Parent support for learning at school entry: Benefits for aggressive children in high-risk urban contexts

Early Childhood Research Quarterly 31 (2015) 9–18

Contents lists available at ScienceDirect

Early Childhood Research Quarterly

Parent support for learning at school entry: Benefits for aggressive children in high-risk urban contexts Rachel M. Abenavoli a,∗ , Mark T. Greenberg b , Karen L. Bierman c a The Pennsylvania State University, Department of Human Development and Family Studies, 315-A Health and Human Development East, University Park, PA 16802, United States b The Pennsylvania State University, Department of Human Development and Family Studies, Bennett Pierce Prevention Research Center, 306 Biobehavioral Health Building, University Park, PA 16802, United States c The Pennsylvania State University, Department of Psychology, 251 Moore Building, University Park, PA 16802, United States

a r t i c l e

i n f o

Article history: Received 7 January 2014 Received in revised form 4 December 2014 Accepted 21 December 2014 Available online 31 December 2014 Keywords: Academic knowledge Executive function Parenting Support for learning

a b s t r a c t Children growing up in urban poverty are at high risk for low achievement across the school years, particularly when they enter school with elevated aggressive–disruptive behavior problems. In general, parent support for child learning is associated with school readiness and school success, but whether it serves as a protective factor for aggressive children in disadvantaged urban contexts is unknown. In this study, 207 urban and predominantly African American children with elevated aggressive–disruptive behavior problems at kindergarten entry (M = 5.94 years, SD = 0.39 years) were followed into first grade. Two dimensions of parent support for learning were assessed: teacher-rated parent school involvement and observed quality of parent teaching behaviors. Cross-lagged analyses indicated that parent support for learning predicted growth in aspects of children’s academic knowledge and executive functioning over time, controlling for children’s prior skills and demographic risk factors. Promoting parent support for learning may be a promising strategy to enhance the school readiness of children at dual risk due to contextual adversity and elevated aggressive–disruptive behavior problems. © 2015 Elsevier Inc. All rights reserved.

Introduction Children who grow up in urban areas characterized by poverty and violence are at high risk for school difficulties. Over one-third enter school lacking the cognitive and/or self-regulation competencies that are needed for positive adjustment and academic success (Ryan, Fauth, & Brooks-Gunn, 2006). Researchers have suggested that family strengths may provide a particularly critical source of resiliency under conditions of adversity (Mistry, Benner, Biesanz, & Clark, 2010). However, parent support for learning rarely has been studied as a protective factor in samples of children at high risk for school readiness delays and school difficulties. The current study addressed this gap in the existing literature by examining predictive links between parent support for learning and child school readiness in a sample of children with elevated aggressive–disruptive behaviors at school entry living in very

∗ Corresponding author. Tel.: +1 914 588 6284. E-mail addresses: [email protected] (R.M. Abenavoli), [email protected] (M.T. Greenberg), [email protected] (K.L. Bierman). http://dx.doi.org/10.1016/j.ecresq.2014.12.003 0885-2006/© 2015 Elsevier Inc. All rights reserved.

disadvantaged urban neighborhoods. Using both observer and teacher ratings of parent support for learning and a longitudinal design that modeled bidirectional parent–child influences across kindergarten and first grade, this study examined change in academic knowledge and executive functioning (EF), two aspects of school readiness that enhance the pace of learning by supporting learning engagement and problem-solving (Blair, 2002). The importance of early academic knowledge and EF for school success Academic knowledge (e.g., literacy and math skills) and EF (e.g., working memory, inhibitory control, and attention set-shifting) are two critical components of school readiness that predict children’s long-term academic trajectories (Blair, 2002). Because learning is a cumulative process, children who enter kindergarten with higher levels of emergent academic knowledge are in a better position to take advantage of learning opportunities and show accelerated academic growth as they progress through school, whereas children with lower academic readiness fall further behind their classmates over time (Duncan et al., 2007; La Paro & Pianta, 2000;

10

R.M. Abenavoli et al. / Early Childhood Research Quarterly 31 (2015) 9–18

Lee & Burkam, 2002). In addition, EF at school entry contributes to the pace of children’s learning both by enhancing their knowledge acquisition directly and by supporting self-regulated and engaged learning behaviors in the classroom (Bierman, Torres, Domitrovich, Welsh, & Gest, 2009; Blair, 2002; McClelland et al., 2007). Longitudinal studies suggest synergistic associations between emergent academic knowledge and EF at school entry, and each makes unique contributions to growth in children’s achievement during the early school years (Welsh, Nix, Blair, Bierman, & Nelson, 2010). Given the developmental sequelae of early academic knowledge and EF, it is of concern that children who exhibit elevated rates of aggressive–disruptive behavior at school entry also often exhibit deficits in these important aspects of school readiness (Bierman et al., 2009; Conduct Problems Prevention Research Group, 1999; Hughes, White, Sharpen, & Dunn, 2000; National Institute of Child Health and Human Development Early Child Care Research Network [NICHD ECCRN], 2004). Furthermore, aggressive behavior problems, delays in academic knowledge, and deficits in EF are all more common among children growing up in poverty, relative to children growing up in more socioeconomically advantaged families (Blair et al., 2011; Dodge, Pettit, & Bates, 1994; NICHD ECCRN, 2005; Raver, Blair, Willoughby, & the Family Life Project Key Investigators, 2013). Urban poverty is associated with a set of adversities that might account for these concurrent risks for school difficulties. For example, aggressive–disruptive behavior problems and poor school readiness (i.e., deficits in academic knowledge and EF) are each linked empirically with low levels of parent educational attainment, single parenthood, family insularity and low levels of social support, and exposure to stress associated with neighborhood violence and family conflict (DeBell, 2008; Dodge et al., 1994; Duncan, Brooks-Gunn, & Klebanov, 1994; Lee & Burkam, 2002; Magnuson, Sexton, Davis-Kean, & Huston, 2009). The role of parent support for learning In the context of these multiple adversities associated with urban poverty, family strengths may be particularly important because they may serve as protective factors that foster resilience and support school readiness among children at risk (Mistry et al., 2010). Certainly, among aggressive children living in high-risk neighborhoods, there is variation in their school readiness at school entry and their later school achievement (Bierman et al., 2009; Mistry et al., 2010). Parent support for child learning may represent a key process that promotes school readiness over time and that, conceivably, could protect children living in urban poverty who show high rates of aggressive–disruptive behavior from experiencing academic difficulties when they enter elementary school. A wealth of empirical evidence links various aspects of parent support for learning with child school readiness. For example, parents who provide greater access to learning activities and materials in their homes have children with better academic knowledge (Bradley, Corwyn, Burchinal, McAdoo, & Garcia Coll, 2001; Gershoff, Aber, Raver, & Lennon, 2007; Mistry et al., 2010), though studies examining associations with EF specifically have produced mixed results (Dilworth-Bart, 2012; Sarsour et al., 2011). In addition, parents’ involvement at school and attitudes regarding education have been associated with children’s early literacy and math achievement (Arnold, Zeljo, Doctoroff, & Ortiz, 2008; Galindo & Sheldon, 2012; Powell, Son, File, & San Juan, 2010). Although the link between parent school involvement and EF has not been examined, involvement has been linked with children’s attention and task persistence as reported by teachers, which may be proxies for EF (Fantuzzo, McWayne, Perry, & Childs, 2004). Observational studies examining parent teaching behaviors, which include responsiveness to the child’s needs, support for child self-direction, explicit teaching or clarification, and cognitive stimulation (Ayoub,

Vallotton, & Mastergeorge, 2011; Mulvaney, McCartney, Bub, & Marshall, 2006; Pianta, Smith, & Reeve, 1991), have shown that these behaviors are linked with children’s early academic knowledge (Chazan-Cohen et al., 2009; Dodici, Draper, & Peterson, 2003; Mulvaney et al., 2006) and EF (Bernier, Carlson, & Whipple, 2010; Hammond, Müller, Carpendale, Bibok, & Liebermann-Finestone, 2012; Hughes & Ensor, 2009; Lengua, Honorado, & Bush, 2007). Although many studies have documented links between parent support for learning and child school readiness, much of this research has focused on representative samples in which parent support for learning is highly correlated with parent education and socioeconomic status (Arnold et al., 2008; Brooks-Gunn & Markman, 2005). More recently, a few studies have focused specifically on families living in poverty to better understand how parent support for learning may promote child school readiness in the context of socioeconomic risk (Chazan-Cohen et al., 2009; Fantuzzo et al., 2004; McWayne, Hampton, Fantuzzo, Cohen, & Sekino, 2004). Furthermore, parenting practices often have been studied as predictors of aggressive–disruptive child behavior at school entry, but this research typically has focused on behavior management practices that increase risk for child aggression (e.g., harsh, inconsistent discipline), rather than support for learning, which might represent a protective factor that reduces the likelihood of emerging academic difficulties when children high in aggressive behavior enter school. Additional research is thus needed to determine whether support for learning has beneficial effects even among children at high risk for school readiness difficulties, such as aggressive children in disadvantaged urban contexts.

The present study The present study took advantage of longitudinal data that were collected in the context of a preventive intervention for children with elevated aggressive–disruptive behavior problems at school entry (Greenberg, Bierman, Nix, & Gatzke-Kopp, 2012). The intervention focused on improving parent management of problem behaviors and promoting positive peer relations and did not have significant effects on the variables studied here (i.e., parent support for learning, child academic knowledge, and child EF). Although intervention outcomes are outside the scope of the current paper, the rich longitudinal dataset made it possible to examine links between parent support for learning and aspects of child school readiness in this high-risk sample, controlling for several concurrent risk factors. Specifically, this study controlled for family income, parent education level, age, and single parent status, which are typically associated with parent support for learning (Burchinal, Vernon-Feagans, Cox, & Key Family Life Project Investigators, 2008; Lengua et al., 2007; NICHD ECCRN, 2005). The current study also controlled for children’s aggression at the start of kindergarten, which could potentially influence both parent support for learning and child school readiness. We addressed the critical question of whether parent support for learning serves as a protective factor promoting child school readiness among children who face multiple risks for school difficulties: those with elevated aggressive–disruptive behavior problems living in highrisk urban contexts. Although prior work has demonstrated the link between parent support for learning and children’s academic knowledge in high-risk, low-income samples (Chazan-Cohen et al., 2009; McWayne et al., 2004; Mistry et al., 2010), the association between parent support for learning and children’s EF has only been documented in relatively low-risk and ethnically homogenous samples (Bernier et al., 2010; Hammond et al., 2012; Hughes & Ensor, 2009). Thus, we extend existing research by examining the effects of two independently measured dimensions of parent support for learning on children’s academic knowledge and EF within

R.M. Abenavoli et al. / Early Childhood Research Quarterly 31 (2015) 9–18

a uniquely high-risk sample of low-income families with children exhibiting aggressive behavior at the start of kindergarten. Parent support for learning primarily has been shown to predict the level of children’s skills at a given point in time; however, we utilized a longitudinal design here in order to examine change in children’s skills, which permits a more rigorous test of these associations. We employed cross-lagged path models to examine concurrent (kindergarten) and predictive (to first grade) relations between parent support for learning and child school readiness outcomes. Using cross-lagged techniques to account for stability in children’s outcomes over time, we examined whether support for learning predicted change in children’s academic knowledge and EF from kindergarten to first grade. This approach reduces potential omitted variable bias and strengthens confidence in the observed associations obtained in this non-experimental study.

Method Participants The current sample was drawn from 10 public elementary schools in one disadvantaged urban school district in central Pennsylvania characterized by high rates of poverty and crime. As part of a broader project on the development and prevention of aggression (Greenberg et al., 2012), a brief teacher-report screening measure was used to identify and recruit two cohorts of children. Information about the study and informed consent forms for the screening procedure were included with other start-of-year school documents sent to parents by the school district. Parents were able to review the study documents and return signed consent forms at the same time as other school documents, which led to a high response rate. About 97% of kindergartners (n = 1182) completed screening procedures following parental consent. Kindergarten teachers rated these children on 10 items describing aggressive and disruptive behaviors (e.g., gets in many fights, breaks rules) from the Authority Acceptance subscale of the Teacher Observation of Classroom Adaptation—Revised (Werthamer-Larsson, Kellam, & Wheeler, 1991), using a 6-point scale from “almost never” to “almost always” (˛ = .95). Children who scored in the top quartile of their class on the screening measure were targeted for recruitment into a “high aggression” sample (n = 304). Families of 68% of these targets consented to participate in the full study (n = 207); families of the other 32% of targets did not wish to participate (n = 38), could not be reached (n = 52), or were deemed ineligible (e.g., as a sibling of a participant; n = 7). There were no significant differences between those who did and did not consent to participate in the full study on aggression screen score, age at screening, gender, or ethnicity. The current sample included 207 children (66% males) and their primary caregivers (89% mothers, 3% fathers, 3% grandmothers, 2% other female relatives, 3% other). Children were African American (n = 151), Hispanic (n = 39), Caucasian (n = 16), and Asian (n = 1). On average, children were about six years old (M = 5.94, SD = 0.39 years) at the kindergarten (K) fall assessment and just over seven years old (M = 7.24, SD = 0.37 years) at the first grade (G1) spring assessment. Over 70% of families reported income levels that fell below the federal poverty line, about 34% of primary caregivers had not graduated from high school, and 61% were single parents. After informed consent was obtained from parents, teachers completed additional assessment measures including the Strengths and Difficulties Questionnaire (SDQ; Goodman, 1997). Based on national norms for this measure (Bourdon, Goodman, Rae, Simpson, & Koretz, 2005), 62% of the children in the sample exhibited medium to high levels of conduct problems (i.e., score of 3 or higher).

11

Measures Following screening procedures, research assistants visited families at home in the fall of K to interview parents and videorecord parent–child interactions (described below). They also visited schools and assessed each child, during which tests of academic knowledge and EF were administered. In addition, teachers completed measures on each student. These procedures were repeated again at the end of G1. Parents and teachers were compensated financially for their time, and children received stickers and small prizes. Parent support for learning Two measures were used to assess parent support for learning in K and G1. Nine items from the original 21-item Parent-Teacher Involvement Questionnaire (Kohl, Lengua, & McMahon, 2000) were selected based on measurement work from a previous project (Kohl et al., 2000) and were rated by teachers in the current study. One item (i.e., parent volunteers at school) showed weak correlations with the other eight items in both K and G1 and thus was dropped from further consideration. Of the remaining eight items, four tapped the parent–teacher relationship (i.e., parents interested in getting to know you; you feel you can be heard by child’s parents; you would feel comfortable talking to child’s parents about concerns or problems with child; child’s parents ask questions or make suggestions about child), and four tapped parent involvement and support for the child’s education (i.e., parents are involved in child’s education; education seems important to this family; parents encourage child’s positive attitude toward education through learning activities; parents and school have same goals for child). Confirmatory factor analyses indicated that this two-factor solution fit adequately in kindergarten, 2 (19, N = 180) = 85.08, p < .05; TLI = .94; CFI = .96; RMSEA = .12; SRMR = .04, and fit well in first grade, 2 (18, N = 163) = 26.48, p = .09; TLI = .99; CFI = .99; RMSEA = .05; SRMR = .03. Thus, the four items tapping parent school involvement and support for the child’s education were averaged to form a scale in K (˛ = .96) and G1 (˛ = .95). Second, the quality of parent teaching behaviors was coded from the videotaped parent–child interactions collected during the home visit. During the visit, parents engaged in three tasks with their children: (1) a book-reading task, in which parents were instructed to use a book with their child as they normally would; (2) a free-play task, in which parents were instructed to follow the child’s lead as they played with a castle and toys; and (3) a puzzle task, in which parents were instructed to help their child complete tangram puzzles (without touching the puzzle pieces themselves). Each task lasted approximately 5 min. Subsequently, these videotapes were coded in the laboratory by trained research assistants (n = 4; 75% female; 100% Caucasian) who were naïve concerning the hypotheses of the current study. Prior to coding study tapes, these assistants had participated in a 2-day training workshop and three weeks of practice coding until they attained a minimum of 80% agreement with a master coder. The master coder was a lead research assistant who had worked closely with the investigative team to establish the initial coding system. After watching each study tape twice (once to familiarize themselves with the interaction, once to code specific forms of language use), coders watched each tape a third time and rated various parenting behaviors using a 5-point scale. The master coder overlapped with assistants on 20% of the tapes to establish inter-rater reliability and prevent observer drift; any disagreements were resolved via discussion or the decision of the master coder. The coding process was monitored by a faculty investigator who held regular meetings with the master coder and research assistants. Three items assessed the quality of parent teaching behaviors (e.g., parent is very encouraging and provides information, explains

12

R.M. Abenavoli et al. / Early Childhood Research Quarterly 31 (2015) 9–18

or clarifies, or offers suggestions or ideas in a way that informs, supports, or expands the scope of child’s understanding or play; parent frequently makes direct attempts to teach new concepts, ideas, information, or vocabulary that may be new and challenging for the child; parent adopts a child-centered interactive style that is conversationally responsive to the child and maintains the topic that the child has initiated). These items were averaged across the three tasks, except the item “parent frequently makes direct attempts to teach,” which was averaged only across the book-reading and puzzle tasks because interrater reliability was poor during the free-play task. The three items then were averaged together to create a total score (K ˛ = .83; G1 ˛ = .86). Coders were within 1 scale point of the master coder 87% (K) to 91% (G1) of the time across these items, and the one-way random effects ICC for the full scale ranged from .66 (K) to .77 (G1). Teacher-rated parent school involvement was correlated with observed parent teaching behaviors, K r = .26 and G1 r = .24, ps < .05. In addition, both indices of support for learning showed moderate stability from K to G1, r = .46 for parent involvement and r = .37 for teaching behaviors, ps < .05. Child academic knowledge Child academic knowledge was assessed directly in K and G1 using two well-validated subtests from the Woodcock–Johnson III Tests of Achievement (Woodcock, McGrew, & Mather, 2001). Early literacy skills were assessed with the Letter-Word Identification subtest, in which children identified up to 76 letters and words of increasing difficulty. Early math skills were assessed with the Applied Problems subtest, in which children were asked to solve up to 39 counting, mathematical, and reasoning problems. Child EF Three tasks assessed child EF. First, working memory was assessed with the Backward Word Span (Davis & Pratt, 1996), in which children were asked to recall a string of up to six words they had just heard in reverse order. Scores were the number of words repeated in the correct backwards order. Secord, inhibitory control was assessed with the Peg Tapping task (Diamond & Taylor, 1996) in K and with the Head–Toes–Knees–Shoulders task (McClelland et al., 2007) in G1. In the Peg Tapping task, children were asked to tap a pencil once when the experimenter tapped twice and to tap twice when the experimenter tapped once. Scores were the proportion of correct trials out of 16 (˛ = .90). In the Head–Toes–Knees–Shoulders task, children were asked to touch their head when the experimenter touched his or her toes and to touch their toes when the experimenter touched his or her head. Similarly, children were asked to touch their knees when the experimenter touched shoulders, and to touch their shoulders when the experimenter touched knees. Scores were the proportion of correct trials out of 20 (˛ = .89). Finally, the research assistant who administered the assessments rated the degree to which the child showed strong task orientation and sustained engagement throughout the assessment (e.g., pays attention to instructions, sustains concentration). Scores were the average of 13 items rated on a 4-point scale (K ˛ = .92; G1 ˛ = .94). This measure, adapted from a previous study (Smith-Donald, Raver, Hayes, & Richardson, 2007), has been used as a proxy for EF in prior work (Bierman, Nix, Greenberg, Blair, & Domitrovich, 2008). Covariates Parents completed a short demographic questionnaire. Four family background items were included as covariates in each model to control for confounding effects. Income-to-needs was calculated as the family’s total income divided by the poverty threshold for a family of that size averaged across K and G1 (e.g., a ratio of .50 indicates that income is 50% of the U.S. poverty level, and a ratio of 2.00 indicates that income is 200% of the poverty level). Parent

education level was calculated as the years of schooling completed averaged across K and G1. Parent age was reported in the fall of K. Single parent status was dummy coded “1” if the respondent reported being unmarried and not in a committed relationship in both K and G1. In addition, five child-level items were included as covariates. Children’s aggression in the fall of K was assessed by teacher-report on seven items from the Teacher Observation of Classroom Adaptation–Revised (Werthamer-Larsson et al., 1991; ˛ = .89). Child gender (1 = male), age in K, and two dummy codes for race/ethnicity (1 = Black, 0 = other; 1 = Hispanic, 0 = other) were also included as covariates. Results Preliminary analyses Inspection of missing data patterns indicated that 183 of 207 total participants (88%) had provided some data (i.e., child direct assessment, teacher-reported school involvement, or observed teaching behaviors) in both K and G1, whereas 24 participants (12%) only provided data in K. There were few significant differences between those who did and did not provide data at both time points across a host of demographic, parent, and child behavior variables. Compared to participants with data in both grades, participants with data in K only had higher K Peg Tapping scores, t(55.69) = 3.75, p < .05, d = 0.59, and were less likely to be Caucasian, 2 (1, N = 207) = 6.54, p < .05, ϕ = 0.18. Full information maximum likelihood (FIML) was used to handle missing data in the path models described below. Descriptive statistics of the raw variables in the fall of K and the spring of G1 are presented in Table 1. Raw means are reported in the table; however, the inhibitory control and task orientation distributions in K and G1 were log-transformed to address negative skew due to ceiling effects, and the income-to-needs and parent age distributions were log-transformed to address positive skew. Correlations among the child school readiness measures are presented in Table 2. Literacy and math skills were strongly and significantly correlated in K, r = .58, and G1, r = .60, ps < .05, and were quite stable from K to G1, r = .75 for literacy and r = .69 for math, ps < .05. Measures of EF were moderately correlated in K, rs ranged from .34 to .36, and G1, rs ranged from .19 to .42, ps < .05. EF indices were moderately stable from K to G1, r = .35 for working memory, r = .24 for inhibitory control, and r = .35 for task orientation, ps < .05. In addition, academic knowledge and EF were moderately and significantly correlated at each grade level, K rs ranged from .21 to .46 and G1 rs ranged from .33 to .59, ps < .05. Simple correlations between the indices of parent support for learning and child school readiness are presented in Table 3. In K, parent school involvement was significantly associated with concurrent literacy skills and working memory, rs ranged from .16 to .26, ps < .05. In G1, school involvement was significantly correlated with concurrent literacy and math skills, rs ranged from .19 to .31, ps < .05. In K, nonsignificant trends suggested that teaching behaviors were marginally associated with concurrent working memory and inhibitory control, rs ranged from .13 to .15, ps < .10. In G1, teaching behaviors were significantly associated with concurrent literacy skills, math skills, inhibitory control, and task orientation, rs ranged from .17 to .24, ps < .05. A nonsignificant trend also suggested that G1 teaching behaviors were marginally associated with working memory, r = .14, p < .10. Longitudinal cross-lagged models Multiple group path analysis was utilized to estimate 10 longitudinal, cross-lagged models (two parent variables crossed with

R.M. Abenavoli et al. / Early Childhood Research Quarterly 31 (2015) 9–18

13

Table 1 Descriptive statistics for study measures in kindergarten and first grade. Variable

K

G1

n Support for learning School involvement Teaching behaviors Academic knowledge Literacy skills Math skills Executive functioning Working memory Inhibitory control Task orientation Covariates Income-to-needs Parent education Parent age Child aggression

M (SD)

Range

n

180 173

2.15 (1.19) 2.35 (0.65)

205 205

16.59 (6.95) 15.63 (4.75)

204 199 203

1.85 (0.76) 0.83 (0.23) 3.20 (0.62)

1–4 0.00–1.00 1.23–4.00

175 203 198 179

0.87 (0.65) 12.14 (1.31) 31.23 (8.06) 3.05 (1.07)

0.14–4.06 6–16 19.08–60.75 1.00–5.71

M (SD)

0.00–4.00 1.00–4.22

163 152

2.01 (1.17) 2.23 (0.68)

1–49 0–29

164 164

30.97 (8.93) 21.96 (4.87)

164 164 164

2.16 (0.76) 0.61 (0.27) 3.05 (0.71)

Range 0.00–4.00 1.00–4.06 8–59 7–35 1–4 0.00–1.00 1.00–4.00

Note. Raw scores are shown for inhibitory control, task orientation, income-to-needs, and parent age; however, log-transformed scores were used in analyses. K, kindergarten; G1, first grade. Relative to females, males had significantly higher K aggression, t(177) = 2.64, d = 0.41, as well as significantly lower K literacy skills, t(203) = 2.85, d = 0.42, K math skills, t(178.12) = 2.05, d = 0.29, K working memory, t(202) = 3.11, d = 0.45, G1 literacy skills, t(162) = 4.28, d = 0.77, G1 inhibitory control, t(162) = 2.43, d = 0.43, and G1 task orientation, t(162) = 2.88, d = 0.50, ps < .05.

Table 2 Correlations among measures of child school readiness. Measures Academic knowledge K 1. Literacy skills 2. Math skills Academic knowledge G1 3. Literacy skills 4. Math skills Executive function K 5. Working memory 6. Inhibitory control 7. Task orientation Executive function G1 8. Working memory 9. Inhibitory control 10. Task orientation

1.

2.

3.

4.

–

5.

6.

7.

–

8.

9.

10.

– .42* .19*

– .40*

–

– .58*

–

.75* .55*

.49* .69*

– .60*

.46* .21* .31*

.45* .33* .43*

.48* .31* .31*

.45* .38* .34*

– .36* .34*

– .36*

.37* .40* .27*

.37* .43* .24*

.50* .45* .33*

.42* .59* .37*

.35* .33* .16*

.21* .24* .16*

.20* .37* .35*

Note. K, kindergarten; G1, first grade. * p < .05. Table 3 Correlations between parent support for learning and child school readiness. School involvement K Academic knowledge K Literacy skills Math skills Academic knowledge G1 Literacy skills Math skills Executive function K Working memory Inhibitory control Task orientation Executive function G1 Working memory Inhibitory control Task orientation

Teaching behaviors K

School involvement G1

Teaching behaviors G1

.26* .08

.04 .05

.21* .16*

.13 .15

.32* .19*

.12 .13

.31* .19*

.20* .24*

.16* .12 .01

.13 .15 −.02

.01 .14 .10

.13 .12 .05

.20* .02 .14

.10 .15 .13

.14 .06 .09

.14 .17* .23*

Note. K, kindergarten; G1, first grade. * p < .05.

five child variables) testing the associations between parent support for learning and child school readiness across K and G1. These cross-lagged models simultaneously tested bidirectional influences from parent to child and child to parent, accounting for the stability in parent support for learning and child school readiness. To control for possible confounding effects, the four variables of interest in each model (parent support for learning and child

school readiness both in K and G1) were each regressed on nine covariates (i.e., income-to-needs, parent education level, parent age, single parent status, child aggression, gender, age, and two dummy codes for race/ethnicity). Covariances among the covariates were also modeled. The general hypothetical cross-lagged model is presented in Fig. 1. All path analyses were conducted with Mplus 7.

14

R.M. Abenavoli et al. / Early Childhood Research Quarterly 31 (2015) 9–18

Fig. 1. K, kindergarten; G1, first grade. Hypothetical longitudinal cross-lagged model tested in 10 sets of analyses (two indicators of parent support for learning crossed with five indicators of child school readiness). Paths a and b represent the stability of parent support for learning and child school readiness, respectively, from K to G1. Path c represents the concurrent association between parent support for learning and child school readiness in the fall of K. Path d represents the extent to which parent support for learning in K predicts change in child school readiness from K to G1, and path e represents the extent to which child school readiness predicts change in parent support for learning. Nine covariates were controlled in all models but are not depicted here: Income-to-needs, parent education level, parent age, single parent status, child aggression, gender, age, and two race/ethnicity dummy codes (Black, Hispanic).

Because the sample included participants who had been randomly assigned to intervention (n = 100; 65% male) or control (n = 107; 67% male) conditions following the K assessment, a series of sub-models was estimated to determine model equivalence across groups and improve model fit for each of the 10 path models. It is important to account for intervention status in these analyses because the intervention could have affected the strength of associations between parenting and school readiness variables. Rather than enter intervention status simply as another covariate, multiple group path analysis was used to test model equivalence across intervention and control groups and examine whether specific associations were or were not invariant across groups. In each set of models, we first estimated an unconstrained model in which all parameters were freely estimated in intervention and control groups. Second, we estimated a model in which the paths of substantive interest (i.e., paths a, b, c, d, and e in Fig. 1) were constrained to be equal across intervention and control groups. As shown in Table 4, the chi-square difference between this constrained model and the unconstrained model was not significant for any of the 10 models, indicating invariance across intervention conditions in the paths of substantive interest. Third, we estimated a further constrained model in which all paths (i.e., the paths of substantive interest as well as paths involving covariates) were constrained to be equal across intervention and control groups. Across all 10 models, modification indices consistently indicated that freeing equality constraints on the following four parameters would improve model fit: (1) the covariance between income-toneeds and parent age, (2) the covariance between income-to-needs and child gender, (3) the covariance between single parent status and child age, and (4) the beta for first grade support for learning (school involvement or teaching behavior) regressed on the Hispanic dummy code. Thus, all 10 final models allowed these four estimates to vary across intervention and control groups. All other paths were constrained to be equal across groups. As shown in Table 4, the chi-square difference between this final constrained model and the previous model constraining only paths of substantive interest was not significant for any of the 10 models, indicating invariance across intervention conditions in all paths (except the four parameters listed above). Table 5 presents the results of the 10 final models.

Literacy skills The fit of the final school involvement model was good, 2 (75, N = 207) = 62.98, p = 0.84; TLI = 1.06; CFI = 1.00; RMSEA = 0.00; SRMR = 0.07. Parent school involvement in K was significantly associated with children’s concurrent literacy skills, r = .26, p < .05. In addition, parent school involvement in K significantly predicted growth (i.e., residual change) in child literacy skills from K to G1, ˇ = .13, p < .05, whereas child literacy skills in K did not predict change in parent school involvement, ˇ = .06, p = .41. The fit of the final teaching behavior model also was good, 2 (75, N = 207) = 66.54, p = 0.75; TLI = 1.04; CFI = 1.00; RMSEA = 0.00; SRMR = 0.07. Parent teaching behaviors in K were not significantly associated with children’s concurrent literacy skills, r = −.01, p = .87. However, parent teaching behaviors in K significantly predicted growth in child literacy skills from K to G1, ˇ = .11, p < .05, whereas child literacy skills in K did not predict change in parent teaching behaviors, ˇ = .08, p = .24. These results suggest that both forms of parent support for learning influenced gains in child literacy skills. Math skills The fit of the final school involvement model was good, 2 (75, N = 207) = 72.50, p = 0.56; TLI = 1.01; CFI = 1.00; RMSEA = 0.00; SRMR = 0.07. Parent school involvement in K was not significantly associated with children’s concurrent math skills, r = .10, p = .19. However, parent school involvement in K significantly predicted growth in child math skills from K to G1, ˇ = .18, p < .05, whereas child math skills in K did not predict change in parent school involvement, ˇ = .08, p = .29. The fit of the final teaching behavior model also was good, 2 (75, N = 207) = 76.14, p = 0.44; TLI = 0.99; CFI = 0.99; RMSEA = 0.01; SRMR = 0.08. Parent teaching behaviors in K were not significantly associated with children’s concurrent math skills, r = .00, p = .97. However, parent teaching behaviors in K significantly predicted growth in child math skills from K to G1, ˇ = .14, p < .05, whereas child math skills in K did not predict change in parent teaching behaviors, ˇ = .08, p = .22. These results suggest that both forms of parent support for learning influenced gains in child math skills. Working memory The fit of the final school involvement model was good, 2 (75, N = 207) = 71.34, p = 0.60; TLI = 1.04; CFI = 1.00; RMSEA = 0.00; SRMR = 0.07. Parent school involvement in K was significantly associated with children’s concurrent working memory, r = .16, p < .05. In addition, parent school involvement in K significantly predicted growth (i.e., residual change) in child working memory from K to G1, ˇ = .17, p < .05, whereas child working memory in K did not predict change in parent school involvement, ˇ = −.01, p = .85. The fit of the final teaching behavior model also was good, 2 (75, N = 207) = 74.29, p = 0.50; TLI = 1.01; CFI = 1.00; RMSEA = 0.00; SRMR = 0.08. Parent teaching behaviors in K were not significantly associated with children’s concurrent working memory, r = .08, p = .33. Parent teaching behaviors in K did not significantly predict growth in child working memory from K to G1, ˇ = .07, p = .38, nor did child working memory in K predict change in parent teaching behaviors, ˇ = .05, p = .38. These results suggest that parent school involvement, but not parent teaching behaviors, influenced gains in child working memory. Inhibitory control The fit of the final school involvement model was good, 2 (75, N = 207) = 69.68, p = 0.65; TLI = 1.07; CFI = 1.00; RMSEA = 0.00; SRMR = 0.07. Parent school involvement in K was not significantly associated with children’s concurrent inhibitory control, r = .08, p = .28. In addition, parent school involvement in K did not predict growth in child inhibitory control from K to G1, ˇ = .03, p = .70, nor

R.M. Abenavoli et al. / Early Childhood Research Quarterly 31 (2015) 9–18

15

Table 4 Chi-square difference tests of model invariance. School involvement models

Literacy skills 2 Difference test 1 2 Difference test 2 Math skills 2 Difference test 1 2 Difference test 2 Working memory 2 Difference test 1 2 Difference test 2 Inhibitory control 2 Difference test 1 2 Difference test 2 Task orientation 2 Difference test 1 2 Difference test 2

Teaching behavior models

2

df

p

2

df

p

2.98 58.36

5 68

0.70 0.79

1.80 62.52

5 68

0.88 0.66

8.44 61.82

5 68

0.13 0.69

5.75 67.79

5 68

0.33 0.48

8.87 61.58

5 68

0.11 0.70

5.78 66.14

5 68

0.33 0.54

4.65 65.00

5 68

0.46 0.58

9.65 70.45

5 68

0.09 0.40

10.83 64.50

5 68

0.05 0.60

6.22 71.43

5 68

0.29 0.36

Note. The 2 Difference test 1 tested whether the unconstrained model fit significantly better than the model constraining paths of substantive interest to be equal across intervention and control groups. The 2 Difference test 2 tested whether the model constraining paths of substantive interest to be equal across groups fit significantly better than the final constrained model in which all paths were constrained to be equal except the: (1) covariance between income-to-needs and parent age, (2) covariance between income-to-needs and child gender, (3) covariance between single parent status and child age, and (4) beta for first grade support for learning (school involvement or teaching behavior) regressed on the Hispanic dummy code.

Table 5 Results from final constrained longitudinal cross-lagged path models.

Literacy skills School involvement Teaching behavior Math skills School involvement Teaching behavior Working memory School involvement Teaching behavior Inhibitory control School involvement Teaching behavior Task orientation School involvement Teaching behavior

a. SFL K → SFL G1

b. SR K → SR G1

ˇ

(SE)

ˇ

(SE)

r

.41* .27*

(.07) (.06)

.74* .77*

(.05) (.05)

.41* .27*

(.07) (.06)

.67* .71*

.41* .27*

(.07) (.06)

.41* .27* .41* .28*

c. SFL K ↔ SR K

d. SFL K → SR G1

e. SR K → SFL G1

(SE)

ˇ

(SE)

ˇ

.26* −.01

(.07) (.08)

.13* .11*

(.06) (.05)

.06 .08

(.07) (.07)

(.05) (.05)

.10 .00

(.07) (.08)

.18* .14*

(.06) (.06)

.08 .08

(.07) (.07)

.25* .28*

(.07) (.07)

.16* .08

(.08) (.09)

.17* .07

(.08) (.08)

−.01 .05

(.07) (.06)

(.07) (.06)

.18* .17*

(.07) (.07)

.08 .08

(.08) (.08)

.03 .11

(.07) (.07)

.07 .04

(.07) (.07)

(.07) (.06)

.34* .36*

(.08) (.08)

−.03 −.06

(.07) (.08)

.18* .18*

(.08) (.07)

.11 .05

(.08) (.07)

(SE)

Note. Each row represents the results of one path model. Letters in column headings correspond to paths depicted in Fig. 1. SFL, support for Learning; SR, school readiness; K, kindergarten; G1, first grade. Standardized coefficients from control group shown above. Parameter constraints were placed on unstandardized parameters, so standardized parameters sometimes differed between intervention and control groups by trivial amounts, which never affected interpretation. * p < .05.

did child inhibitory control in K predict change in parent school involvement, ˇ = .07, p = .33. The fit of the final teaching behavior model also was good, 2 (75, N = 207) = 84.20, p = 0.22; TLI = 0.88; CFI = 0.90; RMSEA = 0.03; SRMR = 0.08. Parent teaching behaviors in K were not significantly associated with children’s concurrent inhibitory control, r = .08, p = .29. Parent teaching behaviors in K did not predict growth in child inhibitory control from K to G1, ˇ = .11, p = .13, nor did child inhibitory control in K predict change in parent teaching behaviors, ˇ = .04, p = .54.

TLI = 0.93; CFI = 0.94; RMSEA = 0.03; SRMR = 0.08. Parent teaching behaviors in K were not associated with children’s concurrent task orientation, r = −.06, p = .45. However, parent teaching behaviors in K significantly predicted growth in child task orientation from K to G1, ˇ = .18, p < .05, whereas child task orientation in K did not predict change in parent teaching behaviors, ˇ = .05, p = .45. These results suggest that both forms of parent support for learning influenced gains in child task orientation. Discussion

Task orientation The fit of the final school involvement model was good, 2 (75, N = 207) = 75.82, p = 0.45; TLI = 0.99; CFI = 0.99; RMSEA = 0.01; SRMR = 0.08. Parent school involvement in K was not significantly associated with children’s concurrent task orientation, r = −.03, p = .68. However, parent school involvement in K significantly predicted growth in child task orientation from K to G1, ˇ = .18, p < .05, whereas child task orientation in K did not predict change in parent school involvement, ˇ = .11, p = .17. The fit of the final teaching behavior model also was good, 2 (75, N = 207) = 83.07, p = 0.24;

Parent support for learning has been identified in prior research as a mechanism through which macro-level risk factors, such as family poverty and parental education, influence children’s school readiness and later achievement (Brooks-Gunn & Markman, 2005; Burchinal et al., 2008; Linver, Brooks-Gunn, & Kohen, 2002). The current study explored variation in parent support for learning within a high-risk, low-income sample of urban, disadvantaged families with children showing elevated aggressive–disruptive behavior in kindergarten. Not only were families in this

16

R.M. Abenavoli et al. / Early Childhood Research Quarterly 31 (2015) 9–18

sample low in income (over 70% below poverty), but they also were primarily single parents (61%) with low educational attainment (34% high school dropouts), and they were living in high-crime neighborhoods where schools were characterized by low levels of student achievement. Understanding variation within this kind of high-risk sample may shed light on specific processes that may promote positive outcomes among children who are at the highest risk for academic failure (Mistry et al., 2010). Indeed, results of the current study suggest that parent support for learning as assessed by both observation of parent–child interaction as well as by teacher report of parent school involvement fosters child academic knowledge and EF, even among very disadvantaged families with children exhibiting elevated aggressive–disruptive behavior problems. Identifying these associations in a uniquely high-risk sample extends prior work that has documented links between parent support for learning and both academic knowledge and EF in other lower risk populations. A strength of this study was the use of cross-lagged models to examine bidirectional associations between independent indices of parent support for learning and child school readiness. This approach provides a more rigorous test of associations over time than simple predictive models because it reduces the chance that omitted variables account for the observed associations over time. That is, it reduces inflated estimates that may occur in crosssectional or longitudinal models that do not account for the initial and co-varying associations among parent and child variables over time (Cole & Maxwell, 2003; MacCallum & Austin, 2000). The role of parent support for learning in a high-risk sample This study explored the degree to which parent support for learning was related concurrently to academic knowledge and EF among high-risk, aggressive children at the start of kindergarten, and using cross-lagged analyses, it also examined predictive relations between parent support for learning and children’s skills from the fall of kindergarten to the spring of first grade. Interestingly, parent support for learning was not consistently associated with child academic knowledge or EF at kindergarten entry: Across the 10 models, there were only two significant concurrent associations between support for learning and child school readiness in kindergarten (i.e., school involvement with literacy skills and working memory). In addition, child school readiness in kindergarten did not predict change in parent support for learning in any of the 10 models. However, parent support for learning significantly predicted growth in child school readiness from kindergarten to first grade in seven of the 10 models. Specifically, both forms of support for learning measured (teacher-rated school involvement and observer-rated teaching behaviors) significantly predicted growth in children’s academic knowledge (i.e., both literacy and math skills). Support for learning also predicted growth in children’s EF, although less consistently across measures: Both teacher-rated school involvement and observed teaching behaviors predicted growth in children’s task orientation, and school involvement predicted growth in working memory. Neither indicator of support for learning significantly predicted growth in inhibitory control. The fact that inhibitory control was indexed by different measures in kindergarten and first grade may have added measurement error and thereby attenuated this association. In models where parent support for learning significantly predicted growth in child school readiness, these longitudinal associations generally were small (ˇs ranged from .11 to .18). Still, these associations are notable given that the models accounted for the high stability of child academic knowledge and EF as well as key covariates associated with urban poverty. Because early literacy, math, and EF skills are highly predictive of later academic success (Blair & Razza, 2007; Duncan et al., 2007; Welsh et al., 2010), these results underscore the importance

of parents’ early support for their children’s learning as a protective factor among children at very high risk for school difficulties. It was surprising that parent support for learning in kindergarten generally was not significantly associated with concurrent academic knowledge or EF, yet predicted growth between kindergarten and first grade. Typically, correlations between variables are strongest when they are measured concurrently and become weaker as the lag between measurements increases. Although the reason for this remains unclear, perhaps there was more “developmental noise” at the start of kindergarten in the measures of academic knowledge and EF. Children varied in their participation in Head Start or other preschool programs prior to kindergarten and were likely maturing at different rates, and these factors may have influenced learning in ways that masked parent contributions. It is possible that this noise was partially reduced as children matured and experienced similar school settings, allowing the effect of parent support for learning to be detected more readily. These reasons are speculative and further research in high-risk populations is warranted. Limitations A few limitations should be noted. This study examined parent support for learning in a unique, high-risk sample because identifying and promoting protective mechanisms within these families can have profound effects on children’s developmental trajectories. The tradeoff, though, is that this sample limits the generalizability of these findings. However, it is worth noting that longitudinal associations documented here are consistent with prior studies in more normative samples. Second, although the longitudinal design and analytic approach enabled a more rigorous test of causal influence than simple longitudinal predictions, cross-lagged designs are not without limitations (Rogosa, 1980). Causal conclusions are still speculative because the data were non-experimental (Selig & Little, 2012). Although covariates such as income-to-needs, parent education level, parent age, and single parent status were controlled in the models, it is possible that the associations observed were caused by an omitted variable, such as parent IQ, shared genes between parents and children, or correlated parenting behaviors like sensitivity or discipline practices. For example, parent school involvement and parent teaching behaviors may be proxies for parent support more generally. In future work, this limitation could be addressed by applying stronger causal inference methods, such as propensity score analyses (Stuart & Rubin, 2007), or by specifically manipulating parent support for learning (e.g., through an intervention targeting school involvement and teaching behaviors), controlling for changes in more general parent warmth and support (e.g., through an active control condition targeting these skills). A third limitation concerns the measurement of parent support for learning. Although the reliability of the teacher-reported school involvement measure was high, the inter-rater reliability of the observer-rated teaching behavior measure was moderate, particularly in the fall of kindergarten. Given the relatively low ICCs, the measure of teaching behaviors may have been subject to more measurement error, and this could have attenuated the associations. It should be noted, however, that results generally were similar across models using observer-rated teaching behaviors or teacher-rated school involvement as the indicator of support for learning, which increases confidence in these findings. Another limitation is that an intervention occurred during the period in which this longitudinal study was conducted. The intervention did not directly target or affect the variables studied here, and tests of equivalence indicated that associations between support for learning and school readiness—the focus of the current study—were equivalent for children in the intervention and control

R.M. Abenavoli et al. / Early Childhood Research Quarterly 31 (2015) 9–18

groups. Hence, the likelihood that the intervention affected these results is very slim. Conclusion Our findings suggest that parent support for learning promotes children’s academic knowledge and EF in the first years of schooling and serves as a protective factor for children affected by high levels of contextual risk and behavioral risk. This is promising news. Low income, low levels of parent education, and other correlated macro-level risk factors can have extreme, long-lasting negative consequences for families and children (Duncan et al., 1994; McLoyd, 1998; NICHD ECCRN, 2005), but they may not be so overpowering that they completely obstruct parents’ ability to support their children’s learning. The results suggest that support for learning may be one promising point of entry for preventive intervention, as there may be opportunities to promote the quantity and quality of support for learning that parents are able to provide to their children and thereby enhance child school readiness, even in contexts of very high risk. Acknowledgements Funding for this project was provided by a grant from the Pennsylvania Department of Health. The work of the first author was also supported by grant R305B090007 from the Institute of Education Sciences. The views expressed are the authors’ and do not represent the granting agencies. The authors thank Robert Nix, Michael Coccia, Jennifer Ford, and the PATHS to Success research assistants for their hard work in managing the data collection and data processing for this project, as well as the teachers, parents, and children who made this study possible. References Arnold, D. H., Zeljo, A., Doctoroff, G. L., & Ortiz, C. (2008). Parent involvement in preschool: Predictors and the relation of involvement to preliteracy development. School Psychology Review, 37, 74–90. Ayoub, C., Vallotton, C. D., & Mastergeorge, A. M. (2011). Developmental pathways to integrated social skills: The roles of parenting and early intervention. Child Development, 82, 583–600. http://dx.doi.org/10.1111/j.1467-8624.2010.01549.x Bernier, A., Carlson, S. M., & Whipple, N. (2010). From external regulation to self-regulation: Early parenting precursors of young children’s executive functioning. Child Development, 81, 326–339. http://dx.doi.org/10.1111/j.1467-8624.2009.01397.x Bierman, K. L., Nix, R. L., Greenberg, M. T., Blair, C., & Domitrovich, C. E. (2008). Executive functions and school readiness intervention: Impact, moderation, and mediation in the Head Start REDI Program. Development and Psychopathology, 20, 821–843. http://dx.doi.org/10.1017/S0954579408000394 Bierman, K. L., Torres, M. M., Domitrovich, C. E., Welsh, J. A., & Gest, S. D. (2009). Behavioral and cognitive readiness for school: Cross-domain associations for children attending Head Start. Social Development, 18, 305–323. http://dx.doi.org/10.1111/j.1467-9507.2008.00490.x Blair, C. (2002). School readiness: Integrating cognition and emotion in a neurobiological conceptualization of children’s functioning at school entry. American Psychologist, 57, 111–127. http://dx.doi.org/10.1037/0003-066X.57.2.111 Blair, C., Granger, D. A., Willoughby, M., Mills-Koonce, R., Cox, M., Greenberg, M. T., et al. (2011). Salivary cortisol mediates effects of poverty and parenting on executive functions in early childhood. Child Development, 82, 1970–1984. http://dx.doi.org/10.1111/j.1467-8624.2011.01643.x Blair, C., & Razza, R. P. (2007). Relating effortful control, executive function, and false belief understanding to emerging math and literacy ability in kindergarten. Child Development, 78, 647–663. http://dx.doi.org/10.1111/j.1467-8624.2007.01019.x Bourdon, K. H., Goodman, R., Rae, D. S., Simpson, G., & Koretz, D. S. (2005). The Strengths and Difficulties Questionnaire: U.S. normative data and psychometric properties. Journal of the American Academy of Child and Adolescent Psychiatry, 44, 557–564. http://dx.doi.org/10.1097/01.chi.0000159157.57075.c8 Bradley, R. H., Corwyn, R. F., Burchinal, M., McAdoo, H. P., & Garcia Coll, C. (2001). The home environments of children in the United States Part II: Relations with behavioral development through age thirteen. Child Development, 72, 1868–1886. http://dx.doi.org/10.1111/1467-8624.t01-1-00383 Brooks-Gunn, J., & Markman, L. B. (2005). The contribution of parenting to ethnic and racial gaps in school readiness. The Future of Children, 15, 139–168. http://dx.doi.org/10.1353/foc.2005.0001

17

Burchinal, M., Vernon-Feagans, L., Cox, M., & Key Family Life Project Investigators. (2008). Cumulative social risk, parenting, and infant development in rural low-income communities. Parenting: Science and Practice, 8, 41–69. http://dx.doi.org/10.1080/15295190701830672 Chazan-Cohen, R., Raikes, H., Brooks-Gunn, J., Ayoub, C., Pan, B. A., Kisker, E. E., et al. (2009). Low-income children’s school readiness: Parent contributions over the first five years. Early Education & Development, 20, 958–977. http://dx.doi.org/10.1080/10409280903362402 Cole, D. A., & Maxwell, S. E. (2003). Testing mediational models with longitudinal data: Questions and tips in the use of Structural Equation Modeling. Journal of Abnormal Psychology, 112, 558–577. http://dx.doi.org/10.1037/0021-843X.112.4.558 Conduct Problems Prevention Research Group. (1999). Initial impact of the Fast Track prevention trial for conduct problems: I. The highrisk sample. Journal of Consulting and Clinical Psychology, 67, 631–647. http://dx.doi.org/10.1037/0022-006X.67.5.631 Davis, H. L., & Pratt, C. (1996). The development of children’s theory of mind: The working memory explanation. Australian Journal of Psychology, 47, 25–31. http://dx.doi.org/10.1080/00049539508258765 DeBell, M. (2008). Children living without their fathers: Population estimates and indicators of educational well-being. Social Indicators Research, 87, 427–443. http://dx.doi.org/10.1007/s11205-007-9149-8 Diamond, A., & Taylor, C. (1996). Development of an aspect of executive control: Development of the abilities to remember what I said and to do as I say, not as I do. Developmental Psychobiology, 29, 315–334. http://dx.doi.org/10.1002/(SICI)1098-2302(199605)29:4<315::AID-DEV2> 3.3.CO;2-C Dilworth-Bart, J. E. (2012). Does executive function mediate SES and home quality associations with academic readiness? Early Childhood Research Quarterly, 27, 416–425. http://dx.doi.org/10.1016/j.ecresq.2012.02.002 Dodge, K. A., Pettit, G. S., & Bates, J. E. (1994). Socialization mediators of the relation between socioeconomic status and child conduct problems. Child Development, 65, 649–665. http://dx.doi.org/10.2307/1131407 Dodici, B. J., Draper, D. C., & Peterson, C. A. (2003). Early parent–child interactions and early literacy development. Topics in Early Childhood Special Education, 23, 124–136. http://dx.doi.org/10.1177/02711214030230030301 Duncan, G. J., Brooks-Gunn, J., & Klebanov, P. K. (1994). Economic deprivation and early childhood development. Child Development, 65, 296–318. http://dx.doi.org/10.2307/1131385 Duncan, G. J., Dowsett, C. J., Claessens, A., Magnuson, K., Huston, A. C., Klebanov, P., et al. (2007). School readiness and later achievement. Developmental Psychology, 43, 1428–1446. http://dx.doi.org/10.1037/0012-1649.43.6.1428 Fantuzzo, J., McWayne, C., Perry, M. A., & Childs, S. (2004). Multiple dimensions of family involvement and their relations to behavioral and learning competencies for urban low-income children. School Psychology Review, 33, 467– 480. Galindo, C., & Sheldon, S. B. (2012). School and home connections and children’s kindergarten achievement gains: The mediating role of family involvement. Early Childhood Research Quarterly, 27, 90–103. http://dx.doi.org/10.1016/j.ecresq.2011.05.004 Gershoff, E. T., Aber, J. L., Raver, C. C., & Lennon, M. C. (2007). Income is not enough: Incorporating material hardship into models of income associations with parenting and child development. Child Development, 78, 70–95. http://dx.doi.org/10.1111/j.1467-8624.2007.00986.x Greenberg, M. T., Bierman, K. L., Nix, R. L., & Gatzke-Kopp, L. M. (2012). PATHS to success: Final report to Department of Health, State of Pennsylvania. University Park, PA: Prevention Research Center, Pennsylvania State University. Retrieved from. http://www.prevention.psu.edu/paths to success project.html Goodman, R. (1997). The Strengths and Difficulties Questionnaire: A research note. Journal of Child Psychology and Psychiatry, 38, 581–586. http://dx.doi.org/10.1111/j.1469-7610.1997.tb01545.x Hammond, S. I., Müller, U., Carpendale, J. I. M., Bibok, M. B., & LiebermannFinestone, D. P. (2012). The effects of parental scaffolding on preschoolers’ executive function. Developmental Psychology, 48, 271–281. http://dx.doi.org/10.1037/a0025519 Hughes, C. H., & Ensor, R. A. (2009). How do families help or hinder the emergence of early executive function? New Directions in Child and Adolescent Development, 123, 35–50. http://dx.doi.org/10.1002/cd.234 Hughes, C., White, A., Sharpen, J., & Dunn, J. (2000). Antisocial, angry, and unsympathetic: Hard-to-manage preschoolers’ peer problems and possible cognitive influence. Journal of Child Psychology and Psychiatry, 41, 169–179. http://dx.doi.org/10.1111/1469-7610.00558 Kohl, G. O., Lengua, L. L., & McMahon, R. J. (2000). Parent involvement in school: Conceptualizing multiple dimensions and their relations with family and demographic risk factors. Journal of School Psychology, 38, 501–523. http://dx.doi.org/10.1016/S0022-4405(00)00050-9 La Paro, K. M., & Pianta, R. C. (2000). Predicting children’s competence in the early school years: A meta-analytic review. Review of Educational Research, 70, 443–484. http://dx.doi.org/10.3102/00346543070004443 Lee, V. E., & Burkam, D. T. (2002). Inequality at the starting gate: Social background differences in achievement as children begin school. Washington, DC: Economic Policy Institute. Lengua, L. J., Honorado, E., & Bush, N. R. (2007). Contextual risk and parenting as predictors of effortful control and social competence in preschool children. Journal of Applied Developmental Psychology, 28, 40–55. http://dx.doi.org/10.1016/j.appdev.2006.10.001

18

R.M. Abenavoli et al. / Early Childhood Research Quarterly 31 (2015) 9–18

Linver, M. R., Brooks-Gunn, J., & Kohen, D. E. (2002). Family processes as pathways from income to young children’s development. Developmental Psychology, 38, 719–734. http://dx.doi.org/10.1037/0012-1649.38.5.719 MacCallum, R. C., & Austin, J. T. (2000). Applications of Structural Equation Modeling in psychological research. Annual Review of Psychology, 51, 201–226. http://dx.doi.org/10.1146/annurev.psych.51.1.201 Magnuson, K. A., Sexton, H. R., Davis-Kean, P. E., & Huston, A. C. (2009). Increases in maternal education and young children’s language skills. Merrill-Palmer Quarterly, 55, 319–350. http://dx.doi.org/10.1353/mpq.0.0024 McClelland, M. M., Cameron, C. E., Connon, C. M., Farris, C. L., Jewkes, A. M., & Morrison, F. J. (2007). Links between behavioral regulation and preschoolers’ literacy, vocabulary, and math skills. Developmental Psychology, 43, 947–959. http://dx.doi.org/10.1037/0012-1649.43.4.947 McLoyd, V. C. (1998). Socioeconomic disadvantage and child development. American Psychologist, 53, 185–204. http://dx.doi.org/10.1037/0003-066X.53.2.185 McWayne, C., Hampton, V., Fantuzzo, J., Cohen, H. L., & Sekino, Y. (2004). A multivariate examination of parent involvement and the social and academic competencies of urban kindergarten children. Psychology in the Schools, 41, 363–377. http://dx.doi.org/10.1002/pits.10163 Mistry, R. S., Benner, A. D., Biesanz, J. C., & Clark, S. L. (2010). Family and social risk, and parental investments during the early childhood years as predictors of low-income children’s school readiness outcomes. Early Childhood Research Quarterly, 25, 432–449. http://dx.doi.org/10.1016/j.ecresq.2010.01.002 Mulvaney, M. K., McCartney, K., Bub, K. L., & Marshall, N. L. (2006). Determinants of dyadic scaffolding and cognitive outcomes in first graders. Parenting: Science and Practice, 6, 297–320. http://dx.doi.org/10.1207/s15327922par0604 2 National Institute of Child Health and Human Development Early Child Care Research Network. (2004). Trajectories of physical aggression from toddlerhood to middle childhood. Monographs of the Society for Research in Child Development, 69, 1–119. http://dx.doi.org/10.1111/j.1540-5834.2004.00311.x National Institute of Child Health and Human Development Early Child Care Research Network. (2005). Duration and developmental timing of poverty and children’s cognitive and social development from birth through third grade. Child Development, 76, 795–810. http://dx.doi.org/10.1111/j.1467-8624.2005.00878.x Pianta, R. C., Smith, N., & Reeve, R. E. (1991). Observing mother and child behavior in a problem-solving situation at school entry: Relations with classroom adjustment. School Psychology Quarterly, 6, 1–15. http://dx.doi.org/10.1037/h0088238 Powell, D. R., Son, S., File, N., & San Juan, R. R. (2010). Parent–school relationships and children’s academic and social outcomes in

public school pre-kindergarten. Journal of School Psychology, 48, 269–292. http://dx.doi.org/10.1016/j.jsp.2010.03.002 Raver, C. C., Blair, C., Willoughby, M., & the Family Life Project Key Investigators. (2013). Poverty as a predictor of 4-year-olds’ executive function: New perspectives on models of differential susceptibility. Developmental Psychology, 49, 292–304. http://dx.doi.org/10.1037/a0028343 Rogosa, D. (1980). A critique of cross-lagged correlation. Psychological Bulletin, 88, 245–258. http://dx.doi.org/10.1037/0033-2909.88.2.245 Ryan, R. M., Fauth, R. C., & Brooks-Gunn, J. (2006). Childhood poverty: Implications for school readiness and early childhood education. In B. Spodek, & O. N. Saracho (Eds.), Handbook of research on the education of children (2nd ed., pp. 323–346). Mahwah, NJ: Erlbaum. Sarsour, K., Sheridan, M., Jutte, D., Nuru-Jeter, A., Hinshaw, S., & Boyce, W. T. (2011). Family socioeconomic status and child executive functions: The roles of language, home environment, and single parenthood. Journal of the International Neuropsychological Society, 17, 120–132. http://dx.doi.org/10.1017/S1355617710001335 Selig, J. P., & Little, T. D. (2012). Autoregressive and cross-lagged panel analysis for longitudinal data. In B. Laursen, T. D. Little, & N. A. Card (Eds.), Handbook of developmental research methods (pp. 265–278). New York, NY: The Guildford Press. Smith-Donald, R., Raver, C. C., Hayes, T., & Richardson, B. (2007). Preliminary construct and concurrent validity of the Preschool Self-Regulation Assessment (PSRA) for field-based research. Early Childhood Research Quarterly, 22, 173–187. http://dx.doi.org/10.1016/j.ecresq.2007.01.002 Stuart, E. A., & Rubin, D. B. (2007). Best practices in quasi-experimental designs: Matching methods for causal inference. In J. Osborne (Ed.), Best practices in quantitative social science (pp. 155–176). Thousand Oaks, CA: Sage Publications. Welsh, J. A., Nix, R. L., Blair, C., Bierman, K. L., & Nelson, K. E. (2010). The development of cognitive skills and gains in academic school readiness for children from low-income families. Journal of Educational Psychology, 102, 43–53. http://dx.doi.org/10.1037/a0016738 Werthamer-Larsson, L., Kellam, S., & Wheeler, L. (1991). Effect of first-grade classroom environment on shy behavior, aggressive behavior, and concentration problems. American Journal of Community Psychology, 19, 585–602. http://dx.doi.org/10.1007/BF00937993 Woodcock, R. W., McGrew, K. S., & Mather, N. (2001). Woodcock–Johnson III tests of achievement. Itasca, IL: Riverside.