Pathways from mothers' early social support to children's language development at age 3
Abstract
The relationships between early maternal social support, maternal psychological well-being, the home learning environment, and children's language skills at age 3 in Korean families were examined. We hypothesized that maternal social support would predict children's language development through its effect on maternal psychological well-being and the home learning environment. Structural equation modeling was employed to analyze the conceptual model, using data collected from 1,725 Korean families who participated in the Panel Study of Korean Children. Mothers' early social support predicted children's language skills via 2 indirect paths. First, greater social support predicted more positively adjusted psychological well-being; maternal psychological well-being was significantly associated with the quality of the home learning environment, which in turn predicted the positive development of language skills in children. Second, greater social support predicted a better home learning environment; again, the home learning environment positively predicted children's linguistic development. The direct paths from social support to language development were not significant. Social support during the child's infancy was observed to be beneficial for stimulating the home learning environment later. A comparison between working mothers versus nonworking mothers yielded no significant differences in the relationships between the variables. The results shed light on the importance of sensitive measures of social support in order to better understand its beneficial effects on mothers and young children.
Highlights
- The relationships between early maternal social support, maternal psychological well-being, the home learning environment, and children's language skills at age 3 in Korean families were examined.
- Structural equation modeling was employed to analyze the conceptual model, using data collected from 1,725 Korean families who participated in the Panel Study of Korean Children.
- Greater social support predicted more positively adjusted psychological well-being; maternal psychological well-being was significantly associated with the quality of the home learning environment, which in turn predicted the positive development of language skills in children.
1 INTRODUCTION
As a child grows, support from family members, friends, and communities helps a mother to cope with parenting stress (Koeske & Koeske, 1990), gain competence in rearing a young child (Warren, 2005), and provide a better environment for the child's development (Burchinal, Follmer, & Bryant, 1996). Social support can be defined as “the social resources that persons perceive to be available or that are actually provided to them by nonprofessionals in the context of both formal support groups and informal helping relationships” (Gottlieb & Bergen, 2010, p. 512). Based on its functional elements, social support can take a range of forms, including instrumental, emotional, informational, companionate, and esteem support (Gottlieb & Bergen, 2010). The structural elements of social support can be conceived of as the social network, which may be formal (professionals) or informal (family), (Shumaker & Brownell, 1984). When parents are supported by individuals in their personal social network, and possess greater emotional and psychological resources, they are more likely to be stimulating in their parenting, which might in turn predict better language development for the child. Especially when mothers work, a social network would be an important factor helping mothers balance work and family (Erdwins, Buffardi, Casper, & O'Brien, 2001), and would eventually promote the child's development. However, few studies have examined the pathways through which mothers' social support predicts children's developmental outcomes.
Several pathways connecting mothers' social support to children's language development are possible. First, there might exist a positive direct effect of mothers' social networks on children's early language. A mother's social network might provide her child with various interactions with people (Cochran, 1993), which might stimulate the child's language acquisition. Second, social support can influence a child's language development indirectly through its effects on the mother's well-being. As one of the contextual sources of parenting (Belsky, 1984), social support might enhance mothers' access to emotional and material assistance, child-rearing controls, and role models, which in turn help them to promote children's language acquisition. Third, social support can influence the more proximal contexts for the child such as the home learning environment. There is a converging evidence that when families provide a home environment with learning, academic, and language stimulation and varied experience (e.g., the child has toys that teach color, size, and shape and is encouraged to learn the alphabet and numbers), children are more likely to demonstrate better language acquisition and usage during their first few years (Bradley, Corwyn, Burchinal, McAdoo, & Coll, 2001; Morrison & Cooney, 2001; Murray & Yingling, 2000; Son & Morrison, 2010). It is important to understand how social support can help mothers provide an adequate home environment that promotes their children's language development.
Despite great emphasis on the importance of social support for mothers with young children, there is a little empirical evidence of the direct and indirect contribution of social support on the children's developmental outcomes. The present study sought to further enhance understanding of how mothers' early social support translates to children's language skills at age 3 in Korean families. Specifically, the current study examined the mediational roles of maternal psychological well-being and the home learning environment. Additionally, the current study investigated the systematic differences between employed mothers and stay-at-home mothers in the relationship between social support and children's language development.
1.1 Social support among Korean mothers with young children
It is possible that social support has different functions depending on the cultural context. Factors associated with ethnicity or culture, such as differences in employment opportunities or the nature of kinship, may result in variations in parents' and children's responses to ecological factors (McLoyd, 1990). Thus, it may not be appropriate to draw conclusions about the role of social support within one culture from research on another population. Similar to the findings in Western societies, in Korean literature, social support has been a strong predictor of parenting stress (Heo & Kim, 2004; Kim & Kang, 2011), parenting efficacy (Moon & Min, 2012), depression (Chun & Ok, 2011), parenting sensitivity (Shin & Hwang, 2008), and children's intellectual and socioemotional outcomes (Rhee, 1997). Employed Korean mothers tend to utilize greater social support compared to nonworking mothers, especially in the form of direct provision of childcare by their own parents, parents-in-law, or relatives (Lee & Chin, 2013).
Meanwhile, Korean mothers have their own issues to deal with. They are faced with a remarkable growth in women's participation in the labor force, a rapid increase in nuclear families, and a lag in the development of systematic and reliable childcare service in the last few decades. In 2015, Korean women's participation in the labor force reached 51.8%; the number of three-generation households has decreased by more than half, from 17.6% in 1980 to 8.2% in 2010 (Statistics Korea, 2016), making informal help from the extended family members less available. These social changes make it challenging for Korean mothers to deal with their numerous responsibilities. Due to the decreasing fertility rate, Korean society is trying to develop parenting-aid policies. One of the most crucial elements would be strengthening mothers' social networks, including family relationships, and providing informal and formal social support for parenting (Song et al., 2009), calling attention to the role of social support in Korean families.
One may assume that in Asian cultures, the exchange of both instrumental and emotional support is more common compared to the more independent Western cultures, where the self is seen as fundamentally separate from others (Markus & Kitayama, 1991); thus, social support would have a greater benefit for mothers. In fact, however, the opposite may be the case. Because Asian cultural norms emphasize maintaining harmony within the social group, any effort to bring personal problems to the attention of others or enlist their help may risk undermining harmony and/or making inappropriate demands on the group. Studies showed that European Americans are more likely to report needing, seeking, and receiving social support than are Asians and Asian Americans (Hsieh, 2000; Shin, 2002; Taylor et al., 2004; Wellisch et al., 1999). Furthermore, social support may even have negative-buffering effects for Asians. When receiving a higher level of social support, the negative association between stress and adjustment became stronger for Chinese students, whereas social support lessened the negative effect of stress for American students (Liang & Bogat, 1994). The present study extends research on mothers' social support, which has mostly focused on mothers in Western countries, to a population of East Asian mothers based on a nationally representative Korean sample.
1.2 Social support and children's language development
Mothers' social support can influence children's language development both directly and indirectly. When the child has contact with a range and variety of persons on a recurring basis, the extended interaction with persons in the mothers' network can provide the child with developmental stimulation (Cochran, 1993). For example, the overall size of the early social networks of mothers with low socioeconomic status was a strong predictor of their children's expressive and receptive language performance at 3 to 4 years of age (Bee et al., 1982). In one Korean study, when mothers perceived greater support from the partner, family members, and friends, their children showed better performance in vocabulary awareness and language understanding skills (Rhee, 1997). In spite of the empirical findings, it is not clear why and how mothers' social support benefits their child's development. Is it the amount of interaction, family structure, or extra help with childcare that predicts better language skills? Moreover, simply having more people around the family may not necessarily provide positive interactions and stimulations. It is also possible that mothers can learn bad parenting practices such as inappropriate discipline or spend excessive time or energy on social networking and pay less attention to their child's needs. Therefore, more investigation of the specific pathways connecting mothers' social support and their child's language skills is needed.
Mothers can learn about appropriate parenting practices from the social network, or they might be better equipped to handle parenting stress when they have support from other people within and outside the family (Bee, et al., 1982; Cochran, 1993; Koeske & Koeske, 1990). Social support, then, can indirectly influence the child's language development through better parenting or well-being of the mother. However, there exist only a few studies investigating the indirect effects of social support. In a concept development study for a theoretical framework of social support, Armstrong, Birnie-Lefcovitch, and Ungar (2005) proposed that a sense of belonging can contribute to a parent's sense of wellness, and assistance with problem solving by friends or family members can contribute to parenting, and as a consequence, their child may demonstrate better developmental outcomes. Although subsets of the paths in this framework have been empirically supported, the entire picture in this hypothesis remains to be tested in the literature. The current study attempted to set up a greater picture including possible links between social support, maternal well-being, home learning environment, and children's linguistic development.
1.3 Social support and mothers' psychological well-being
According to the social support theory, social support can function as a stress mediator or as a coping resource (i.e., a buffering effect) that can protect parents from the negative effects of stress, while having its own direct positive effect on well-being (i.e., a main effect; Cohen & Wills, 1985). The relationship between social support and maternal psychological well-being has been well established. Social support of mothers with young children predicts less depressive mood (Cairney, Boyle, Offord, & Racine, 2003; Haslam, Pakenham, & Smith, 2006) and parenting stress (Koeske & Koeske, 1990), as well as greater competence in parenting (Haslam et al., 2006; Warren, 2005).
However, the research has been mostly focused on families under stress or crisis, including mothers with an infant (Burchinal et al., 1996; Cutrona & Troutman, 1986), mothers with a child with a chronic disease or disability (Bromley, Hare, Davison, & Emerson, 2004; Dyson, 1997; Horton & Wallander, 2001; Weiss, 2002), single mothers (Ceballo & McLoyd, 2002; Simons, Beaman, Conger, & Chao, 1993), and mothers with low socioeconomic status (Burchinal et al., 1996; McLoyd, Jayaratne, Ceballo, & Borquez, 1994). Considering social support can exert positive effects on an individual's well-being, parents can benefit from having supportive friends and family members within their social network even when they are not under particular stress (cf. Cohen & Wills, 1985). For example, Weiss (2002) compared mothers of typically developing children, children with autism, and children with mental retardation. There were group differences in depression, anxiety, somatic complaints, and burnout; however, social support was equally predictive of depression and parenting efficacy for all mothers. The current study focused on the main effect of social support by investigating mothers of healthy, normally developing children.
1.4 Social support and home learning environment
It is well known that social support from family members and friends can influence mothers' parenting practices (Izzo, Weiss, Shanahan, & Rodriguez-Brown, 2000; Macphee, Fritz, & Miller-Heyl, 1996; McLoyd, 1990). Greater social support predicts sensitive mother–infant interaction (Crnic, Greenberg, Robinson, & Ragozin, 1984), greater stimulation of infants (Adamakos et al., 1986), and more effective parenting strategies (Ceballo & McLoyd, 2002). However, the variables studied have been limited to the relational and behavioral aspects of parenting. Relatively, less attention has been paid to the overall adequacy and richness of the home learning environment.
The overall quality of the home environment reflects an array of parenting practices throughout the house and the daily lives of the family. This includes proximal parenting behaviors, such as providing educational interactions and activities, as well as distal practices of making learning materials available at home (Bradley & Caldwell, 1995). Because the home learning environment measure, by its nature, assesses the processes of parenting within a child's immediate environment (Bradley, Caldwell, Rock, Hamrick, & Harris, 1988), it is likely to be affected by the parents' psychological and emotional resources, as well as by their physical and financial resources. Parenting knowledge and strategies are learned and strengthened by the kinds of help and assistance provided by informal and formal social network members (Bronfenbrenner, 1979; Cochran, 1993). Although a body of research has reported that a family's socioeconomic status (Bradley et al., 2001; Davis-Kean, 2005; Guo & Harris, 2000; Linver, Brooks-Gunn, & Kohen, 2002; Votruba-Drzal, 2003) or family structure (Baharudin & Luster, 1998; Bradley & Caldwell, 1984; Evans, Maxwell, & Hart, 1999) are closely related to the home learning environment, much less is known about the contribution of contextual factors. A few studies showed that a larger social network for mothers or greater perception of social support predicted better quality in the home environment (Burchinal et al., 1996; Chang, 2016). However, these studies did not examine the mediating role of the home learning environment in the relationship between social support and a child's development.
Meanwhile, there is more direct evidence that greater psychological well-being of parents is related to a more prepared and developmentally stimulating home environment. Mothers were more likely to provide a better quality home environment when they were less depressed (Son & Morrison, 2010), more satisfied with their marriage, and had higher levels of self-esteem (Baharudin & Luster, 1998). A supportive home environment, again, predicted better academic achievement (Baharudin & Luster, 1998) and language development (Son & Morrison, 2010). These findings suggest that there might be a path connecting mothers' psychological well-being with children's development through the mediation of the home learning environment. The current study examined how social support plays a role in predicting the overall home learning environment, both directly and indirectly through its effect on psychological well-being.
Mothers' employment may be one of the contextual factors that affect the quality of the home learning environment (Belsky, 1984). Some studies reported a better home learning environment for employed mothers (Lleras, 2008; Son & Morrison, 2010). However, when family size and welfare use were controlled, the gap diminished, suggesting that employment status itself does not solely explain differences in home environment quality (Lleras, 2008). In addition to differences in the home learning environment quality, it might be useful to understand whether there is any difference in the extent to which social support and the home learning environment influence the outcomes for children depending on the mothers' employment status.
1.5 Employed versus nonemployed mothers
One can argue that the relationships between mothers' social support and language development would be weaker for the children of employed mothers, because the amount of time the mother and child spend together is limited. A shift of some of the locus of influence on children's development from mothers to the alternative caregiver or the childcare setting might explain the attenuated effects of family predictors. The empirical findings are, however, inconclusive. In a study comparing the predictive power of family factors between full-time childcare (which correlated highly, but not perfectly, with mothers' employment) and maternal care, Howes (1990) revealed that family structure and parental involvement were stronger predictors of verbal IQ for the children of maternal care. In the National Institute of Child Health and Human Development Early Child Care Research Network's study (NICHD ECCRN, 1998), weaker associations were found for the full-time childcare group compared to the mother-care group in the evaluation of mothers' sensitivity and positive play and children's expressive and receptive vocabulary at 36 months. However, maternal depression was more strongly related to poor language skills for children in full-time mothercare.
Social support might have different meanings for the psychological well-being of working versus nonworking mothers. On the one hand, it is possible that nonworking mothers benefit from communication and sharing with other people; otherwise, they might feel isolated and stressed. Hall, Williams, and Greenberg (1985) reported that the lack of a social network predicted depressive symptoms among only unemployed mothers. On the other hand, even though working mothers have a social network at work, emotional and practical support from relatives and friends could be more crucial when juggling work and family. In fact, one Korean study found that social support was a stronger predictor of parenting stress for employed mothers than for nonemployed mothers (Lee & Chin, 2013). In spite of all of the accumulated evidence on the importance of social support for working mothers (Benin & Keith, 1995; Parasuraman, Greenhaus, & Granrose, 1992), it is still not clear from the previous literature whether social support has greater benefits for working mothers compared to nonworking mothers.
1.6 Present study
This study extends the earlier research in three respects. First, it attempts to understand the greater picture, connecting maternal social support, psychological well-being, the home learning environment, and children's language development at age 3 by including all relationships in one model. This study focuses on the influence of early social support from friends and extended family members on the language development of typically developing 3-year-old children. A longitudinal approach was adopted to investigate any lagged effects of early social support. It was assumed that the parental support system from the social network may in fact increase in importance over time in later toddler and preschool years, when the parents' interactions within that support system are more capable and frequent (Crnic et al., 1984). However, considering the importance of social support on mothers' well-being and parenting during their children's infancy (e.g., Crnic et al., 1984; Haslam et al., 2006; Moon & Min, 2012; Warren, 2005), it is also possible that a mother's support experience when their child is an infant can have a lingering effect on later well-being, home learning environment, or language development. Second, by using the data for Korean mothers from the Panel Study on Korean Children (PSKC), this study attempts to provide empirical evidence of the role of maternal social support in a different cultural context than those of Western countries. Finally, this study compares the patterns and strengths of the pathways between employed versus unemployed mothers to investigate whether the effects of social support on psychological well-being, the home environment, and the child's language skills are any stronger for mothers working outside the home.
In the present study, PSKC data for children aged 1 to 3 years were used to examine the direct links between mothers' social support and children's language development and indirect links via the mediation of (a) maternal psychological well-being, (b) the home learning environment, and (c) both maternal psychological well-being and the home learning environment. This study hypothesized that mothers who perceived greater support from their extended family members, friends, or community would exhibit better psychological well-being, as indicated by greater self-efficacy, less parenting stress, and lower levels of depressed mood. We hypothesized that greater social support would predict a better home learning environment directly, as well as indirectly through its positive effect on maternal well-being. Home learning environments would predict better language skills in children. The current study also tested the possible direct effect of social support on better language skills in children. By adapting a longitudinal approach, we examined the lingering effects of early social support. We expected that mothers' early social support when their child is an infant would predict a greater psychological well-being and a better home learning environment 2 years later. Although previous findings are not consistent, the current study hypothesized that social support would be more strongly related to the other variables for working mothers compared to nonworking mothers. The conceptual model is depicted in Figure 1.
2 METHOD
2.1 Data and participants
Original data on 2,150 children during the first 3 years of their life were gathered through the PSKC. In 2008, the PSKC recruited mothers who gave birth in hospitals in Korea, using a stratified sampling technique. With the exception of those that fit the exclusion criteria (i.e., the mother could not communicate in Korean, the mother's postpartum health was poor, the infant or the mother had a serious illness, the infant was put up for adoption, the infant was from a multiple pregnancy, or the mother was aged 18 years or less), and those who refused to participate in the study, 2,562 households that met the requirements to participate in the study were contacted. Of the potential households, 2,150 households were sampled. From these, 1,725 families comprised the analysis sample of this study, with complete information from a 36-month assessment for the home learning environment and language development. About 51% of the children were male; the children were 38.27 (standard deviation [SD] = 1.47) months old on average, and about 99% of the mothers were married, with an average of 14.36 (SD = 1.97) years of education. Compared to children who were excluded from the study as a result of missing data (N = 425), the children from the analysis sample were not significantly different with regard to gender (child = male 48.5% vs. 51.6%, t = −1.10, n.s.), mothers' years of education (14.33 vs. 14.28 years, t = .01, n.s.), marital status (married 99% vs. 99%, t = .01, n.s.), and family monthly income (₩3,195,300 vs. ₩3,282,000, t = −.99, n.s.). However, mothers who dropped out of the study were slightly older (M = 31.79, SD = 3.88) than the mothers who stayed (M = 31.24, SD = 3.67, t = −2.70, p < .01).
2.2 Procedures
The data were collected during home visits when the child was 1 and 3 years of age. The home visit was scheduled from June to October in 2009 for the age 1 assessment, and from June to November in 2011 for the age 3 assessment. The home visit was scheduled by calling the families in advance. About a week before each home visit, a family background questionnaire was sent to the families and collected at the time of the home visit. Two research staff members visited the families at their home, and the home visit lasted about 45–90 min. At each assessment, the mothers completed questionnaires regarding their level of social support, depression, parenting stress, and self-efficacy. The Early Childhood Home Observation for Measurement of the Environment (EC-HOME; Caldwell & Bradley, 2003) was conducted by assessors extensively trained in its administration. The interviewer visited each child's home for the 36-month-assessement and completed the 55-item EC-HOME inventory. In this study, the EC-HOME was administered through direct observation and a semi-structured interview with the mother. Training of the assessors was conducted by the Korean Institute of Child Care and Education for 2 weeks. The full EC-HOME takes approximately 10–45 min. Children's language skills were assessed during the 36-month visit using the Receptive and Expressive Vocabulary Test (REVT; Kim, Hong, & Kim, 2009). The assessors were trained by the developers of the REVT. The duration of the assessment was about 30–60 min.
2.3 Measures
2.3.1 Social support
Maternal social support was measured using the social support questionnaire developed by Lee and Ok (2001) when the child was aged 1 and 3. The questionnaire included 12 items that assessed the mothers' emotional, instrumental, informative, and social support from family members who did not live in the same household, friends, the neighborhood, or the community. The items were assessed using a 5-point scale (1 = “not at all” to 5 = “all the time”), and the sum of the 12 items was calculated for the analysis. The measure had high internal reliability (α = .91 at age 1 and α = .93 at age 3). The mean score was 46.67 (SD = 7.65) at age 1 and 45.73 (SD = 7.65) at age 3; scores ranged from 12 to 60 at ages 1 and 3.
2.3.2 Maternal psychological well-being
The latent factor of maternal psychological well-being was represented by three indicators, namely the mother's self-efficacy, depression, and parenting stress. The mother's self-efficacy was measured using 4 items from the Pearlin Self-Efficacy Scale (Pearlin & Schooler, 1978). The Korean version of the measure has been widely used in Korean literature (Lim, Ko, Kim, & Park, 2008; Noh, 2014; Shin & Ahn, 2014). The mothers completed a series of 5-point scaled items at the 12- and 36-month interviews. Items included “I cannot control the things happening around me” and “There are some problems I cannot solve by myself whatsoever.” Higher scores on each item indicated more negative self-efficacy, and the sum of the recoded scores was used for the analysis. There was a possible range of 4 to 20 for the self-efficacy score. The self-efficacy measure had high internal consistency (α = .88 at age 1 and α = .82 at age 3). The mean scores were 14.77 (SD = 2.81) at age 1 and 16.84 (SD = 2.75) at age 3.
Maternal depression was assessed using the Korean version of the K6, Kessler et al. (2002) 6-item scale of depression. This scale includes items such as “during the past 30 days, about how often did you feel so depressed that nothing could cheer you up?” The items were assessed using a 5-point Likert scale, the sum of which was used in the analysis. Validation studies have suggested that the K6 can be a valid and reliable screening tool of psychological distress for Korean populations (Kim, 2011; Min & Lee, 2015). The reliability of the K6 for this study was high, with a Cronbach's alpha of .91 at both ages 1 and 3. The mean score was 11.47 (SD = 4.19) at age 1 and 11.86 (SD = 4.50) at age 3; scores ranged from 6 to 30.
For us to assess the mothers' parenting stress, a subscale of “parenting distress” was taken from the parenting stress scale developed by Kim and Kang (1997), based on the parenting stress index-short form (Abidin, 1990). The items from the parenting stress index-short form were translated and modified to better represent the parenting experiences of Korean parents. The parenting distress subscale of the parenting stress scale demonstrated a moderate correlation (r = .47) with the general health questionnaire (Goldberg, Morris, Simmons, Fowler, & Levinson, 1990), indicating acceptable validity (Kim & Kang, 1997). The scale was composed of 11 items, such as “my life was happier before my child was born” and “there are times when I want to run away from my child.” The items were assessed using a 5-point Likert scale, the sum of which was used in the analysis; the internal consistency of the parenting stress scale in this study was α = .88 at age 1 and α = .84 at age 3. The mean scores were 27.30 (SD = 6.45) at age 1 and 30.62 (SD = 7.07) at age 3; the scores ranged from 11 to 55.
2.3.3 Home learning environment
As indicators of the home learning environment, four selected subscales from the original version of the home observation for measurement of the environment (HOME) inventory (Caldwell & Bradley, 2003) were used. These four subscales were learning materials, language stimulation, academic stimulation, and variety of experience. These scales tap the stimulation of the cognitive aspects of a child's development and are closely related to children's cognitive skills, whereas the other scales purport to measure social aspects (i.e., responsiveness and warmth) and physical aspects (i.e., organization of the physical environment) of the home environment (Bradley et al., 2001; Menaghan & Parcel, 1991). In previous studies, the four scales were used to indicate the quality of the home learning environment (e.g., Klebanov, Brooks-Gunn, & Duncan, 1994; Son & Morrison, 2010). The items were rated dichotomously (1 = “Yes” and 0 = “No”), and the total scores of each subscale were used as indicators of the latent factors of the home learning environment. Learning materials included 11 items that assessed the toys, books, and activities directed toward the cognitive development of the child (e.g., child has toys that help teach the names of animals). Language stimulation included seven items that assessed the amount and variety of verbal stimulation provided by the parents (e.g., parent uses complex sentence structure and vocabulary). Academic stimulation was composed of five items and measured whether the parent encouraged the child to learn basic reading concepts, colors, and numbers (e.g., child is encouraged to learn to read a few words). Variety of experience included nine items that referred to child enrichment opportunities and choices in daily life (e.g., child is taken on an outing by the family). The reliability and validity of the HOME have been proven to be adequate within a Korean sample (Jang, 1984; Kim & Jo, 2002; Kim & Kwak, 2007; Lee, 1992; Lee & Jang, 1982). For example, Kim and Kwak (2007) examined the test–retest reliability of the Korean translated EC-HOME for 49 families out of 294 families in the study sample (about 16%). The families were revisited within 1 month after the first visit. The correlations between the two assessments ranged from .76 to .97. The subscales showed an internal consistency ranging from .54 to .84, and correlated with the child's language and literacy skills. The EC-HOME items also report a very good level of inter-rater reliability in the literature ranging from .87 to .90 (Bradley & Caldwell, 1979; Elardo & Bradley, 1981; Wang, 2014). In the present sample, the Cronbach's alpha for the four subscales of the HOME ranged from .51 to .63. The mean for learning materials was 8.89 (SD = 1.21), and that for language stimulation was 6.44 (SD = .85). The mean scores were 4.58 (SD = .85) for academic stimulation and 7.58 (SD = 1.31) for variety.
2.3.4 Children's language skills
The child outcome of language development was represented by expressive and receptive vocabulary skills, assessed using the full version of the REVT (Kim et al., 2009). The REVT was developed to assess receptive and expressive vocabulary abilities from infancy to adulthood. The items were developed based on Korean vocabulary and language development literature, as well as from pilot studies. The REVT assesses the understanding of spoken words and vocabulary acquisition. Each of the receptive and expressive versions is composed of 185 items. At the 36-month assessment, children started with the first item; the items became increasingly difficult, and testing continued until the participant made 6 errors in 8 consecutive items. Each subject's responses were first scored according to the established scoring criteria for the REVT, and raw scores and standard scores were derived. The standard scores were included in the analysis. The split-half reliability of the REVT calculated by the Spearman–Brown formula was .94. The 2-week test–retest reliability of a subset was .84 or higher (Kim et al., 2009). The mean score of standardized expressive vocabulary was 29.89 (SD = 13.75), and that of receptive vocabulary was 33.13 (SD = 15.90).
2.3.5 Controls
Several demographic factors were included in the models as covariates to rigorously examine the predictability of social support, the psychological well-being of mothers, and the home learning environment on child language development. These background factors have also been reported to influence the mothers' social support (Koeske & Koeske, 1990; Schulz et al., 2006) and psychological well-being (McLoyd, 1990; Mistry, Lowe, Benner, & Chien, 2008), as well as the home learning environment (Baharudin & Luster, 1998; Bradley et al., 2001; Davis-Kean, 2005; Evans et al., 1999; Menaghan & Parcel, 1995; Son & Morrison, 2010). The covariate variables included the child's gender (a dummy variable indicating boy), the child's age, the mother's age, maternal education (years of completed education), the family's average monthly income, the total size of the household, and the mother's average working hours per day. These variables were measured during the 36-month assessment using a questionnaire. The descriptive statistics for all study and control variables are presented in Table 1.
| M | SD | N | |
|---|---|---|---|
| Controls | |||
| Child = male (%) | 51. 4 | 1,725 | |
| Child's age (months) | 38.27 | 1.47 | 1,725 |
| Mother's age (years) | 33.79 | 3.68 | 1,702 |
| Mother's education (years) | 14.36 | 1.97 | 1,718 |
| Family average income (10,000 won/month) | 398.42 | 245.74 | 1,643 |
| Total household size | 4.05 | .85 | 1,725 |
| Mother's average working hours (/day) | 2.89 | 4.03 | 1,722 |
| Social support age 1 | 46.47 | 7.65 | 1,649 |
| Social support age 3 | 45.73 | 7.65 | 1,675 |
| Mother's psychological well-being age 1 | |||
| Self-efficacy | 14.77 | 2.81 | 1,648 |
| Depression | 11.47 | 4.19 | 1,648 |
| Parenting stress | 27.30 | 6.45 | 1,655 |
| Mother's psychological well-being age 3 | |||
| Self-efficacy | 16.84 | 2.75 | 1,674 |
| Depression | 11.86 | 4.50 | 1,674 |
| Parenting stress | 30.62 | 7.07 | 1,674 |
| Home learning environment | |||
| Learning materials | 8.98 | 1.21 | 1,725 |
| Language stimulation | 6.44 | .85 | 1,725 |
| Academic stimulation | 4.58 | .85 | 1,725 |
| Variety | 7.58 | 1.31 | 1,725 |
| Children's language skills | |||
| Expressive vocabulary | 29.89 | 13.72 | 1,725 |
| Receptive vocabulary | 33.13 | 15.90 | 1,725 |
2.3.6 Analysis plan
To examine the relations depicted in the conceptual model (see Figure 1), we utilized structural equation modeling (SEM) as the major analysis technique, with Amos 22 (Arbuckle, 2013). SEM allows us to test hypotheses about both structural and measurement relationships and to evaluate entire models, including direct and indirect (i.e., mediating) effects, with a single model (Kline, 2005). With SEM, this study attempted to examine the relationships between the predictors (i.e., social support at ages 1 and 3), the mediators (i.e., maternal psychological well-being at ages 1 and 3 and the home learning environment at age 3), and the children's language development at the age of 3, and to estimate the total and decomposition effects (i.e., direct and indirect).
Social support was entered as an independent predictor variable, and psychological well-being, the home learning environment, and the language development of the child were represented by latent factors. Background variables were controlled by allowing paths from these variables to social support and all other latent factors in the model. For all of the paths, standardized path coefficients were reported that assessed the direct effect and that could be used as an effect size, r (Kline, 2005). Missing data were handled using full information maximum likelihood (Kline, 2005).
3 RESULTS
3.1 Preliminary analysis
The intercorrelations between the variables are shown in Table 2. Social support at ages 1 and 3 was significantly associated with all of the indicators of maternal well-being and the home learning environment. Social support at age 1 had a significant, albeit weak, association with the child's expressive and receptive vocabulary. Social support at age 3 was significantly related to expressive vocabulary, but not to receptive language at age 3. Social support at both ages 1 and 3 was consistently associated with the indicators of psychological well-being at both time points, as well as with all indicators of the home learning environment. The indicators of maternal psychological well-being at ages 1 and 3 showed significant associations with the home learning environment variables, with the exception of one correlation between self-efficacy at age 3 and the home academic stimulation. There was no direct association between the psychological well-being variables and the child's language test scores except for a negative correlation between parenting stress at age 1 and the child's expressive vocabulary. As expected, all of the home learning environment indicators were significantly associated with the child's expressive vocabulary scores.
| 1. | 2. | 3. | 4. | 5. | 6. | 7. | 8. | 9. | 10. | 11. | 12. | 13. | 14. | 15. | 16. | 17. | 18. | 19. | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1. Child = male | |||||||||||||||||||
| 2. Mother's age | -.03 | ||||||||||||||||||
| 3. Mother's education | .01 | .10*** | |||||||||||||||||
| 4. Family income | −.01 | .15*** | .28*** | ||||||||||||||||
| 5. Household size | −.06* | .14*** | −.06* | .05 | |||||||||||||||
| 6. Working hours | −.04 | .07** | .08** | .28*** | .03 | ||||||||||||||
| 7. Social support 1 yr | .01 | −.04 | .15*** | .04 | −.06* | −.01 | |||||||||||||
| 8. Social support 3 yr | −.05* | −.01 | .19*** | .05* | −.05 | −.02 | .60*** | ||||||||||||
| 9. Self-efficacy 1 yr | −.02 | .05* | .08** | .05* | −.04 | .03 | .30*** | .22*** | |||||||||||
| 10. Depression 1 yr | −.03 | −.04 | −.09*** | −.07** | −.02 | −.03 | −.29*** | −.23*** | −.55*** | ||||||||||
| 11. Parenting stress 1 yr | −.10 | −.05* | −.12*** | −.11*** | .00 | −.04 | −.28*** | −.22*** | −.47*** | .51*** | |||||||||
| 12. Self-efficacy 3 yr | .00 | .00 | .02 | .01 | −.05* | .00 | .21*** | .14*** | .43*** | −.34*** | −.29*** | ||||||||
| 13. Depression 3 yr | .03 | −.01 | −.07** | −.07** | .01 | −.04 | −.32*** | −.22*** | −.37*** | .47*** | .33*** | −.53*** | |||||||
| 14. Parenting stress 3 yr | .04 | −.03 | −.09*** | −.08** | .01 | −.06* | −.30*** | −.21*** | −.34*** | .37*** | .60*** | −.39*** | .47*** | ||||||
| 15. Learning matrl. | −.10* | .08 | .19*** | .14*** | −.03 | .04 | .12*** | .14*** | .10*** | −.10*** | −.10*** | .07** | −.07** | −.11*** | |||||
| 16. Language stiml. | −.09* | −.03 | .08** | .06* | −.10*** | .02 | .08* | .09*** | .08** | −.10*** | −.11*** | .06* | −.06* | −.11*** | .34*** | ||||
| 17. Academic stiml. | −.07*** | −.02 | .08** | .05 | −.09*** | .01 | .08* | .12*** | .06* | −.08** | −.10*** | .03 | −.07** | −.09*** | .36*** | .46*** | |||
| 18. Variety | −.09** | .03 | .09*** | .05* | .03 | .06* | .09*** | .07** | .07** | −.07** | −.09*** | .07** | −.08** | −.11*** | .34*** | .35*** | .27*** | ||
| 19. Expressive voca | −.10*** | .04 | .18*** | .04 | −.08** | −.01 | .08** | .07** | .04 | −.03 | −.06** | .01 | −.01 | −.04 | .13*** | .15*** | .15*** | .11*** | |
| 20. Receptive voca | −.08** | .01 | .06** | −.02 | .02 | .00 | .04 | .05* | .04 | −.01 | −.01 | −.01 | .01 | −.02 | .07** | .03 | .07** | .05 | .52*** |
- * p < .05.
- ** p < .01.
- *** p < .001.
3.2 Structural equation modeling
To examine the effect of mothers' early social support on the child's language development, we tested the first model with only social support at age 1 as an independent variable (Figure 2). The tested model fits the data reasonably well, χ2(73, N = 1725) = 176.220, p < .001, Comparative Fit Index (CFI) = 972, Incremental Fit Index (IFI) = 973, and Root Mean Square Error of Approximation (RMSEA) = .029. Consistent with the expectation, greater social support predicted a higher level of psychological well-being (β = .40, p < 001). Early social support (β = .09, p < .05) and psychological well-being (β = .14, p < .01) of mothers also predicted provision of a better home learning environment for their child 2 years later. When the home environment provided greater educational and cognitive stimulation and various activities, the child showed better language skills at age 3 (β = .19, p < .001). However, in contrast to the hypothesis, the direct association from the mother's early social support to the child's language skills was not significant.
The second model including fuller predictions from social support and psychological well-being at age 3 was statistically tested (Figure 3). As before, the model fits the data at an acceptable level, χ2(130, N = 1725) = 693.58, CFI = .917, IFI = .918, and RMSEA = .050. First, both social support and psychological well-being were stable over time. Social support at age 1 significantly predicted social support at age 3 (β = .58, p < .001), and psychological well-being at age 1 was also strongly associated with psychological well-being at age 3 (β = .81, p < .001). As hypothesized, mothers' social support positively predicted maternal psychological well-being (β = .40, p < .001 at age 1; β = .32 and p < .001 at age 3) as indicated by greater self-efficacy, fewer depressive symptoms, and a lower level of parenting stress. When the mothers exhibited better psychological well-being, they were more likely to provide a better home learning environment for their child (β = .14, p < .001). Social support at age 3 also directly predicted the home learning environment (β = .07, p < .05), suggesting that mothers with greater social support were more likely to provide their child with learning materials, language, and academic stimulation, as well as various out-of-home experiences. As expected, a better home learning environment predicted greater language skills in the children (β = .19, p < .001); however, there was no direct prediction regarding how social support (β = .04, ns) or maternal well-being (β = −.03, ns) affected the child's language skills at 36 months.
The addition of a more proximal indicator of social support did not account for the direct effect of mothers' early social support on the home learning environment. Social support during the child's infancy remained significant even after the indirect effect through more proximal social support was taken into account (β = .10, p < .01), indicating a lagged effect of early social support. We did not observe a lagged effect from early social support on psychological well-being at age 3 after controlling for well-being at child age 1. Rather, it seemed to predict later psychological well-being and the home learning environment via its effect on psychological well-being at age 1. A statistical test of indirect effects using the bootstrapping approach yielded significant indirect effects of social support at age 1 on psychological well-being at age 3 (standardized coefficient = .52, p < 01) and on the home learning environment at age 3 (standardized coefficient = .08, p < .05).
3.3 Group comparison
Within the sample used in the current study, 631 mothers (36.58%) were employed, and 1,094 mothers were unemployed or on leave. Compared to the nonworking mothers, the working mothers were older (34.78 vs. 33.56 years old, t = 3.36, p < .001), had more education (14.60 vs. 14.22 years, t = 3.88, p <. 001), and had a higher monthly family income (₩4,886,400 vs. ₩3,469,300; t = 11.70; p < .001). The working mothers typically worked an average of 7.92 hr per day (SD = 2.12). No significant group difference was found in the child's gender, the family monthly income, and the household size. The working mothers showed a lower level of depression when their children were aged 3 (11.56 vs. 12.03, t = −2.1, p < .05) and parenting stress when their children were aged 1 and 3 (26.85 vs. 27.57, t = −2.27, p < .05 at age 1; 30.05 vs. 30.95, t = −2.5, p < .05 at age 3). They also had a higher HOME variety score (7.67 vs. 7.53, t = 2.21, p < .05) than the nonworking mothers. No significant group differences were found in social support, self-efficacy, learning materials, language stimulation, or academic stimulation. The expressive and receptive language scores did not differ according to the mother's employment status (Table 3).
| Working (N = 631) | Nonworking (N = 1094) | t | |||
|---|---|---|---|---|---|
| M | SD | M | SD | ||
| Controls | |||||
| Mother's age | 34.18 | 3.58 | 33.56 | 3.72 | 3.36*** |
| Mother's education | 14.60 | 2.04 | 14.22 | 1.92 | 3.88*** |
| Family average income | 488.64 | 303.60 | 346.93 | 187.13 | 11.70*** |
| Total household size | 4.08 | .91 | 4.04 | .82 | .88 |
| Mother's working hours | 7.93 | 2.12 | 0.00 | 0.00 | 123.43*** |
| Social support | 45.80 | 7.47 | 45.68 | 7.75 | .30 |
| Psychological well-being | |||||
| Self-efficacy | 16.90 | 2.64 | 16.84 | 2.81 | .44 |
| Depression | 11.56 | 4.32 | 12.04 | 4.59 | −2.10* |
| Parenting stress | 30.05 | 6.81 | 30.95 | 7.21 | −2.51* |
| Home learning environment | |||||
| Learning materials | 9.05 | 1.06 | 8.94 | 1.28 | 1.75 |
| Language stimulation | 6.47 | .78 | 6.43 | .88 | 1.00 |
| Academic stimulation | 4.58 | .82 | 4.58 | .87 | .12 |
| Variety | 7.67 | 1.26 | 7.53 | 1.33 | 2.21* |
| Children's language skills | |||||
| Expressive vocabulary | 29.70 | 13.19 | 30.01 | 14.03 | −.45 |
| Receptive vocabulary | 33.12 | 15.79 | 33.13 | 15.98 | −.02 |
- Note: Degrees of freedom for the t test varied from 1,672 to 1,723, depending on the number of respondents for each measure.
- * p < .05.
- *** p < .001.
Before carrying out a between-group comparison of working versus nonworking mothers, a test of the measurement invariance assumption was conducted to ensure that the relationships of the latent variables with each of their indicators were identical across groups (Cheung & Rensvold, 2002; Widaman & Reise, 1997). For us to examine this assumption, intercepts, factor loadings, and uniqueness values for the measures of all of the latent constructs in the SEM were constrained to be identical across the groups. The restrictions resulted in well-fitted models for all multiple group analyses completed (all CFIs ≥ .912; all TLIs ≥ .890; all RMSEAs ≤ .049; and χ2 ratios ≤ 5.014), and all constrained models were not significantly different from the unconstrained model, indicating that the assumption was met. Systematic differences in the pathways between maternal social support, maternal psychological well-being, the home learning environment, and language development were tested by constraining the paths equally across the employment groups. The group comparison between working vs. nonworking mothers yielded no significant differences in the structural paths, Δχ2(11) = 97.301, p = .760. In sum, working mothers and nonworking mothers did not differ in the relationships between social support, maternal psychological well-being, the home learning environment, and children's language skills.
3.4 Mediations
We hypothesized that social support would predict children's language skills in complicated ways. Our study model examined the predictability of maternal social support, including a direct effect on the children's language skills and three indirect effects (a) through maternal psychological well-being, (b) through the home learning environment, and (c) through both maternal psychological well-being and home learning environment. To examine whether these mediational paths were statistically supported, we calculated and tested the indirect effects.
Because AMOS does not yield a statistical test for each indirect effect in the model, a z score was produced by using the product of the coefficient approach (see Hayes, Preacher, & Myers, 2011; Taylor, MacKinnon, & Tein, 2008). First, the indirect path from social support at age 3 to children's language skills, via only maternal psychological well-being, was not significant (z = −1.15, ns). Next, the indirect effect of social support at age 3 to home learning environment to child language skills was significant (z = 2.190, p < .05). Finally, the two-step mediation of social support at age 3 to psychological well-being to home learning environment to child language skills was also found to be significant (z = 3.812, p < .001).
Using the bootstrapping method, the total indirect effect was calculated. In the model with all three of the mediations tested above, the total indirect effect of social support at age 1 was significant (standardized coefficient = .026, p < .05), but that of social support at age 3 was not significant (standardized coefficient = .006, n.s.). When only two significant mediations (i.e., social support → home learning environment → language skills and social support → psychological well-being → home learning environment → language skills) were included in the model, the test yielded significant total indirect effects from social support at age 1 (standardized coefficient = .021, p < .05) and at age 3 (standardized coefficient = .020, p < .01) on the child's language. These results indicate that mothers' early social support when their children are aged 1 predicts the children's language skills later at age 3 through the multiple links of social support, psychological well-being, and home learning environment as measured at age 3.
3.5 Alternative models
To determine whether the pathways from social support to children's language skills in our model were reasonably proposed, we hypothesized different ways in which social support may work in its associations with the psychological well-being and the home learning environment and compared our study model with two alternative SEM models. The first alternative model included only direct paths from social support, psychological well-being, and the home learning environment to language skills. This model does not suggest a mediating role for the home environment. The second alternative model included only indirect paths from social support and psychological well-being on language development through the home learning environment, without an indirect effect via psychological well-being and a two-step indirect effect via psychological well-being, and the home learning environment. This model does not hypothesize that psychological well-being would function as a mediator between social support and the home learning environment or between social support and the child's language skills.
The study model with indirect effects showed a better model fit, χ2(693.581, df = 130), compared to the alternative model with only direct effects, χ2(134) = 1051.117, Δχ2(4) = 357.536, and p < .001. These results support the mediational role of maternal psychological well-being and the home learning environment. Another alternative model that did not include the indirect effects of social support on the home environment and on language development via psychological well-being was analyzed. In this model, social support and psychological well-being directly predicted the home learning environment, rather than suggesting a route from social support to maternal psychological well-being and direct paths from social support and psychological well-being to the child's language development. The model fit was χ2(134) = 1033.33, Δχ2(4) = 339.749, and p < .001, significantly worse than the original model, suggesting the mediating role of psychological well-being in the relationship between social support and parenting.
4 DISCUSSION
By exploring the influences of Korean mothers' social support on their psychological well-being, the home learning environment, and their typically developing children's language skills, this study suggests the importance of mothers' social support as a contextual resource for parenting and child development. The findings provided empirical evidence that social support can not only have positive effects on the home learning environment but can also indirectly enhance the home environment through its positive effect on maternal psychological well-being. The home learning environment was positively related to the child's language skills. The findings also empirically confirm the framework of social support proposed by Armstrong et al. (2005), in which they suggested there would be multiple pathways from mothers' social support to the development of children. Moreover, the current study contributed to the existing literature by adding information on the effects of maternal social support in an East Asian culture. Unlike previous findings that seeking social support for personal problems is considered inappropriate in Asian cultural contexts because emphasis is placed on maintaining harmony within the social group (Taylor et al., 2004), and social support may even cause stress among Asians (Liang & Bogat, 1994), Korean mothers and children in the current study seemed to benefit from receiving social support through enhanced psychological well-being and a better home learning environment. There were no systemic differences in the magnitudes of the links among the variables between working and nonworking mothers.
4.1 The effect of social support on psychological well-being and the home learning environment
First, the findings of this study supported the prediction that mothers' social support was significantly related to their psychological well-being. When the mothers in the sample thought they were receiving emotional, instrumental, social, and informational support from extended family members, friends, and community, they tended to function better psychologically; they reported greater self-efficacy, fewer depressive symptoms, and a lower level of parent stress. These findings are highly consistent with previous reports, in which social support demonstrated beneficial effects on maternal psychological well-being (Ceballo & McLoyd, 2002; Crnic et al., 1984; Cutrona & Troutman, 1986; Haslam et al., 2006). The significant relationship between maternal social support and the home learning environment suggested by several studies (Baharudin & Luster, 1998; Burchial et al., 1996; Son & Morrison, 2010) was also confirmed in this study. When mothers perceived higher levels of social support, they were more likely to provide a developmentally adequate and more stimulating learning environment throughout the house. Interestingly, mothers' social support during the child's infancy had a lingering effect on the home learning environment 2 years later, emphasizing the importance of help and support from the social network when the child is very young.
4.2 From social support to the child's language skills
The current study did not find a significant direct effect of social support on children's language skills. Although Bee et al. (1982) found a direct association between maternal social support and both the receptive and expressive vocabulary skills of children at 36 months, we only found a weak association between social support and expressive vocabulary, and no significant correlation with receptive language skills. This weak association was not detected once the mediating variables and controls were included in the analytic model. The direct effect of social support might have suggested that people in the mother's social network could provide some actual interaction with the child and promote the child's language skills (Burchinal et al., 1996). It is possible that more direct indicators of social support such as the size of the social network or the amount of time spent with other people can better reflect the amount of interaction and stimulation as well as children's language skills.
Rather, the current study revealed two significant pathways through which social support indirectly predicts the outcomes for children. First, greater social support for mothers predicted a better home learning environment, which, in turn, predicted their children's language skills. Second, social support was related to better psychological well-being for mothers, which predicted a more satisfactory home learning environment for children; again, the home environment significantly predicted the children's language development. This finding is in agreement with previous studies on social support and well-being (Cutrona & Troutman, 1986; Haslam et al., 2006; Jackson, 1998), social support and the home environment (Burchinal et al., 1996), and the home learning environment and children's linguistic development (Bradley et al., 2001; Morrison & Cooney, 2001); at the same time, it extends the previous findings by connecting these links within a wider picture. The findings address the importance of a social support system for mothers with young children, not only for mothers' well-being but also for the positive and healthy development of children.
Interestingly, the indirect route from social support to children's language skills via only maternal psychological well-being was not significant. The finding suggests that social support influences children's language development through the mediation of the home environment, rather than solely as a function of the mothers' psychological well-being. In this study, neither social support nor maternal well-being directly predicted children's language development. In other words, the adequacy of the home learning environment, indicated by age-appropriate learning materials, language and academic stimulation, and the variety of the child's experiences, plays an important role as a proximal context representing the children's language development, which can convey the somewhat distal aspects of parenting (i.e., mothers' social network and psychological health). Thus, the findings imply that the information and support the mother receives from her social network and enhanced psychological well-being help her to provide more stimulation and richer physical settings that are developmentally appropriate for her child. By conducting statistical comparisons of alternative models with the pathways arranged differently, this study confirmed the robustness of the hypothesized model. Moreover, even after controlling for a series of structural variables, we found a moderate predictive power of maternal social support and psychological well-being for the home learning environment.
4.3 Working versus nonworking mothers
To examine whether social support would have some universal positive influence in the lives of mothers and children regardless of their situation, this study attempted to examine whether the process in which social support predicts children's language skills differed as a function of maternal employment. No significant differences in the pattern or the strength of the paths among the variables were observed. It seems that despite the presence of social support or lack thereof, which might affect a working mother's well-being greatly because they are under stress due to managing multiple roles (e.g., Parasuraman et al., 1992), the beneficial effects of social support might be just as significant for nonworking mothers. The findings also suggest that the contextual and psychological aspects of parenting influence the mother's parenting practices and the child's development in a similar manner, whether or not the mother works outside the home or stays at home. This supports previous studies that found no differences in the family process according to the mother's employment status (Chang, 2013; NICHD ECCRN, 1998).
4.4 Limitations and future directions
In spite of the contributions of this study, several limitations deserve mention. First, the current study attempted to represent mothers' early experience with social support by including information on social support and psychological well-being when their children were aged 1 and 3. However, social support itself is a contextual factor and is not necessarily assumed to be stable over time (Crnic et al., 1984). Thus, increases or decreases in social support can be included as predictors of maternal well-being, the home environment, and children's development. Second, this study used a single measure, comprising the total score of the mothers' reports of how much instrumental, emotional, informational, and social support they were receiving. The form or source of the social support (i.e., mother, friends, or community) was not specified. Because it is possible that some aspects of social support, such as the amount and type of support, the size of the network, or the level of satisfaction with the support, might be differentially related to parenting (e.g., Weinraub & Wolf, 1983), further investigation that includes various factors related to social support is needed. For example, mothers of infants might need direct contact with their own mother or friends who can provide emotional support and actual help with childcare, whereas mothers with school-aged children might need information about the school life of the child and might receive more help from the mothers of the child's friends or even from an online community. Another limitation to this study is that SEM itself does not prove causal relations among the variables. There may be other causal directions (e.g., bidirectional relations between social support and well-being), additional relations, and the roles of variables not included in the model. Finally, this study is based on a nationally representative sample of Korean mothers. Because the culture of social networks and parenting is different from that in Western countries, the findings provide a culturally diverse addition to social support and home environment studies; however, one needs to be cautious in generalizing the findings. The present study leaves some intriguing questions to be answered in future research. For example, cross-cultural studies on social support, well-being, and the home learning environment may help in understanding the ecology of the process model of parenting. In particular, differences in family structure, the relationship between a parent and an adult child, and childcare policy might affect the role of social support in parenting and the home environment within diverse cultural contexts.
