Formation of Educational Expectations of Lower Socioeconomic Status Children

Abstract

The purpose of this study was to investigate the mediation effects of children’s cognitive and noncognitive traits on the relationship between dropout mothers’ traits and their children’s educational expectations and to examine the interaction effects of dropout mothers’ General Education Development (GED) on children’s traits and educational expectations. The data were drawn from a series of National Longitudinal Survey Data. This study demonstrated the effect of mothers’ self-esteem on children’s educational expectations, which were mediated through children’s cognitive ability and self-esteem after controlling for the mothers’ cognitive ability, self-esteem, and socioeconomic status. Furthermore, mothers’ GED attainment moderated the relationship between children’s self-esteem and educational expectations. These findings provide better understanding of the intergenerational connections between cognitive and noncognitive traits and educational expectations.

Keywords

educational expectations intergenerational transmission self-esteem dropout mothers’ GED

A considerable share of children’s characteristics are directly predetermined by interactions with their parents, while parent traits indirectly permeate children’s lives through social selection and allocation processes such as education and jobs (Duncan, Kalil, Mayer, Tepper, & Payne, 2005; Farkas, 2003; Jencks et al., 1979). Children’s educational expectations condition actions or efforts to accomplish their expected educational goals and thereby have a high correlation with educational attainment (Berndt & Miller, 1990; Spenner & Featherman, 1978). This study sought to investigate the intergenerational link in cognitive and noncognitive ability underlying children’s educational expectations and to examine the effect of mothers’ GED on children’s educational expectations, drawing on a series of National Longitudinal Survey Data.

Studies of children’s social mobility and development have focused on the transmission of intellectual and psychological traits across generations (Coopersmith, 1967; Duncan et al., 2005). Earlier, Coopersmith (1967) found that a mother’s and her children’s sense of self-worth are positively linked. Studies have depicted low self-esteem as a source of dysfunction in social life and high self-esteem as a source of advantage in the labor market (Heckman & LaFontaine, 2006; Rosenberg & Owens, 2001; Wilson & Portes, 1975). Furthermore, cognitive ability, a point of significant relation between children and parents, is considered the most salient factor in school and workplace performance (Farkas, England, Vicknair, & Kilbourne, 1997; Ganzach, 2000). Given that self-esteem and cognitive ability are powerful predictors of school achievement and labor market success, connectedness in these traits between parents and children implies that a parental deficit could prevent children from moving forward (Elder, 1994).

Cognitive and noncognitive traits have been placed at the center of differential educational attainment structuring social stratification. Understanding how children from low socioeconomic status (SES) families form their educational expectations may be a prerequisite to raising their educational attainment. However, less is known about how children’s and parents’ connectedness on these two traits works to form children’s educational expectations, especially lower SES children (Bowles & Gintis, 2000; Farkas, 2003; Jencks et al., 1979). Studies suggest that intergenerational connections on these traits may be due to SES and thus may be neutralized within the same SES (Midgett, Ryan, Adams, & Corville-Smith, 2002). Even so, not much attention has been given to variation within the same SES group. By focusing on dropout mothers and their children, findings from this study of intergenerational connections on cognitive and noncognitive traits and their effects on educational expectations can influence the development of effective social interventions aimed at increasing the educational attainment of low-SES children.

General Education Development (GED) is a prevalent way for dropout mothers to obtain a high school degree. However, studies have confined the effects of GED acquisition to recipients’ benefit through further education or employment and have underplayed the effects of obtaining a GED on children. Studies suggest that parents’ additional education changes their attitudes toward schooling and enhances their skill and knowledge on parenting (Elder, 1994; Taylor, Clayton, & Rowley, 2004). Dropout mothers who enhance their academic socialization skills through obtaining a GED may protect their children from low achievement due to low cognitive and noncognitive ability. In this respect, the effect of children’s traits on educational expectations is likely to depend on mother’s additional education, which may play a role in buffering children’s traits in a fragile environment.

Effects of Maternal Self-Esteem on Children’s Self-Esteem and Cognitive Ability

Earlier studies have compared the self-esteem of parents and their children across all SES levels and generalized the positive relationship between mothers’ and children’s self-esteem (see, e.g., Coopersmith, 1967). Bengtson, Biblarz, and Roberts (2002) listed three routes through which parents pass their self-esteem and values on to their children: “status inheritance, social learning, and parent-child relationships (solidarity)” (p. 26). Duncan et al. (2005) suggested four mechanisms for relating traits between parents and children: “genetic inheritance, socioeconomic resources, parenting, and role model/identity” (p. 26). With regard to modeling or the social learning process, Bandura (1968) suggested that since the self is formed via social interaction and parents’ behaviors are a critical element in their children’s social modeling, parents are considered to be the most important factor in the development of children’s self-esteem. With regard to solidarity, Bengtson et al. (2002) suggested two mechanisms through which parental support affects children’s self-value: direct effect and mediating effect between SES and children’s feeling of self-worth. In looking at the role of SES, monetary reward for parental traits in the labor market affects a parent’s capacity to invest in children’s psychological development (Bengtson et al., 2002; Duncan et al., 2005). In addition to these environmental approaches, behavioral genetic studies have suggested a genetic influence on the transmission of self-esteem (McGuire et al., 1999; Neiss, Sedikides, & Stevenson, 2003; Turkheimer, 2000). Duncan et al. (2005) evaluated genetic effects in intergenerational transmission as trivial but not ignorable.

Few studies have been conducted on the effects of maternal self-esteem on children’s cognitive ability, especially for adolescents of lower SES. Instead, literature has demonstrated a negative influence of mothers’ low self-esteem on children’s development through authoritative parenting styles (Aunola, Nurmi, Onatsu-Arvilommi, & Pulkkinen, 1999) and abuse and neglect (Oates & Forrest, 1985), which negatively influence the development of children’s cognitive ability. According to Surkan et al.’s (2008) neurodevelopment study of infants, lower maternal self-esteem decreased an infant’s mental ability due to the mother’s weak interaction with her infants, stemming from her inability to handle stress.

Effects of Children’s Self-Esteem and Cognitive Ability on Their Educational Outcomes

Few studies have been conducted of how children’s self-esteem influences their educational expectations. However, studies have documented the relationship between self-esteem and educational attainment, which are strongly related to educational expectations (Coopersmith, 1967). Rosenberg, Schooler, and Schoenbach (1989) explained the causality with the self-fulfilling characteristics of human behaviors to attain equilibrium with perceived self-worthiness. That is, high levels of self-worthiness lead individuals to take action to accomplish their beliefs about themselves. Other studies have challenged the causality between self-esteem and educational outcomes (Bachman & O’Malley, 1977; Midgett et al., 2002; Rubin, Dorle, & Sandidge, 1977). That is, family background or academic ability may compound these relationships. Midgett et al. (2002) demonstrated that after controlling for family background, the assumed relationship had disappeared. Gagne and St. Pere (2001) found that motivation does not predict better performance after controlling for cognitive ability, whereas Leeson, Ciarrochi, and Heaven (2008) showed that personality, including self-esteem, is related to academic achievement, independent of cognitive ability. Contrary to inconsistent research results on the relationship between self-esteem and educational outcomes, it has been well documented that children’s cognitive ability is the most salient determinant of their educational expectations and attainment (Gagne & St. Pere, 2001).

Interactive Effects of Mothers’ GED Achievement on Children’s Educational Outcomes

Mothers’ educational level is related to her children’s educational attainment, especially with lower SES children (Haveman & Wolfe, 1995; Magnuson, 2007). Studies have shown that this relationship stems from an educated mother’s increased levels of human, financial, social, and cultural capital (Coleman, 1988; Duncan & Brooks-Gunn, 1997). By providing a better home learning environment, stimulating cognitive development, or passing on beneficial traits, educated mothers directly or indirectly promote their children’s achievement in school (Taylor et al., 2004).

GED credentials are considered to be the equivalent of a high school diploma. Many GED studies have focused on employment and the wage effects of GED certification. Studies have demonstrated that a GED’s economic benefits are due to self-selection by ability, indicative only of the signal effect of GED acquisition (Cameron & Heckman, 1993; Heckman & LaFontaine, 2006). That is, in the labor market the GED credential signals that GED holders have better cognitive abilities than other school dropouts—a GED indicates higher cognitive ability than that found in uncertified dropouts. Other studies have asserted a positive effect of GED acquisition on economic outcomes, especially for women (Murnane, Willett, & Boudett, 1995). A focal point of this discrepancy among studies is whether preparation for the GED tests requires enough educational activities to enhance human capital (Heckman & LaFontaine, 2006; Murnane et al., 1995).

Ganzach (2000) asserted that the assumed linear relationship between cognitive ability and educational outcome is affected by a parent’s factors. That is, children’s positive educational outcomes are increased by either their own high cognitive ability or a parent’s high educational level. Ganzach showed that the relationship between cognitive ability and educational expectation in children depends on the level of the mother’s education. Thus, children’s cognitive ability has a greater influence on their educational expectations when the mother has lower rather than higher levels of education.

Research Questions for This Study

The first question asked in this study was whether within lower SES families, mothers’ self-esteem levels influenced children’s educational expectations as mediated through children’s cognitive ability and self-esteem. Does children’s cognitive ability mediate the relationship between mother’s self-esteem and children’s educational expectations after controlling for mother’s cognitive ability, children’s self-esteem, and SES? The possible influence of mothers’ self-esteem on their children’s cognitive ability has not been considered in previous studies of intergenerational transmission of individual attributes. Furthermore, this study hypothesized that children’s self-esteem would mediate the association between mothers’ self-esteem and children’s educational expectations after controlling for mother’s cognitive ability, children’s cognitive ability, and SES.

The second question was whether a mother’s GED influences the relationship discussed in the first study question. Does mothers’ GED send a more positive signal to their children about their educational expectations compared with uncertified mothers? If so, does mothers’ GED attainment offset the negative effects of their children’s low cognitive ability? This study explored the compensating effects of mothers’ GED achievement on their children’s low cognitive ability related to the children’s educational outcomes. Furthermore, this study hypothesized that mothers’ GED acquisition would increase desire for social mobility and provide enough investment in their children to offset the negative effects of low self-esteem on children’s educational outcomes.

Method

Data and Sample

This study used data collected from biological mothers and their children to minimize biases from nonpersistent parent figures. The data were derived from the National Longitudinal Survey of Youth 79 (NLSY79), the Children of National Longitudinal Survey of Youth (CNLSY), and the Young Adults of National Longitudinal Survey of Youth (YANLSY). CNLSY and YANLSY respondents are the biological children of female respondents in NLSY79, where 6,283 females aged 14 to 22 in 1979 were sampled (Bureau of Labor Statistics, 2006). Since 1986, CNLSY has collected data from the children of female respondents in the NLSY79. Children aged 15 years and older in CNLSY have been separated into YANLSY since 1994.

Samples consisted of GED-certified mothers’ 185 children and uncertified mothers’ 266 children. When the dependent variable, educational expectations of children ages 14 or 15, was measured, the mean age of the mothers was 36.16 years (SD = 3.30). To construct a sample, the following filtering process was used. First, the mother samples were selected from cross-sectional data in NLSY79 and had equal sampling weights. Second, the mother samples were constructed from respondents whose highest degree was less than high school and who did not formally graduate from high school. Thus, the mother samples were composed of two groups: GED-certified mothers and uncertified mothers. Third, GED-certified mothers were restricted to those who received their GEDs before 1987, when the mother’s self-esteem was measured in NLSY79, to control for effect of GED attainment on self-esteem. Fourth, from among the selected mothers, children fitting the study variables were selected. Fifth, the oldest sibling was selected.

Variables and Measures

Table 1 shows the variables and measures selected to address this study’s questions. Figure 1 depicts the measures for each variable after removing items with low factor loadings and adjusting for errors of correlated items.

Table 1.

Variables and Measures.

Variable	Indicator
Self-esteem (mother and child)	1. I am a person of worth
	2. I have a number of good qualities
	3. I am inclined to feel that I am a failure^a
	4. I am as capable as others
	5. I feel I do not have much to be proud of^a
	6. I have a positive attitude
	7. I am satisfied with myself
	8. I wish I had more self-respect^a
	9. I feel useless at times^a
	10. I sometimes think I am no good at all^a
Cognitive ability (mother)	1. Arithmetic reasoning
	2. Mathematical knowledge
	3. Word knowledge
	4. Paragraph comprehension
Cognitive ability (child)	1. Math
	2. Reading recognition
	3. Reading comprehension
Educational expectations	1. The highest year of school that respondents would (child) like to complete

Negatively worded items.

Figure 1.

Measurement and structural model.

Educational expectations

YNLSY asked respondents aged 14+ two questions relating to their educational prospects: one related to educational aspirations and the other related to educational expectations. This study used educational expectations as a dependent variable. Educational expectations measured the amount of schooling that respondents expected to achieve, reflecting on the reality of their situation. This was collected at age 14 or 15 from one item asking for “the highest year of school that respondents would like to complete.” Berndt and Miller (1990) demonstrated that educational expectations formed around this age are consistent and reliable and motivate the accomplishment of these children’s educational goals, implying that educational expectation is a good precedent for educational attainment. Factor loading and residual for this measure were fixed at 1 and 0, respectively.

Self-esteem

NLSY79 and CNLSY used the same Rosenberg self-esteem scales. This allowed us to use the same items to measure both mothers’ and children’s self-esteem. The items were scaled at four levels: strongly disagree (1), disagree (2), agree (3), and strongly agree (4). The mothers’ self-esteem was measured in 1987 when the mothers were between the ages of 22 and 28. For children’s self-esteem, items measured at the age of 14 or 15 were used. However, the literature showed that including all 10 items in a model leads to biases relating to item wording, that is, negatively or positively, even though including negative items was intended to remove response biases (Horan, DiStefano, & Motl, 2003). Diverse methods have been suggested in the literature. Marsh (1996) suggested that Rosenberg’s self-esteem scales are composed of one trait and the biased effects of negatively worded items should be removed or identified. Horan et al. (2003) suggested two solutions for seemingly two-factor constructs. One is to place a constraint by setting errors for negatively worded items to be correlated (Kenny & Kashy, 1992). The other is to define the effects of negative wording as a “response style trait” separated from self-esteem content traits (Billiet & McClendon, 2000). Both these methods commonly remove negative wording effects, but the latter allows for identification of wording effects (Horan et al., 2003). Meanwhile, Benson and Hocevar (1985) showed that young students or respondents with lower verbal ability have difficulty responding to negatively stated items. Since the samples considered in this study were young students and their mothers with less than a high school education, seemingly two-factor traits are more likely due to response style traits. This study used the former model, that is, “correlated uniqueness” (Kenny & Kashy, 1992, p. 168), to make the models parsimonious.

Cognitive ability

NLSY79 measured respondents’ cognitive ability in 1980 using the Armed Forces Qualification Test. This study used four test scores measuring mother’s verbal and mathematical abilities: arithmetic reasoning, mathematical knowledge, word knowledge, and paragraph comprehension. Children’s cognitive ability was obtained from a math and reading subset test of the Peabody Individual Achievement Test. Three raw scores measured at the age of 12 or 13 were used: math, reading recognition, and reading comprehension.

Group distinction: GED mother and dropout mother

The term GED mother measured whether a mother had achieved her GED credential. This study included mothers who dropped out of school but achieved their GED before 1987, when the mother’s self-esteem test was administered. Dropout mother represents mothers who achieved neither their high school diploma nor a GED.

Identification for measures

To address identification issues, factor structure and loadings were examined by applying a scree test of SPSS and common factor analysis. Items showing low factor loading were removed from the analysis, such as the following indicators: Nos. 8 and 9 for mother’s self-esteem and No. 1 for children’s self-esteem. Then, researchers determined whether factor loadings were distributed evenly and whether each factor loading was greater than .40. With regard to child’s cognitive ability, there was a high correlation (r = .72) between reading recognition and reading comprehension. This study used math and reading comprehension only to address empirical underidentification.

Analytical Strategy

Data analysis was conducted using structural equation modeling (SEM) and involved LISREL 8.80 (du Toit & du Toit, 2001; Jöreskog & Sörbom, 1996). SEM provides an advantage over the examination of measurement error and latent variables, especially measurement of self-esteem. Furthermore, the SEM approach allows for multigroup comparison to test the interaction effects of mothers’ GED certification on children’s educational expectations.

To evaluate model fit, this study used chi-square, root mean square error of approximation (RMSEA; Browne & Cudeck, 1993), Rho (Bentler & Bonett, 1980; Tucker & Lewis, 1973), and comparative fit index (CFI; Bentler, 1990). As good model fit indices, the cited references suggested the following values: RMSEA values less than .05, Rho values greater than .95, and CFI values greater than .96. Hu and Bentler (1999) provided slightly lower criteria for a good fit than the studies referred to above: RMSEA values less than .06, Rho values greater than .95, and CFI values greater than .95.

Figure 1 shows the measurement and structural model (hereafter called Mediation Model) reflecting analytical strategies to address the first research question in this study. Paths related to major research questions are marked with bold lines. To test the second research question, whether the size of the path coefficients in the Mediation Model differed due to a mother’s GED certification (hereafter called Interaction Effect Model), a three-step strategy for two-group interaction was employed (Jaccard & Wan, 1996). First, model fits for each group were tested separately. Second, we sought to determine whether factor loading across groups was invariant. Third, differences in path coefficients across groups were tested by placing equality constraints on coefficients.

Results

Descriptive Statistics

Table 2 presents descriptive statistics, zero-order correlations among variables, and simple comparisons of means between mothers’ subgroups. The results for correlations¹ showed a positive relationship between mothers’ and children’s cognitive ability (r = .19, p < .01), a positive relationship between mothers’ self-esteem and their children’s cognitive ability (r = .21, p < .01) and self-esteem (r = .12, p < .05), a positive relationship between children’s cognitive ability and self-esteem (r = .25, p < .01), and a positive relationship between children’s cognitive ability (r = .27, p < .01) and self-esteem (r = .18, p < .001) and educational expectations. Results for the t test showed that the GED mothers’ group exhibited greater cognitive ability and self-esteem and their children had greater cognitive ability and educational expectations than did the uncertified dropout mothers’ group and its children. However, there was no significant difference in children’s self-esteem between GED mothers’ group and the dropout mothers’ group (p > .05).

Table 2.

Descriptive Statistics, Correlation Among Variables, and Mean Difference Between GED Mother Group (n = 190) and Dropout Mother Group (n = 266).

		Correlations				Group Difference
Variable	M (SD)	1	2	3	4	GED	Dropout	t
1. Cognitive ability (mother)	46.44 (9.34)					47.97 (9.66)	45.39 (8.99)	2.80**
2. Self-esteem (mother)	3.09 (0.37)	.02				3.14 (0.40)	3.02 (0.36)	3.51***
3.Cognitive ability (child)	47.58 (9.95)	.19**	.21**			50.72 (10.63)	45.47 (8.89)	5.09***
4. Self-esteem (child)	3.08 (0.41)	.09	.12*	.25**		3.11 (0.42)	3.07 (0.40)	1.93
5.Educational expectation (child)	13.75 (2.28)	.02	.00	.27**	.18***	14.1 (2.15)	13.51 (2.33)	2.58**

p < .05. **p < .01. ***p < .001.

Factor Loadings and Model Fit for Mediation Model

Table 3 presents the completely standardized factor loading of indicators on latent variables. All factor loadings were significant and substantial in size. Factor loadings for self-esteem for negatively worded indicators, on which residual correlations were placed, did not differ considerably from those of positively worded indicators. This indicates that modeling correlating uniqueness for negatively worded indicators served to successfully specify the latent construct of self-esteem.

Table 3.

Completely Standardized Factor Loadings.

Indicator No.	Cognitive Ability (Mother)	Self-Esteem (Mother)	Cognitive Ability (Child)	Self-Esteem (Child)	Educational Expectations (Child)
Cognitive ability (mother)
1	.77
2	.92
3	.88
4	.72
Self-esteem (mother)
1		.70
2		.72
4		.71
6		.64
7		.57
3		.43
5		.40
10		.41
Cognitive ability (child)
1			.72
3			.80
Self-esteem (child)
2				.64
4				.61
6				.68
7				.69
3				.55
5				.51
8				.44
9				.46
10				.56
Educational expectations (child)
1					1.00

The chi-square value for the Mediation Model was significant, χ²(230) = 352.35, p < .001. Given the chi-square test’s sensitivity in a large sample, model fit was evaluated with other fit indices. Values for Rho and CFI were .960 and .967, respectively, indicating a good model fit (Bentler, 1990; Bentler & Bonett, 1980; Hu & Bentler, 1999). The value for RMSEA1 was .034, and the value for RMSEA2,² which was adjusted for sample size due to missing data, was .036. These RMSEA values were indicative of a good model fit (Browne & Cudeck, 1993).

Results for Mediation Model

Unstandardized coefficients for the latent variable Mediation Model are presented in Table 4 and Figure 2. The path model explained 12.6% of variances in children’s educational expectations, 10.9% of children’s cognitive ability, and 2.2 % of children’s self-esteem.

Table 4.

Latent Variable Mediation Model.

	Cognitive Ability (Mother)	Self-Esteem (Mother)	Cognitive Ability (Child)	Self-Esteem (Child)
Beta
Cognitive ability (child)
b	0.257	4.122
se	0.070	1.415
t	3.686	2.914
Self-esteem (child)
b	0.003	0.123
se	0.003	0.061
t	1.052	2.011
Educational exp (child)
b	−0.022	−0.494	0.093	0.909
se	0.017	0.357	0.022	0.386
t	−1.241	−1.384	4.165	2.356
Total effect
Educational exp (child)
b	0.005	0.003
se	0.017	0.350
t	0.313	0.010
Indirect effect
Educational exp (child)
b	0.027	0.497
se^a	0.009	0.156
t	3.057	2.579

Note. Educational exp = educational expectations.

Standard errors were calculated with the Sobel test.

Figure 2.

Mediation model.

The coefficient for total effect of mothers’ self-esteem on their children’s educational expectations was .003. The mediated effects and residual direct effect had different signs. The paths indicating the influence of mothers’ self-esteem on their children’s educational expectations took one of three routes. The mediated effect was composed of two paths via children’s cognitive ability or self-esteem, and the indirect effects of mother’s self-esteem on children’s educational expectations was significant (b = 0.497, se = 0.156, t = 2.579). As for the first path mediated by children’s cognitive ability, the effect of a mother’s self-esteem on her children’s cognitive ability was significant (b = 4.122, se = 1.415, t = 2.914). The effects of children’s cognitive ability on children’s educational expectations were significant (b = 0.093, se = 0.022, t = 4.165). Based on the joint significance test from MacKinnon, Lockwood, Hoffman, West, and Sheets (2002), the path coefficient for the indirect effects of mothers’ self-esteem on their children’s educational expectations via the children’s cognitive ability was significant, and the size of the coefficient was .383 (= 4.122 * .093).³ Second, the indirect effect of mothers’ self-esteem on their children’s educational expectations via the children’s self-esteem was significant; the path coefficient for the effects of mothers’ self-esteem on children’s self-esteem was .123 (se = 0.061, t =2.011), and the path coefficient for the effects of children’s self-esteem on children’s educational expectations was .909 (se = 0.386, t = 2.356) according to the joint significance test. The coefficient for the indirect effect of the mothers’ self-esteem on children’s educational expectations via children’s self-esteem was .112 (= .123 * .909). Third, the residual direct effect was insignificant (b = −0.494, se = 0.357, t = −1.384).

Results for Two-Group Interaction Effect Model

Table 5 shows procedures and results from the investigation into the interaction effect of mothers’ GED certification on their children. Step 1 in Table 5 showed that each model fit the data well. For the GED mother group (Model 1-1), fit indices were χ²(230) = 299.96, p < .01; RMSEA = .043; Rho = .950; CFI = .950. For the dropout mother group (Model 1-3), fit indices were χ²(230) = 293.61, p < .001; RMSEA = .034; Rho = .940; CFI = .950.

Table 5.

Model Fit Indices for Interaction Effect.

Step	Model	χ²	df	p	RMSEA	Rho	CFI
1
	Model 1-1 (GED)^a	299.96	230		.043	.940	.950
	Independence 1-1	1643.46	276		.164
	Model 1-2 (Drop)^b	293.61	230		.034	.940	.950
	Independence 1-2	1543.11	276		.158
2
	Model 2-1 (two-group)	593.57	460		.036	.950	.959
	Independence 2-1	3786.29	552		.161
	Model 2-2 (LY = IN)^c	608.18	479		.035	.954	.960
	Model 2-1 vs. Model 2-2	14.61	19	p = .747	.001	.006	.001
3
	Model 3-1	616.56	485
	Model 2-2 vs. Model 3-1	8.38	6	p = .211
	Model 3-2 [Cognitive (C) → Edu. (C)]^d	617.46	486
	Model 3-1 vs. Model 3-2	0.90	1	p = .342
	Model 3-3 [Esteem (C) → Edu. (C)]^d	621.55	486
	Model 3-1 vs. Model 3-3	4.99	1	p = .025
Final model		617.46	486		.035	.954	.959

Note. C = child.

GED mother.

Dropout mother.

Invariant factor loading between two groups.

Equality constraints were placed on these paths.

In Step 2, the two-group model (Model 2-1), which was simultaneously tested for each group’s model fit, also showed a good fit to the data. Fit indices were χ²(460) = 593.57, p < .001; RMSEA = .036; Rho = .950; CFI = .959. The fit indices of Model 2-2, in which patterns of factor structure such as factor loading, factor correlation, and errors were constrained to be invariant across two groups, showed a good fit, χ²(479) = 680.18, p < .001; RMSEA = .035; Rho = .954; CFI = .960. The comparison of Model 2-1 and Model 2-2, which investigated structural similarity between the two groups, revealed that there was no difference in the pattern of factor structure between the groups, χ²(19) = 14.61, p = .747. The same factor loadings between the two groups did not decrease or worsen model fit, which allowed this study to proceed in testing the difference in the magnitude of path coefficients for the two groups.

In Step 3, a test was performed to see which path coefficients were different or the same between the GED mother group and dropout mother group. In Model 3-1, paths except for two paths—“children’s cognitive ability → children’s educational expectation” and “children’s self-esteem → children’s educational expectations”—were set to be invariant across the two groups. The model fit was about the same as that for Model 2-2, χ²(6) = 8.38, p = .211, indicating that coefficients for other paths with the exception of the two paths in question were the same between the two groups. Models 3-2 and 3-3 tested the difference in coefficients by placing equality constraints on the regression paths between the groups. Test results showed that the path—“children’s cognitive ability → children’s educational expectations”—did not differ between the GED mother group and dropout group (Model 3-2), χ²(1) = .90, p = .342. However, Model 3-3 showed that placing an equality constraint on the path—“children’s self-esteem → children’s educational expectations”—between the two groups worsened the model fit significantly compared to Model 3-1, χ²(1) = 4.99, p = .025, indicating that the path to children’s educational expectations from the children’s self-esteem differed between the two groups. This p value was the same as the adjusted α value by a Bonferroni correction (.05/2 = .025). Given that the Bonferroni correction is conservative, this study considered the effect of children’s cognitive ability on their educational expectation to be significantly different between the two groups.

Figure 3 shows the differences in the magnitude of the path coefficients from children’s self-esteem to children’s educational expectations between the GED mother group and dropout mother group. For the dropout mother group, a one-unit increase in children’s self-esteem indicated a 1.48 year increase in child’s educational expectations and ability—its relationship was significant (t = 2.970). However, the coefficient for the GED mother group was insignificant (b = 0.031, t = 0.040). Results for other paths were similar to those for the Mediation Model shown in Table 4 and Figure 2.

Figure 3.

Interaction effect model.

Summary of Findings

Children’s cognitive ability and self-esteem mediated the relationship between mothers’ self-esteem and children’s educational expectations after controlling for mothers’ cognitive ability, self-esteem, and SES. Furthermore, mothers’ GED attainment moderated the relationship between children’s self-esteem and educational expectations.

Discussion

This study provided a picture of intergenerational transmission by considering cognitive ability and self-esteem simultaneously in accounting for the formation of lower SES children’s educational expectations. Two research questions are discussed here: (a) the mediating effect of children’s cognitive ability and self-esteem on the relationship between mother’s self-esteem and children’s educational expectations and (b) the interacting effect of mother’s GED attainment on children’s traits and educational expectations. Then, study limitations and suggestions for future research are discussed.

Mediation Effect

This study showed that even in a lower SES group, mother’s self-esteem exerted a salient influence on their children’s cognitive ability and self-esteem—key mechanisms in forming children’s educational expectations, even after considering mother’s cognitive ability and SES. Results contribute to those from previous studies in three ways. First, this study demonstrated a cross-trait influence between mothers’ self-esteem and children’s cognitive ability, which has not been examined in previous studies, in explaining children’s educational expectations. Higher self-esteem among low-SES mothers serves to enhance their children’s cognitive ability. These results insinuated that, to some degree, mother’s self-esteem was related to resource investment or academic socialization for the next generation (Sewell & Shah, 1968; Taylor et al., 2004). The virtue of self-worthiness could reinforce investment in that generation. Thus, children’s high cognitive ability, as influenced by mother’s high self-esteem, resulted in children’s high educational expectations. Second, the directions of the self-esteem effect were the same for mothers and children in accounting for children’s educational expectations. Low-SES children’s self-esteem was positively related to mother’s self-esteem, which was consistent with findings from Coopersmith’s (1967) study, which predicted an intergenerational transmission of self-esteem. Mruk (2006) listed the following sources of self-esteem: acceptance, virtue, influence, achievements, and competence. This positive link between low-SES mothers and their children indicated that mothers with higher self-esteem may be more skilled in practicing and encouraging self-esteem than mothers with lower self-esteem. Meanwhile, similar to Rosenberg et al.’s (1989) results, this study indicated that children base their educational expectations on their feelings of self-worth, as influenced by their mothers. The formation of expectations could be cultivated or driven by their motivation to maintain or protect self-esteem, as suggested by Rosenberg et al. (1989). Third, the residual direct effects of mother’s self-esteem on children’s educational expectations were insignificant, indicating that the Mediation Model for children’s self-esteem and cognitive ability explained the relationship between a mother’s self-esteem and her children’s educational expectations. This corresponded with results from Kaplan, Liu, and Kaplan (2001), which denied the direct influence of parents’ self-esteem on children’s educational outcomes. Furthermore, the mediated and direct effect of mothers’ self-esteem on children’s educational expectations was in opposite directions. “Inconsistent mediation” (MacKinnon, Krull, & Lockwood, 2007, p. 175) through children’s cognitive ability and self-esteem cancelled out the direct effect of mothers’ self-esteem on children’s educational expectations. Thus, the total effect of mother’s self-esteem was insignificant.

Two-Group Interaction Effect

However, the effect of children’s self-esteem on educational expectations depended on mother’s attainment of a GED; the interaction analyses for mothers’ subgroups revealed that the relationship between children’s self-esteem and educational expectations differed according to the mother’s attainment of a GED. The effect of self-esteem on educational expectations was not significant among children whose mothers had attained their GED. On the other hand, the self-esteem of children whose mothers had not achieved a GED affected those children’s educational expectations significantly and positively. This implies that GED-certified mothers are more proficient in providing a source of self-esteem or encouraging children’s self-esteem than are non-GED-certified counterparts. For this reason, low self-esteem is a risk factor for children’s educational expectations: mother’s GED indicates strong resilience—a factor that acts to mitigate children’s disadvantages due to life in vulnerable environments. Thus, mother’s GED attainment is a protective factor for children with low self-esteem.

This study revealed that mother’s GED achievement exerts a positive effect on the next generation, beyond a GED holder’s labor market advantage—the focus of previous studies. Furthermore, since the self-esteem of non-GED mothers’ children is not related to educational expectations, findings from this study fit with those from other studies that have suggested a positive correlation between self-esteem and educational outcomes due to SES and cognitive ability (Baumeister, Campbell, Kreuger, & Vohs, 2003; Midgett et al., 2002). Thus, this result may provide some clues to the inconsistent results reported in a previous study on self-esteem and educational outcomes. Unlike Ganzach’s (2000) result, which showed a greater positive connection between cognitive ability and educational expectations for less-educated mothers’ children, no interaction was found between mothers’ GED and children’s cognitive ability as these relate to educational expectations. Mothers’ GED attainment is not enough to offset the negative effects of children’s low cognitive ability. This may indicate that GED attainment does not guarantee that academic socialization or investment activity will compensate for children’s low cognitive ability.

Limitations and Suggestions for Future Study

This study had some limitations. First, children’s educational expectations were used as dependent variables since many respondents were still under “school age,” that is, age 25. Even though the CNLSY asked children about their educational expectations, which differ from educational aspirations, the educational attainment of low-SES children is vulnerable to the influence of family income. Further study is needed to confirm that the relationships described above may be generalized to educational attainment. Second, educational expectations were measured with a single item, which may lead to measurement error. Therefore, future studies may release measurement error by constructing multiple items or applying longitudinal design. Third, while this study focused on relationships among personal variables such as cognitive and noncognitive traits, other environmental variable such as parenting style may intervene in such relationships. Therefore, attention to intervening processes that affect cognitive and noncognitive ability is also needed. Fourth, potential biases in the CNLSY, which sampled young mothers’ children (Mayer, Duncan, & Kalil, 2004), should be considered.

Footnotes

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biography

Kyung-Nyun Kim obtained a PhD from the Workforce Education and Development Program at The Pennsylvania State University. He is deputy director at the Korean Ministry of Education and research fellow at the Korean Research Institute for Vocational Education & Training. His research focuses on the relationship between education, work, and social mobility.

References

Aunola

Nurmi

Onatsu-Arvilommi

Pulkkinen

(1999). The role of parents’ self-esteem, mastery-orientation and social background in their parenting styles. Scandinavian Journal of Psychology, 40, 307-317.

Bachman

J. G.

O’Malley

P. M.

(1977). Self-esteem in young men: A longitudinal analysis of the impact of educational and occupational attainment. Journal of Personality and Social Psychology, 35, 365-380.

Bandura

(1968). A social learning interpretation of psychological dysfunctions. In London

Rosenhan

D. L.

(Eds.), Foundations of abnormal psychology (pp. 293-344). New York, NY: Holt, Rinehart & Winston.

Baumeister

R. F.

Campbell

J. D.

Kreuger

J. I.

Vohs

K. D.

(2003). Does high self-esteem cause better performance, interpersonal success, happiness or healthier lifestyles? Psychological Science in the Public Interest, 4(1), 1-44.

Bengtson

V. L.

Biblarz

T. J.

Roberts

E. L.

(2002). How families still matter: A longitudinal study of youth in two generations. New York, NY: Cambridge University Press.

Benson

Hocevar

(1985). The impact of item phrasing on the validity of attitude scales for elementary school children. Journal of Educational Measurement, 22, 231-240.

Bentler

P. M.

(1990). Comparative fit indexes in structural models. Psychological Bulletin, 107, 238-246.

Bentler

P. M.

Bonett

D. G.

(1980). Significance tests and goodness of fit in the analysis of covariance structures. Psychological Bulletin, 88, 588-606.

Berndt

T. J.

Miller

K. E.

(1990). Expectancies, values, and achievement in junior high school. Journal of Educational Psychology, 82, 319-326.

10.

Billiet

McClendon

(2000). Modeling acquiescence in measurement models for two balanced sets of items. Structural Equation Modeling, 7, 608-628.

11.

Bowles

Gintis

(2000). Does schooling raise earnings by making people smarter? In Arrow

Bowles

Durlauf

(Eds.), Meritocracy and economic inequality (pp. 118-136). Princeton, NJ: Princeton University Press.

12.

Browne

M. W.

Cudeck

(1993). Alternative ways of assessing model fit. In Bollen

K. A.

Long

J. S.

(Eds.), Testing structural equation models (pp. 136-162). Newbury Park, CA: Sage.

13.

Bureau of Labor Statistics. (2006). NLSY 79. Retrieved from http://www.bls.gov/nls/nlsy79.htm

14.

Cameron

S. V.

Heckman

J. J.

(1993). The nonequivalence of high school equivalents. Journal of Labor Economics, 11, 1-47.

15.

Coleman

J. S.

(1988). Social capital in the creation of human capital. American Journal of Sociology, 94, 95-120.

16.

Coopersmith

(1967). The antecedents of self-esteem. San Francisco, CA: Freeman.

17.

Duncan

G. J.

Brooks-Gunn

(1997). Consequences of growing up poor. New York, NY: Russell Sage Foundation.

18.

Duncan

Kalil

Mayer

S. E.

Tepper

Payne

M. R.

(2005). The apple does not fall far from the tree. In Bowles

Gintis

Groves

M. O.

(Eds.), Unequal chances: Family background and educational success (pp. 23-79). New York, NY: Russell Sage Foundation.

19.

du Toit

(2001). Interactive LISREL: User’s guide. Lincolnwood, IL: Scientific Software International.

20.

Elder

G. H.

Jr. (1994). Time, human agency, and social change: Perspectives on the life course. Social Psychology Quarterly, 57, 4-15.

21.

Farkas

(2003). Cognitive skills and noncognitive traits and behaviors in stratification processes. Annual Review of Sociology, 29, 541-562.

22.

Farkas

England

Vicknair

Kilbourne

B. S.

(1997). Cognitive skill, skill demands of jobs, and earnings among young European American, African American, and Mexican American workers. Social Forces, 75, 913-938.

23.

Gagne

St. Pere

(2001). When IQ is controlled, does motivation predict achievement? Intelligence, 30, 71-100.

24.

Ganzach

(2000). Parents’ education, cognitive ability, educational expectations and educational attainment: Interactive effects. British Journal of Educational Psychology, 70, 419-441.

25.

Graham

J. W.

Hofer

S. M.

(2000). Multiple imputation in multivariate research. In Little

T. D.

Schnabel

K. U.

Baumert

(Eds.), Modeling longitudinal and multiple-group data: Practical issues, applied approaches, and specific examples. (pp. 201-218). Hillsdale, NJ: Erlbaum.

26.

Haveman

Wolfe

(1995). The determinants of children’s attainments: A review of methods and findings. Journal of Economic Literature, 33, 1829-1878.

27.

Heckman

J. J.

LaFontaine

(2006). Bias-corrected estimates of GED returns. Journal of Labor Economics, 24, 661-700.

28.

Horan

P. M.

DiStefano

Motl

R. W.

(2003). Wording effects in self-esteem scales: Methodological artifact or response style? Structural Equation Modeling, 10, 435-455.

29.

Bentler

P. M.

(1999). Cutoff criteria for fit indices in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1-55.

30.

Jaccard

Wan

C. K.

(1996). LISREL analyses of interaction effects in multiple regression. Newbury Park, CA: Sage.

31.

Jencks

Bartlett

Corcoran

Crouse

Eaglesfield

Jackson

. . . Williams

(1979). Who gets ahead? The determinants of economic success in America? New York, NY: Basic Books.

32.

Jöreskog

K. G.

Sörbom

(1996). LISREL 8: User’s reference guide. Chicago, IL: Scientific Software International.

33.

Kaplan

D. S.

Liu

Kaplan

H. B.

(2001). Influence of parent’s self-feelings and expectations on children’s academic performance. Journal of Educational Research, 94, 360-370.

34.

Kenny

D. A.

Kashy

D. A.

(1992). Analysis of the multitrait multimethod matrix by confirmatory factor analysis. Psychological Bulletin, 112, 165-172.

35.

Leeson

Ciarrochi

Heaven

P. C. L.

(2008). Cognitive ability, personality, and academic performance in adolescence. Personality & Individual Differences, 45, 630-635.

36.

MacKinnon

D. P.

Krull

J. L.

Lockwood

C. M.

(2000). Equivalence of the mediation, confounding and suppression effect. Prevention Science, 1, 173-181.

37.

MacKinnon

D. P.

Lockwood

C. M.

Hoffman

J. M.

West

S. G.

Sheets

(2002). A comparison of methods to test mediation and other intervening variable effects. Psychological Methods, 7, 83-104.

38.

Magnuson

(2007). Maternal education and children’s academic achievement during middle childhood. Developmental Psychology, 43, 1497-1512.

39.

Marsh

H. W.

(1996). Positive and negative self-esteem: A substantively meaningful distinction or artifactors? Journal of Personality and Social Psychology, 70, 810-819.

40.

Mayer

S. E.

Duncan

Kalil

(2004). Like mother, like daughter? SES and the intergenerational correlation of traits, behaviors and attitudes (Harris School Working Paper Series 04.15). Chicago, IL: Harris School of Public Policy, University of Chicago.

41.

McGuire

Manke

Saudino

Reiss

Hetherington

E. M.

Plomin

(1999). Perceived competence and self-worth during adolescence: A longitudinal behavioral genetic study. Child Development, 70, 1283-1296.

42.

Midgett

J. D.

Ryan

B. A.

Adams

G. R.

Corville-Smith

(2002). Complicating achievement and self-esteem: Considering the joint effects of child characteristics and parent-child interactions. Contemporary Educational Psychology, 27, 132-143.

43.

Mruk

(2006). Self-Esteem research, theory, and practice: Toward a positive psychology of self-esteem (3rd ed.). New York, NY: Springer.

44.

Murnane

R. J.

Willett

J. B.

Boudett

K. P.

(1995). Do high school dropouts benefit from obtaining a GED? Education Evaluation and Policy Analysis, 17, 133-147.

45.

Neiss

M. B.

Sedikides

Stevenson

(2006). Genetic influences on level and stability of self-esteem. Self and Identity, 5, 247-266.

46.

Oates

R. K.

Forrest

(1985). Self-esteem and early background of abusive mothers. Child Abuse and Neglect, 9, 89-93.

47.

Rosenberg

Owens

T. J.

(2001). Low self-esteem people: A collective portrait. In Owens

T. J.

Stryker

Goodman

(Eds.), Extending self-esteem theory and research: Sociological and psychological currents (pp. 400-436). New York, NY: Cambridge University Press.

48.

Rosenberg

Schooler

Schoenbach

(1989). Self-esteem and adolescent problems: Modeling reciprocal effects. American Sociological Review, 54, 1004-1018.

49.

Rubin

R. A.

Dorle

Sandidge

(1977). Self-esteem and school performance. Psychology in the Schools, 14, 503-507.

50.

Sewell

W. H.

Shah

V. P.

(1968). Social class, parental encouragement and educational aspirations. American Journal of Sociology, 13, 559-572.

51.

Spenner

K. I.

Featherman

(1978). Achievement ambitions. Annual Review of Sociology, 4, 373-420.

52.

Surkan

P. J.

Schnaas

Wright

R. J.

Téllez-Rojo

M. M.

Lamadrid-Figueroa

. . . Wright

R. O.

(2008). Maternal self-esteem, exposure to lead, and child neurodevelopment. Neurotoxicology, 29, 278-285.

53.

Taylor

L. C.

Clayton

J. D.

Rowley

S. J.

(2004). Academic socialization: Understanding parental influences on children’s school-related development in the early years. Review of General Psychology, 8, 163-178.

54.

Tucker

L. R.

Lewis

(1973). A reliability coefficient for maximum likelihood factor analysis. Psychometrika, 38, 1-10.

55.

Turkheimer

(2000). Three laws of behavior genetics and what they mean. Current Directions in Psychological Science, 9, 160-164.

56.

Wilson

K. K.

Portes

(1975). The educational attainment process: Results from a national sample. American Journal of Sociology, 81, 343-363.