University of Nebraska - Lincoln

DigitalCommons@University of Nebraska - Lincoln
Open Access Theses and Dissertations from the College of Education and Human Sciences 11-16-2012 Education and Human Sciences, College of (CEHS)

Family Environment and School Environment as Predictors for Physical Aggression in Low-Income Children
Xiaoyu Li
University of Nebraska-Lincoln, whulucy@gmail.com

Follow this and additional works at: http://digitalcommons.unl.edu/cehsdiss Part of the Pre-Elementary, Early Childhood, Kindergarten Teacher Education Commons
Li, Xiaoyu, "Family Environment and School Environment as Predictors for Physical Aggression in Low-Income Children" (2012). Open Access Theses and Dissertations from the College of Education and Human Sciences. Paper 164. http://digitalcommons.unl.edu/cehsdiss/164

This Article is brought to you for free and open access by the Education and Human Sciences, College of (CEHS) at DigitalCommons@University of Nebraska - Lincoln. It has been accepted for inclusion in Open Access Theses and Dissertations from the College of Education and Human Sciences by an authorized administrator of DigitalCommons@University of Nebraska - Lincoln.

FAMILY ENVIRONMENT AND SCHOOL ENVIRONMENT AS PREDICTORS FOR PHYSICAL AGGRESSION IN LOW-INCOME CHILDREN

by

Xiaoyu Li

A THESIS

Presented to the Faculty of The Graduate College at the University of Nebraska In Partial Fulfillment of Requirements For the Degree of Master of Science

Major: Child, Youth, & Family Studies

Under the Supervision of Professor Soo-Young Hong

Lincoln, Nebraska November, 2012

FAMILY ENVIRONMENT AND SCHOOL ENVIRONMENT AS PREDICTORS FOR PHYSICAL AGGRESSION IN LOW-INCOME CHILDREN

Xiaoyu Li, M.S. University of Nebraska, 2012

Adviser: Soo-Young Hong The purpose of the current study was to examine the unique and collective contributions of child’s own characteristics, their family environment and school environment to the development of child physical aggression at Grade 5. This study was based on Bronfenbrenner’s Process-Person-Context-Time model (Bronfenbrenner & Morris, 2006). Children’s gender and their aggression at age 3 were included as person characteristics. Family environment (primary caregiver’s ethnicity, maternal education, home warmth, physical punishment, exposure to violence, family conflict, and parentchild dysfunctional interaction) and early child care experience measured by whether the child was in child care at both age 3 and age 4 were included in the microsystem of the bioecological model. Percentage of free or reduced lunch at school level was conceptualized as the exosystem factor in the bioecological model. The current study used the data from the Early Head Start Research and Evaluation Project (EHSREP). Participants of this study were 690 children (340 girls), followed longitudinally from age 3 to Grade 5, representing multiple races and ethnicities (White, 44.3%, Black, 24.6%, Hispanic 25.9%, other races 5.1%). Results suggested that early aggression at age 3

predicted later aggression at Grade 5. Home warmth was a marginally significant protective factor for children’s aggression, whereas physical punishment, violence exposure, family conflict and parent-child dysfunctional interaction were risk factors for children’s aggression. Child’s experience in formal child care significantly predicted his/her higher aggression at Grade 5 and this effect was maintained with all predictors included in the hierarchical regression model. Results also indicated that school poverty at Grade 5 was not a significant predictor of children’s aggression at Grade 5. The moderation of home warmth for the relations between physical punishment and child’s aggression was not significant. Limitations of the current study, directions for future research, and implications for intervention are also discussed.

ACKNOWLEDGEMENTS

I would like to thank Dr. Helen Raikes for your continuous support and inspiration of my thesis writing during the past year. I appreciate the opportunity to participate in the secondary data analysis of the Early Head Start Research and Evaluation Project. Without your encouragement and patience, I wouldn’t believe that I can complete this big project of thesis writing. Thank you so much, Dr. Raikes. I would also like to thank Dr. Soo-Young Hong and Dr. Greg Welch for their assistance for my thesis writing as well. I appreciate the suggestions of theoretical knowledge and writing instructions from Dr. Soo-Young Hong. I am grateful for Dr. Greg Welch’s guidance in data analysis. I would also like to thank Chaorong Wu and Houston Lester for their advices in data analysis. I also appreciate my good friend Lu Gong’s company during the extended writing hours. Last but not least, I would like to thank my fiancé, Xuejian Li. Following his steps, I came to US to pursue a Master Degree. His support, along with my other family members’ support is the foundation of this thesis.

i TABLE OF CONTENTS Page Chapter 1: Introduction ....................................................................................................... 1 Conceptual Framework: Bioecological Model of Human Development ................... 2 Aggressive Behavior and Family and School Environment Predictors ..................... 7 Physical Aggression versus Relational Aggression ............................................ 7 Person Characteristics ......................................................................................... 8 Child’s Aggressive Behavior at Age 3........................................................ 8 Gender. ........................................................................................................ 9 Microsystem Factors at Home Related To Aggression ...................................... 9 Primary Caregiver’s Ethnicity .................................................................... 9 Maternal Education ................................................................................... 10 Parental Warmth ....................................................................................... 10 Physical Punishment ................................................................................. 12 Exposure to Violence ................................................................................ 13 Family Conflict ......................................................................................... 15 Dysfunctional Parent-Child Interaction .................................................... 17 Microsystem Factor at School Related to Aggression ...................................... 18 Amount of Formal Child Care .................................................................. 18 Exosystem Factor at School Related to Aggression ......................................... 20

ii School Poverty .......................................................................................... 20 The Current Study .................................................................................................... 21 Research Questions and Hypotheses ........................................................................ 22 Research Question 1.......................................................................................... 22 Hypothesis 1.............................................................................................. 22 Research Question 2.......................................................................................... 22 Hypothesis 2.............................................................................................. 22 Hypothesis 3.............................................................................................. 23 Hypothesis 4.............................................................................................. 23 Hypothesis 5.............................................................................................. 23 Research Question 3.......................................................................................... 23 Hypothesis 6.............................................................................................. 23 Research Question 4.......................................................................................... 23 Hypothesis 7.............................................................................................. 23 Chapter 2: Method ............................................................................................................ 24 Participants ........................................................................................................ 24 Measures ........................................................................................................... 25 Demographic Information. ........................................................................ 25 Child Aggressive Behavior ....................................................................... 25 Home Warmth ........................................................................................... 26

iii Physical Punishment ................................................................................. 28 Witnessed Violence .................................................................................. 28 Family Environment Conflict ................................................................... 29 Parent-Child Dysfunctional Interaction .................................................... 29 Experience of Formal Child Care ............................................................. 30 School Poverty .......................................................................................... 31 Procedures ......................................................................................................... 31 Data Analysis Plan ............................................................................................ 31 Data Attrition ............................................................................................ 31 Statistics Model......................................................................................... 35 Assumptions for Regression ..................................................................... 37 Chapter 3: Results ............................................................................................................. 40 Preliminary Analysis ......................................................................................... 40 Demographic Information ......................................................................... 40 Correlational Analysis....................................................................................... 43 Child Gender ............................................................................................. 45 Maternal Education ................................................................................... 45 Home Warmth ........................................................................................... 46 Negative Family Environment Factors ..................................................... 46 Witnessing Violence ................................................................................. 47

iv Hierarchical Regression Examining the Ecological Model .............................. 47 Block 1 and Hypothesis 1 ......................................................................... 52 Block 2 ...................................................................................................... 52 Block 3 and Hypothesis 2 ......................................................................... 52 Block 4 and Hypothesis 3 ......................................................................... 53 Block 5 and Hypothesis 4 ......................................................................... 53 Block 6 and Hypothesis 5 ......................................................................... 54 Block 7 and Hypothesis 6 and Hypothesis 7 ............................................ 54 Chapter 4: Discussion ....................................................................................................... 56 Person Characteristics ....................................................................................... 56 Early to Later Aggression ......................................................................... 56 Microsystem Context ........................................................................................ 57 Home warmth............................................................................................ 57 Physical Punishment and Violence Exposure ........................................... 58 Home Warmth as a Moderator .................................................................. 60 Family Conflict and Parent-Child Dysfunctional Interaction ................... 60 Experience of Formal Child Care ............................................................. 62 Exosystem Factor .............................................................................................. 63 School Poverty .......................................................................................... 63 Limitations and Future Research ...................................................................... 64

v Implications for Interventions ........................................................................... 66 References ......................................................................................................................... 68

vi List of Tables Page Table 1 Table 2 Table 3 Table 4 Table 5 Table 6 Independent Samples t Test for Data Attrition ....................................33 Cross-tab and Chi-Square Test for Primary Caregiver’s Race ............34 Cross-tab and Chi-Square Test for Child’s Gender ...........................34 Cross-tab and Chi-Square Test for Maternal Education ......................35 Collinearity Test for the Hierarchical Regression Model ....................39 Descriptive Statistics of Categorical Variables and Binary Variables ..............................................................................................41 Table 7 Table 8 Descriptive Statistics of Continuous Variables ...................................42 Descriptive Statistics and Frequency of Percent Time in Formal Child Care at Both Ages 3 and 4, Imputed ..........................................43 Table 9 Correlations among Variables..............................................................44

Table 10 Hierarchical Regression Analysis for Aggressive Behavior................49

vii

List of Graphs Graph 1 Graph 2 Distribution of the Original Aggressive Behavior at Grade 5 .............36 Distribution of the Root Square Form of Aggressive Behavior at Grade 5 .............................................................................................36 Graph 3 Graph 4 Normal P-P Plot of Regression Standardized Residual .......................37 Scatterplot of Regression Deleted Residual against Predicted Value ....................................................................................................38

viii

List of Figures Figure 1 The Model of the Current Study ............................................................7

1

Chapter 1: Introduction Aggressive behavior predicts maladjustment in multiple areas for children and adolescents, such as poor peer relations, low prosocial behavior, school dropout, violence, and delinquency (Card, Stucky, Sawalani, & Little, 2008; Kokko, Tremblay, Lacourse, Nagin, & Vitaro, 2006). Research evidence indicates that early aggressive behaviors of children, parenting and out-of-home experiences may all contribute to children's aggressive behavior problems (Brame, Nagin, & Tremblay, 2001; Broidy et al., 2003; Li, Putallaz, & Su, 2011; NICHD Early Child Care Research Network, 2003). Many studies have investigated how the familial factors influence children’s aggressive behavior. Family conflict, physical punishment and harsh parenting have shown to be significantly related to children’s aggressive behavior (Alink, Mesman, Zeijl, Stolk, Juffer, Bakermans-Kranenburg, & Ijzendoorn, 2009; Tanaka, Raishevich, & Scarpa, 2010; Weiss, Dodge, Bates, & Pettit, 1992). There is controversy in the literature regarding whether parental warmth moderates between physical punishment and early childhood aggression (Alink et al., 2009; Stacks, Oshio, Gerard & Roe, 2009). Children’s early child care experience is associated with children’s aggression in subsequent education experience (NICHD Early Child Care Research Network, 2003). Though many research studies have investigated the relation between family environment factors and children’s aggressive behavior, few studies have examined the impact of family environment, early childhood education and subsequent school context simultaneously on children’s aggression. The purposes of this study are four-fold: (1) to examine whether early childhood aggression predicts later mid-childhood aggression; (2) to investigate which family factors constitute risk factors and protective factors for children’s aggressive

2 behavior and examine whether parental warmth moderates the relation between physical punishment and aggressive behavior; (3) to examine whether formal early childhood education experience predicts later aggression and (4) how percentage of free and reduced lunch impacts children’s aggressive behavior as well. Conceptual Framework: Bioecological Model of Human Development This study will adopt the bioecological model and the Process-Person-ContextTime (PPCT) model of human development proposed by Bronfenbrenner (Bronfenbrenner & Morris, 2006) as the conceptual framework. In the bioecological model, development is framed as continuity and change in the biopsychological characteristics of human beings, both in the individual level and in the environment (Bronfenbrenner & Morris, 2006). One of the principal characteristics of the bioecological theory is the emphasis of the interaction between the individual and his or her environment. There are four major properties of the bioecological model and they are process, person, context, and time. These four properties constitute the basis of the Process-Person-Context-Time (PPCT) model. The following are the illustrations of the concepts. Process is the core of the model, which refers to the interactions between the individual and environment. Proximal processes are daily interactions of children with environment, which happen “on a fairly regular basis, over an extended period of time.” (Bronfenbrenner & Morris, 2006, p.798). Proximal processes are considered as the primary mechanisms of human development. Some of the proximal processes examples could be: mother-child interactions, early child care experience in center, playing with a peer from child care, reading for the child on a daily basis and so on. Participation in

3 those proximal processes over time helps children to develop the competence, knowledge and skill to engage in such activities both with others and on their own (Bronfenbrenner & Morris, 2006). Person in the PPCT model refers to the biological, cognitive, and behavioral characteristics of the developing person. According to Brofenbrenner and Morris (2006), there are three types of Person characteristics which are most influential in shaping the future development via their capacity to impact the direction of proximal processes through the life course. Those three types of person characteristics are: dispositions, bioecological resources and demand. Dispositions direct proximal processes in a certain developmental domain and maintain their operation. Ability, experience, knowledge in the bioecological resources are necessary for the effective functioning of proximal processes. Demand characteristics may invite or discourage interactions from the social environment. In the bioecological model, person characteristics are acting as influential factors for proximal processes and outcomes of those processes. Children’s gender and their behavioral characteristics (e.g. earlier aggression at age 3) are person characteristics which could influence the proximal processes and children’s development. According to the literature, boys tend to display higher physical aggression, while girls tend to have higher relational aggression (Barth, Dunlap, Dane, Lochman, & Wells, 2004; Crick & Grotpeter 1995). Also, children’s behavioral characteristics affect children’s development. Many research studies have indicated that there is moderate stability of aggression across time (Olweus, 1979). Some trajectory studies have shown that around 4-10% of children follow chronic physical aggression from early childhood to adolescence (Broidy et al., 2003).

4 Context refers to the nested systems of children’s environment from immediate to more distant. There are four levels of systems in the context and they are: microsystem, mesosystem, exosystem and macrosystem (Brofenbrenner, 1979). Detailed explanations of those four nested systems are as follows. A microsystem emphasizes an individual’s role and relations in the immediate setting containing the individual. Microsystem is defined as “a pattern of activities, roles, and interpersonal relations experienced by the developing person in a given face-to-face setting with particular physical and material features and containing other persons with distinctive characteristics of temperament, personality, and systems of belief”(Bronfenbrenner, 2005, p. 148). Bronfenbrenner (1979) described that an individual would go through a series of “ecological transitions” as a “person’s position in the ecological environment is altered as the result of a change in role, setting, or both” (p.26). For instance, with the development of a child, he or she is introduced to caregivers other than the mother: day care, preschool, peers, and finally the school systems. All of these interactions might symbolize “ecological transitions” for the developing child since the child assumes new roles and establish new relations with the immediate environment and increasingly distant environments (White & Klein, 2008, p. 259). In this study, there are several variables in the microsystems at home: maternal education level, primary caregiver’s race and ethnicity, home warmth, physical punishment/spanking, exposure to domestic violence, family conflict, and parent-child dysfunctional interaction. There is one variable in the microsystem at school: whether the child attended formal child care at both age 3 and age 4.

5 Mesosystem was defined by Bronfenbrenner (1979) “as a set of interrelations between two or more settings in which the developing person becomes an active participant” (p. 209). Mesosystem involves the interconnection between two or more microsystems. It happens when the same person participates in activities in more than one setting, for instance, “when a child spends time both at home and at the day care center” (p. 209). When this occurs, it involves the incidence of “ecological transition.” Many hypotheses pertaining to the mesosystem emphasize the compatibility of the role requirement across different settings and the smooth transition across different settings. Parent involvement at school and parent-teacher relationships may help children to have smooth transitions from home to school. “An exosystem has been defined as consisting of one or more settings that do not involve the developing person as an active participant but in which events occur that affect, or are affected by, what happens in that setting” (Bronfenbrenner, 1979, p. 237). One example of this exosystem for a child is the relation between the home and the parent’s workplace. If the mother assumes great responsibilities and pressure from work, her pressure and lack of time to spend with families would influence her relationship with her children. Social reforms, financial upheaval and new laws are also examples of exosystem factors which could influence a child’s life indirectly. In this study, school poverty measured by percentage of enrolled students eligible for free or reduced lunch is an exosystem factor. Though a student doesn’t necessarily play a role in the school financial system, school poverty could influence the child indirectly via the quality of teachers and quality of school environment.

6 The last dimension is the macrosystem, which “refers to the consistency observation within a given culture or subculture in the form and content of its constituent micro-, meso-, and exosystems, as well as any belief systems of ideology underlying such consistencies” (Bronfenbrenner, 1979, p. 258). Culture is one example of macrosystem. As we know, parenting practices vary from culture to culture. Thus macrosystem could also influence a child’s development indirectly via parenting, cultural beliefs and social environment. Time is also a critical element in the study of behavioral development. In the PPCT model, person characteristics interact with context to form processes, which take place over time. There are three successive levels of time: mirco-, meso-, and macro(Bronfenbrenner & Morris, 2006). “Microtime refers to continuity versus discontinuity in ongoing episodes of proximal process. Mesotime is the periodicity of these episodes across broader time intervals, such as days and week. Finally, macrotime focuses on the changing expectations and events in the larger society, both within and across generations (p. 796)”. Please refer to Figure 1 for the model of this study, which was based on the Process-Person-Context-Time model and the Bioecological Model.

7 Figure 1 The Model of the Current Study

Aggressive Behavior and Family and School Environment Predictors Physical Aggression versus Relational Aggression Aggression has usually been defined as behaviors to cause harm for others though specific definitions varied over the years (Crick & Grotpeter, 1995). There are also different ways for classifications of aggression depending whether it is overt or covert, direct or indirect, verbal or nonverbal. In this study, I am going to differentiate physical aggression from relational aggression. Physical aggression is a more direct form of aggression, which intends to cause or threaten physical harm; while relational aggression is a more indirect form of aggression that involves inflicting harm on someone through

8 hostile manipulation of relationships, such as intentionally excluding a peer from social plans, spreading rumors about a peer, and insulting or hurting a peer through words (Archer & Coyne, 2005). The literature indicates that boys have higher physical aggression than girls, while girls display higher relational aggression than boys (Barth et al., 2004; Crick & Grotpeter, 1995). This study will mainly focus on physical aggression and examine children’s individual characteristics, family environment, and school environment as predictors for children’s physical aggression at Grade 5. Person Characteristics Child’s Aggressive Behavior at Age 3. Many research efforts have been devoted to study the developmental trajectories of aggression from early childhood to adolescence or early adulthood. Research indicates that child’s aggression peaks around 2 to 3.5 years and then reduces steadily afterwards (Tremblay & Nagin, 2005). Broidy et al. (2003) carried out a six-site study in three countries (Canada, New Zealand, United States) to examine the relationship between childhood disruptive behaviors and adolescent delinquency. For most sites, children’s physical aggression was assessed repeatedly from age 6 to age 15. The developmental trajectories across sites and across countries were similar for boys in that vast majority within the samples follow a low-decreasing aggression or no aggression pattern from early childhood to adolescence (around 70%), while only around 4-10% of the sample followed a chronic physical aggression trajectory from early childhood to adolescence. A group of Chinese scholars also employed Nagin and Trembley (1999)’s group-based semi-parameter method to examine the developmental trajectory of 1618 Chinese Middle school students’ aggressive behavior between the ages of 9 to 12 with the outcome measured annually (Chen, Zhang, Ji, Chen,

9 Wei, & Chang, 2011). The results of this study paralleled the Broidy et al’s study (2003) in that 68.7% of children were in the membership of no aggression, 26.8% of children were in the membership of low-decreasing group, while only 4.5% children were chronically high aggressors. These two studies indicate that there is some continuity in children aggression from early to later stages. Gender. Boys had significantly higher aggression than girls across early childhood to adulthood (Barth et al., 2004; Card, Stucky, Sawalani, & Little, 2008; Colder, Mott, Levy, & Flay, 2000; Stacks et al., 2009). Also, the boys were more likely than girls to be in “high increaser” membership of physical aggression across years in trajectory studies (Côté, Vaillancourt, LeBlanc, Nagin, & Tremblay, 2006; Underwood, Beron, & Rosen, 2009). The gender difference might be due to the social and developmental influence. Girls’ problem behavior was primarily channeled to internalizing behavior due to socialization, whereas boys’ aggression was more accepted in the socialization (Keenan, & Shaw, 1997). Microsystem Factors at Home Related To Aggression Primary Caregiver’s Ethnicity. Literature indicates that physical punishment is more common in African American parents than European parents (Deater-Deckard, Dodge, Bates, & Pettit, 1996; Straus & Stewart 1999). Research also demonstrates that physical punishment predicts children’s aggressive behavior (Erath, Bierman, & Conduct Problems Prevention Research Group, 2006; Taylor, Manganello, Lee, & Rice, 2010). Thus, primary caregiver’s ethnicity could influence child’s aggressive behavior indirectly via parenting, such as physical punishment.

10 Maternal Education. Low maternal education was a risk factor for child’s aggression in early childhood (Benzies, Keown, & Magill-Evans, 2009; Côté et al., 2007). Two trajectory studies demonstrated that low maternal education (not finishing high school) increased the children’s likelihood of falling into high physical aggression profile (Côté et al., 2006; Harachi, Fleming, White, Ensminger, Abbott, Catalano, & Haggerty, 2006). There hasn’t been much research to investigate the mechanisms between maternal education and child’s aggression. Maternal education might influence parenting beliefs and parenting practices, which, in turn, influence children’s social emotional behavior.

Parental Warmth. Parental warmth is mainly constructed to measure parents’ emotional availability, sensitivity, affection, and support for their child. There are some inconsistencies in the literature as to whether parental warmth is associated with children’s externalizing behavior. Some research indicates that parental warmth doesn’t predict aggressive behavior directly in a sample of 1 to 5 years old children (Alink et al., 2009; Stacks et al., 2009); while other research finds that parental warmth is significantly negatively associated with children’s externalizing behavior (White & Renk, 2012; Chen, Wu, Chen, Wang, & Cen, 2001). Furthermore, in a longitudinal study which tracks children’s aggressive behavior over multiple years, low home warmth increases children’s potential to fall into a “high-increasers” trajectory instead of other lower-risk trajectories, such as “low-stable”, and “medium-desisters” trajectories (Underwood et al., 2009). The literature also shows contradictory evidence in regards to whether home warmth moderates the relation between negative discipline and aggression in early

11 childhood. Alink et al (2009) examined the relations among maternal sensitivity, negative discipline and aggression in a sample of children (age ranges from 13.58 to 41.91 months) with high scores on child externalizing behavior. This study confirmed that maternal sensitivity moderated the effect of negative discipline on children’s aggressive behavior. More negative discipline was associated with child aggression one year later, but only when mothers had low-sensitivity for their children. In contrast, Stacks et al. (2009)’s study didn’t find this moderating effect of parental warmth. Stacks et al. also tested whether parental warmth moderated the negative impact of spanking on child aggression in a low-income sample of children (longitudinally measured at 14, 24 and 36 months). The results of this study didn’t confirm the moderating role of parental warmth. The inconsistency results may be due to the sample difference since Alink et al (2009) studied this moderating effect in a higher-risk sample for aggression while Stacks et al. (2009) studied this effect in a low-income sample. It is important for future research to investigate more about the mechanism underlying the moderating effect of maternal sensitivity/home warmth. A randomized controlled trial study provides some insight into the relation between home warmth and children’s aggression by examining children’s cortisol response (O’ Neal, Brotman, Huang, Gouley, Kamboukos, Calzada, & Pine, 2010). This study recruited 92 children (mean age = 48 months) at risk for antisocial behavior and assigned 47 of them randomly into an intervention group. The intervention mainly offered home visits to improve parenting skills and encouraged parents to use nonharsh, consistent and appropriate disciplinary strategies to promote children’s social competence. This intervention significantly increased parents’ warmth, which in turn was

12 associated with increases in cortisol response. The cortisol level negatively predicted children’s aggression. Therefore, cortisol level was a mediator between the intervention effects and children’s aggression. Post hoc analysis indicated that changes in cortisol accounted for 69% of the intervention effect on child aggression. This study provided better understanding of the complex nature of relations among the family environment, neurobiological regulation of stress and developmental psychopathology. Moreover, this study also demonstrated evidence that home warmth could help children to improve their stress response system, which in turn helped children deal with social challenging situation relevant with later aggressive behavior. Physical Punishment. Spanking and other corporal punishment in families have been disputed practices in the United States. According to a study, 94% of American parents spank their children by the time they are 3 or 4 years old (Straus & Stewart, 1999). Straus and Kantor (1994, p.543) defined corporal punishment as following: “the use of physical force with the intention of causing a child to experience pain but not injury for the purposes of correction or control of the child’s behavior.” In Hicks-Pass (2009)’s review article of corporal punishment, corporal punishment has been associated with aggression, and higher probability of violence in intimate relationships at a later age. Many studies indicated that spanking was related to later aggressive behaviors for young children (Erath, Bierman, & Conduct Problems Prevention Research Group, 2006; Taylor, Manganello, Lee, & Rice, 2010; Weiss, Dodge, Bates, & Pettit, 1992). Spanking at 3 years old predicted the child’s increased risk for higher levels of aggression when the child was 5 years of age after controlling the child’s initial level of aggression and 8 potential parenting risk confounders (Taylor et al., 2010). Two studies

13 also found that harsh parenting (including slapping and spanking) was associated with elevated aggressive behavior in sample of kindergarteners (Weiss, et al., 1992; Stormshak, Bierman, McMahon, Lengua, & Conduct Problems Prevention Research Group, 2000). Weiss et al. (1992) suggested that the path from harsh parenting to children’s aggression in school context was partially mediated by maladaptive social information processing. Some studies indicated that physical punishment could interact with other parenting features, such as maternal sensitivity and emotional support. Both maternal sensitivity and emotional support were found to be moderators between physical discipline and children’s aggressive behavior in that children only demonstrated high aggression in the context of low maternal sensitivity and low emotional support (Alink et al., 2009; McLoyd & Smith, 2002). Exposure to Violence. Exposure to violence is a widespread phenomenon according to a study of a United States sample, which indicates that around one-third of all children are victim of some form of violence and approximately 90 percent witness violence at least once during their lifetimes (Richters & Martinez, 1993). Many retrospective studies indicate that exposure to violence during childhood predicted later aggressive behavior in young adults (Henning, Leitenberg, Coffey, Bennett & Jankowski, 1997; Milletich, Kelley, Doane, & Pearson, 2010; Bailey & Coore-Desai, 2011). In Henning et al. (1997) study, 203 of 1452 in a survey sample reported that they had witnessed at least one incident of physical aggression between their parents. Compared with the group of young adults who never observed any physical aggression between their parents, the group of young adults exposed to interparental violence not only had significantly higher level of psychological distress but also higher externalizing behavior

14 problems. A meta-analytic review study demonstrated a significant association between exposure and child problems (d=-0.29) in correlational studies (Kitzmann, Gaylord, Holt, & Kenny, 2003). This meta-analysis also indicated that witnesses of violence had significantly worse outcomes relative to nonwitnesses, such as increased risk for psychological or interpersonal problems later. A longitudinal prospective study also confirmed that prior exposure to violence significantly predicted subsequent elevated aggression, normative beliefs about aggression, and aggressive fantasy (Guerra, Huesmann & Spindler, 2003). Moreover, this study also suggested that social cognition supporting aggression partially mediated the association between earlier violence exposure and later aggressive behaviors. Besides the social cognition as a mechanism to understand this relationship between violence exposure and subsequent aggression, Massachusetts Coalition of Battered Women Service Groups (1995) provided some other possible mechanisms for this association: “(1) violence is an appropriate way to resolve conflicts; (2) violence is a part of family relationships; (3) the perpetrator of violence in intimate relationships often goes unpunished; and (4) violence is a way to control other people” (Osofsky, 2003, p. 165). Research also indicates compounding effects of physical punishment and violence exposure. Hughes, Parkinson, & Vargo (1989) demonstrated that children who experienced both physical punishment and violence exposure had the most externalizing behavior; children who were exposed to violence but not received physical punishment were rated as having less externalizing behavior; while children who were nonexposed and non-receivers of physical punishment had the least externalizing behavior. This study illustrated that both physical punishment and violence exposure were risk factors for

15 externalizing behavior and those two factors have cumulative effects. Since there are few prospective literatures studying how earlier exposure to violence influences later aggressive, this current study will contribute to the literature in the use of longitudinally prospective design. Family Conflict. Family conflict is a strong familial factor related to children’s aggression in the literature (Andreas & Watson, 2009; Tanaka et al., 2010; Cummings, Goeke-Morey, & Papp, 2004； Li et al., 2011). Family conflict was directly associated with children’s aggression in studies with middle childhood children (Cummings et al., 2004; Li et al., 2011). Tanaka et al. (2010)’s study found that family conflict was related to increased proactive aggression in children with high levels of anxiety in a sample of children from 7 to 13 years. In a longitudinal trajectory study of children’s aggression from 2nd grade to 8th grade, family conflict was a risk factor for children’s higher likelihood of falling into the high aggression trajectory (Harachi et al., 2006). All of these studies confirm that family conflict was associated with increased aggression and a risk factor for children’s externalizing behavior. Additionally, positive family environment featured by high cohesion and low conflict reduced the aggressive behaviors of children with high aggressive beliefs in childhood (Andreas & Watson, 2009). Taken together, the literature indicates that positive family environment reduces children’s aggression while negative family environment increases children’s aggressive behavior. The following two studies provide some understanding of the mechanisms of how family conflict/marital conflict influences children’s aggressive behavior. Cummings et al. (2004) examined how marital conflict was associated with children’s immediate aggressive responding, which, in turn, predicted children’s

16 aggression. Parents’ tactics and emotionality were important elements during the process of marital conflict. Parents’ destructive conflict tactics and negative emotionality were associated with a higher possibility of aggression in children. By contrast, constructive behaviors and positive emotionality were associated with a lower probability of children’s aggressive responses. Cummings et al. (2004)’s study not only confirmed family conflict as a risk factor for children’s externalizing behavior, but also provided deeper understanding of the process context, such as, specific tactics and emotionality, which influenced children’s aggressive behavior. Parents’ constructive tactics and positive emotionality would help to reduce the negative effects of marital conflict on children’s externalizing behavior. Interparental conflict, parenting behaviors and children’s overt and relational aggression were examined in Li et al. (2011)’s study. This study indicated that interparental conflict could directly and indirectly predict children’s physical and relational aggression. Parental overt conflict was positively associated with boys’ aggression, but not girls’ aggression, and this effect was mediated through paternal coercive control. In addition, the associations between maternal overt conflict style and maternal coercive control with boys and girls were marginally significant. This study suggests that interparental conflicts could impact parent’s overt and covert coercive control for children, which in turn is related to children’s aggression. Thus this study provides some understanding for the mechanism of how family conflict influences children’s aggression via the function of parenting. Most of these studies focus on studying relations between family conflict and aggressive behavior for children in their middle childhood to early adolescence (age

17 range from 7-19 years old). Few studies have investigated how family conflict influences the aggression in young preschoolers and children in early childhood. The current study will contribute to the body of knowledge by studying children from age 3 to fifth Grade. Moreover, since this current study measures children’s family conflict repeatedly at both age 3 and fifth grade, it will investigate which age period is more sensitive and vulnerable for children’s exposure to family conflict. Dysfunctional Parent-Child Interaction. Parent-child interaction is also an important element in the family process. A large body of research has indicated that negative parent-child interaction is related to children’s aggression. Eichelsheim et al. (2010) examined parent-adolescent relationship and adolescents’ aggression and delinquency in two Dutch samples. This study found that negative parent-child interaction, mainly featured by conflict and antagonism, was strongly associated with adolescents’ aggression in both samples and in both genders. A prospective intergenerational study also found strong associations between child’s externalizing behavior and the mutual parent-child attachment relationship, which mainly measured parent-child closeness, and child’s admiration of parents (Brook, Lee, Finch, & Brown, 2012). This intergenerational study followed the participants from early adolescence (mean age around 14 years old) to middle adulthood (mean age approximately 32.3 years old), thus being able to capture three generations: G1, the participants’ parents; G2, the participants; and G3 the offspring of the participants. This study demonstrated a bidirectional relationship between G3 child’s externalizing behavior and G2/G3 mutual parent-child attachment. Dysfunctional parent-child attachment subtypes, including avoidance and anxiety attachment, were also related to physical aggression in young

18 adults. Female young adults were more likely to be physically aggressive when they had higher levels of avoidance attachment to their mothers and higher levels of anxiety attachment levels with their fathers (Williams & Kennedy, 2012). Microsystem Factor at School Related to Aggression Amount of Formal Child Care. Majority mothers in the United States choose to return to work after giving birth to a child before the child turns 1 year old. According to the Bureau of Labor Statistics (2000), 58% of all mothers with infants under 1 year old are in the labor force. Thus, extensive non-maternal child care becomes a routine for most families. The effects of such extensive and continuous non-maternal child care have called the attention of parents, policy-makers, and developmentalists (Belsky, 2001). Several studies have examined the effects of quantity, type and timing of external child care on problem behavior and socioemotional adjustment for children in ensuring years of school (Averdijk, Besemer, Eisner, Bijleveld & Ribeaud, 2011; Loeb, Bridges, Bassok, Fuller, & Rumberger, 2007; NICHD ECCRN, 2003). Accumulative quantity of nonmaternal care has been found to be associated with children’s problem behavior and aggression (Averdijk, et al., 2011; NICHD ECCRN, 2003; Loeb, et al., 2007). Results from the National Institute of Child Health & Human Development (NICHD) study indicate that accumulative amount of time spent in any type of nonmaternal care arrangement across the first 4.5 years of life is predictive of children’s externalizing problems and conflicts with adults at 54 months of age and in kindergarten as reported by mothers, caregivers, and/ or teachers; and that these effects remain even after controlling several plausible mediators of the quantity of child care effects—including quality, type, instability of child care, maternal sensitivity and other

19 family factors (NICHD ECCRN, 2003). Averdijk et al. (2011) also confirmed that accumulative external child care over the life course is associated with children’s aggressive behavior and other types of non-aggression problem behaviors at age 7 in the context of Swiss sample; and this effect is due to group-based external child care rather than individual childcare; and also this effect remains even after controlling relevant child’s characteristics at birth and family risk factors, including negative parenting, parental conflict, maternal depression, parental education and etc. Using the Early Childhood Longitudinal Study, Kindergarten Class of 1998-99 (ECLS-K) data, Loeb et al. (2007) also found that the quantity of experience in center child care has negative effects on children’s externalizing behaviors even after controlling plausible family risk factors. In contrast, some studies didn’t confirm the relationship between amount of external childcare and children’s externalizing behavior. Bacharach and Baumeister (2003) found that quantity of external care was not associated with severe externalizing behavior among kindergarten children beyond selection factors. Borge, Rutter, Cote & Tremblay (2004) also intended to differentiate the social selection and social causation regarding the effect of early childcare on physical aggression. This study found that for 2to 3- year old Canadian sample, aggression was significantly more common in children under the maternal care setting than in those attending group day-care. Moreover, there is strong social selection effect in that there is significantly higher proportion of high-risk families choosing homecare than group day care. Another issue of the literature is that the effect of early childcare on children’s externalizing behavior fades out overtime. Sammons et al. (2007) found that children

20 who enrolled into center-based childcare below age 2 had higher level of antisocial behavior at entry into primary school than children who stayed at home in an England sample contest. However, this relationship disappeared when children were 10 years old. However, another study in the United States setting found the early child care effects sustained till 6th Grade (Belsky, et al., 2007). Since there are some controversies in literature regarding the effect of external child care on children’s externalizing behaviors, more studies are needed to testify this relationship with a different sample. Also, the literature indicates the negative effect of childcare on children’s externalizing behavior fades out over time. Therefore it is also important to employ longitudinal data and investigate whether the negative effect of childcare on child’s social emotional adjustment maintains when children are 10 years or older in a different sample. Exosystem Factor at School Related to Aggression School Poverty. School financial resource level has been associated with both students’ academic performance and their behavioral functioning as well. High level of school poverty was negatively correlated with students’ academic outcomes and positively correlated with students’ social behavior adjustment (Battistich, Solomon, Kim, Watson & Schaps, 1995). Classroom climate and school poverty (measured by percentage of students qualifying for free or reduced cost lunch) were examined collectively for their roles in predicting children’s aggression in a sample of elementary students (Thomas, Bierman, Thompson, Powers, & Conduct Problems Prevention Research Group, 2008). This study found that low-quality classroom contexts and school poverty were related to increased level of student aggressive-disruptive behavior at

21 school, even though only low-quality classroom contexts contributed unique variance for the prediction of aggression in the regression model (Thomas et al., 2008). Another empirical study found school poverty predicted unique variance in children’s aggression for the urban African American and European children, but not the rural students. (Thomas et al., 2006). More research is needed to understand the mechanisms under which school-level poverty influences students’ aggressive behavior.

The Current Study

This study examined the unique and collective contributions of microsystem factors (home warmth, physical punishment, violence exposure, family conflict, parentchild dysfunctional interaction, whether the child attended formal care at both age 3 and age 4), and exosystem factor (percentage of children enrolled eligible for free or reduced lunch at school level at Grade 5) to children’s aggressive behavior at Grade 5, controlling child’s early aggression at age 3, child’s gender, primary caregiver’ ethnicity, and maternal education in a sample of low-income children across 17 research sites in United States. This study was based on Bronfenbrenner’s bioecological and Process-PersonContext-Time models, mainly examining how the microsystem and exosystem influence children’s aggression at Grade 5. This study would contribute to the literature in understanding the mechanisms underlying children’s aggressive behavior by including both the family environment and school context factors in a longitudinally prospective design. Since children’s aggression was measured repeatedly at both age 3 and Grade 5, this study investigated the continuity of aggression across time and how early aggression related to later aggression. Moreover, the family environment and school context factors

22 were also measured across time from age 3 and Grade 5. Therefore, this study examined how early and concurrent home and school environment predicted aggression at Grade 5. The current study would add to the literature in understanding the moderating role of home warmth for the relation between physical punishment and children’s aggression since there is inconsistency in the literature (Alink et al., 2009; Stacks et al., 2009). Also, this study would contribute to the literature in testing the longitudinal effect of formal early child care experience on children’s aggressive behavior since there were also controversies regarding the longitudinal effect of early child care experience (Belsky et al., 2007; Sammons et al, 2007). The specific research questions and hypotheses are as follows: Research Questions and Hypotheses Research Question 1 Is there strong a correlation between early aggression at age 3 and later aggression at Grade 5? Hypothesis 1. Early aggression at age 3 will positively correlate with later aggression and predict later aggression in Grade 5. Research Question 2 Which home factors constitute risk factors for children’s aggression and which home factors are protective factors for children’s aggressive behavior? Hypothesis 2. Home warmth across age 3, age 5 and Grade 5 will be negatively correlated with children’s aggression and predict lower aggression for children in Grade 5.

23 Hypothesis 3. Physical punishment at age 3, whether children were spanked last week at age 5, whether child witnessed violence at age 5, and child’s exposure to domestic violence at Grade 5 would predict higher aggression significantly in Grade 5. Hypothesis 4. Home warmth would moderate the relationship between physical discipline and children’s aggression. Hypothesis 5. Family conflict at age 3 and Grade 5, parent-child dysfunctional interaction at age 3 and Grade 5 would predict higher aggression significantly in Grade 5. Research Question 3 Does early child care experience at age 3 and age 4 predict aggression for children in Grade 5? Hypothesis 6. Early child care experience would positively predict children’s aggression in Grade 5. Research Question 4 Does school poverty measured by enrolled students eligible for free or reduced lunch predict child’s aggression at Grade 5? Hypothesis 7. Percentage of free or reduced lunch would predict higher aggressive behavior for children in Grade 5.

24 Chapter 2: Method Participants

This study used data from the Early Head Start Research and Evaluation Project (EHSREP) which includes 17 research sites throughout the United States. A total of 3001 children and their families were enrolled into this project when they were 12 months old or younger. Half of the children were assigned randomly into the birth to three intervention program and half of the children were assigned into the control group. EHSREP collected data and implemented parent interviews and assessments of children when they were 14, 24, and 36 months of age. Data were collected again in the spring before children were eligible to enter kindergarten, roughly two years after the end of Early Head Start services for children in the program (Chazan-Cohen et al., 2007). Then children’s social-emotional, cognitive, family environment and school environment data were collected when they were in Grade 5 in order to test the longitudinal effects of Early Head Start. The sample for this study was 690 (340 girls, 49.2%) children using the listwise deletion method. Those children had no missing data on all the dependent and independent variables across three time points age 3, age 5 and Grade 5. This sample represented multiple race and ethnicity (White, 44.3%, Black, 24.6%, Hispanic 25.9%, other races 5.1%). Around half the sample had been enrolled in the EHS intervention group (n=327, 47.4%).

25 Measures Demographic Information. Child’s gender, race and ethnicity and maternal education were collected in the baseline phase, when the EHSREP was initiated. There were four categories of race and ethnicity, and they were White, Black, Hispanic, and other ethnicities. Maternal education was coded into 3 levels of categories. Mother who had less than 12 years of education (not finishing high school) was coded 1, who had 12 years of education (finished high school) was coded into 2, who has more than 12 years of education was coded into 3. Child Aggressive Behavior. Children’s aggressive behavior at 36 months was measured by parent report using the aggressive behavior scale score on the Achenbach System of Empirically Based Assessment (ASEBA) Child Behavior Checklist for ages 1 ½ -5 years (Achenbach & Rescorla, 2000). The ASEBA has 39 items at 36 months, of which 19 items measure the incidence of aggressive behavior in children. The following are some of the items: can’t stand waiting and wants everything now; is defiant; destroys things belonging to family or other children; gets in many fights; gets hurt a lot, is accident-prone; has angry moods; physically attacks people; screams a lot; and has temper tantrums or a hot temper. On each of the 19 items, parents were asked to report how true they are for their child. The scores are coded as following: 0=not true, 1=sometimes or somewhat true, 2=very or often true. Scores on this subscale can range from 0, if parents report all of the behavior problems are not true for their child, to 38, if the parents report that all of the behavior problems are very true or often true for their child. Internal consistency for this scale at 36 months is reported to be 0.88 by study authors Love et al. (2002).

26 Children’s aggressive behavior at Grade 5 was measured by Child Behavior Checklist (CBCL) (Achenbach & Rescorla, 2001). Similar with ASEBA, CBCL is also reported by parents. Eighteen items in CBCL measure the incidence of child behavior and emotional problems. A sample of the items includes: argues a lot; cruelty, bullying, or meanness to others; demands a lot of attention; destroys his/her own things; disobedient at school; gets in many fights; screams a lot; sudden changes in mood or feelings; and temper tantrums or hot temper. On each of the 18 items, parents were asked to report how true they are for their child. The scores are coded as following: 0=not true, 1=sometimes or somewhat true, 2=very or often true. Subscale scores ranges from 0 if all the behaviors are never observed by the parents; to 36 if all the behavioral problems are often observed by parents. If more than 8 items were missing, then subscale score was set to missing. If data was missing for 8 or fewer items, the values of the missing items were imputed with the means of the non-missing items for that person. Home Warmth. Home warmth was measured repeatedly at the following time points of child’s age: 36 months, age 5 and Grade 5. Home warmth measures responsive and supportive parenting behavior observed by the interviewer during the home visit using the subscale of warmth from the HOME measurement (Caldwell, & Bradley, 1984). At 36 months, interviewer observes the following three parent behaviors during the home visit: (1) parent's voice conveys positive feeling to child, (2) parent spontaneously praises child's qualities twice during visit; (3) parent caresses, kisses, or cuddles child during visit. The score can range from 0, if none of the positive behaviors were observed, to 3, if all of the behaviors were observed. The internal consistency of the

27 HOME warmth was reported to be 0.72 at age 36 months for 1794 families participating in the EHSRE study in study by Love, Kisker, Ross, Schochet & Brooks-Gunn (2002). Home warmth at age 5 was also measured by the subscale of home warmth from HOME (Caldwell, & Bradley, 1984).There are 6 items in the age 5 home warmth subscale: (1) caregiver converses twice with child twice; (2) caregiver answers questions or requests, (3) caregiver responds verbally to talking, (4) caregiver spontaneously praises child twice, (5) caregiver caresses kisses cuddle child once, (6) caregiver lets child show off. Interviewer observed whether those 6 responsive and supportive parenting behaviors exhibited by parent during visit. Scores can range from 0, if none of the positive behaviors were observed, to 6, if all of the behaviors were observed. Home warmth at Grade 5 was also measured by the subscale of home warmth from HOME (Caldwell, & Bradley, 1984). There are 9 items in the Grade 5 home warmth subscale: (1) parent talks twice to child during visit (beyond correction and introduction), (2) parent answers one of child’s questions or requests verbally, (3)parent encourages child to contribute to the conversation during the visit, (4) parent helps child demonstrate some achievement during visit or mentions a particular skill, strength, or achievement, (5) parent spontaneously praises child’s behavior or qualities twice during visit, (6) parent uses some term of endearment or some diminutive for child’s name when talking about or to him/her at least twice during visit, (7) parent’s voice conveys positive feeling when speaking of or to child. (8) parent caresses, kisses, or cuddles child once during visit, (9) parent shows some positive emotional responses to praise of child by visitor. Interviewer made observation of those 9 responsive and supportive parenting behaviors

28 during visit. Score can range from 0, if none of the positive behaviors were observed, to 9, if all of the behaviors were observed. Physical Punishment. Physical punishment at 36 months during the past week was reported by parents. The parent was asked whether he/she spanked the child during the past week, and if so, how many times. This variable was coded as binary. If parent did not spank the child, then this variable was coded as 0. If parents did spank the child in the previous week, then it was coded as 1. If parents didn’t respond, then it was coded as missing. Whether child was spanked during past week when the child was 5 years old was also reported by parents to indicate whether they spanked their child during past week. The coding was the same as the variable “physical punishment at age 3.” If parent did not spank the child, then this variable was coded as 0. If parents did spank the child in the previous week, then it was coded as 1. If parents didn’t respond, then it was coded as missing. Witnessed Violence. When children were age 5, parents reported whether child had witnessed violence in the past year. If child had witnessed violence during the past year, then this variable was coded as 1. If child hadn’t witnessed violence during the past year, then this variable was coded as 0. Exposure to domestic violence at Grade 5 was also reported by parents. The parent responded to the following question: “In the past year, has [child’s name] been a witness to domestic violence?” It was coded as a binary score. If child had been exposed to domestic violence, then this variable was coded as 1. If child hadn’t been exposed to domestic violence, then this variable was coded as 0.

29 Family Environment Conflict. Family environment conflict was measured repeatedly; one is when child was 36 months, and the other is when child was in Grade 5 using the Family Environment Scale. The Family Environment Scale measures the social environments of families in 10 key dimensions, which include family relationships (cohesion, expressiveness, and conflict); features of personal development that could be endorsed by families (for instance, achievement orientation; independence); and maintenance of the family system (organization and control) (Moos & Moos, 1994). Only one dimension—family conflict was assessed in this study. Family conflict measures the magnitude of open expression of anger and aggression and generally conflictual interactions in the family. Parents report on a 4-point scale. Sample items include "we fight a lot", and "we hardly ever lose our tempers." Some item scores were reverse-coded so that a score of 4 indicates higher conflict in the family environment. There are 5 items in this scale. The average score of the 5 items was calculated to be the subscale score. Thus the score of family conflict ranges from 1 to 4. Any case with a missing item is set to be missing for the scale score. According to Moos and Moos (1981), internal consistency for the Conflict scale was good (Cronbach’s alpha = .75). Also, the test– retest reliability was .85 after 2 months and .76 after 12 months. Parent-Child Dysfunctional Interaction. Parent-Child Dysfunctional Interaction was measured repeatedly; one was when child was 36 months and the other was when child was at Grade 5. Parent-child dysfunctional interaction was measured by Parenting Stress Index – Short Form (Abidin, 1995). Parent-child dysfunctional interaction measures parent’s perception that the child does not meet the expectations of the parent, and interactions with the child are not reinforcing the parent. The parent might feel that

30 the child abuses or refuses the parent or the parent feels frustrated or estranged from the child. The parent responded his/her agreement or disagreement to the following sample statements "Your child rarely does things for you that make you feel good," and "Most times you feel that your child does not like you and does not want to be close to you," and "Your child seems to smile less than most children." Item responses are coded on a 5-point scale, with 5 indicating high levels of parent-child dysfunctional interaction. Scores on the 12-item subscale can range from 12 to 60. Missing data were imputed for this variable. Parent-Child Dysfunctional Interaction at Grade 5 was measured again using the same measurement Parenting Stress Index – Short Form (Abidin, 1995). The difference from the 36-month Parent-Child Dysfunctional Interaction measurement is that only 6 items were assessed in this Grade 5 variable. Item responses are still coded on a 5-point scale, with 5 indicating high levels of parent-child dysfunctional interaction. Scores on the 6-item subscale can range from 6 to 30. Experience of Formal Child Care. Percentage of time in formal program at ages 3 and 4 was reported by parent to indicate child’s experience in formal child care program. Originally, the parent reported whether his/her child was in formal child care program when the child was at age 3 and age 4 or not. If the child didn’t attend any formal child care program during that period, then this variable was coded as 0; if the child participated in formal child care during the time period, then it was coded as 1.Thus it was a dichotomous variable. Since imputation was used to deal with missing data, some children were imputed to have certain amount of child care experience between 0 and 1. In the current sample of this study (n=690), 395 children had no childcare experience and

31 they had 0 as the score for this variable, and 279 child had childcare experience and they had 1 as the score for this variable. Sixteen children were imputed to have certain amount of child care experience between 0 to 1. Please refer to Table 8 for the frequency distribution of this variable. School Poverty. School poverty was measured by the percentage of enrolled students eligible for free or reduced lunch at Grade 5. Score of this index ranges from 0 percent to 99.65% (mean=61.8%) in this sample.

Procedures One of the primary research coordinators provided access to the usage of EHS dataset necessary for this study. I signed the confidentiality form to agree with the usage of the EHS data for the research purpose only. The Institutional Review Board at University of Nebraska-Lincoln approved this study. Data Analysis Plan Data Attrition. There was a considerable amount of data attrition in this study due to the longitudinal nature of the dataset. The amount of the sample of this study 690 was around 23% of the amount of the original 3001 participants in the EHSREP, which tracked child’s development from birth to Grade 5. Listwise deletion method was adopted to deal with the missing data so that all 690 children had complete data in all variables of this study. Independent samples t test and Chi-square test were performed to compare the difference between the participants with missing data and the current sample with complete data. The results from independent samples t test indicated that there was not significant difference between the individuals with missing data and the current sample in

32 the core measurements. The current sample (with complete data) had significantly higher home warmth at age 3 (t=-2.60, p