Evaluating the Effects of Child Savings Accounts Program Participation on Parental Well-Being

Abstract

Objectives: Using baseline and second wave data, the study evaluated the impact of child savings accounts participation on parenting stress, personal mastery, and economic strain with N = 381 lower income parents who decided to join and those who did not join in a child development savings account program. Methods: Structural equation modeling for multiple group and time models was performed. Results: The degree of parenting stress at the first measurement occasion (M = 2.02, standard deviation [SD] = 0.34) was significantly different, Δχ²(2, n = 381) = 28.30, p <.001, (at the .017 level; Bonferroni correction) from the second measurement occasion (M = 1.93, SD = 0.34); however, the effect size was in the small-to-medium range (d = .25). Conclusions: Participation in such programs does not negatively affect the well-being of parents. Implications are directed toward practice, policy, and research.

Keywords

child development accounts college education economic strain parenting strain personal mastery

Introduction

Universal accounts established at birth and designed to provide structure and support for future asset accumulation and personal development are known as child development accounts or child savings accounts (Curley & Sherraden, 2000; Sherraden, 1991). Sherraden (1991) theorized that owning such assets has positive effects on children and families. His asset effects theory suggested that assets are positively related with, among others, marriage and family stability, general satisfaction, and increased individual self-efficacy for parents. Wolff (2002) contended that asset-poor children were more likely to be vulnerable to lifelong poverty.

Empirical findings show that participation in child savings accounts programs have positive effects on children and parents. Families with the same levels of income but differing assets typically had different levels of well-being and outcomes for their children (Wolff, 2002). Asset-poor households typically had fewer opportunities to invest in geographic mobility, education, home ownership, or in small businesses (Sherraden, 2003). For instance, children who participated in a 4-year quasi-experimental child savings program scored significantly higher on a financial literacy test than comparison group students (Sherraden, Johnson, Guo, & Elliot, 2009). For single mothers, assets affect maternal educational expectations, which, in turn, lead to greater academic achievement for their children (Zhan & Sherraden, 2003).

The study reported here comes from analysis of longitudinal data in a large quasiexperiment that is part of a national multimethod, multiyear demonstration of child savings accounts in the United States. This policy, practice, and research initiative is called Saving for Education, Entrepreneurship, and Downpayment (SEED). One method is a demonstration and impact assessment site at a large community-based agency in Michigan (Adams, 2008; Adams, Williams Shanks & Marks, 2008). Institutional mechanisms and outreach program services were built into this program to help parents open accounts and save for their preschool children’s higher education in Head Start programs.

Flowing from the assets effects theory, this study evaluated SEED program participation effects on the parental well-being outcomes of parenting stress, personal mastery, and economic strain. The focus was on parents who decided to open accounts for their children and those who did not; data were collected at baseline and the second (final) occasions. The study is organized into five main sections. The first section describes the SEED program intervention, while section two reviews literature on the well-being outcomes or constructs of interest among lower income households. Section three describes the method while section four presents the results. The final section is a discussion on the implications of the study findings.

Description of the SEED Program Intervention

SEED research team members from the University of Kansas (KU), the University of Michigan, and the Center for Social Development (CSD) at Washington University designed the SEED impact assessment and developed an extensive survey instrument so that outcomes such as school readiness, parenting practices, and family social and economic well-being for the treatment group can be measured against comparison group outcomes at the end of the initiative (Beverly & Williams Shanks, 2004). Of particular interest were parenting behaviors such as reading with their children and monitoring television time; children’s school readiness and early academic achievement; parent and child expectations and plans regarding college; a number of other age-appropriate measures of social and economic well-being for SEED participants and their families; as well as parents’ well-being outcomes including parenting stress, personal mastery, and economic strain.

Research team members developed matched pairs of preschool centers with similar enrollment and demographic characteristics including poverty rates, racial and ethnic composition, and proportion of one-parent families. One preschool center from each of the seven matched pairs was randomly assigned to a treatment condition and the other to a comparison condition. Using this research design and survey instrument, parents of children in the SEED treatment and comparison groups were interviewed by RTI International in fall 2004 (Marks & Rhodes, 2005). Follow-up data using the same technique as in the baseline data were collected in fall 2008. Human subject protection protocols were approved by the Institutional Review Boards (IRBs) at KU, the University of Michigan, and RTI International (Beverly & Williams Shanks, 2004).

SEED staff provided various social services in an effort to help parents enroll and save in the program. These efforts included various group and in-person orientation programs to parents about SEED; home visits by SEED staff; telephone and e-mail contacts to parents; and financial education classes. Similar to opening a regular savings account through a bank or a financial institution, the decision by parents to enroll was the first step toward saving and building assets for their children (Beverly & Williams Shanks, 2004). Like staff in other SEED programs across the country, the staff in this study found the recruitment phase of the parents to enroll in the program to be very challenging. Some parents expressed interest about enrolling in the program, completed the necessary administrative documents but still failed to enroll in SEED (Okech, 2010).

When parents decided to enroll and open accounts, an initial $800 through the funders was deposited into their accounts. An additional progressive incentive included $200 deposit from the State of Michigan for families with household incomes less than $80,000 a year. Further, personal deposits made by parents and others on behalf of children with SEED accounts were matched dollar for dollar during the 4 years of the initiative. After the end of the program in 2008, personal deposits can still be made but will no longer be matched. If the maximum amount that the initiative could match was saved on behalf of a child with a SEED account, there could be up to $3,400 in her account at the end of the 4-year period.

Parenting Stress, Personal Mastery, and Economic Strain Among Lower Income Families

It is assumed that most assets are built through institutional mechanisms (Howard, 1997; Oliver & Shapiro, 2006; Seidman, 2001; Sherraden, 1991; 2005) normally unavailable to lower income households. According to Sherraden (1991, 2005) and Beverly and colleagues (2008), similar structures, if provided to lower income families, might help reduce income and asset gaps that currently exist. These features, if implemented contextually, may lessen the economic and psychological stresses of lower income parents who are striving to ensure the long-term well-being of their children. Among the outcomes that may be affected by participation in child savings accounts include parenting stress, personal mastery, and economic strain.

Some literature on parental well-being describe the belief in one’s control over social and economic issues such as economic strain, personal mastery, and parenting stress as being key to optimal functioning. Economic strain affects the physical, social, and psychological well-being of adults in a variety of different arenas. Among lower income households, economic strain can significantly affect the functioning of parents, which in turn may impact the parent–child relationships (McLoyd, 1990) and how parents plan for children’s future well-being. Economic burden is also detrimental to health and well-being across the life course (Kahn & Pearlin, 2006) and is related to increases in depression (Dennis, Parke, Coltrane, Balcher, & Borthwick-Duffy, 2003) and to conflicts among family members and also among work place colleagues (Schieman, & Young, 2011). Dennis, Parke, Coltrane, Balcher, and Borthwick-Duffy (2003) found that economic pressure could decrease the emotional functioning of parents. Persistent economic strain may negatively impact on children’s emotional and social functioning (McLoyd, 1990; Takeuchi, Williams, & Adair, 1991) which may in turn impact on the well-being of parents.

McLoyd (1990) found that economic strain diminished the capacity for positive parenting and affected children’s behavior, in part, through the behavior of the parents to children. A recent study by Mistry, Lowe, Benner, and Chien (2008) found that lower income depressed mothers had difficulty managing their children’s behaviors while those who reported feeling more efficacious were more confident in monitoring and managing their children’s activities. In addition, mothers who had economic strain in affording desired goods for their households felt stressed, depressed, and less efficacious. These findings were somewhat different from a study by Crnic, Gaze, and Hoffman (2005) which found no evidence to support the notion that parent behavior mediates the association between parenting stress and child outcomes.

Personal mastery is crucial when dealing with pressures as it moderates the effects of other stresses on well-being (Spencer & Patrick, 2009). Therefore, the self-concepts of mastery and self-esteem are important personal resources for dealing with stresses. People with higher socioeconomic status, for example, have higher levels of mastery and self-esteem (Schieman, Pudrovska, Pearlin, & Ellison, 2006). However, effective coping methods or mastery are unequally distributed in society, with men, those more educated, and the affluent making greater use of these efficacious mechanisms (Pearlin & Schooler, 1978). Evidently, there is an underlying relationship between mastery, economic strain, and stress; mastery may moderate the effects of economic strain, which in turn may affect the parent–child relationship. The question then asked is how does parental well-being change over time in a child savings accounts program?

Participating in a long-term program to save for a child’s education like SEED can impact the well-being of lower income parents. Parents, may, for instance, experience pressures as they think and save for the future of their children. These pressures may, in turn, impact other areas of well-being, including how they relate with their children or their sense of mastery. Alternatively, participation in such programs, as postulated by Shobe and Page-Adams (2001), leads to positive psychological outcomes, including future orientation, which may in turn, positively affect parenting and mastery because of a belief by parents that the programs may enhance their children’s future chances of attaining higher education.

Research Questions and Hypothesis

As a result of the less-than-expected enrollment and participation levels in SEED among parents who had a chance to enroll, it became pivotal to assess the differences between those who joined and those who did not join the program. Figure 1 presents the flow of participants through the quasiexperiment to the decision to enroll in the program. The purpose of this study was to compare the effects of SEED program participation on parental well-being between lower income parents who joined and those who did not join a child savings accounts program. Groups here refer to those parents who opened accounts and those who did not while measurement occasions refer to baseline and second or final wave of the program. Specifically, we were interested in answering the following three research questions:

Figure 1.

Flow of participants through each stage—SEED baseline impact assessment survey (N = 381).

Do parenting stress, personal mastery, and economic strain have the same meaning across both measurement occasions and both groups?

Does participation in SEED affect the relationships between parenting stress, personal mastery, and economic strain across both measurement occasions and groups?

Does participation in the program affect the levels of parenting stress, personal mastery, and economic strain across both measurement occasions and groups?

Based on an earlier study by Okech, Little, Williams Shanks, and Adams (2011), we expected equivalence of the indicators and constructs across account holders and nonholders at baseline. Second, we expected program participation to have no effect on the relationships among parenting stress, personal mastery, and economic strain between the groups and across the measurement occasions. Finally, we did not expect significant differences on the degree (means) of parenting stress, personal mastery, and economic strain across both groups and measurement occasions.

Method

Sample Description

A total of N = 790 study participants were interviewed through a computer-assisted telephone interviewing for the baseline study. Of these, n = 381 (48.2%) were assigned to the treatment condition and were offered the opportunity to open SEED accounts for their preschool children, and n = 409 (51.8%) were the comparison group (Marks & Rhodes, 2005). Parents in the treatment group received program social services aimed at helping them join SEED and open accounts. Of the n = 381 parents in the treatment group, n = 235 (62%) decided to join the program by opening accounts while n = 146 (38%) did not. Of the 381 parents in the treatment group, n = 235 (62%) decided to join the program by opening accounts while n = 146 (38%) did not. These account holders and nonholders are the focus of comparison in this study.

Descriptive results at baseline showed that among account holders, there were 91% female, 51% White, 61% not married, 47% had some college or higher level of education, and the age range was 19–56, mean (M) = 30.75, standard deviation (SD) = 7.95. Further, 16.2% owned their homes and only 16.2% had household annual incomes of $35,000 or higher. Among those who did not open accounts, there was 90% female, 47% White, 63% married, 26% had some college or higher level of education, and the age range was 20–58, M = 29.90, SD = 7.93. Further, 20% owned their homes, while only 10% had household annual incomes of $35,000 or higher. Both groups were similar in most sociodemographic characteristics except for the education variable. Table 1 presents selected sociodemographic variables and program social services extended to the parents to join SEED.

Table 1.

Differences by Enrollment Status—SEED Baseline Impact Assessment Survey^a

	Did Not Enroll (n = 146)		Enrolled (n = 235)
Independent Variables	n	%	n	%
Individual and household-level variables
Education
Less than high school	56	38.4	61	25.9
High school or GED	52	35.5	67	28.5
Some college, voc or trade sch.	29	19.9	77	32.8
2-Year college degree or more	9	6.6	30	12.8
Household income
<$5,000	18	12.3	21	8.9
$5,000–$9,999	25	17.1	33	14.0
$10,000–$14,999	26	17.8	36	15.3
$15,000–$19,999	34	23.3	57	24.3
$20,000–$24,999	17	11.6	28	11.9
$25,000–$34,999	12	8.2	22	9.4
$35,000–$49,999	7	4.8	23	9.8
>$50,000	7	4.8	15	6.4
Receipt of means-tested assistance
Yes	97	66.0	166	70.6
No	49	34.0	69	29.4
Age at baseline	Range = 20–58 M = 29.90; SD = 7.93		Range = 19–56 M = 30.75; SD = 7.95
Telephone contacts
No phone contact	104	71.2	114	48.5
One phone contact	20	13.7	63	26.8
Two or more phone contacts	22	15.1	58	24.7
Mail contacts
No mailings	119	81.5	140	59.6
One or more mailings	27	18.5	95	40.4
In-person contacts
No in-person contact	119	81.5	85	36.2
One in-person contact	18	12.3	93	39.6
Two or more in-person contacts	9	6.2	57	24.3
Program orientation attendance
Did not attend orientation	102	69.9	36	15.3
Attended face-to-face orientations	28	19.2	145	61.7
Attended group orientations	16	11.0	54	23.0

Note. SEED = Saving for Education, Entrepreneurship, and Downpayment.

^a N = 381.

Study Measures

The parenting stress variables correlate with general levels of stress and were adapted from the Parenting Stress scale by Berry and Jones (1995) and the Psychological Assessment Resources (PAR) Parenting Stress Index by Abidin (1992) and Morrison, Zaslow, & Dion (1998). They are thought to be valid and reliable measures of parenting stress, useful for both male and female respondents that capture four factors: parental rewards, stressors, sense of control, and satisfaction (Berry & Jones, 1995). The personal mastery variables are the seven items introduced by Pearlin and Schooler (1978) as utilized in the Administration for Children and Families’ Head Start Family and Child Experiences (FACES) parent survey. The economic strain variables were adapted from the Iowa Youth and Families Project (IYFP) as created by Rand Conger and colleagues and utilized across several research samples to test a family stress model of economic hardship (Conger & Conger, 2002; Conger et. al, 2002, 1994). The variables were transformed and re-coded to 1–4, indicating strongly disagree to strongly agree. The terms latent constructs or factors are used synonymously with the outcome variables used in this study.

Analytic Procedures

The multiple imputation technique (Enders, 2010; Graham, 2009; Graham, Cumsille, & Elek-Fisk, 2003; Graham, Taylor, Olchowski, & Cumsille, 2006) was used to manage the missing data. In order to retain optimal power (Collins, Schafer, & Kam, 2001; Graham, Olchowski, & Gilreath, 2007; Schafer & Graham, 2002) and under the assumption of a missing at random (MAR) mechanism and multivariate normality, 100 imputations (data sets) were created. The imputation process included the three outcomes (factors) of parenting stress, personal mastery, and economic strain.

A multiple group confirmatory factor analysis (CFA) was then used to evaluate program effects. Using maximum likelihood, LISREL 8.80 statistical package (Jöreskog, & Sörbom, 2007) was used. To determine whether the factors had the same meaning across measurement occasions and groups, the following sequence of steps were used: (1) 2-group, 2-occasion configural model assessment, (2) test of equal factor loadings (i.e., weak measurement invariance), and (3) test of equal intercepts (i.e., strong measurement invariance). To assess the significance of each comparison, we evaluated whether (a) the root mean square error of approximation (RMSEA) value of the nested model fell within the RMSEA confidence interval of the comparison model (Little, 1997) and (b) the change in CFI was ≤.01 (Cheung & Rensvold, 2002).

We were also interested in determining whether participation in SEED affected the relationships among the three factors across measurement occasions and groups. These tests are only appropriate when weak invariance is established. Lastly, we were also interested in comparing measurement occasion and groups on factor means, which is only appropriate with the establishment of the same meaning among the factors. We assessed these relationships using the following sequence of steps: (1) test of factor variance and covariance equality, (2) test of factor variance equality, (3) test of latent correlations, and (4) test of the equality of factor means. To assess the significance of these steps, a chi-square difference test was used (Chen, Sousa, & West, 2005; French & Finch, 2006; Hancock & Mueller, 2006; Kline, 2010; Little, 1997).

Scale reliability was assessed with the composite reliability for congeneric measures model (CRCMM) method because it does not assume equal factor loadings, and thus, a better lower bound estimate of reliability than Cronbach’s α (Raykov, 1997). Among participants that were account holders, the reliability index was .72 and .67 for the parenting stress indicators at the first and second measurement occasions, respectively; .69 and .71 for the personal mastery indicators at the first and second measurement occasion, respectively; and .83 and .85 for the economic strain indicators at the first and second measurement occasion, respectively. Among the participants who were account nonholders, the reliability index was .67 and .53 for the parenting stress indicators at the first and second measurement occasion, respectively; .85 and .69 for the personal mastery indicators at the first and second measurement occasion, respectively; and .87 and .88 for the economic strain indicators at the first and second measurement occasion, respectively. Overall, the scales are reliable because the items have a relatively high internal consistency.

Measurement Model

A 3-factor measurement model with two measurement occasions was specified in which “My family has enough money to afford the kind of home that we need” (economic strain [ES]1, T2ES1), “My family has enough money to afford the kind of clothes that we need” (ES2, T2ES2), “My family has enough money to afford the kind of furniture or household equipment that we need” (ES3, T2ES3), “My family has enough money to afford the kind of car that we need” (ES4, T2ES4), “My family has enough money to afford the kind of food that we need” (ES5, T2ES5), “My family has enough money to afford the kind of medical care that we need” (ES6, T2ES6), and “My family has enough money to afford the kind of leisure and recreational activities we need” (ES7, T2ES7) loaded on the latent variable of economic strain for the first and second measurement occasions respectively, that is, T2 refers to the second measurement occasion.

Furthermore, the personal mastery latent variable was composed of the following manifest indicators measured at two occasions: “There is no way I can solve some of the problems I have” (personal mastery [PM]1, T2PM1), “I feel that I am being pushed around in life” (PM2, T2PM2), “I have little control over the things that happen to me” (PM3, T2PM3), “I can do anything I set my mind to” (PM4, T2PM4), “I feel helpless in dealing with the problems of life” (PM5, T2PM5), “What happens to me in the future depends on me” (PM6, T2PM6), and “There is little I can do to change the important things in my life” (PM7, T2PM7).

Lastly, the latent variable of parenting stress was composed of the following indicators: “I am happy with my role as a parent” (parenting stress [PS]1, T2PS1), “In my role as a parent/caregiver, I often find that I have too little time for myself” (PS2, T2PS2), “As a parent, I enjoy the time I spend with my child(ren)” (PS3, T2PS3), “I feel overwhelmed with responsibilities of being a parent” (PS4, T2PS4), “As a parent/caregiver, I am able to find a balance between the many demands for my time and energy” (PS5, T2PS5), “As a parent/caregiver, I often find that my life is much more work than pleasure” (PS6, T2PS6), “I am satisfied as a parent caregiver” (PS7, T2PS7), and “I often feel tired, worn out or exhausted by the responsibilities of being a parent/caregiver” (PS8, T2PS8). T2 refers to the second measurement occasion.

Results

The Meaning of Parenting Stress, Personal Mastery, and Economic Strain

Following the steps mentioned above, results showed that parenting stress, personal mastery, and economic strain have the same meaning across both measurement occasions and both groups. Apart from the chi-square test, we incorporated several alternative measures of fit. Alternative measures of model fit do not assume that the model will exactly fit the data and are among the most commonly used tools to assess fit (Browne & Cudeck, 1993; Rigdon, 1998). The specific alternative fit indices of interest were the RMSEA (Steiger, 1990), the Nonnormed Fit Index (NNFI; Bentler & Bonett, 1980), and the Comparative Fit Index (CFI; Bentler, 1990). Acceptable RMSEA values are less than or equal to .08 (Brown, 2006), while values equal to or greater than .90 are considered acceptable for the NNFI and the CFI (Brown, 2006; Kline, 1993; Reise, Widaman, & Pugh, 1993). As Table 2 shows, the value of the RMSEA, NNFI, and CFI show acceptable levels, χ²(1,844, n = 381) = 4,470, p < .001, RMSEA = .067 (.065; .070), NNFI = 0.88, CFI = 0.88. As Table 2 shows, we established the equivalence of the indicator means simultaneously across both measurement occasions and groups, suggesting that at a given level of parenting stress, personal mastery, and economic strain among participants with and without an account measured at both occasions would have the same observed scores on an indicator. In Table 2, certain models are evaluated with the RMSEA model test, while certain other models are evaluated with the χ² difference test. Also, each nested model contains its constraints, plus the constraints of all previous, tenable models.

Table 2.

Fit Indices for the Nested Sequence in the Two Occasion Confirmatory Factor Analysis—SEED Baseline Impact Assessment Survey^a

								RMSEA 90% CI
Model	χ²	df	p	Δ χ²	Δ df	p	RMSEA	Lower	Upper	NNFI	CFI	Tenable
Configural invariance	4,254.60	1,730	<.001	---	---	---	0.069	0.066	0.071	0.876	0.887	---
Loading invariance^b	4,366.50	1,787	<.001	---	---	---	0.068	0.065	0.070	0.878	0.884	Yes
Intercept invariance^b	4,470.60	1,844	<.001	---	---	---	0.067	0.065	0.070	0.879	0.882	Yes
Homogeneity of variances/covariances (omnibus)^c	4,505.80	1,868	<.001	35.20	24	>.05	0.067	0.064	0.070	0.880	0.882	Yes
Equality of correlations (omnibus)^c	4,495.10	1,865	<.001	24.50	21	>.05	0.067	0.064	0.070	0.880	0.882	Yes
Latent mean invariance (omnibus)^c	4,508.30	1,853	<.001	37.70	9	<.001	0.067	0.065	0.070	0.878	0.881	No
Latent mean invariance (omnibus between groups)^c	4,477.27	1,850	<.001	6.67	6	>.05	0.067	0.065	0.070	0.880	0.882	Yes
Latent mean invariance (omnibus across Time)^c	4,505.10	1,850	<.001	34.50	6	<.001	0.067	0.065	0.070	0.878	0.881	No
Latent mean invariance (time with PS and T2PS)^c	4,498.90	1,846	<.001	28.30	2	<.001	0.067	0.065	0.070	0.878	0.881	No
Latent mean Invariance (time with PM and T2PM)^c	4,472.80	1,846	<.001	2.20	2	>.05	0.067	0.064	0.070	0.879	0.882	Yes
Latent mean invariance (time ES and T2ES)^c	4,474.80	1,846	< .001	4.20	2	>.05	0.067	0.064	0.070	0.879	0.882	Yes

Note. SEED = Saving for Education, Entrepreneurship, and Downpayment; CI = confidence interval; RMSEA, root mean square error of approximation; NNFI = Nonnormed Fit Index; CFI = Comparative Fit Index.

^a N = 381. ^b Evaluated with the RMSEA model test. ^c Evaluated with the χ² difference test.

Relationships Among Parenting Stress, Personal Mastery, and Economic Strain

Our second research question asked whether participation in SEED affected the relationships between parenting stress, personal mastery, and economic strain across both measurement occasions and groups. As noted in Table 2, the omnibus test of both equal variances and covariances across time and groups was not significant, Δχ²(24, n = 381) = 35.20, p > .05, indicating that parenting stress, personal mastery, and economic strain did not differ across either measurement occasion or group membership. The correlation matrix of the outcome variables is reported in Table 3 and illustrates that parenting stress and personal mastery correlate around .60 within each measurement occasion; parenting stress and economic strain correlate around −.30 within each measurement occasion; and personal mastery and economic strain correlate around −.14 within each measurement occasion.

Table 3

Correlations, Means, and Standard Deviations Between Latent Constructs for Both Measurement Occasions—SEED Baseline Impact Assessment Survey^a

	PS	PM	ES	T2PS	T2PM	T2ES
PS	1.00
PM	0.60	1.00
ES	−0.28	−0.13	1.00
T2PS	0.51	0.41	−0.22	1.00
T2PM	0.34	0.53	−0.12	0.58	1.00
T2ES	−0.25	−0.08	0.50	−0.32	−0.21	1.00
M	2.02	1.79	2.48	1.93	1.77	2.47
SD	0.34	0.44	0.59	0.34	0.44	0.59

Note. SEED = Saving for Education, Entrepreneurship, and Downpayment; PS = parenting stress; PM = personal mastery; ES = economic strain; T2PS = parenting stress at second occasion; T2PM = personal mastery at second occasion; T2ES = economic strain at second occasion.

^a N = 381.

Levels of Parenting Stress, Personal Mastery, and Economic Strain

The next research question asked whether participation in SEED affected the groups’ average levels (i.e., means) of parenting stress, personal mastery, and economic strain across measurement occasions. As noted in Table 2, the latent means of parenting stress, personal mastery, and economic strain were not similar, Δχ²(3, n = 381) = 37.70, p < .001, across time and groups. These results suggest that the hypothesis of mean equivalence across time and groups is rejected. That is, at least one of the three means changed over time, between groups, or both. The latent means and standard deviations are included in Table 3. Follow-up tests indicated evidence for a main effect of time relating to parenting stress. The latent mean of parenting stress at the first measurement occasion (M = 2.02, SD = 0.34) was significantly higher, Δχ²(2, n = 381) = 28.30, p < .001 (at the .017 level; Bonferroni correction) from the latent mean of parenting stress at the second measurement occasion (M = 1.93, SD = 0.34). The latent effect size was in the small-to-medium range (d = .25). All other latent means did not significantly change over time (see Tables 2 and 3). Results are based on the effects-coding method of identification, and the latent correlations and standard deviations were based on phantom variables.

Discussion

The purpose of this study was to determine whether SEED program participation affected parenting stress, personal mastery, and economic strain between account openers and nonopeners at the beginning and end of a child savings accounts program among lower income parents. The results suggest that there was no significant difference in the meaning as well as the relationships of these outcomes across time and groups. However, a mean difference was observed between parenting stress at the first and second measurement occasions, suggesting some time effect on this variable. Given these results, we concluded that SEED participants on average reported more parenting stress at the first measurement occasion than at the second occasion.

The significant difference in the degree of parenting stress over time is interesting; it shows that at the beginning of program, the variances were more homogeneous because individuals were more similar with regard to their level of parenting stress. By the end of the program, however, individual differences were more pronounced because some parents reported more parenting stress than they did at the beginning of the program while others did not. This is reflected in the input matrix where the standard deviation increased in magnitude from the beginning to the end of the program. It is possible that other unmeasured household and environmental factors could have contributed to decreased parenting stress. Clearly, the finding of limited program effects on parents’ well-being is not causal; it should not be linked exclusively to participation in the child savings accounts program.

Among lower income parents for instance, economic strain may affect parental stress (Cain & Combs-Orme, 2005; Dennis et al., 2003; Elder, Jr., Eccles, Ardelt, & Lord, 1995). Another plausible factor may have to do with increased demands and responsibilities on parents, including additional children in a household and the general emotional, physical, and logistical demands that children place on parents. On the other hand, the fact that economic strain was not affected over time is also interesting. Among those who opened accounts and therefore, saving, one would expect that their economic strain may have been affected as a result of the pressures and sacrifice to save. This was not the case and probably points to the fact that probably most of the savings were lower.

However, a comprehensive account monitoring report by the CSD at Washington University on the accounts at the SEED project reported in this article noted that at the end of the project in December 31, 2008, the total accumulation in all the SEED accounts, including program incentives from SEED partners, in all accounts was $734,042 (Loke, Clancy, & Zager, 2009). This is clearly a significant amount and only time and future studies may show the trend of the accounts in the years to come as the time for college education draws near for the children of these parents. Apart from the CSD report, the final SEED program report by Adams and colleagues (2010) is also informative. Among the lessons documented in this report was that (1) families used innovative strategies to save despite having serious financial resources; (2) household and financial obligation factors made saving difficult; (3) program features facilitated savings, especially among participants with lower incomes; and (4) a strong relationship with community-based agencies and their staff increased the motivation to save.

The services provided by SEED staff members were expected to make it easier for families to enroll in the program. However, fewer than expected parents in the experimental group decided to join the program. From a programming perspective, even more time and money may need to be devoted to social services that help those with particularly difficult economic challenges to save and build assets for their children’s futures. Particularly, strategies that build the trust of lower income families with financial institutions aimed at their welfare and that of their children may be critical. Specifically, responsive program orientations may prove critical in providing stable foundations for future fruitful engagements between the families and agencies that are involved (Okech, 2010).

An ideal way to implement child savings accounts would be a universal approach where all children receive accounts when they are born. However, many western countries, including the United States, are falling back into more neoliberal socioeconomic policies with calls for reduced public funding. This trend was exacerbated by the recession and lingering volatile economic outlook. This raises some pertinent questions: What really is the most effective way to ensure that lower income youth in the United States attain college education? What policy proposals are more likely to be acceptable to law makers and the public within the current economic and political environments? Could child savings accounts perpetuate intergenerational poverty if children from higher income families save more than their lower income counterparts? In other words, how do we, as in Sherraden’s (1988) earlier words, “rethink social welfare” in a manner that meets the needs of lower income populations and ensures the economic development of their children within the context of increased neoliberal state welfare policies?

The view that there are limits to acceptable inequality is fundamental to most societies and value systems. Both poverty rates and gaps have increased because of the recent recession and college is becoming inaccessible to more children from lower income families. Social work practice builds on the capacities of individuals and communities to overcome social and economic challenges. Taking an institutional approach in asset building is consistent with social work values of justice and equity as articulated in our professional code of conduct, which speaks, in part, to the professional mandate of supporting structures and policies that promote equality (National Association of Social Workers, 1999). Further, asset-building policies and programs seem to be one good way that social work can reclaim the progressive era spirit, emphasizing social and economic well-being, with a priority on serving especially vulnerable individuals, families, groups, and communities. When lower income families own assets and their children attain college education, both income and asset gaps will reduce, and the families may ultimately experience some of the positive asset effects described earlier in this study.

Footnotes

The correlation items tables with means and standard deviations that may be used for replicating the findings presented here are available upon request from the author.

Acknowledgments

Deborah Adams, associate professor at the University of Kansas School of Social Welfare and SEED research principal investigator, always provided invaluable help in understanding assetbuilding programming and policy design. Finally, the author thanks colleagues at the University of Kansas Department of Quantitative Psychology including Professor Todd D. Little and Mr. Waylon J. Howard for help in applying Structural equation modeling (SEM) methods in a systematic manner. I thank the two anonymous reviewers who offered helpful critiques to the original manuscript.

Notes

References

Abidin

. (1992). The determinants of parenting behavior. Journal of Clinical Child Psychology, 21, 407–412. Retrieved from http://www.jccap.net

Adams

(2008). SEED pre-school demonstration and impact assessment. Research brief, SEED research at KU. Lawrence, KS: University of Kansas School of Social Welfare.

Adams

Boshara

Clancy

Cramer

Friedman

Howard

Zager

. (2010). Lessons from SEED: A national demonstration of Child Development Accounts. (Report, November 2010). Retrieved from SEED Partners website: http://cfed.org/assets/pdfs/SEEDSynthesis.pdf

Adams

Williams Shanks

T. R.

Marks

(2008). SEED pre-school demonstration and impact assessment: Findings to date. Research brief, SEED research at KU. Lawrence, KS: University of Kansas School of Social Welfare.

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.

Berry

J. O.

Jones

W. H.

(1995). The parental stress scale: Initial psychometric evidence. Journal of Social and Personal Relationships, 12, 463–472. doi: 10.1177/0265407595123009

Beverly

S. G.

Sherraden

Zhan

Shanks

T. R.

Nam

Cramer

(2008). Determinants of asset building. St. Louis, MO: Center for Social Development, Washington University.

Beverly

S. G.

Williams Shanks

(2004). SEED impact assessment survey: Baseline instrument in english and Spanish. Research instrument, SEED research at KU. Lawrence, KS: University of Kansas School of Social Welfare.

10.

Brown

T. A.

(2006). Confirmatory Factor Analysis for applied research. New York: The Guilford Press.

11.

Browne

M. W.

Cudeck

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

K. A. B. J. S.

. (Ed.), Testing structural equation models (pp. 136–162). Newbury Park, CA: SAGE.

12.

Cain

D. S.

Combs-Orme

. (2005). Family structure effects on parenting in African American family. Journal of Sociology and Social Welfare, 32, 19–40. Retrieved from http://www.wmich.edu

13.

Chen

Sousa

West

(2005). Teacher's corner: Testing measurement invariance of second-order factor models. Structural Equation Modeling: A Multidisciplinary Journal, 12, 471–492.

14.

Cheung

G. W.

Rensvold

R. B.

(2002). Evaluating goodness-of-fit indexes for testing measurement invariance. Structural Equation Modeling, 9, 233–255.

15.

Collins

L. M.

Schafer

J. L.

Kam

C. M.

(2001). A comparison of inclusive and restrictive strategies in modern missing data procedures. Psychological methods, 6, 330–351.

16.

Conger

R. D.

Conger

K. J

. (2002). Resilience in Midwestern families: Selected findings from the first decade of a prospective, longitudinal study. Journal of Marriage and the Family, 64, 361–373. doi: 10.1111/j.1741-3737.2002.00361.x

17.

Conger

R. D.

X. J.

Elder

G. H.

Jr. Lorenz

F. O.

Simons

R. L

. (1994). Economic stress, coercive family process, and developmental problems of adolescents. Child Development, 65, 541–561. doi: 10.111/j.1467-8624.1994.tb00768.x

18.

Conger

R. D.

Wallace

L. E.

Sun

Simons

R. L.

McLoyd

V. C.

Brody

G. H

. (2002). Economic pressure in African American families: A replication and extension of the family stress model. Developmental Psychology, 38, 179–193. doi: 10.1037//0012-1649.38.2.179

19.

Crnic

K. A.

Gaze

Hoffman

. (2005). Cumulative parenting stress across the preschool period: Relations to maternal parenting and child behavior at age 5. Infant and Child Development, 14, 117–132. doi: 10.1002/icd.384

20.

Curley

Sherraden

(2000). Policy lessons from children’s savings accounts. Child Welfare, 79, 661–687.

21.

Dennis

J. M.

Parke

R. D.

Coltrane

Balcher

Borthwick-Duffy

S. A

. (2003). Economic pressure, maternal depression, and child adjustment in Latino families: An exploratory study. Journal of Family and Economic Issues, 24, 183–202. doi: 10.1023/A:1023706424725

22.

Elder

G. H.

Jr. Eccles

J. S.

Ardelt

Lord

. (1995). Inner-city parents under economic pressure: Perspectives on the strategies of parenting. Journal of Marriage and the family, 57, 771–784. Retrieved from http://www.jstor.org/stable/353931

23.

Enders

C. K.

(2010). Applied missing data analysis. New York, NY: Guilford Press.

24.

French

B. F.

Finch

W. H.

(2006). Confirmatory factor analytic procedures for the determination of measurement invariance. Structural Equation Modeling, 13, 378–402.

25.

Graham

J. W.

(2009). Missing data analysis: Making it work in the real world. Annual Review of Psychology, 60, 549–576.

26.

Graham

J. W.

Cumsille

P. E.

Elek-Fisk

(2003). Methods for handling missing data (Vol. 2). New York, NY: Wiley.

27.

Graham

J. W.

Olchowski

A. E.

Gilreath

T. D.

(2007). How many imputations are really needed? Some practical clarifications of multiple imputation theory. Prevention Science, 8, 206–213.

28.

Graham

J. W.

Taylor

B. J.

Olchowski

A. E.

Cumsille

P. E.

(2006). Planned missing data designs in psychological research. Psychological Methods, 11, 323–343.

29.

Hancock

Mueller

(2006). Structural equation modeling: A second course. New York, NY: The Guilford Press.

30.

Howard, C. (1997). The hidden welfare state: Tax expenditures and social policy in the United States. Princeton, NJ: Princeton University Press

31.

Jöreskog

K. G.

Sörbom

(2007). LISREL 8.80. Chicago, IL: Scientific Software International.

32.

Kahn

Pearlin

L. I.

(2006). Financial strains and health across the life course and health among older adults. Journal of Health and Social Behavior, 47, 17–31. doi: 10.1177/002214650604700102

33.

Kline

(1993). An east guide to factor analysis. New York , NY: Routledge.

34.

Kline

(2010). Principles and practice of structural equation modeling. New York, NY: The Guilford Press.

35.

Little

(1997). Mean and covariance structures (MACS) analyses of cross-cultural data: Practical and theoretical issues. Multivariate Behavioral Research, 32, 53–76.

36.

Loke

Clancy

Zager

(2009). Account monitoring research at Michigan SEED (CSD Research report No. 09-62. Center for Social Development. St. Louis, MO).

37.

McLoyd

V. C

. (1990). The impact of economic hardship on black families and children: Psychological distress, parenting, and socioemotional development. Child Development, 61, 311–346. doi: 10.1111/j.1467-8624.1990.tb02781.x

38.

Marks

Rhodes

(2005). Saving for education, entrepreneurship and downpayment wave 1 survey–Fall 2004. North Carolina: RTI International.

39.

Mistry

R. S.

Lowe

E. D.

Benner

A. D.

Chien

. (2008). Expanding the family economic stress model: Insights from a mixed-methods approach. Journal of Marriage and Family, 70, 196–209. doi: 10.1111/j.1741-3737.2007.00471.x

40.

Morrison

D. R.

Zaslow

M. J.

Dion

M. R.

(1998). Completing the portrayal of parenting: Behavior with survey measures of parenting. In Zaslow

Eldred

(Eds.), Parenting behavior in a sample of young mothers in poverty: Results of the new chance observational study (pp. 25-53). New York, NY: Manpower Demonstration Research Corporation.

41.

National Association of Social Workers. (1999). Code of Ethics of the National Association of Social Workers. Washington, DC: NASW Press.

42.

Okech

(2010). Enrollment decisions in a child development Accounts program for low-income families. (2011). Journal of Family and Economic Issues, 32, 400–410. doi: 10.1007/s10834-010-9234-z

43.

Okech

Little

T. D.

Williams Shanks

T. R.

Adams

. (2011). Self-efficacy and joining a savings program for children’s education. Research on Social Work Practice, 21, 442–451. doi: 10.1177/1049731510389079

44.

Oliver

M. L.

Shapiro

T. M.

(2006). Black wealth/White wealth: A new perspective on racial inequality (2nd ed.). New York, NY: Rutledge.

45.

Pearlin

L. I.

Schooler

. (1978). The structure of coping. Journal of Health and social Behavior, 9, 2–21. Retrieved from http://www.jstor.org/stable/2136319

46.

Raykov

. (1997). Scale reliability, Cronbach's coefficient alpha, and violations of essential tau-equivalence with fixed congeneric components. Multivariate Behavioral Research, 33, 343–363.

47.

Reise

S. P.

Widaman

K. F.

Pugh

R. H.

(1993). Confirmatory factor analysis and item response theory: Two approaches for exploring measurement invariance. Psychological Bulletin, 114, 552–566.

48.

Rigdon

E. E

. (1998). Structural equation modeling. In Marcoulides

G. A.

(Ed.), Modern methods for business research (p. 251–294). Mahwah, NJ: Lawrence Erlbaum.

49.

Schafer

J. L.

Graham

J. W.

(2002). Missing data: Our view of the state of the art. Psychological Methods, 7, 147–177.

50.

Schieman

Pudrovska

Pearlin

L. I.

Ellison

C. G.

(2006). The sense of divine control and psychological distress: Variations across race and socioeconomic status. Journal for the Scientific Study of Religion, 45, 529–549.

51.

Schieman

Young

(2011). Economic hardship and family-to-work conflict: The importance of gender and work conditions. Journal of Family and Economic Issues, 32, 46–61. doi: 10.1007/s10834-010-9206-3

52.

Seidman

(2001). Assets and the tax code. In Shapiro

T. M.

Wolff

E. N.

(Eds.). Assets for the poor: Benefits and mechanisms of spreading asset ownership (pp. 324–356). New York, NY: Russell Sage Foundation.

53.

Sherraden

(1988). Rethinking social welfare: Toward assets. Social Policy, 18, 37–43.

54.

Sherraden

(1991). Assets and the poor: A new American welfare policy. Armonk, NY: M.E. Sharpe, Inc.

55.

Sherraden

(2003). Assets and the social investment state. In Paxton

(Ed.). Equal shares? Building a progressive and coherent asset-based welfare policy (pp. 28–41). Bristol, UK: The Policy Press.

56.

Sherraden

(2005). Assets and public policy. In Sherraden

(Ed.), Inclusion in the American dream: Assets, poverty, and public policy (pp. 3–19). New York, NY: Oxford University Press.

57.

Sherraden

S. M.

Johnson

Guo

Elliot

(2009). Financial capability in children: Effects of participation in school-based financial education and savings program. Journal of Family & Economic Issues, 32, 385–399. doi: 10.1007/s10834-010-9220-5

58.

Shobe

Page-Adams

. (2001). Assets, future orientation, and well-being: Exploring and extending Sherraden’s framework. Journal of Sociology and Social Welfare, 28, 109–127. Retrieved from http://www.wmich.edu/hhs/newsletters_journals/jssw/

59.

Spencer

S. M.

Patrick

J. H.

(2009). Social support and personal mastery as protective resources during emerging adulthood. Journal of Adult Development, 16, 191–198. doi: 10.1007/s10804-009-9064-0

60.

Steiger

(1990). Structural model evaluation and modification: An interval estimation approach. Multivariate Behavioral Research, 25, 173–180.

61.

Takeuchi

D. T.

Williams

D. R.

Adair

R. K

. (1991). Economic stress in the family and children’s emotional and behavioral problems. Journal of Marriage and the Family, 53, 1031–1041. Retrieved from http://www.jstor.org/stable/353006

62.

Wolff

E. N.

(2002). Top heavy: The increasing inequality of wealth in America and what can be done about it. New York, NY: The New Press.

63.

Zhan

Sherraden

(2003). Assets, expectations, and children’s educational achievement in female-headed households. Social Service Review, 77, 191–211.