years.

Developmental Timing and the Enduring Associations of Early Experiences

Although a central and enduring theme in developmental science, the research examining effects associated with early experience is limited in several respects. First, there is lack of consistency regarding the period of time represented by ‘early experience.’ Whereas some investigators have considered the cumulative effects associated with the first five years of life (Campbell et al., 2001; Vandell et al, 2010), others have emphasized the preschool period with a focus on three and four year olds (Diamond, Barnett, Thomas, & Munro, 2007; Schweinhart, 1993). Still others have emphasized the importance of children’s experiences in the first three years (Fraley et al., 2013; Hirsh-Pasek et al., 2015; Longo et al., 2017; Lupien, McEwen, Gunnar, & Heim, 2009; Raby et al., 2015). In the current paper, we focus on children’s interactions with their mothers and caregivers in the first three years with follow-up analyses that include interactions in the later preschool period.

The focus on the first three years was motivated, in part, by evidence from neuroscience that brain development may be particularly susceptible to serve-and-return exchanges that occur when infants and toddlers interact with adults who respond to their vocalizations, cries, smiles, and interests (Fox, Levitt, & Nelson, 2010; Shonkoff, 2010). In addition, exposure to language-rich environments during the first three years has been identified as critical for language development, and differences in language input and conversational turn-taking have been related to later differences in children’s vocabulary and reading skills (Gilkerson et al., 2018; Hirsh-Pasek et al., 2015), as well as children’s brain activation in the Broca’s area (Romeo et al., 2018). A final motivation for the current study is a closer examination of persistence versus fadeout of effects associated with early experiences. Although there are influential reports of enduring impacts of stimulating-responsive experiences during early childhood into adulthood for low-income children (Heckman, Pinto, & Savelyev, 2013; Schweinhart, 1993), other studies have found early childhood effects fadeout shortly after children move on to formal schooling (Bailey, Duncan, Odgers, & Yu, 2017). Identifying conditions under which early experiences are carried forward or fadeout are needed to advance both developmental theory and inform practice.

Current Study

To summarize, the current study extends prior research in several ways. First, we examine if early stimulating and responsive interactions with mothers and caregivers have additive or synergistic relations with children’s vocabulary and mathematics skills from age 4 ½ to 15 years. Second, we ask if early stimulating-responsive interactions are associated with children’s mathematics skills and vocabulary in substantively different ways at five time points from early childhood to adolescence (i.e., age 4 ½ years, grades 1, 3, and 5, age 15). Specifically, we examine if relations between early interactions and later vocabulary or mathematics competencies are persistent or fade over time, or are apparent at some ages but not others. Third, we consider if associations for the first three years and later outcomes are attenuated after including mother-child and caregiver-child interactions at age 4 ½ years. Because of the study’s correlational design, we include a number of covariates that may account for the associations between early life experiences and later vocabulary and mathematics skills, including child gender and ethnicity, maternal vocabulary, maternal education, family income, household structure, maternal depression, parenting stress, separation anxiety, and maternal health.

Method

Participants

The current study uses data from the SECCYD, an archived dataset available at https://www.icpsr.umich.edu/icpsrweb/ICPSR/series/00233. Participants were recruited during hospital visits conducted with mothers shortly after the birth of a child in 1991 at 10 locations in the United States (Little Rock, AR; Irvine, CA; Lawrence, KS; Boston, MA; Hickory, NC; Philadelphia, PA; Pittsburgh, PA; Charlottesville, VA; Seattle, WA; Madison, WI). During selected 24-hr intervals, all women giving birth (N = 8,986) were screened for eligibility. Of those families, 3,142 were excluded due to a priori criteria such as not English speaking or plans to move within the next 3 years. At a follow-up telephone interview at 2 weeks, 1,353 could not be contacted or refused to participate. Families were randomly selected among the remaining pool of eligible participants. A total of 1,364 families were recruited, completed a home interview at 1 month, and became the study participants. Overall, this constituted a 52% response rate from the original approach to families in the hospital to successful recruitment in the study.

At recruitment, 26% of the mothers had no more than a high school education; 21% had incomes no greater than 200% of the poverty level; and 22% were of minority race or ethnicity (i.e., not non-Hispanic White). See Table 1 for sample descriptive statistics. For more details about the sample and sample recruitment, see NICHD ECCRN (2005a). The current study received IRB approval at University of California, Irvine: Study of Early Child Care and Youth Development, HS# 2006-5347.

Table 1.

Child, Family, and Childcare Covariates Included in the Analyses

N Mean SD Range
Percent Male 1,364 51.69%
Percent White 1,364 80.43%
Percent Black 1,364 12.90%
Percent Hisp 1,364 4.69%
Percent Other 1,364 1.98%
Maternal Educationa 1,363 14.23 2.51 7 – 21
Maternal agea 1,364 28.11 5.63 18 – 46
Parenting stressb 1,275 50.23 9.90 26 – 83
Maternal PPVTc 1,167 99.01 18.35 40 – 159
Separation anxietyd 1,300 65.42 12.72 29 – 104.33
Maternal Depressione 1,303 9.26 6.93 0 – 43
Income-to-needse 1,302 3.62 2.87 0.14 – 22.47
Father in the homee 1,305 0.82 0.35 0 – 1
Maternal Healthe 1,305 3.17 0.56 1 – 4
CC hours 6 864 34.10 15.22 1 – 99
CC hours 15 909 34.69 16.02 1 – 100
CC hours 24 922 35.14 15.47 0 – 100
CC hours 36 996 34.00 16.80 1 – 148
GP CCe 1,074 0.12 0.28 0 – 1
Home CCe 1,074 0.43 0.43 0 – 1
Center CCe 1,074 0.24 0.36 0 – 1
Open in a new tab

Note. CC is childcare. GP is grandparent.

a

From the 1 month assessment wave

b

From the 6 month assessment wave

c

From the 36 month assessment wave

d

Average of 6, 15, and 24 month assessment wave

e

Average of 6, 15, 24, and 36 month assessment wave

Procedures

The current study utilizes measures from the SECCYD that were collected when children were 1 month to 15 years. We first describe our two primary independent variables of interest: stimulating-responsive mother-child interactions and stimulating-responsive caregiver-child interactions. We then describe our two primary dependent variables of interest: vocabulary and mathematics achievement. Third, we present the covariates included in the analyses.

Mother-child interactions.

Observational assessments of mother-child interactions were conducted when the study children were 6, 15, 24, and 36 months old. At age 6 and 15 months, the videotaped interactions occurred in the home. At age 24 and 36 months, the videotaped interactions occurred in a lab setting (for more detailed descriptions of these procedures, see NICHD ECCRN, 1999, 2000b). At each age, the videotaped interactions were 15 minutes in duration. At age 6 months, the interaction was divided into two segments: 7 – 8 minutes where the mother chose toys to use while interacting with the child, and 7 – 8 minutes where the mother was given a standard set of toys (e.g., rattle with faces). At the age 15, 24, and 36 month observations, mothers were asked to provide their child with toys from 3 bags in a set order. At age 15 and 24 months, the toys were a storybook (differed at 15 and 24 months) in the first bag, a toy stove in the second bag, and a toy house in the third bag. At age 36 months, the toys included washable markers in the first bag, dress-up clothes in the second bag, and Dublo blocks in the third bag. Mothers were told that this part of the visit would last 15 minutes and that their child should spend some time with the toys in each bag, starting with the first bag. They were not explicitly told to play with their child nor were they told how much time their child was to spend with each toy.

All of the video-taped mother-child interactions were coded at a single site by teams of five to six coders who received extensive training and met regularly to maintain reliability. At 6, 15, and 24 months, the coders used 4-point scales to rate four aspects of maternal behavior (maternal stimulation, sensitivity to nondistress, positive regard, and intrusiveness [reverse coded]) that were summed to create a composite variable representing stimulating-responsive mother-child interaction at each of these three ages. At 36 months, 7-point scales were used by coders to rate four aspects of maternal behavior (maternal stimulation, supportive presence, hostility [reverse coded], and respect for autonomy), which were summed. Alphas ranged from .70 to .82 across the four ages (Hirsh-Pasek & Burchinal, 2006). Previous studies have found these ratings of stimulating-responsive mother-child interaction to be related to children’s cognitive and language skills in early childhood and middle childhood (Hirsch-Pasek & Burchinal, 2006; NICHD ECCRN, 2005b).

Caregiver-child interactions.

At 6, 15, 24, and 36 months, children who experienced nonparental care on a routine basis for 10 or more hours a week were observed in that care setting for two half-day visits at each age period (resulting in up to eight half-day visits) per child. Within each half-day visit, observers conducted two 30-minute observations using the Behavioral Frequency portion of the Observation Record of the Caregiving Environment (ORCE) for a total of four 30-minute observations distributed over two days at each age period that the study child was in routine nonparental care (NICHD ECCRN, 1996, 2000b).

The behavioral frequency codes were scored using a time-sample procedure in which observers observed for 30 seconds and then recorded for 30-seconds. For each 30-second record period, coders marked if the following caregiver behaviors had occurred during the previous observed interval: ask questions, speak positively to the child, respond to child vocalization or talk, positive talk, other talk, reads aloud to child, cognitive stimulation, social stimulation, and teaches academic skills (at the 24 and 36 month waves). At each age period (e.g., 6 months), the frequency of the specific behaviors (e.g., ask questions) were summed over the 30-minute observation, standardized, and then summed to create a composite measure of stimulating-responsive caregiver-child interaction (for means, standard deviations and ranges, see Table 2). Because each behavior contributing to the composite was standardized before summing, negative and positive values for each scale were contributing to the composite scores and mean scores could differ from 0 (unlike if standardization happened after the summing of the specific behaviors). Standardized Cronbach’s alphas were .78 at 6 months, .82 at 15 months, .78 at 24 months, .75 at 36 months, and .68 at 54 months.

Table 2.

Mother-Child Interactions, Caregiver-Child Interactions, and Performance on Later Competencies

N Mean SD Range
MCI 6 months 1,272 11.81 2.15 4 – 16
MCI 15 months 1,240 12.06 1.99 4 – 16
MCI 24 months 1,172 12.07 2.27 4 – 16
MCI 36 months 1,161 21.65 3.84 5 – 28
CCI 6 months 487 −0.09 2.60 −5.71 – 7.82
CCI 15 months 520 −0.37 6.87 −17.40 – 28.67
CCI 24 months 553 −0.25 5.72 −22.80 – 21.75
CCI 36 months 594 −0.44 5.35 −14.49 – 23.93
VOC 54 months 1,060 100.24 15.03 10 – 143
VOC grade 1 1,020 105.46 15.57 44 – 151
VOC grade 3 1,014 105.47 14.80 34 – 152
VOC grade 5 992 103.09 14.78 29 – 155
VOC age 15 889 99.93 14.77 34 – 158
MATH 54 months 1,053 102.94 15.63 41 – 153
MATH grade 1 1,023 110.80 17.14 46 – 163
MATH grade 3 1,013 115.05 15.00 30 – 153
MATH grade 5 993 109.31 13.54 37 – 156
MATH age 15 887 102.92 14.22 48 – 168
Open in a new tab

Note. MCI is mother-child interaction score. CCI is caregiver-child interaction score. VOC is the Woodcock-Johnson Picture Vocabulary score. MATH is the Woodcock-Johnson Applied Problems score.

Childcare observers were trained at a central location and certified as reliable coders using both live and videotaped assessments before conducting the live behavioral observations in the field (NICHD ECCRN, 2000b). Additionally, there were checks every three to four months for observer drift. Inter-observer agreement with the master coders ranged from .86 to .99. Prior research has found significant relations between the Behavioral Frequency observations of stimulating-responsive interactions with caregivers and children’s cognitive and language performance at 36 and 54 months (NICHD ECCRN, 2000a; NICHD ECCRN, 2003). The global ratings of the ORCE have been linked to children’s achievement outcomes at age 15 (Vandell et al., 2010) and at the end of high school (Vandell, Burchinal, & Pierce, 2016). The purpose of the current study is to determine if the behavioral ratings of the ORCE extend to children’s language and mathematics skills in middle childhood and adolescence, and to what extent they moderate associations of early mother-child interactions and these outcomes.

Vocabulary.

Vocabulary skills were assessed using the Picture Vocabulary Subtest from the Woodcock-Johnson Psycho-Educational Battery-Revised (WJ-R; McGrew, Werder, & Woodcock, 1991; Woodcock & Johnson, 1989) at 4 ½ years, 1st grade, 3rd grade, 5th grade, and age 15 years. The subtest measures both receptive and expressive vocabulary and is reliable and valid for the age ranges used in the current study (McGrew et al., 1991). The standardized achievement scores were used at each wave, thus a mean score of 100 and standard deviation of 15 were expected in a normative sample, which were close to the observed values in the study sample (see Table 2).

Mathematics.

Mathematics achievement was assessed using the Applied Problems Subtest from the Woodcock Johnson Tests of Achievement (McGrew et al., 1991; Woodcock & Johnson, 1989), which was individually administered at age 4 ½ years, 1st grade, 3rd grade, 5th grade, and age 15 years. As with the Picture Vocabulary Subtest, this instrument is valid and reliable for this age range (McGrew et al., 1991). The subtest assesses children’s ability to solve mathematical problems that include basic counting, addition, subtraction, and multiplication primarily through word problems read to the child. The standardized achievement scores were used at each wave, with scores close to the normative sample expected mean of 100 and standard deviation of 15 (see Table 2). For vocabulary and mathematics, we used the standardized achievement scores because we were interested in how early interactions related to overall performance in these domains rather than growth in these areas.

Child, family, and childcare covariates.

Child, family, and childcare covariates were included in all of the models (see Table 1 for the descriptive statistics on all covariates). Child-level covariates were: gender (1 = male, 0 = female) and race/ethnicity (dummy codes for Black, Hispanic, and Other with White as the reference group). Mother-level covariates were: maternal education (years of completed schooling), maternal age at the 1-month visit, parenting stress measured at the 6-month home visit using the Parenting Stress Index (Abidin, 1983), maternal vocabulary measured at the 36-month visit using the Peabody Picture Vocabulary Test, and maternal self-reports of separation anxiety (Hock, DeMeis, & McBride, 1988) collected at the 6, 15, and 24 month visits, maternal depression measured using the Center for Epidemiological Studies Depression Scale (Radloff, 1977) at the 6, 15, 24, and 36 month waves, and maternal health at the 6, 15, 24, and 36 month waves. Additional family level covariates were the mean income-to-needs ratio collected at the 6, 15, 24, and 36 month visits, and the proportion of visits in which there is a father present in the household at the 6, 15, 24, and 36 month waves (1 = father present, 0 = father not present). Childcare covariates included: the mean number of hours in all types of child care at the 6, 15, 24, and 36 month waves, and the primary type of child care across the 6, 15, 24, and 36 month waves (i.e., grandparent, child care home, or center-based care).

To simplify our statistical models and form more reliable measurement, when child, family, and childcare covariates existed at multiple waves across the first three years, we used the average of all waves. For example, this was done to control for overall levels of family income, rather than to control for family income at 6, 15, 24, and 36 month waves specifically. However, for hours in childcare, we controlled for each specific wave because of its close relation to the independent variable of interest: caregiver-child interactions. More information on all the child, family, and childcare covariates are available in the SECCYD technical manuals that are available at https://www.icpsr.umich.edu/icpsrweb/ICPSR/series/00233.

Analytic Plan

To address the research questions of additive and synergistic relations between stimulating-responsive mother-child and caregiver-child interactions during the first three years and children’s vocabulary and mathematics achievement from age 4 ½ to 15 years, a structural equation modeling approach is utilized. Two latent variables for early interactions are considered: stimulating-responsive mother-child interactions measured at four time points during the first three years of life (i.e., 6, 15, 24, and 36 months) and stimulating-responsive caregiver-child interactions measured at four time points during the first three years (i.e., 6, 15, 24, and 36 months). Two latent variables for child outcomes are also examined: child vocabulary measured at 5 time points from age 4 ½ to 15 years and child mathematics achievement measured at 5 time points from age 4 ½ to 15 years. A latent variable interaction was added to the models to test whether mother-child and caregiver-child stimulating-responsive interactions have synergistic relations on later vocabulary and mathematics.

The structural equation modeling framework has the advantage of modeling the shared variance among observed indicator variables for the constructs (i.e., no measurement error in the latent variables). Another advantage of this approach is full information maximum likelihood estimation (FIML), which allows the use of all available data to inform the model estimates, including control variables.

In addition to using latent variables to examine overall associations between stimulating-responsive mother-child and caregiver-child interactions on children’s vocabulary and mathematics achievement from age 4 ½ to 15 years, we conduct additional analyses to determine if effects associated with mother-child and caregiver-child interactions are different at specific time points (for example, at 4 ½ years, but not later; or up to grade 5, but not later). We elected to use this approach rather than linear or piecewise linear growth curve models to best address our specific research question of whether associations remain stable, become weaker with time, or are related sporadically. Importantly, these models were not constrained in any way that would force associations between the mother-child or caregiver-child interactions and later mathematics and vocabulary outcomes to change in a parametric way across time. This allows for the associations for the first three years to be assessed in terms of “persistence” versus “fadeout” with regard to magnitude of estimates that are not restricted. Analyses that specify linear growth curve models and piecewise linear growth curve models for the outcomes are available by request from the first author.

In a final set of models, we add the age 4 ½ year mother-child and caregiver-child interaction variables to examine if they explain additional variance in the outcomes above and beyond the first three years alone, and if the statistical significance for the first three years variables are altered once the age 4 ½ year variables are included.

All descriptive analyses and data management were conducted using Stata 14 (StataCorp, 2015), and all structural equation modeling was done using Mplus 8 (Muthén & Muthén, 2017). The maximum likelihood estimator with robust standard errors was used to deal with non-normality and non-independence of observations (Muthén & Muthén, 2017). This provides more conservative estimations of statistical significance by providing a correction to standard errors due to non-normality of distribution. Additionally, the cluster option was used to indicate that observations came from 10 research sites, such that observations within site were not independent, and therefore standard errors were adjusted.

Tables 1 and 2 provide the number of observations for each measure (for a correlation matrix of all study variables, see Table S1). All models used a full information maximum estimation (FIML) approach to deal with missing data. Given the longitudinal nature of the study design (i.e., children followed from birth through age 15), some participants were lost to attrition. Specifically, missing data ranged from 22.29% at 54 months to 34.60% at age 15 years on the outcomes, though only 15.25% of participants provided no outcome data (of the original 1,364 sample). In other words, outcome data were provided by 1,156 of the 1,364 total participants included in anlayses. In Table S2 we examine all covariates as predictors of whether children were missing data from the study at each wave for the outcomes (54 months to age 15; columns 1-5) and if children did not have any outcome data (column 6). Only two covariates were strongly related to providing no outcome data (i.e., ps < .001), such that children were less likely to have any outcome data if they had younger, less educated mothers. However, because FIML uses covariates to inform model estimates, we consider the missing at random a reasonable assumption and a better alternative to listwise deletion. That is, the FIML approach reduces bias associated with missing data for any reason by using all the available information to provide the most likely model estimates (Acock, 2012). FIML typically performs similar to multiple imputation approaches and outperforms model estimates using listwise deletion (Acock, 2005).

Results

Measurement Models

Analyses followed a sequence of three steps, with additional models provided in the Supplemental Materials. First, a confirmatory factor analysis approach was used to model latent factor variables for mother-child stimulating-responsive interactions, caregiver-child stimulating-responsive interactions, vocabulary, and mathematics to examine measurement model fit indices prior to the inclusion of covariates (see Table 3). All models statistically fit well in terms of their CFI and SRMR (Kline, 2005). As expected, the models with fewer degrees of freedom had poorer fit in terms of their RMSEA (Kenny, Kaniskan, & McCoach, 2015), however, the full measurement model had a relatively well-fitting RMSEA (Kline, 2005). All loadings on the latent factors were of satisfactory level (only one loading fell below .60, i.e., caregiver-child interaction at 6 months), suggesting each wave of measurement shared substantial variance with the underlying construct. All loadings for vocabulary and mathematics were especially large, which is evidence of relatively substantial shared variance in performance over time in these skills (loadings ranged from .71-.90).

Table 3.

Key Fit Statistics from Measurement Models

N (df)X2 CFI RMSEA SRMR T1 λ T2 λ T3 λ T4 λ T5 λ
MCI 1,306 (2) 17.85*** 0.99 0.08 0.02 .60 .66 .64 .75 –
CCI 848 (2) 19.21*** 0.94 0.10 0.04 .41 .63 .80 .60 –
VOC 1,156 (5) 83.40*** 0.98 0.12 0.02 .73 .82 .88 .90 .88
MATH 1,155 (5) 59.52*** 0.98 0.10 0.02 .71 .84 .85 .88 .80
Full Model 1,313 (129) 556.46*** 0.95 0.05 0.05 – – – – –
Open in a new tab

Note. MCI is mother-child interaction score. CCI is caregiver-child interaction score. VOC is the Woodcock-Johnson Picture Vocabulary score. MATH is the Woodcock-Johnson Applied Problems score. For MCI and CCI, T1 is 6 months, T2 is 15 months, T3 is 24 months, and T4 is 36 months. For VOC and MATH T1 is 54 months, T2 is grade 1, T3 is grade 3, T4 is grade 5, and T5 is age 15. Full Model included MCI, CCI, VOC, and MATH in a single analysis with all latent variables correlated.

In the full measurement model, mother-child and caregiver-interactions were correlated at r = .36, p < .001. This suggests that children who have more stimulating-responsive interactions with mothers have more stimulating-responsive interactions with caregivers on average (and vice versa). However, it also suggests these are separable constructs that are not tapping the same construct, which could be the case if both types of interactions were strongly driven by the child as opposed to the mother or caregiver. In summary, the measurement models are statistically consistent with the theoretical conceptualization that there are underlying latent factors that account for stimulating-responsive mother-child interactions and stimulating-responsive caregiver-child interactions from child age 6 to 36 months and underlying latent factors that account for vocabulary and mathematics scores from child age 4 ½ to 15 years.

Additive Associations of Stimulating-Responsive Interactions with Mothers and Caregivers on Later Vocabulary and Mathematics

Once the measurement models were established to be well-fitting, a structural equation model was run to estimate associations between stimulating-responsive mother-child interactions and stimulating-responsive caregiver-child interactions during the first 3 years of life on children’s vocabulary and mathematics achievement between 4 ½ and 15 years (see Figure 1; χ2(409) = 1013.47, p < .001; RMSEA = .03; CFI = .95; SRMR = .03; BIC = 201898). This model included the child, family, and childcare covariates and estimated the associations for latent factors of stimulating-responsive mother-child interactions and stimulating-responsive caregiver-child interactions predicting latent factors of vocabulary and mathematics achievement between 4 ½ and 15 years.

Figure 1. Associations between Early Mother-Child and Caregiver-Child Interactions and Children’s Later Vocabulary and Mathematics Skills.

Figure 1.

Open in a new tab

Standard errors were adjusted based on site location. MCI is mother-child interaction score. CCI is caregiver-child interaction score. VOC is the Woodcock-Johnson Picture Vocabulary score. MATH is the Woodcock-Johnson Applied Problems score.

* p < .05.

** p < .01.

*** p < .001.

Stimulating-responsive mother-child interactions during the first 3 years had significant positive associations with children’s later vocabulary (β = .20, p < .001) and mathematics achievement (β = .26, p < .001), and stimulating-responsive caregiver-child interactions had a significant positive association with mathematics (β = .11, p = .045). To interpret the standardized effects specifically, the model estimates for a standard deviation increase in stimulating-responsive interaction with mothers during the first 3 years, the predicted standard deviation increase for vocabulary was .20 and for mathematics was .26. For a standard deviation increase in caregiver-child interactions during the first 3 years, the predicted standard deviation increase for mathematics achievement was .11. See Table S3 for the estimates of the covariates.

Persistence versus Fadeout of Associations.

A series of additional models were tested because of the assumptions the primary analyses made for how mother-child interactions and caregiver-child interactions relate to later vocabulary and mathematics. First, we estimated models that included specific paths between latent mother-child interactions and caregiver-child interactions for each specific wave of vocabulary and mathematics at age 4 ½ years, 1st grade, 3rd grade, 5th grade, and age 15 years, rather than modeled on the overall latent factors of vocabulary and mathematics. This was done to check if stimulating-responsive mother-child interactions and caregiver-child interactions during the first three years had substantially different effects on child outcomes at different child ages (if, for example, the associations faded over time).

In general, the associations between stimulating-responsive mother-child interactions and caregiver-child interactions during the first three years and subsequent vocabulary and mathematics skills indicated persistent associations that did not diminish as quickly as would be predicted from a purely autoregressive standpoint of early skills influencing later skills (see Figure 2). Although the largest associations were typically observed at 4 ½ years, other associations were equivalent at later ages. For example, associations between stimulating-responsive interactions with mothers and child vocabulary were similar at age 4 ½ years and 1st grade, and similar at 3rd grade and 5th grade. For mathematics, associations were larger at age 15 than 5th grade. Stimulating-responsive caregiver-child interactions were more highly associated with mathematics skills at 3rd grade than 1st grade.

Figure 2. Associations between Early Mother-Child and Caregiver-Child Interactions and Later Vocabulary and Mathematics at Each Wave.

Figure 2.

Open in a new tab

Standard errors were adjusted based on site location. MCI is mother-child interaction. CCI is caregiver-child interaction. VOC is the Woodcock-Johnson Picture Vocabulary. MATH is the Woodcock-Johnson Applied Problems.

* p < .05.

** p < .01.

*** p < .001.

Including Age 4 ½ Stimulating-Responsive Interactions.

Second, models were tested in which we included measures of stimulating-responsive mother-child and caregiver-child interactions at 4 ½ years. This was done to test if the mother-child and caregiver-child interaction variables at the 4 ½ year wave explained additional variation in the outcomes beyond the first three years. The pattern of significant results was unchanged for the birth through three variables, though stimulating-responsive mother-child interactions at child age 4 ½ years was also significantly associated with mathematics achievement. All other associations for the 4 ½ year old variables were non-significant.

Sensitivity checks.

Two additional models were run to examine whether using latent variables to model mother-child interactions and caregiver-child interactions influenced the estimates. The observed pattern of significant associations was unaffected when only examining mother-child interactions at 36 months and caregiver-child interactions at 24 months (the largest loading indicator variables for each construct; see Figure S1). Additionally, when examining average mother-child interactions and caregiver-child interactions over the first three years, the effect was no longer significant between average caregiver-child interactions and mathematics (see Figure S2).

Synergistic Associations of Mother-Child and Caregiver-Child Interactions

Next, we examined if effects associated with early mother-child interaction and later outcomes were moderated by early caregiver-child interaction. To address this question, a statistical interaction between the latent mother-child interactions and latent caregiver-child interactions was added to the model. The statistical interaction between stimulating-responsive experiences with mothers and caregivers on children’s mathematics achievement was statistically significant, β = −.18, p = .007 (see Figure 3; BIC = 201894; and see Table S4 for the estimates of child, family, and childcare covariates). The dim lines in Figure 3 represent the main effects in the model, but should be interpreted within the context of the statistical interaction effect in the model.

Figure 3. The Moderation of Associations between Early Mother-Child and Caregiver-Child Interactions and Later Competencies.

Figure 3.

Open in a new tab

Standard errors were adjusted based on site location. MCI is mother-child interaction score. CCI is caregiver-child interaction score. VOC is the Woodcock-Johnson Picture Vocabulary score. MATH is the Woodcock-Johnson Applied Problems score. Dim lines represent main effects in the model; however, these should be interpreted within the context of the interaction effects.

* p < .05.

** p < .01.

*** p < .001.

To aid interpretation of this synergistic effect, a graph showing the predicted effect on standard deviation changes from average mother-child interactions and caregiver-child interactions on mathematics are shown for four hypothetical groups: 1) low (i.e., −1 standard deviation) maternal stimulation and low caregiver stimulation, 2) high (i.e., +1 standard deviation) maternal stimulation and low caregiver stimulation, 3) low maternal stimulation and high caregiver stimulation, and 4) high maternal stimulation and high caregiver stimulation (see Figure 4). Children low on both maternal and caregiver stimulating-responsive interactions performed substantially worse (.61 standard deviations) than children average on mother-child interactions and caregiver-child interactions (i.e., the x-axis in Figure 4), respectively. The statistical interaction between maternal and caregiver responsiveness-stimulation on children’s vocabulary was not significant (see Figure 3), although the main effect of caregiver responsiveness-stimulation on vocabulary did become statistically significant in the model that included the statistical interaction.

Figure 4. Predicted Standardized Effects for Later Competencies based on Hypothetical Groups of MCI and CCI.

Figure 4.

Open in a new tab

The values are calculated based on standardized results from Figure 3 (i.e., includes the interaction). ‘Low’ is 1 standard deviation below the mean and ‘High’ is 1 standard deviation above the mean. MCI is mother-child interaction score. CCI is caregiver-child interaction score. VOC is the Woodcock-Johnson Picture Vocabulary score. MATH is the Woodcock-Johnson Applied Problems score.

Persistence versus Fadeout of Synergistic Associations.

Next, we examined the synergistic effects of maternal and caregiver stimulating-responsive interactions on each time period (age 4 ½ years, grade 1, grade 3, grade 5, age 15 years) separately. For both vocabulary and mathematics, the synergistic associations between stimulating-responsive interactions with mother and caregiver were most negative for grade 3 and grade 5 when examining the outcomes at each wave (see Figure S3). Thus, although there was some evidence of a diminishing effect when examining the additive effects, there was no such evidence with the interaction term.

Post-Hoc Analyses: Does Excluding Maternal Vocabulary Alter Results?

After we found stronger associations with children’s later mathematics skills than vocabulary skills, we examined if our pattern of estimates would change if maternal vocabulary was excluded as a control variable. For the first model with additive effects only, the main effects on mathematics remained larger than the main effects on vocabulary, consistent with the model that included maternal vocabulary. For the second model with the synergistic effect, the main effect for mother-child interactions did become larger for vocabulary than the main effect for mother-child interactions predicting mathematics (though this model included a statistically significant interaction term making it challenging to interpret the main effect). In general, the pattern of results remained substantively similar across both specification and are available by request from the first author.

Discussion

The current study examines the extent to which stimulating-responsive interactions with mothers and nonparental caregivers during the first three years of life are associated with children’s later vocabulary and mathematics achievement. A structural equation modeling approach was used to test both additive and synergistic relations between early mother-child and caregiver-child interactions and children’s later skills. More stimulating-responsive interactions with both mothers and with caregivers were associated with higher mathematics achievement. More stimulating-responsive interactions with mothers during the first three years were associated with children’s larger vocabularies through age 15 years.

Our findings extend prior research that has examined children’s experiences during the first three years of life as setting the stage for later development. Prior work has identified early maternal sensitivity as having enduring associations with social outcomes through adolescence (Fraley et al., 2013) and into adulthood (Raby et al., 2015). Importantly, our findings suggest significant associations of early mother-child interactions are maintained controlling for caregiver-child interactions, as well as a host of child, mother, and family covariates, including maternal vocabulary.

The caregiver-child interactions aspect of the study fits within broader work on the role of early childhood education for long-run cognitive and language outcomes. A contribution of the current study is connecting the specific behavioral frequency ratings observed during the first three years of life with these outcomes. For instance, these behavioral frequency codes were associated with child outcomes at ages 3 (NICHD ECCRN, 2000a) and 4 ½ years (NICHD ECCRN, 2003), but this is the first study to our knowledge that examined associations between the frequency counts of stimulating-responsive caregiver behaviors and much later skills. Behavioral codes for stimulating-responsive interactions with caregivers may be an important tool for practitioners and policymakers interested in understanding early childhood education quality. Unlike broad, global assessments, these measures are easier to code and use by practitioners (NICHD ECCRN, 2001; Pellegrini, 2004). They also provide an easy to interpret description of the specific experiences of individual children in the classroom.

In addition to additive relations in which both mother-child and caregiver-child interactions were related to later mathematics achievement, a moderated relation was found that was consistent with the dual risk hypothesis. Specifically, children who had few stimulating-responsive interactions with their mothers and with their caregivers had substantially worse mathematics achievement. These findings are consistent with the work of Nelson and colleagues (2007) who found that very low levels of stimulating-responsive interactions during the first three years were exacerbated when interventions were not put into place to provide supplementary enriching care experiences.

A second question motivating the current study was whether the associations between early mother-child and caregiver child interactions with mathematics and vocabulary outcomes faded over time. Evidence consistent with some fadeout but also persistence was found. Relations between stimulating-responsive interactions in the first three years and children’s vocabulary and mathematics achievement were higher at 4 ½ years than at later ages, but relations between these early stimulating-responsive interactions and later functioning continued to be significant through age 15 years. Indeed, there were instances of larger associations between mother-child and caregiver-child interactions at later ages than earlier waves.

In general, the associations between stimulating-responsive interactions during the first three years and later outcomes were larger for mathematics than vocabulary. This was unanticipated because we had expected that early stimulating experiences, which were generally focused on language, would be more important for later vocabulary. Additionally, the smaller effects found between early interactions and vocabulary were not explained by removing maternal vocabulary as a control variable.

Follow-up analyses examined if the inclusion of children’s experiences in the later preschool period at age 4 ½ years accounted for later vocabulary or mathematics achievement. The inclusion of the later mother-child and caregiver-child experiences at 4 ½ years did not alter the significant associations between experiences in the first three years and later outcomes, providing support for the importance of the very early experiences with respect to later vocabulary and mathematics. At the same time, stimulating-responsive interactions with mothers at 4 ½ years did predict later mathematics above and beyond the earlier experiences of children. Taken together, these findings suggest the importance of children’s experiences during the first three years for later vocabulary and mathematics, with later experiences also potentially playing a role.

Potential Implications for Early Childhood Interventions and Policies

The findings from this large study of normative development are consistent with early childhood intervention studies (e.g., Campbell et al., 2001) and neuroscience research (Grantham-McGregor et al., 2007; Nelson et al., 2014) that focused on low-income and at-risk samples. In particular, the findings underscore the potential value of stimulating-responsive interactions with mothers during the first three years of life for children’s later vocabulary and mathematics achievement, as well the independent value of stimulating-responsive interactions with caregivers during the first three years in relation to adolescents’ mathematics achievement. These findings suggest that interventions aimed at both mothers and caregivers could have additive effects on children’s skills. Interventions would ideally seek to boost the quality of interactions with both mothers and caregivers. The current study also identifies a group of children who were at the highest risk for lower mathematics achievement: the negative effects associated with a dearth of stimulating-responsive mother-child interactions were exacerbated and magnified when children also lacked stimulating-responsive interactions with caregivers. In circumstances in which intervening with mothers is not feasible, providing high quality experiences with caregivers may be particularly important.

The current findings also highlight the need to consider the timing of early childhood interventions. Although many contemporary interventions are focused on prekindergarten programs serving 4- and 5-year-olds, the current study identifies the potential importance of the first three years for the development of vocabulary and mathematics skills. The possible long-run effects on cognitive and language outcomes may be amplified by starting earlier and by targeting the quality of interactions with mothers and/or caregivers.

Limitations and Future Directions

Several study limitations must be noted. First, in this non-experimental study, issues of child effects and omitted variable bias may account for the obtained findings. Children who had more advanced language and social skills may have elicited more and better responsive and stimulating interactions with their mothers and nonparental caregivers. Future research would benefit from exploring potential bidirectional relations among children’s early skills and responsive-stimulating interactions with mothers and nonparental caregivers. The associations found in this study also may be due to unobserved characteristics of the children, parents or caregivers that led to higher quality mother-child and caregiver-child interactions. Furthermore, the stronger associations found for mathematics achievement compared to later vocabulary may reflect the fact that our measures of early stimulating-responsive interactions did not directly measure the conversational turn-taking which has been closely associated with later language outcomes (Gilkerson et al., 2018; Romeo et al., 2018).

A potential limitation with comparing the estimates between mother-child and caregiver-child interactions is that the measures were assessed in different ways. The mother-child interactions involved short semi-structured play situation with observers, whereas the caregiver-child interactions were observed for longer periods and in the natural setting. Thus, although standardized estimates are presented to eliminate differences due to raw scaling, differences in how the measures were collected could contribute to the magnitudes of the observed associations and conclusions should not be drawn about which type of interaction is more important. Additionally, children’s child care experiences probably change considerably during the first three years and we placed our focus on overall levels rather than this variability. Future research could better inform issues related to specific timing of experiences and to what extent within-child variability in caregiver-child interactions relate to developmental outcomes. Furthermore, we did not have item level data for our outcomes to assess longitudinal invariance of the measures, thus our findings could be influenced by measurement characteristics that change over time. Finally, although we included many child, family, and child care covariates, it will be important for future research to explore additional confounds, with the strongest evidence being findings from experimental studies.

Conclusions

This study finds evidence for both additive and synergistic associations between mother-child and caregiver-child interactions during the first three years and children’s mathematics achievement and vocabulary from early childhood through adolescence. More stimulating-responsive interactions with mothers were associated with higher mathematics and vocabulary achievement, with statistically significant associations persisting through age 15 years. More stimulating-responsive interactions with caregiver also predicted children’s later mathematics. Consistent with a dual risk hypothesis, children who received little stimulation from both their mothers and caregivers during the first three years were at heightened risk for poorer mathematics achievement. Taken together, these finding underscore the potential importance of children’s experiences in the first three years in helping to address academic achievement gaps.

Supplementary Material

Supplemental Material 1
Supplemental Material 2

Acknowledgments

A cooperative agreement (5 U10 HD027040) between the study investigators, including Deborah Lowe Vandell, and the Eunice Kennedy Shriver National Institute of Child Health and Human Development supported the design and data collection of the Study of Early Childcare and Youth Development (SECCYD) from birth through age 15 years. Work on this paper was supported by grants from the Charles Stewart Mott Foundation, the Institute of Education Sciences, U.S. Department of Education, Grant R305B120013 to the University of California, Irvine (PI: Greg Duncan) and the Eunice Kennedy Shriver National Institute of Child Health & Human Development of the National Institutes of Health under Award Number P01HD065704 (PI: Greg Duncan). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health, the Institute of Education Sciences, or the U.S. Department of Education.

References

  1. Abidin RR (1983). Parenting Stress Index Manual. Charlottesville, VA: Pediatric Psychology Press. [Google Scholar]
  2. Acock AC (2005). Working with missing values. Journal of Marriage and Family, 67, 1012–1028. Doi: 10.1111/j.1741-3737.2005.00191.x [DOI] [Google Scholar]
  3. Acock AC (2012). What to do about missing values In Cooper H, Camic PM, Long DL, Panter AT, Rindskopf D & Sher KJ (Eds.), Data analysis and research publication. APA handbook of research methods in psychology, Vol. 3. (Vol. 3, pp. 27–50). Washington, DC: American Psychological Association. [Google Scholar]
  4. Ayoub C, Vallotton CD, & Mastergeorge AM (2011). Developmental pathways to integrated social skills: The roles of parenting and early intervention. Child Development, 82(2), 583–600. [DOI] [PubMed] [Google Scholar]
  5. Bailey D, Duncan GJ, Odgers CL, & Yu W (2017) Persistence and fadeout in the impacts of child and adolescent interventions, Journal of Research on Educational Effectiveness, 10:1, 7–39, doi: 10.1080/19345747.2016.1232459 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bradley RH, Caldwell BM, Rock SL, Ramey CT, Barnard KE, Gray C, … Johnson DL (1989). Home environment and cognitive development in the first 3 years of life: A collaborative study involving six sites and three ethnic groups in North America. Developmental Psychology, 25(2), 217–235. [Google Scholar]
  7. Burchinal MR, Lowe Vandell D, & Belsky J (2014). Is the prediction of adolescent outcomes from early child care moderated by later maternal sensitivity? Results from the NICHD study of early child care and youth development. Developmental psychology, 50(2), 542–553. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Burchinal M, Magnuson K, Powell D, & Hong SS (2015). Early childcare and education In Lerner RM (Ed.in Chief) & Bornstein M& Leventhal T(Vol. Eds.), Handbook of child psychology and developmental science: Vol. 4. Ecological settings and processes (7th ed., pp. 223–267). Hoboken, NJ: Wiley; Doi: 10.1002/9781118963418.childpsy406 [DOI] [Google Scholar]
  9. Burchinal MR, Roberts JE, Nabors LA, & Bryant DM (1996). Quality of center child care and infant cognitive and language development. Child Development, 67, 606–620. [PubMed] [Google Scholar]
  10. Burchinal M, Vernon-Feagan L, Cox M, and Key Family Life Project Investigators (2008). Cumulative social risk, parenting, and infant development in rural low-income communities. Parenting Science and Practice. 8(1), 41–69 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Campbell FA, Pungello EP, Miller-Johnson S, Burchinal M, & Ramey CT (2001). The development of cognitive and academic abilities: growth curves from an early childhood educational experiment. Developmental Psychology, 37, 231–242. Doi: 10.1037/0012-1649.37.2.231 [DOI] [PubMed] [Google Scholar]
  12. Chugani HT, Behen ME, Muzik O, Juhász C, Nagy F, & Chugani DC (2001). Local brain functional activity following early deprivation: a study of postinstitutionalized Romanian orphans. Neuroimage, 14, 1290–1301. [DOI] [PubMed] [Google Scholar]
  13. Dearing E, McCartney K, & Taylor BA (2009). Does higher quality early child care promote low-income children’s math and reading achievement in middle childhood? Child Development, 80:5, 329–1349. [DOI] [PubMed] [Google Scholar]
  14. Diamond A, Barnett WS, Thomas J, & Munro S (2007). Preschool program improves cognitive control. Science, 318:5855, 1387–1388. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Duncan GJ, & Sojourner AJ (2013). Can intensive early childhood intervention programs eliminate income-based cognitive and achievement gaps? Journal of Human Resources, 48(4), 945–968. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Fraley RC, Roisman GI, & Haltigan JD (2013). The legacy of early experiences in development: Formalizing alternative models of how early experiences are carried forward over time. Developmental Psychology, 49, 109–126. Doi: 10.1037/a0027852 [DOI] [PubMed] [Google Scholar]
  17. Fox SE, Levitt P, & Nelson CA III (2010). How the timing and quality of early experiences influence the development of brain architecture. Child Development, 81:1, 28–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Gilkerson J, Richards JA, Warren SF, Oller DK, Russo R, & Vohr B (2018). Language experience in the second year of life and language outcomes in late childhood. Pediatrics, e20174276 Doi: 10.1542/peds.2017-4276 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Golinkoff RM, & Hirsh-Pasek K (2000). How babies talk: The magic and mystery of language in the first three years of life. New York: Penguin. [Google Scholar]
  20. Grantham-McGregor S, Cheung YB, Cueto S, Glewwe P, Richter L, Strupp B, & International Child Development Steering Group. (2007). Developmental potential in the first 5 years for children in developing countries. The Lancet, 369(9555), 60–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Heckman J, Pinto R, & Savelyev P (2013). Understanding the mechanisms through which an influential early childhood program boosted adult outcomes. American Economic Review, 103:6, 2052–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Heckman JJ, Holland ML, Makino KK, Pinto R, & Rosales-Rueda M (2017). An analysis of the Memphis Nurse-Family Partnership program National Bureau of Economic Research Working Paper No. 23610. Doi: 10.3386/w23610 [DOI] [Google Scholar]
  23. Hirsh-Pasek K & Burchinal M (2006). Mother and caregiver sensitivity over time: Predicting & language and academic outcomes with variable- and person-centered approaches. Merrill-Palmer Quarterly, 52, 449–485. [Google Scholar]
  24. Hirsh-Pasek K, Adamson LB, Bakeman R, Owen MT, Golinkoff RM, Pacel A, Yust PKS & Suma K (2015). The contribution of early communication quality to low-income children’s language success. Psychological Science, 26, 1071–1083. [DOI] [PubMed] [Google Scholar]
  25. Hock E, DeMeis D, & McBride S (1988). Maternal separation anxiety: Its role in the balance of employment and motherhood in mothers of infants In Gottfried AE (Ed), Maternal Employment and Children’s Development. Springer Science: New York. [Google Scholar]
  26. Kenny DA, Kaniskan B, & McCoach DB (2015). The performance of RMSEA in models with small degrees of freedom. Sociological Methods & Research, 44, 486–507. [Google Scholar]
  27. Kline RB (2005). Principles and practice of structural equation modeling (2nd ed.). New York: Guilford. [Google Scholar]
  28. Li W, Farkas G, Duncan GJ, Burchinal MR, & Vandell DL (2013). Timing of high-quality child care and cognitive, language, and preacademic development. Developmental Psychology, 49, 1440–1451. doi: 10.1037/a0030613 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Lupien SJ, McEwen BS, Gunnar MR, & Heim C (2009). Effects of stress throughout the lifespan on the brain, behaviour and cognition. Nature Review Neuroscience, 10, 434–445. [DOI] [PubMed] [Google Scholar]
  30. McGrew KS, Werder JK, & Woodcock RW (1991). WJ-R Technical Manual. Allen, TX: DLM. [Google Scholar]
  31. Miller E, Farkas G, Vandell DL, & Duncan G (2014). Do the effects of Head Start vary by parental pre-academic stimulation? Child Development, 85, 1385–1400. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Muthén LK, & Muthén BO (1998-2017). Mplus user’s guide (8th ed.). Los Angeles, CA: Muthén & Muthén. [Google Scholar]
  33. Nelson CA, Fox N,A & Zeanah CH (2014). Romania’s abandoned children. Harvard University Press. [Google Scholar]
  34. Nelson CA III, Zeanah CH, Fox NA, Marshall PJ, Smyke AT, & Guthrie D (2007). Cognitive recovery in socially deprived young children: The Bucharest Early Intervention Project. Science, 318: 5858, 1937–1940. DOI: 10.1126/science.1143921 [DOI] [PubMed] [Google Scholar]
  35. NICHD Early Child Care Research Network. (1996). Characteristics of infant child care: Factors contributing to positive caregiving. Early Childhood Research Quarterly, 11(3), 269–306. [Google Scholar]
  36. NICHD Early Child Care Research Network. (1999). Child care and mother-child interaction in the first 3 years of life. Developmental Psychology, 35, 1399–1413. doi: 0012-1649/99/$3.00 [PubMed] [Google Scholar]
  37. NICHD Early Child Care Research Network. (2000a). The relation of child care to cognitive and language development. Developmental Psychology, 71, 960–980. doi: 0009-3920/2000/7104-0015 [DOI] [PubMed] [Google Scholar]
  38. NICHD Early Child Care Research Network. (2000b). Characteristics and quality of child care for toddlers and preschoolers. Applied Developmental Science, 4(3), 116–135. [Google Scholar]
  39. NICHD Early Child Care Research Network. (2001). A new guide for evaluating child care quality. Zero to Three, 21(5), 40–47. [Google Scholar]
  40. NICHD Early Child Care Research Network. (2003). Does quality of child care affect child outcomes at age 4½? Developmental Psychology, 39, 451–469. doi: 10.1037/0012-1649.39.3.451 [DOI] [PubMed] [Google Scholar]
  41. NICHD Early Child Care Research Network. (Eds.). (2005). Child care and child development: Results from the NICHD Study of Early Child Care and Youth Development. New York: Guilford. [Google Scholar]
  42. NICHD Early Child Care Research Network. (2005b). Predicting individual differences in attention, memory, and planning in first graders from experiences at home, child care, and school. Developmental Psychology, 41(1), 99–114. [DOI] [PubMed] [Google Scholar]
  43. Pellegrini AD (2004). Observing children in their natural worlds: A methodological primer. Mahwah NJ: Lawrence Erlbaum [Google Scholar]
  44. Phillips DA, & Shonkoff JP (Eds.). (2000). From neurons to neighborhoods: The science of early childhood development. National Academies Press. [PubMed] [Google Scholar]
  45. Raby KL, Roisman GI, Fraley RC, & Simpson JA (2015). The enduring predictive significance of early maternal sensitivity: Social and academic competence through age 32 years. Child Development, 86, 695–708. doi: 10.1111/cdev.12325 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Radloff LF (1977). The CES-D Scale: A self-report for research in the general population. Applied Psychological Measurement, 1, 385–401. [Google Scholar]
  47. Rafferty Y, Griffin KW, & Lodise M (2011). Adolescent motherhood and developmental outcomes of children in Early Head Start: The influence of maternal parenting behaviors, well-being, and risk factors within the family setting. American Journal of Orthopsychiatry, 81(2), 228–245. [DOI] [PubMed] [Google Scholar]
  48. Romeo RR, Leonard JA, Robinson ST, West MR, Mackey AP, Rowe ML, & Gabrieli JD (2018). Beyond the 30-Million-Word Gap: Children’s Conversational Exposure Is Associated with Language-Related Brain Function. Psychological science, 29, 700–710. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Rutter M (1987). Psychosocial resilience and protective mechanisms. American Journal of Orthopsychiatry, 57, 316–331. [DOI] [PubMed] [Google Scholar]
  50. Schweinhart LJ (1993). Significant benefits: The High/Scope Perry Preschool Study through age 27. Monographs of the High/Scope Educational Research Foundation, No. 10. [Google Scholar]
  51. Shonkoff JP, Garner, and The Committee On Psychosocial Aspects Of Child And Family Health, Committee On Early Childhood, Adoption, And Dependent Care, And Section On Developmental And Behavioral Pediatrics. (2012). The Lifelong Effects of Early Childhood Adversity and Toxic Stress. Pediatrics, 129(1), D1–D12, doi: 10.1542/peds.digest129 [DOI] [PubMed] [Google Scholar]
  52. StataCorp. (2015). Stata Statistical Software: Release 14. College Station, TX: StataCorp LP. [Google Scholar]
  53. Vandell DL, Belsky J, Burchinal M, Steinberg L, Vandergrift N, & NICHD Early Child Care Research Network. (2010). Do effects of early child care extend to age 15 years? Child Development, 81, 737–756. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Vandell DL, Burchinal M, & Pierce KM (2016). Early child care and adolescent functioning at the end of high school: Results from the NICHD Study of Early Child Care and Youth Development. Developmental Psychology, 52, 1634–1645. doi: 10.1037/dev0000169 [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Vernon-Feagans L, Bratsch-Hines ME, & Family Life Project Key Investigators. (2013). Caregiver–child verbal interactions in child care: A buffer against poor language outcomes when maternal language input is less. Early Childhood Research Quarterly, 28(4), 858–873. doi: 10.1016/j.ecresq.2013.08.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Woodcock RW, & Johnson MB (1989). Woodcock-Johnson Psycho-Educational Battery – Revised. Allen, TX: DLM. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Material 1
NIHMS1047342-supplement-Supplemental_Material_1.xlsx (17.1KB, xlsx)
Supplemental Material 2
NIHMS1047342-supplement-Supplemental_Material_2.pdf (331.9KB, pdf)

RESOURCES