View in EDS HTML Full Text PDF Full Text

Understanding Associations between Low-Income Mothers' Participation in Education and Parenting

Saved in:
Bibliographic Details
Title: Understanding Associations between Low-Income Mothers' Participation in Education and Parenting
Language: English
Authors: Harding, Jessica F., Morris, Pamela A., Hill, Jennifer
Source: Journal of Research on Educational Effectiveness. 2017 10(4):704-731.
Availability: Routledge. Available from: Taylor & Francis, Ltd. 530 Walnut Street Suite 850, Philadelphia, PA 19106. Tel: 800-354-1420; Tel: 215-625-8900; Fax: 215-207-0050; Web site: http://www.tandf.co.uk/journals
Peer Reviewed: Y
Page Count: 28
Publication Date: 2017
Document Type: Journal Articles
Reports - Research
Education Level: Preschool Education
Descriptors: Correlation, Low Income, Mothers, Scores, Parenting Styles, Educational Attainment, Mother Attitudes, Prediction, Disadvantaged Youth, Preschool Education, Preschool Children, Child Development, Statistical Analysis, Family Characteristics, Cultural Activities, Reading Materials, Family Environment, Randomized Controlled Trials, Regression (Statistics)
DOI: 10.1080/19345747.2016.1266536
ISSN: 1934-5747
Abstract: Maternal education is one of the strongest predictors of children's academic outcomes. One possible explanation for this is that more highly educated mothers more frequently engage in parenting practices that may promote children's later cognitive development; however, most of this evidence is correlational. This study uses Head Start Impact Study data (N = 1,953) to explore whether low-income mothers' participation in education affects their parenting practices and beliefs. Principal scores, which predict maternal educational participation based on covariates, were used for analysis. Principal score matching was used to identify mothers who we predicted participated in education because their children were randomly assigned to Head Start. We compared these mothers' outcomes to those of mothers we predicted would have participated in education if they were assigned to Head Start. For these mothers, participation in maternal education was associated with children watching fewer hours of TV, having more types of printed media at home, and more frequent participation in cultural activities, but it was not associated with a host of other parenting outcomes. Changing parenting is one potential pathway by which maternal educational participation may influence children's later academic outcomes.
Abstractor: As Provided
Number of References: 86
Entry Date: 2017
Accession Number: EJ1159786
Database: ERIC
Full text is not displayed to guests.
FullText Links:
  – Type: pdflink
    Url: https://content.ebscohost.com/cds/retrieve?content=AQICAHj0k_4E0hTGH8RJwT4gCJyBsGNe_WN95AvKlDbXJGqwxwGl5gARaHMTJXQNpB6fcFHzAAAA4zCB4AYJKoZIhvcNAQcGoIHSMIHPAgEAMIHJBgkqhkiG9w0BBwEwHgYJYIZIAWUDBAEuMBEEDLb4Ons6ka3-tr_klgIBEICBm7HptD1_wN-Ez0xsooFvfG-djqYDXkir1J9JDR-Iao5lDziFCWYmiPZDYQh5Uv1KQV_yJeUJFtxyc6BfVmK8IwL_Kxoe_iQ03aaQ0-B0ra7r12F3EMQ1X5z7W2BRnkqH_gIDlbtWcIaVtpnA2faZXDAgdZHb5jAOu7QydnP2tSFsj-i3XRpKg2fvSCOqGjfLTtC6mzHLlo-Ti7Wl
Text:
  Availability: 1
  Value: <anid>AN0126281570;[5ew9]01oct.17;2019Feb08.14:54;v2.2.500</anid> <title id="AN0126281570-1">Understanding Associations Between Low-Income Mothers' Participation in Education and Parenting. </title> <p>Maternal education is one of the strongest predictors of children's academic outcomes. One possible explanation for this is that more highly educated mothers more frequently engage in parenting practices that may promote children's later cognitive development; however, most of this evidence is correlational. This study uses Head Start Impact Study data (N = 1,953) to explore whether low-income mothers' participation in education affects their parenting practices and beliefs. Principal scores, which predict maternal educational participation based on covariates, were used for analysis. Principal score matching was used to identify mothers who we predicted participated in education because their children were randomly assigned to Head Start. We compared these mothers' outcomes to those of mothers we predicted would have participated in education if they were assigned to Head Start. For these mothers, participation in maternal education was associated with children watching fewer hours of TV, having more types of printed media at home, and more frequent participation in cultural activities, but it was not associated with a host of other parenting outcomes. Changing parenting is one potential pathway by which maternal educational participation may influence children's later academic outcomes.</p> <p>Keywords: parenting; maternal education; low-income; intergenerational; parenting practices</p> <hd id="AN0126281570-2">Education and Parenting</hd> <p>Maternal education is one of the strongest predictors of children's academic outcomes throughout development. Higher levels of maternal education have been associated with children's early language production, cognitive achievement, grade-point averages, and high school completion (Choi, Raley, Muller, & Riegle-Crumb, [<reflink idref="bib9" id="ref1">9</reflink>]; Dollaghan et al., [<reflink idref="bib14" id="ref2">14</reflink>]; Sirin, [<reflink idref="bib68" id="ref3">68</reflink>]). Quasi-experimental analyses have also shown that increases in low-income mothers' education are significantly related to young children's language comprehension and vocabulary skills (Gennetian, Magnuson, & Morris, [<reflink idref="bib22" id="ref4">22</reflink>]; Harding, Morris, & Hughes, [<reflink idref="bib25" id="ref5">25</reflink>]; Magnuson, [<reflink idref="bib47" id="ref6">47</reflink>]; Magnuson, Sexton, Davis-Kean, & Huston, [<reflink idref="bib48" id="ref7">48</reflink>]). Although questions remain about the causal nature of these relations, the majority of evidence does suggest a causal role for maternal education, particularly at the low end of the educational distribution (Holmlund, Lindahl, & Plug, [<reflink idref="bib32" id="ref8">32</reflink>]). Maternal education likely matters for children via associations with family structure, employment, and income, but it may also matter if mothers who are more highly educated interact with their children differently than mothers who have lower education (Augustine, [<reflink idref="bib2" id="ref9">2</reflink>]; Harding et al., [<reflink idref="bib25" id="ref10">25</reflink>]). Indeed, higher maternal education has been related to mothers more frequently engaging in cognitively stimulating activities (Suizzo & Stapleton, [<reflink idref="bib73" id="ref11">73</reflink>]) and spending more time in developmentally salient child-rearing tasks (Kalil, Ryan, & Corey, [<reflink idref="bib39" id="ref12">39</reflink>]).</p> <p>Despite the potential importance of parenting as a mechanism by which maternal education can influence children's outcomes, research exploring the association of maternal education with parenting is limited in three key ways. First, most research treats maternal education as a static characteristic and explores attained level of education, yet many low-income mothers participate in education, making it important to understand how educational participation shapes outcomes for children and families (Perna, Fester, & Walsh, [<reflink idref="bib57" id="ref13">57</reflink>]). Second, research has focused on the association of maternal education with a relatively narrow set of parenting outcomes, including maternal language use and the quality of the home environment. Yet research suggests that education increases individuals' general skills, abilities, and effectiveness (Mirowsky & Ross, [<reflink idref="bib51" id="ref14">51</reflink>]). If these skills are employed in interactions with children, maternal education may shape a wider range of parenting behaviors than is typically examined. Third, there is little causal evidence for the relationship between maternal education and parenting. Correlational evidence suggests the importance of maternal education for parenting, but a number of factors, such as mothers' academic ability and their perceptions of the value of education, may contribute to mothers' decisions to complete additional schooling. If these factors are related to parenting, they could drive the observed associations between maternal education and parenting.</p> <p>This research addresses whether low-income mothers' participation in education influences a constellation of different parenting practices and beliefs in the context of a study of children's developmental outcomes after random assignment to Head Start (Head Start Impact Study, HSIS; U.S. Department of Health and Human Services, [<reflink idref="bib75" id="ref15">75</reflink>]). Random assignment to Head Start has been associated with increases in maternal education (Sabol & Chase-Lansdale, [<reflink idref="bib64" id="ref16">64</reflink>]), and we leverage this experimentally induced increase by using principal scores to identify a group of mothers who we predicted participated in education because their children were assigned to Head Start whose outcomes were compared to those of a group of mothers who we predicted would have participated in education if their children were assigned to Head Start. Comparing the parenting of similar mothers who did and did not participate in education can help us to understand whether participation in education influences mothers' parenting and can illuminate a pathway by which increasing maternal education may have long-lasting influence on family functioning and children's later academic outcomes.</p> <hd id="AN0126281570-3">Maternal Education and Parenting</hd> <p>Prior theoretical work suggests that maternal education is related to a broad set of proximal and distal parenting practices that influences children's cognitive development, academic achievement during schooling, and educational attainment (Harding et al., [<reflink idref="bib25" id="ref17">25</reflink>]). In brief, this work posits that engaging in education may increase mothers' human, cultural, and social capital which, in turn, can increase mothers' engagement in parenting practices related to each form of capital. Education may increase maternal human capital, defined as mothers' own language and cognitive skills (Kingston, Hubbard, Lapp, Schroeder, & Wilson, [<reflink idref="bib42" id="ref18">42</reflink>]). Indeed, "cross-cultural research on the cognitive consequence of schooling has shown that school experience affects proficiency in a variety of specific verbal skills [and] that these skills tend to be retained into adulthood" (LeVine, LeVine, & Schnell, [<reflink idref="bib44" id="ref19">44</reflink>], p. 21). Mothers can then employ these skills in interactions with their children (Becker, [<reflink idref="bib4" id="ref20">4</reflink>]; Coleman, [<reflink idref="bib11" id="ref21">11</reflink>]); for example, engaging in more frequent and higher quality shared-book reading and art and science activities (Suizzo & Stapleton, [<reflink idref="bib73" id="ref22">73</reflink>]; Torr, [<reflink idref="bib74" id="ref23">74</reflink>]). Education may also increase mothers' cultural capital, or their ability to employ socially valued language and behaviors to gain educational benefits for their children (Bourdieu, [<reflink idref="bib6" id="ref24">6</reflink>]). As LeVine, LeVine, Schnell-Anzola, Rowe, and Dexter ([<reflink idref="bib45" id="ref25">45</reflink>]) describe it, by internalizing norms through experiences in the classroom, mothers learn to act adaptively in bureaucratic contexts. For instance, mothers with higher levels of education may be extremely effective in managing their children's school experiences and involvement in activities such as music and art (Lareau, [<reflink idref="bib43" id="ref26">43</reflink>]). Lareau ([<reflink idref="bib43" id="ref27">43</reflink>]) outlines how "concerted cultivation"-type parenting practices benefit children because they fit with the norms of dominant society, even if these parenting practices are not inherently "better" than others. Finally, education may increase maternal social capital because it provides mothers with opportunities to create social networks containing individuals who also have higher levels of education (Lin, [<reflink idref="bib46" id="ref28">46</reflink>]; McPherson, Smith-Lovin, & Cook, [<reflink idref="bib50" id="ref29">50</reflink>]). Mothers can then use these connections to gain information and access to opportunities to structure high-quality educational experiences for their children (Cochran & Brassard, [<reflink idref="bib10" id="ref30">10</reflink>]).</p> <p>Most prior research has explored relations between mothers' attained level of education and parenting. The few studies that explore changes in maternal education suggest that increases in maternal education are significantly associated with the quality of the home environment provided (Magnuson, [<reflink idref="bib47" id="ref31">47</reflink>]; Magnuson et al., [<reflink idref="bib48" id="ref32">48</reflink>]) and more frequent involvement in children's schooling (Crosnoe & Kalil, [<reflink idref="bib12" id="ref33">12</reflink>]; Domina & Roksa, [<reflink idref="bib15" id="ref34">15</reflink>]). In addition, past research has typically explored relations between maternal education and parenting practices that involve direct interactions between mothers and children. Fewer studies have explored the relationship between maternal education and more distal contextual mechanisms, such as differences in maternal social networks, even though these may be relevant for children's academic outcomes. This research will explore associations between participation in maternal education and a broad set of parenting outcomes to capture the varied ways that mothers may structure their children's environments to promote their later cognitive development (although data limitations hinder our ability to explore the full set of parenting outcomes that are of interest).</p> <p>As well as being limited in scope, most research exploring relations between maternal education and parenting is correlational, making it difficult to determine whether maternal education has a causal influence on parenting or whether this association is due to confounding factors that are related to both educational attainment and parenting, such as income or beliefs about the importance of education. Randomized controlled trials (RCTs) of programs that help mothers complete additional education are the best way to provide unbiased estimates of the causal effects of maternal education. However, only a few RCTs have examined such programs, and they have not typically focused on parenting outcomes. A notable exception to this was the evaluation of the Even Start Family Literacy program, which provided child care, adult education services, and parenting education to families. Families who were randomized to receive Even Start reported participating in some adult education at twice the rate of those in the control group, but they did not engage at high levels and their literacy skills did not improve. Perhaps because of these low levels of engagement or the uneven quality of the services provided, Even Start families did not show significant improvements on parent–child interactions compared to families in the control group (St. Pierre, Ricciuti, & Rimdzius, [<reflink idref="bib70" id="ref35">70</reflink>]). Given the lack of RCTs of programs specifically intended to improve maternal education, the current research uses a quasi-experimental analysis technique that leverages mothers' participation in education when they were randomly assigned to Head Start to extend correlational research that demonstrates significant relations between maternal education and parenting.</p> <hd id="AN0126281570-4">Head Start as a Context for Exploring Maternal Education and Parenting</hd> <p>In the absence of programs aimed at improving parental education, RCTs of early childhood education programs that increase maternal education, even if this is not the primary purpose of these programs, hold promise as a way to estimate the effects of maternal education on parenting, provided certain assumptions are accepted. Mothers whose children attend early childhood education programs may participate in education more frequently than other mothers because access to high-quality early childhood education addresses some of the challenges that low-income mothers face in completing education. For instance, staff at these programs may encourage and support mothers to participate in education, the time that children spend in child care may allow mothers to pursue education, and seeing their children learn may motivate mothers to engage in additional education themselves (Sommer et al., [<reflink idref="bib69" id="ref36">69</reflink>]). Head Start, in particular, has an official two-generation mandate to provide intensive family service supports to engage and improve outcomes for low-income parents. Since its inception, Head Start's theory of change has included a focus on involving parents in their children's education and supporting parents to improve their socioeconomic circumstances and parenting. Current Head Start program performance standards reflect this focus, with a specific performance standard related to working with parents to set goals for themselves and their family (Office of Head Start, [<reflink idref="bib54" id="ref37">54</reflink>]). Although other early childhood education programs may include some parental outreach, most other programs are more similar to K–12 school systems in focusing on the child to promote early childhood development in contrast to Head Start's holistic approach. In the HSIS, random assignment to Head Start significantly increased the likelihood that children enrolled in an early childhood education center by 50 percentage points, increased the likelihood that they enrolled in a high-quality center (Friedman-Krauss, Connors, & Morris, [<reflink idref="bib20" id="ref38">20</reflink>]), and, most important for this analysis, increased mothers' educational attainment (Sabol & Chase-Lansdale, [<reflink idref="bib64" id="ref39">64</reflink>]). This study will leverage the experimentally induced impact on mothers' educational participation to explore associations between parenting and participation in maternal education when children are in Head Start.</p> <p>As well as providing a useful analytic tool, the experimental study of Head Start is a particularly interesting context for exploring whether participation in maternal education predicts parenting given current interest in two-generation education programs that aim to improve families' employment and economic opportunities by supporting mothers' participation in education while their children attend early childhood education programs (Chase-Lansdale & Brooks-Gunn, [<reflink idref="bib8" id="ref40">8</reflink>]). Given that the current research explores the effects of participation in maternal education when children were assigned to Head Start, it can contribute to understanding the potential of such two-generation education programs to shape parenting practices that likely influence children's later cognitive development.</p> <hd id="AN0126281570-5">The Present Study</hd> <p>In this research, the HSIS is used to answer the following research question: what is the effect of participation in maternal education on a constellation of parenting outcomes for low-income mothers of young children? Both short and longer term outcomes are explored to test whether educational participation predicts parenting while mothers are participating in education, and one year after mothers participated in education. The challenge in answering this question is that mothers self-select into participating in additional education, so we cannot simply compare the outcomes of children across mothers with different educational participation. Because mothers who participate in education may be more advantaged and educationally able, simply comparing their parenting to mothers who had the opportunity but did not engage in education could inflate estimates of the effect of education on parenting. Instead, we frame the question using a principal stratification framework in which study participants are assigned to predicted strata defined by both observed educational participation and the educational participation that would have been observed if they had been assigned to a different arm of the randomized intervention (Frangakis & Rubin, 2002). Within this framework, we use principal scores, which predict maternal educational participation based on covariates, to match mothers from both randomization arms who would have made the same educational participation choices as one another across the Head Start treatment and control conditions (Hill, Brooks-Gunn, & Waldfogel, [<reflink idref="bib28" id="ref41">28</reflink>]; Jo & Stuart, [<reflink idref="bib37" id="ref42">37</reflink>]; Jo, Stuart, MacKinnon, & Vinokur, [<reflink idref="bib38" id="ref43">38</reflink>]; Stuart & Jo, [<reflink idref="bib72" id="ref44">72</reflink>]). Within each of these strata we consider the effect of Head Start to shed light on the role of maternal educational participation as a mechanism that can affect parenting. Of particular interest is the group of mothers who would have participated in education after random assignment to Head Start but who would not have participated if not assigned to Head Start. For these women, under a specific set of assumptions that are discussed later, differences in parenting may be attributed to the effect of education. We conduct a number of sensitivity analyses, including comparing these results to more traditional mediation analyses and focus on results that are consistent across multiple specifications. This research extends prior work by using a novel quasi-experimental method to explore the causal relationship between participation in maternal education when children are in Head Start and parenting practices that have been associated with children's later academic development. Although estimating causal effects can require strong assumptions, we prefer to be clear about our inferential goal and describe the assumptions required in detail so the reader can make an informed choice about whether a causal versus descriptive interpretation of the results is merited.</p> <hd id="AN0126281570-6">Method</hd> <p></p> <hd id="AN0126281570-7">Data and Procedures</hd> <p>The data for this study come from the HSIS, a randomized controlled trial of the impacts of Head Start on children's school readiness (U.S. Department of Health and Human Services, [<reflink idref="bib75" id="ref45">75</reflink>]). The full HSIS sample includes data on 4,440 3- or 4-year-old children who applied and qualified to attend one of 378 Head Start centers in 22 states in the 2002–2003 school year. Families who applied to these Head Start centers received information about the HSIS, including procedures, potential benefits, and study incentives. Families were then randomly assigned to the offer of enrollment in the Head Start center they applied to or to the control group, in which they could seek out other child care (including care at other Head Start centers).[<reflink idref="bib1" id="ref46">1</reflink>] Families in the 3-year-old cohort were only assigned to one year of Head Start so families assigned to either treatment group could enroll at the Head Start center they applied to once children were 4 years old. Following the random assignment of families within centers, study personnel met with parents to further explain the study and gain parents' informed consent for the duration of the study. Data was collected between fall 2002 (baseline) and spring 2006 through annual parent interviews, teacher reports of children, and direct child assessments conducted by trained data collectors. This research uses data from: baseline (fall 2002), one year after random assignment ("year one," spring 2003), and two years after random assignment ("year two," spring 2004). For full information about procedures, see the technical report (U.S. Department of Health and Human Services, [<reflink idref="bib76" id="ref47">76</reflink>]).</p> <hd id="AN0126281570-8">Sample</hd> <p>The 3-year-old sample was used for analyses because initial analyses found that the significant treatment effect of Head Start on mothers' participation in education was driven by this group. The full 3-year-old sample includes 2,447 families. To ensure accurate information on the key predictor of interest, the 1,953 families (80% of the full sample) who had information on maternal participation in education in year one were included in the analysis. Compared to those in the analysis sample, mothers who were excluded from the sample were more disadvantaged: they were less likely to be married, less likely to have at least some college education, and more likely to have been born outside of the United States. Although this limits generalizability, the key predictor was not imputed because we were using principal-score-based methods and the same variables would have been used to impute the predictor and to create groups of mothers who are matched based on that predictor, resulting in double-matching. Other studies that have used multiple imputation and propensity score methods did not impute the predictor variable for this reason (Harding, [<reflink idref="bib24" id="ref48">24</reflink>]; Hill, Waldfogel, Brooks-Gunn, & Han, [<reflink idref="bib30" id="ref49">30</reflink>]).</p> <hd id="AN0126281570-9">Measures</hd> <p></p> <hd id="AN0126281570-10">Participation in Education</hd> <p>In year one, mothers were asked whether they had attended or enrolled in any courses from a school, college, or university since September the previous year (0 = "did not participate," 1 = "participated").</p> <hd id="AN0126281570-11">Parenting Outcomes</hd> <p>Self-reported parenting practices in year one were measured concurrent with mothers' reports of participating in education during the previous year, and parenting practices in year two were measured one year after mothers' reported participation in education. We conducted a confirmatory factor analysis on the six multi-item parenting scales using MPlus (v. 6.11; Muthen & Muthen, [<reflink idref="bib53" id="ref50">53</reflink>]) to determine that scale items loaded together consistently to represent parenting constructs. This factor analytic work was conducted because many of the parenting constructs were potentially highly related to one another and we wanted to ensure that items that were originally intended to measure a particular construct, such as cultural activity participation, were not too highly related to another potentially similar construct, such as parental involvement in the early childhood setting. We modeled all items as categorical variables using the robust weighted least squares (WLSMV) estimator because it does not assume normally distributed variables and provides the best option for modeling categorical or ordered data (Brown, [<reflink idref="bib7" id="ref51">7</reflink>]; Rhemtulla, Brosseau-Liard & Savalei, [<reflink idref="bib59" id="ref52">59</reflink>]). Based on this analysis, two items that had factor loadings below.3 on their intended scale were excluded to ensure that items adequately reflected the construct they were intended to represent (Estabrook & Neale, [<reflink idref="bib18" id="ref53">18</reflink>]). Model fit statistics indicated that the model fit the data acceptably according to commonly used fit criteria, χ<sups>2</sups>(<reflink idref="bib1" id="ref54">1</reflink>,<reflink idref="bib209" id="ref55">209</reflink>) = 5,595, <emph>p</emph> <.001, RMSEA =.045, CFI =.92 (Hu & Bentler, [<reflink idref="bib33" id="ref56">33</reflink>]). Mean scores were created for all multi-item scales. We chose not to model the outcomes as latent factors because principal scoring methods have not been developed for use with latent outcomes and we did not want to use weighted or regression scored variables as these are sample-specific (DiStefano, Zhu, & Mindrila, [<reflink idref="bib13" id="ref57">13</reflink>]). A recent simulation study showed that although weighted maximum likelihood and regression methods outperformed sum scoring methods, sum scoring methods performed adequately under similar conditions to this study (Estabrook & Neale, [<reflink idref="bib18" id="ref58">18</reflink>]). Means, standard deviations, and alpha coefficients for the outcomes are shown in Table 1.</p> <p>Table 1. Means, standard deviations, and Cronbach's alpha coefficients for baseline covariates and parenting outcomes, overall and for predicted subgroups.</p> <p> <ephtml> <table><thead><tr><td /><td>Compliers</td><td>Always takers</td><td>Never takers</td><td>Total</td><td /></tr><tr><td>Variables (range)</td><td><italic>M</italic></td><td><italic>SD</italic></td><td><italic>M</italic></td><td><italic>SD</italic></td><td><italic>M</italic></td><td><italic>SD</italic></td><td><italic>M</italic></td><td><italic>SD</italic></td><td>α</td></tr></thead><tbody><tr><td>Mother characteristics</td><td /><td /><td /><td /><td /><td /><td /><td /><td /></tr><tr><td> Age (20–64)</td><td char=".">30.41</td><td char=".">5.67</td><td char=".">30.59</td><td char=".">6.07</td><td char=".">31.89</td><td char=".">6.52</td><td char=".">31.50</td><td char=".">6.39</td><td /></tr><tr><td> Less than high school diploma (0,1)</td><td char=".">0.30</td><td char=".">0.46</td><td char=".">0.22</td><td char=".">0.42</td><td char=".">0.40</td><td char=".">0.49</td><td char=".">0.37</td><td char=".">0.48</td><td /></tr><tr><td> High school diploma/GED/vocational training (0,1)</td><td char=".">0.57</td><td char=".">0.50</td><td char=".">0.64</td><td char=".">0.48</td><td char=".">0.52</td><td char=".">0.50</td><td char=".">0.54</td><td char=".">0.50</td><td /></tr><tr><td> Some college or higher (0,1)</td><td char=".">0.13</td><td char=".">0.33</td><td char=".">0.14</td><td char=".">0.35</td><td char=".">0.08</td><td char=".">0.27</td><td char=".">0.09</td><td char=".">0.29</td><td /></tr><tr><td> Married (0,1)</td><td char=".">0.44</td><td char=".">0.50</td><td char=".">0.46</td><td char=".">0.50</td><td char=".">0.48</td><td char=".">0.50</td><td char=".">0.47</td><td char=".">0.50</td><td /></tr><tr><td> Works full time (0,1)</td><td char=".">0.29</td><td char=".">0.46</td><td char=".">0.32</td><td char=".">0.47</td><td char=".">0.37</td><td char=".">0.48</td><td char=".">0.35</td><td char=".">0.48</td><td /></tr><tr><td> Works part time (0,1)</td><td char=".">0.25</td><td char=".">0.43</td><td char=".">0.15</td><td char=".">0.36</td><td char=".">0.16</td><td char=".">0.37</td><td char=".">0.17</td><td char=".">0.38</td><td /></tr><tr><td> Does not work (0,1)</td><td char=".">0.46</td><td char=".">0.50</td><td char=".">0.53</td><td char=".">0.50</td><td char=".">0.47</td><td char=".">0.50</td><td char=".">0.48</td><td char=".">0.50</td><td /></tr><tr><td> Teen mother (0,1)</td><td char=".">0.21</td><td char=".">0.41</td><td char=".">0.25</td><td char=".">0.43</td><td char=".">0.16</td><td char=".">0.36</td><td char=".">0.18</td><td char=".">0.38</td><td /></tr><tr><td> Born in USA (0,1)</td><td char=".">0.68</td><td char=".">0.47</td><td char=".">0.73</td><td char=".">0.45</td><td char=".">0.68</td><td char=".">0.46</td><td char=".">0.69</td><td char=".">0.46</td><td /></tr><tr><td> CES-D depression (0–3)</td><td char=".">0.57</td><td char=".">0.57</td><td char=".">0.51</td><td char=".">0.52</td><td char=".">0.55</td><td char=".">0.56</td><td char=".">0.55</td><td char=".">0.56</td><td char=".">.87</td></tr><tr><td> KFASTS literacy score (55–130)</td><td char=".">82.26</td><td char=".">16.42</td><td char=".">88.85</td><td char=".">16.10</td><td char=".">84.25</td><td char=".">15.91</td><td char=".">84.78</td><td char=".">16.03</td><td /></tr><tr><td> Self-reported health (1–4)</td><td char=".">2.68</td><td char=".">1.05</td><td char=".">2.78</td><td char=".">1.00</td><td char=".">2.64</td><td char=".">1.04</td><td char=".">2.67</td><td char=".">1.03</td><td /></tr><tr><td>Household characteristics</td><td /><td /><td /><td /><td /><td /><td /><td /><td /></tr><tr><td> Father has less than a high school diploma (0,1)</td><td char=".">0.31</td><td char=".">0.46</td><td char=".">0.31</td><td char=".">0.46</td><td char=".">0.40</td><td char=".">0.49</td><td char=".">0.38</td><td char=".">0.49</td><td /></tr><tr><td> Father has high school diploma/GED (0,1)</td><td char=".">0.43</td><td char=".">0.50</td><td char=".">0.41</td><td char=".">0.49</td><td char=".">0.41</td><td char=".">0.49</td><td char=".">0.41</td><td char=".">0.49</td><td /></tr><tr><td> Father has some college or more (0,1)</td><td char=".">0.26</td><td char=".">0.44</td><td char=".">0.28</td><td char=".">0.45</td><td char=".">0.19</td><td char=".">0.40</td><td char=".">0.21</td><td char=".">0.41</td><td /></tr><tr><td> Father lives in household and works (0,1)</td><td char=".">0.41</td><td char=".">0.50</td><td char=".">0.31</td><td char=".">0.48</td><td char=".">0.41</td><td char=".">0.49</td><td char=".">0.40</td><td char=".">0.49</td><td /></tr><tr><td> Father lives in household and does not work (0,1)</td><td char=".">0.09</td><td char=".">0.28</td><td char=".">0.08</td><td char=".">0.27</td><td char=".">0.08</td><td char=".">0.27</td><td char=".">0.08</td><td char=".">0.27</td><td /></tr><tr><td> Father does not live in household (0,1)</td><td char=".">0.51</td><td char=".">0.50</td><td char=".">0.55</td><td char=".">0.50</td><td char=".">0.51</td><td char=".">0.50</td><td char=".">0.52</td><td char=".">0.50</td><td /></tr><tr><td> Primary home language is English (0,1)</td><td char=".">0.66</td><td char=".">0.48</td><td char=".">0.77</td><td char=".">0.42</td><td char=".">0.72</td><td char=".">0.45</td><td char=".">0.72</td><td char=".">0.45</td><td /></tr><tr><td> Number of children in household (1–6)</td><td char=".">2.59</td><td char=".">1.31</td><td char=".">2.37</td><td char=".">1.09</td><td char=".">2.71</td><td char=".">1.25</td><td char=".">2.65</td><td char=".">1.25</td><td /></tr><tr><td> Income to needs ratio (0–4.74)</td><td char=".">0.90</td><td char=".">0.56</td><td char=".">0.89</td><td char=".">0.56</td><td char=".">0.81</td><td char=".">0.54</td><td char=".">0.83</td><td char=".">0.55</td><td /></tr><tr><td> Family receives TANF (0,1)</td><td char=".">0.14</td><td char=".">0.35</td><td char=".">0.19</td><td char=".">0.40</td><td char=".">0.15</td><td char=".">0.36</td><td char=".">0.16</td><td char=".">0.36</td><td /></tr><tr><td> Family experienced economic difficulty (0,1)</td><td char=".">0.31</td><td char=".">0.46</td><td char=".">0.37</td><td char=".">0.48</td><td char=".">0.37</td><td char=".">0.48</td><td char=".">0.37</td><td char=".">0.48</td><td /></tr><tr><td> Family experienced poor housing (0,1)</td><td char=".">0.06</td><td char=".">0.16</td><td char=".">0.04</td><td char=".">0.12</td><td char=".">0.05</td><td char=".">0.14</td><td char=".">0.05</td><td char=".">0.14</td><td /></tr><tr><td> Urban locality (0,1)</td><td char=".">0.87</td><td char=".">0.34</td><td char=".">0.86</td><td char=".">0.34</td><td char=".">0.82</td><td char=".">0.39</td><td char=".">0.83</td><td char=".">0.38</td><td /></tr><tr><td> Month that interview was conducted (1–5)</td><td char=".">1.99</td><td char=".">1.03</td><td char=".">2.21</td><td char=".">1.05</td><td char=".">2.05</td><td char=".">1.02</td><td char=".">2.06</td><td char=".">1.03</td><td /></tr><tr><td>Child characteristics</td><td /><td /><td /><td /><td /><td /><td /><td /><td /></tr><tr><td> Male (0,1)</td><td char=".">0.34</td><td char=".">0.47</td><td char=".">0.50</td><td char=".">0.50</td><td char=".">0.50</td><td char=".">0.50</td><td char=".">0.49</td><td char=".">0.50</td><td /></tr><tr><td> Hispanic (0,1)</td><td char=".">0.27</td><td char=".">0.45</td><td char=".">0.32</td><td char=".">0.47</td><td char=".">0.33</td><td char=".">0.47</td><td char=".">0.33</td><td char=".">0.47</td><td /></tr><tr><td> Black (0,1)</td><td char=".">0.40</td><td char=".">0.49</td><td char=".">0.40</td><td char=".">0.49</td><td char=".">0.37</td><td char=".">0.48</td><td char=".">0.38</td><td char=".">0.48</td><td /></tr><tr><td> White (0,1)</td><td char=".">0.19</td><td char=".">0.39</td><td char=".">0.26</td><td char=".">0.44</td><td char=".">0.26</td><td char=".">0.44</td><td char=".">0.25</td><td char=".">0.44</td><td /></tr><tr><td> Other (0,1)</td><td char=".">0.14</td><td char=".">0.35</td><td char=".">0.01</td><td char=".">0.10</td><td char=".">0.04</td><td char=".">0.20</td><td char=".">0.04</td><td char=".">0.21</td><td /></tr><tr><td> Age in months (75–101)</td><td char=".">84.50</td><td char=".">4.08</td><td char=".">84.40</td><td char=".">3.90</td><td char=".">84.26</td><td char=".">4.09</td><td char=".">84.31</td><td char=".">4.03</td><td /></tr><tr><td> Maternal reported health (1–4)</td><td char=".">3.34</td><td char=".">0.85</td><td char=".">3.21</td><td char=".">0.90</td><td char=".">3.19</td><td char=".">0.91</td><td char=".">3.20</td><td char=".">0.90</td><td /></tr><tr><td> Self regulation (1–4)</td><td char=".">2.91</td><td char=".">0.86</td><td char=".">2.96</td><td char=".">0.83</td><td char=".">2.92</td><td char=".">0.86</td><td char=".">2.93</td><td char=".">0.85</td><td /></tr><tr><td> Exact age at testing in weeks (169–290)</td><td char=".">215.20</td><td char=".">17.83</td><td char=".">215.24</td><td char=".">17.20</td><td char=".">214.33</td><td char=".">17.99</td><td char=".">214.55</td><td char=".">17.76</td><td /></tr><tr><td> Testing language was Spanish (0,1)</td><td char=".">0.22</td><td char=".">0.42</td><td char=".">0.19</td><td char=".">0.39</td><td char=".">0.23</td><td char=".">0.42</td><td char=".">0.22</td><td char=".">0.42</td><td /></tr><tr><td> Peabody Picture Vocabulary Test-III (129–361)</td><td char=".">229.70</td><td char=".">36.82</td><td char=".">237.48</td><td char=".">37.19</td><td char=".">228.44</td><td char=".">37.12</td><td char=".">229.52</td><td char=".">37.04</td><td /></tr><tr><td> Letter-word identification (250–374)</td><td char=".">297.89</td><td char=".">20.67</td><td char=".">296.27</td><td char=".">22.10</td><td char=".">293.16</td><td char=".">23.08</td><td char=".">293.93</td><td char=".">22.71</td><td /></tr><tr><td>Parenting outcomes in year one</td><td /><td /><td /><td /><td /><td /><td /><td /><td /></tr><tr><td> Frequency of cognitively stimulating activities (1–6)</td><td char=".">3.58</td><td char=".">1.11</td><td char=".">3.68</td><td char=".">1.17</td><td char=".">3.60</td><td char=".">1.09</td><td char=".">3.62</td><td char=".">1.11</td><td char=".">0.92</td></tr><tr><td> Hours of TV watched (0–9)</td><td char=".">3.09</td><td char=".">1.90</td><td char=".">2.69</td><td char=".">1.35</td><td char=".">3.00</td><td char=".">1.73</td><td char=".">2.96</td><td char=".">1.69</td><td /></tr><tr><td> Types of printed media in the home (0–1)</td><td char=".">0.64</td><td char=".">0.24</td><td char=".">0.71</td><td char=".">0.21</td><td char=".">0.63</td><td char=".">0.23</td><td char=".">0.65</td><td char=".">0.23</td><td char=".">0.58</td></tr><tr><td> Beliefs about the importance of early learning (1–5)</td><td char=".">4.16</td><td char=".">0.51</td><td char=".">4.27</td><td char=".">0.48</td><td char=".">4.21</td><td char=".">0.53</td><td char=".">4.22</td><td char=".">0.52</td><td char=".">0.76</td></tr><tr><td> Frequency of participation in cultural activities (0–1)</td><td char=".">0.46</td><td char=".">0.23</td><td char=".">0.49</td><td char=".">0.24</td><td char=".">0.45</td><td char=".">0.22</td><td char=".">0.46</td><td char=".">0.23</td><td char=".">0.46</td></tr><tr><td> Involvement with educational institutions (0–4)</td><td char=".">2.31</td><td char=".">1.05</td><td char=".">2.19</td><td char=".">1.00</td><td char=".">2.10</td><td char=".">0.88</td><td char=".">2.14</td><td char=".">0.92</td><td char=".">0.73</td></tr><tr><td> Perceived support (0–3)</td><td char=".">1.27</td><td char=".">0.73</td><td char=".">1.21</td><td char=".">0.70</td><td char=".">1.22</td><td char=".">0.72</td><td char=".">1.22</td><td char=".">0.72</td><td char=".">0.36</td></tr><tr><td>Parenting outcomes in year two</td><td /><td /><td /><td /><td /><td /><td /><td /><td /></tr><tr><td> Frequency of cognitively stimulating activities (1–6)</td><td char=".">4.03</td><td char=".">1.02</td><td char=".">4.02</td><td char=".">1.06</td><td char=".">4.00</td><td char=".">1.06</td><td char=".">4.02</td><td char=".">1.06</td><td char=".">0.92</td></tr><tr><td> Hours of TV watched (0–9)</td><td char=".">2.43</td><td char=".">1.49</td><td char=".">2.37</td><td char=".">1.30</td><td char=".">2.70</td><td char=".">1.54</td><td char=".">2.63</td><td char=".">1.51</td><td /></tr><tr><td> Types of printed media in the home (0–1)</td><td char=".">0.69</td><td char=".">0.22</td><td char=".">0.73</td><td char=".">0.20</td><td char=".">0.66</td><td char=".">0.23</td><td char=".">0.68</td><td char=".">0.23</td><td char=".">0.59</td></tr><tr><td> Beliefs about the importance of early learning (1–5)</td><td char=".">4.26</td><td char=".">0.54</td><td char=".">4.28</td><td char=".">0.51</td><td char=".">4.24</td><td char=".">0.53</td><td char=".">4.25</td><td char=".">0.53</td><td char=".">0.79</td></tr><tr><td> Frequency of participation in cultural activities (0–1)</td><td char=".">0.53</td><td char=".">0.23</td><td char=".">0.55</td><td char=".">0.22</td><td char=".">0.49</td><td char=".">0.23</td><td char=".">0.50</td><td char=".">0.23</td><td char=".">0.41</td></tr><tr><td> Involvement with educational institutions (0–4)</td><td char=".">2.25</td><td char=".">0.92</td><td char=".">2.23</td><td char=".">0.91</td><td char=".">2.17</td><td char=".">0.84</td><td char=".">2.19</td><td char=".">0.86</td><td char=".">0.74</td></tr><tr><td> Perceived support (0–3)</td><td char=".">1.24</td><td char=".">0.74</td><td char=".">1.33</td><td char=".">0.76</td><td char=".">1.26</td><td char=".">0.74</td><td char=".">1.27</td><td char=".">0.74</td><td char=".">0.39</td></tr></tbody></table> </ephtml> </p> <p>8 <emph>N</emph> = 1953.</p> <p>The following set of measures were considered to represent mothers' human capital because they were expected to change as mothers acquired skills and knowledge that they could then use to structure their children's environments.</p> <p> <emph>Frequency of cognitively stimulating activities:</emph> Mothers reported how frequently they or someone in their household participated in 20 language or math activities with their child Monday through Friday between 8 a.m. and 6 p.m. (e.g., "discuss new words"; 1 = <emph>never</emph>, 5 = <emph>every day</emph>).</p> <p> <emph>Types of printed media in the home:</emph> Mothers were asked whether or not they had seven types of printed media in the home (e.g., "newspapers"), and whether or not they had taken any books home from the library in the past month.</p> <p> <emph>Hours of TV watched:</emph> Mothers reported how many hours children usually watch TV in their home each day.</p> <p> <emph>Beliefs about the importance of early learning:</emph> Mothers were asked to what extent they agreed with eight items about children's preparation for school (e.g., "attending preschool is very important for success in kindergarten"; 1 = <emph>strongly disagree</emph>, 5 = <emph>strongly agree</emph>).</p> <p>The following set of measures were considered to represent mothers' cultural capital as they were expected to change through mothers gaining more familiarity with cultural institutions and confidence in navigating bureaucratic settings.</p> <p> <emph>Frequency of participation in cultural activities</emph>: Mothers reported whether or not anyone in their family had participated in six activities with the child in the last month (e.g., "visited an art gallery, museum, or historical site").</p> <p> <emph>Involvement with educational institutions:</emph> Mothers reported how often they had participated in four activities at their child's care setting since the beginning of the year (e.g., "attended parent–care provider conferences"; 1 = <emph>not yet</emph>, 5 = <emph>at least once a week</emph>). Only the 75% of families who reported their child attended child care were included in these analyses (year one: <emph>N</emph> = 1,512; year two: <emph>N</emph> = 1,577).</p> <p>The following measure was considered to represent mothers' social capital; although it does not capture the educational level of mothers' social networks, it was the only potential measure available in the data.</p> <p> <emph>Perceived instrumental support:</emph> Mothers were asked how helpful "friends," "professional help givers like counselors or social workers," and "religious or social group members" were (0 = <emph>not applicable</emph>, 1 = <emph>not very helpful</emph>, 3 = <emph>very helpful</emph>).</p> <hd id="AN0126281570-12">Baseline Demographic Covariates</hd> <p>A large set of baseline variables that could be related to mothers' participation in education and parenting (that is, potential confounders) were included in analyses. Covariates were from the baseline parent interview and baseline child assessment. Means and standard deviations for all covariates for the full sample are shown in Table 1. <emph>Maternal covariates</emph> included dummy codes for: employment status (employed full time, employed part time, not employed), education (less than a HS diploma, a HS diploma or GED, some college or vocational training, an associate's degree or higher); indicators for whether the mother was: born in the United States, married, a teen mother for the birth of this child; mothers' age; maternal literacy scores (Kaufman Functional Academic Skills Test, K-FAST; Kaufman, [<reflink idref="bib40" id="ref59">40</reflink>]); self-reported health (1 = <emph>poor</emph>, 5 = <emph>excellent</emph>); and depression (mean of 12-item Center for Epidemiologic Studies—Depression scale, CES-D; Seligman, [<reflink idref="bib67" id="ref60">67</reflink>]). <emph>Familial covariates</emph> included dummy codes for: fathers' education (less than a HS diploma, a HS diploma or GED, some college or vocational training, an associate's degree or higher), whether the father is present in the household and whether they are employed at the time of the interview (father lives in household and works at least part time, father lives in household and does not work, father does not live in household); indicators for: primary home language is English, the family reports at least one economic difficulty, the family receives TANF, the family reports inadequate housing, the family lives in an urban locale; income-to-needs ratios created using 2002 U.S. Census poverty thresholds based on income and family size (<1 = "below federal poverty line," >1 = "above federal poverty line"); the number of children in the household; and the month the interview was conducted. In addition, 23 state indicators based on geocoding of the Head Start center children applied to were included. <emph>Child covariates</emph> included dummy codes for: child race/ethnicity (Hispanic, Black, White, Other); indicators for: male gender, testing language was English; child's age in months; parent reports of the child's health (1 = <emph>poor</emph>, 5 = <emph>excellent</emph>); exact age at testing in weeks; self-regulation (mean of 4-item scale of the assessor's reports of children from the Leiter Revised Sustained Attention Task; Roid & Miller, [<reflink idref="bib60" id="ref61">60</reflink>]); scores on the letter–word identification subtask from the Woodcock Johnson (WJ-III; Woodcock, McGrew, & Mather, [<reflink idref="bib78" id="ref62">78</reflink>]) and IRT scores for the Peabody Picture Vocabulary Test—Third Edition (PPVT; Dunn & Dunn, [<reflink idref="bib17" id="ref63">17</reflink>]).</p> <hd id="AN0126281570-13">Missing Data</hd> <p>The full analytic sample included the 1,953 families who had information on maternal participation in education in year one (see the "Sample" section for more information). Information on child race, child gender, home language, and state was available for almost the full analytic sample (<.01% missing). Certain baseline covariates, including maternal literacy score, father education, income-to-needs ratios, whether the father was in the household, and direct child assessments had high rates of missing data (12%–20%). All families had outcome data in year one, but seven percent of the sample were missing outcome data in year two. Multiple imputation using chained equations in the mi package in R (v. 3.1.0) was used to impute five complete data sets (Gelman, Hill, Su, Yajima, & Pittau, [<reflink idref="bib21" id="ref64">21</reflink>]). To verify that the imputations converged, the models fit the observed data well, and the distributions of the imputed data were similar to the observed data, plots were examined. Primary analyses were conducted on a single imputed data set that was selected at random.[<reflink idref="bib2" id="ref65">2</reflink>] We chose to conduct the primary analyses, descriptive analyses, and assumption tests in a single data set because methods to combine these statistics across multiple data sets are not always well developed. For instance, we wanted to use <emph>H<subs>T</subs></emph> tests (discussed later) to compare Cohen's <emph>d</emph> effect sizes across groups of mothers in the different predicted strata given the substantial differences in the size of these groups. To ensure that the results were not being driven by the choice of a single data set and to account for the uncertainty of the predicted values, we also present sensitivity analyses that display the results from the principal-score analyses conducted in and combined across the five imputed data sets using Rubin's rules to account for the uncertainty in the imputed data sets (Rubin, [<reflink idref="bib62" id="ref66">62</reflink>]).</p> <hd id="AN0126281570-14">Analytic Approach</hd> <p></p> <hd id="AN0126281570-15">Potential Outcomes Framework</hd> <p>This research draws on the potential outcomes framework, in which every person is assumed to have two potential outcomes—one outcome that would be realized under assignment to control and one outcome that would be realized under assignment to treatment (Rubin, [<reflink idref="bib61" id="ref67">61</reflink>]). Let <emph>T<subs>i</subs></emph> represent the treatment received by participant <emph>i, i</emph> = 1, ..., N. Those with <emph>T<subs>i</subs></emph> = 1 are treated, and those with <emph>T<subs>i</subs></emph> = 0 are untreated. Let <emph>Y<subs>i</subs></emph>(0) be the outcome if <emph>T<subs>i</subs></emph> = 0, and <emph>Y<subs>i</subs></emph>(<reflink idref="bib1" id="ref68">1</reflink>) be the outcome if <emph>T<subs>i</subs></emph> = 1. The individual causal effect is the difference between these two potential outcomes for participant <emph>i, Y<subs>i</subs></emph>(<reflink idref="bib1" id="ref69">1</reflink>) – <emph>Y<subs>i</subs></emph>(0). The challenge in identifying causal effects is that we can never observe both potential outcomes for any individual. However in a randomized experiment we can unbiasedly estimate the average causal effect, E[<emph>Y</emph>(<reflink idref="bib1" id="ref70">1</reflink>) − <emph>Y</emph>(0)] (the average of all of the individual-level causal effects) because we can use the average of the treated outcomes as an unbiased estimate of the average of <emph>Y</emph>(<reflink idref="bib1" id="ref71">1</reflink>) and we can use the average of the control outcomes as an unbiased estimate of the average of <emph>Y</emph>(0).</p> <p>In the current application, the situation is more complicated because the key causal variable, educational participation, can also be affected by the randomized treatment. Thus we need to consider potential outcomes with respect to both maternal educational participation (which is considered a mediator), <emph>M<subs>i</subs></emph>(0) and <emph>M<subs>i</subs></emph>(<reflink idref="bib1" id="ref72">1</reflink>), <emph>and</emph> parenting, <emph>Y<subs>i</subs></emph>(0) and <emph>Y<subs>i</subs></emph>(<reflink idref="bib1" id="ref73">1</reflink>). For individuals who do not receive the treatment, <emph>M<subs>i</subs></emph>(<reflink idref="bib1" id="ref74">1</reflink>) and <emph>Y<subs>i</subs></emph>(<reflink idref="bib1" id="ref75">1</reflink>) are missing. Likewise, for individuals who receive the treatment, <emph>M<subs>i</subs></emph>(0) and <emph>Y<subs>i</subs></emph>(0) are missing. We can use this framework to define the mediator based on the set of potential outcomes <emph>M<subs>i</subs></emph>(<reflink idref="bib1" id="ref76">1</reflink>), <emph>M<subs>i</subs></emph>(0). This is helpful because, unlike observed maternal education, we can condition on these educational potential outcomes without inducing bias (Frangakis & Rubin, 2002).</p> <p>Specifically, in this scenario, mothers' educational potential outcomes represent: (a) whether the mother would have participated in education, given random assignment to Head Start and (b) whether the mother would have participated in education, given random assignment to the control condition. These questions are used to define three "educational participation strata" of mothers. Borrowing language from Angrist, Imbens, and Rubin ([<reflink idref="bib1" id="ref77">1</reflink>]) that has become standard, "compliers" are mothers who would not engage in education when assigned to control, but who would participate in education when assigned to Head Start. Conceptually, these mothers may require the extra support and encouragement offered by Head Start to participate in education. "Always takers" are mothers who would participate in education regardless of whether or not they were randomly assigned to Head Start or the control. These mothers may be particularly motivated around education, or had been enrolled in education before the study began. "Never takers" are mothers who would not engage in education regardless of whether or not they were randomly assigned to Head Start. These mothers may not participate in education because they have stable jobs or have had prior negative educational experiences.[<reflink idref="bib3" id="ref78">3</reflink>] There are compliers, always takers and never takers in the Head Start and control groups; to distinguish them, participants are called "treatment compliers" or "control compliers" (Table 2).[<reflink idref="bib4" id="ref79">4</reflink>]</p> <p>Table 2. Visual representation of labels used to refer to groups.</p> <p> <ephtml> <table><thead><tr><td /><td /><td>Would they have participated in education if they were assigned to the alternate group?</td></tr></thead><tbody><tr><td /><td /><td>Yes</td><td>No</td></tr><tr><td>Did they participate in education given their current random assignment status?</td><td>Yes</td><td>Control always taker<xref ref-type="fn" rid="t2fn0001" /></td><td>Control defier<xref ref-type="fn" rid="t2fn0002" /></td></tr><tr><td /><td /><td>Treatment always taker</td><td>Treatment complier</td></tr><tr><td /><td>No</td><td>Control complier</td><td>Control never taker</td></tr><tr><td /><td /><td>Treatment defier<xref ref-type="fn" rid="t2fn0002" /></td><td>Treatment never taker<xref ref-type="fn" rid="t2fn0001" /></td></tr></tbody></table> </ephtml> </p> <ulist> <item>20001 These groups can be observed.</item> <item>20002 These groups are assumed to not exist.</item> </ulist> <hd id="AN0126281570-16">Identifying Subgroups</hd> <p>The challenge with causal inference is that we only observe each mother's educational participation under one condition; however, educational strata are defined by what the study participant would do in each of the arms. Thus, we have to predict membership in educational strata. We can observe the educational participation stratum of two groups of mothers: those who did not participate in education despite being randomized to Head Start (treatment never takers; <emph>M<subs>i</subs></emph>(<reflink idref="bib1" id="ref80">1</reflink>) = 0 and <emph>M<subs>i</subs></emph>(0) = 0) and those who participated despite being randomized to control (control always takers; <emph>M<subs>i</subs></emph>(0) = 1 and <emph>M<subs>i</subs></emph>(<reflink idref="bib1" id="ref81">1</reflink>) = 1). However, among those who participated in education within the treatment group, we do not know which mothers only participated because of random assignment to Head Start (treatment compliers; <emph>M<subs>i</subs></emph>(<reflink idref="bib1" id="ref82">1</reflink>) = 1 and <emph>M<subs>i</subs></emph> (0) = 0) and which mothers would have participated anyway (treatment always takers; <emph>M<subs>i</subs></emph>(<reflink idref="bib1" id="ref83">1</reflink>) = 1 and <emph>M<subs>i</subs></emph>(0) = 1). Similarly, among those who did not participate in education within the control group, we do not know which mothers would have participated if they were assigned to treatment (control compliers; <emph>M<subs>i</subs></emph>(0) = 0 and <emph>M<subs>i</subs></emph>(<reflink idref="bib1" id="ref84">1</reflink>) = 1) and which mothers would not have participated anyway (control never takers; <emph>M<subs>i</subs></emph>(0) = 0 and <emph>M<subs>i</subs></emph>(<reflink idref="bib1" id="ref85">1</reflink>) = 0) (Jo, [<reflink idref="bib36" id="ref86">36</reflink>]).</p> <p>To identify group membership for mothers whose educational participation stratum was unobserved, we make predictions about stratum membership based on covariate information. Following Hill et al. ([<reflink idref="bib28" id="ref87">28</reflink>]) we call the predicted probabilities associated with these strata "principal scores" to reflect their use within the principal stratification framework (Frangakis & Rubin, 2002). Principal scores facilitate the comparison of subjects who are as similar as possible on background characteristics by using a scalar summary of these values. The key difference between principal scores and propensity scores within a principal stratification framework is that principal scores build a prediction model for stratum membership within a given randomization arm and then use that to make predictions in the other. Propensity score models, on the other hand, predict subgroups by building a model for stratum membership using observations across both arms. Both propensity and principal scores have been used in experimental contexts to create subgroups that are defined by a mediating variable such as dosage, with group membership predicted from baseline covariates (Moulton et al., [<reflink idref="bib52" id="ref88">52</reflink>]). Applications of these approaches have explored the effects of: Job Corps, a training program for disadvantaged youth (Schochet & Burghardt, [<reflink idref="bib66" id="ref89">66</reflink>]), early childhood intervention programs (Hill et al., [<reflink idref="bib28" id="ref90">28</reflink>]), and a job search intervention (Jo et al., [<reflink idref="bib38" id="ref91">38</reflink>]). See Stuart and Jo ([<reflink idref="bib72" id="ref92">72</reflink>]) for a more complete technical introduction to this approach. The current research uses principal scores to identify both the always takers and the compliers (see the analytic steps outlined below). The distinction between always takers and compliers is particularly relevant in the present case because the group of compliers is small relative to the always takers. Had these groups been examined together, the effects for the compliers would have been subsumed in the overall group (Jo et al., [<reflink idref="bib38" id="ref93">38</reflink>]).[<reflink idref="bib5" id="ref94">5</reflink>]</p> <hd id="AN0126281570-17">Step 1</hd> <p>To identify the always takers in the Head Start treatment group, three steps were performed. <emph>Step 1a</emph>: Based on the mothers who participated in education even though they were not assigned to Head Start (i.e., the observed control always takers), a probit regression model was fit to calculate the likelihood of participating in education (<emph>M<subs>i</subs>)</emph> for those assigned to the control group using the following equation: Probit(<emph>Mi) = <emph>β</emph><subs>0</subs></emph> + <emph>∑<emph>β</emph><subs>1</subs>X<subs>i</subs></emph> + ϵ<emph><subs>i</subs></emph>. Here <emph>X<subs>i</subs></emph> represents the extensive set of baseline child, mother, and family characteristics that could influence mother <emph>i</emph>'s likelihood of participating in education. <emph>Step 1b</emph>: The coefficients from this regression model and the values of mothers' demographic characteristics were used to predict principal scores for mothers in the Head Start treatment condition. Thus, each mother had a score that predicted how likely she would be to participate in education had she been assigned to the control group. <emph>Step 1c</emph>: Caliper matching without replacement was used to identify matches for the control always takers from mothers who participated in education in the Head Start group. Matching without replacement means that once an observation has been selected as a match, it cannot be selected again. Caliper matching includes mothers as matches if their principal scores fall within a set range of the score of the mother for whom a match is sought. In this case, mothers in the Head Start treatment group were included as matches for the mothers who participated in education in the control group if they had principal scores within a range of.003.[<reflink idref="bib6" id="ref95">6</reflink>] This matching procedure identified a group of mothers in the Head Start treatment group who participated in education who were predicted to be treatment always takers based on their similarity to the observed control always takers. The other mothers who participated in education in the Head Start treatment group were thus assumed to be treatment compliers.</p> <hd id="AN0126281570-18">Step 2</hd> <p>To identify the compliers in the control group, three steps were performed. <emph>Step 2a</emph>: A probit regression analysis using the same equation as in Step 1a was conducted; however, now the likelihood of participating in education was modeled in the Head Start group (with the predicted always takers excluded), rather than in the control group. <emph>Step 2b</emph>: The coefficients from this regression model and the values of mothers' demographic characteristics were used to predict principal scores for mothers in the control condition who did not participate in education. Thus, each mother had a score that predicted how likely she would be to participate in education had she been assigned to Head Start. <emph>Step 2c</emph>: One-to-one matching, with replacement, was used to identify matches for mothers who increased their education in the Head Start group from those in the control group based on their principal scores. This matching procedure identified a group of mothers in the control group who did not participate in education who were predicted to be control compliers based on their similarity to the predicted treatment compliers. The remaining control group mothers who did not participate in education were assumed to be never takers.</p> <hd id="AN0126281570-19">Estimates of Stratum-Specific Effects</hd> <p>The final analysis equations within each stratum took the form: <emph>Y<subs>i</subs></emph> = β<subs>0</subs> + <emph><emph>β</emph><subs>1</subs></emph>Z<emph><subs>i</subs></emph> + ∑ <emph><emph>β</emph><subs>2</subs>X<subs>i</subs></emph> + ϵ<subs>i</subs>. Here the parenting outcomes (<emph>Y<subs>i</subs></emph>) are a function of random assignment to Head Start (<emph><emph>β</emph><subs>1</subs></emph>Z<emph><subs>i</subs></emph>) and the confounding covariates used to estimate the principal scores (∑<emph><emph>β</emph><subs>2</subs>X<subs>i</subs></emph>). These analyses were run across the matched groups of mothers who were predicted to be in each stratum. The primary analysis of interest was the regression conducted in the group of predicted compliers, which estimates the complier average causal effect. This analysis estimates the effect of participation in maternal education because all the mothers in the Head Start group participated in education, whereas all mothers in the control group did not. Weights were used to indicate how many times a control mother was used as a match for a treatment mother; Head Start treatment group members were assigned weights of 1. All covariates were included in the regression analyses to further reduce bias and improve the precision of the estimates (Rubin & Thomas, [<reflink idref="bib63" id="ref96">63</reflink>]).</p> <p>The same regression analyses were then conducted within the matched always and never takers groups (calculating the never-taker average causal effect and always-taker average causal effect; see Jo & Stuart, [<reflink idref="bib37" id="ref97">37</reflink>]). These analyses help us to better understand the extent to which the results for the compliers may be due to mechanisms other than participation in maternal education. If random assignment to Head Start is associated with the parenting of those in the always and never takers groups, this suggests that access to Head Start is affecting parenting through other mechanisms given that these mothers have the same educational participation but differ in assignment to Head Start. <emph>H<subs>T</subs></emph> tests, which test the weighted sum of squares of the Cohen's <emph>d</emph> effect size estimates for the groups around the weighted mean effect using a chi-squared distribution, were then used to determine whether the coefficients for the compliers, always takers, and never takers differed significantly from one another (Greenberg, Meyer, & Wiseman, [<reflink idref="bib23" id="ref98">23</reflink>]).</p> <hd id="AN0126281570-20">Assumptions of the Principal-Score Approach</hd> <p>The primary assumption is that there is ignorability of selection into participation in maternal education. The principal stratification approach helps to address selection by conditioning on potential outcomes, which are unbiased. Moreover, because of random assignment we can assume that there are mothers in the control group who are comparable to mothers who participated in education because their children were assigned to Head Start (Hill et al., [<reflink idref="bib28" id="ref99">28</reflink>]). However, we still have to assume that the covariates are sufficient to predict both the maternal education potential outcomes and the parenting potential outcomes. A vast number of covariates, including mothers' cognitive skills, income, and marital status, were included in analyses to try to address this assumption. Nonetheless, because participation in education was not randomized, there are other plausible variables that could be associated with both mothers' decisions to enroll in additional education and their parenting. Indeed, the assumption of ignorability is untestable. Notably, a simulation study of principal-score techniques found that when the assumption of ignorability based on covariates was not met, the quality of the estimates decreased but the inferences based on these analyses remained generally correct, particularly when the covariates were strongly predictive of stratum membership (Jo & Stuart, [<reflink idref="bib37" id="ref100">37</reflink>]). This analysis also assumes ignorability of the treatment assignment with respect to both the educational and parenting outcomes. This is satisfied due to the initial randomization. Next, this analysis assumes that the mothers who fall into the compliers, always takers, and never takers groups can be identified (another way of framing this issue is that we are fitting the correct principal-score model). We will explore this descriptively by examining whether the principal-score models adequately predict participation in maternal education and whether the matching procedures result in good balance between groups. Another assumption relevant to estimating causal effects is the stable unit treatment value assumption, which requires that one person's treatment assignment does not affect another person's outcome. This assumption is likely met as the sample is spread across a number of Head Start centers in 22 states, so it is unlikely that one mother's participation in education affects the parenting of another mother.</p> <hd id="AN0126281570-21">Traditional Mediation Analysis Approach</hd> <p>Given the novelty of the method employed, and the assumptions required to accept these estimates as causal, more traditional mediation analyses were also conducted to explore the extent to which participation in maternal education mediated the effects of random assignment to Head Start on parenting. Techniques developed by Imai and colleagues were used to predict parenting based on random assignment to Head Start and whether mothers participated in education (Hicks & Tingley, [<reflink idref="bib27" id="ref101">27</reflink>]; Imai, Keele, & Tingley, [<reflink idref="bib34" id="ref102">34</reflink>]; Imai, Keele, Tingley, & Yamamoto, [<reflink idref="bib35" id="ref103">35</reflink>]). Like the commonly used linear structural equation modeling approach (Baron & Kenny, [<reflink idref="bib3" id="ref104">3</reflink>]), this method decomposes the effect of random assignment to Head Start on parenting into the indirect effect, which occurs through participation in maternal education, and the direct effect, which includes other potential mediators of the relationship. Crucially, the mediation analyses rely on a similar assumption to the principal-score approach: ignorability of the mediator. However, in comparison to the principal-score approach, the mediation approach blends estimates of the effects for the always takers and compliers and calculates the average causal mediation effect. The principal-score approach focuses on the compliers, since this is the only group whose educational participation was actually affected by the randomized intervention so this group can provide empirical evidence about both counterfactual states. Moreover, due to the randomization, we expect that the composition of principal strata and the relationship between principal stratum membership and covariates will be the same across treatment groups. If we can use this covariate information to identify the compliers, estimates of differences in parenting can be unbiased given that these mothers were induced to participate in education <emph>because of</emph> random assignment to Head Start.</p> <hd id="AN0126281570-22">Results</hd> <p></p> <hd id="AN0126281570-23">Treatment Effect of Head Start on Participation in Maternal Education</hd> <p>As a first step, the proportion of mothers who participated in education in the Head Start treatment and control conditions was compared to assess the effect of random assignment to Head Start on participation in education in the year following random assignment. Eighteen percent of mothers in the control group participated in education compared to 24% of mothers in the Head Start group. The additional six percent of participants in the Head Start group are the treatment compliers. This is the primary group of interest because these mothers were induced into educational participation through access to Head Start.</p> <hd id="AN0126281570-24">Identifying Subgroups</hd> <p></p> <hd id="AN0126281570-25">Balance of Groups</hd> <p>In each matching procedure used to identify groups of mothers in different strata, different principal-score model specifications were used to obtain the best balance across groups in means and standard deviations for all covariates (results available from the first author). Rules of thumb suggest that the absolute values of the standardized mean difference between matched groups should be less than 0.1 and that the ratio of standard deviations should be between 0.9 and 1.1 (McCormick, Cappella, O'Connor, Hill, & McClowry, [<reflink idref="bib49" id="ref105">49</reflink>]). Both matching procedures improved the balance of the matched groups compared to the unmatched groups, such that after matching, only a small number of the 62 covariates included in the models did not meet this criteria. The matching also decreased the maximum standardized difference between groups on covariate means from.25 for the unmatched groups to.17–.22 for the matched groups and the maximum deviation in standard deviation ratios from.14–.17 to.13–.14. Overall, the average standardized differences for the matched mothers were.03–.04 and the average differences in the standard deviation ratios were.06–.07, which fall under the rule-of-thumb criteria suggested. These results indicate matched groups were similar in the distribution of observed covariates.</p> <hd id="AN0126281570-26">Accuracy of Predictive Models</hd> <p>To explore how well each model predicted the likelihood of educational participation descriptively, the percentage of mothers who participated in education in each decile of principal scores was plotted. A lower principal score indicates less likelihood of participating in education, thus mothers with low principal scores should have lower rates of educational participation, and vice versa (Kemple, Snipes, & Bloom, [<reflink idref="bib41" id="ref106">41</reflink>]). Appendix B shows the percentage of Head Start treatment and control participants who participated in education within principal-score deciles that represent how likely mothers would be to participate in education if they were in the control condition. As illustrated, mothers who had low predicted probabilities of participating in education infrequently participated in education.[<reflink idref="bib7" id="ref107">7</reflink>] For the second principal-score model, mothers who had low predicted probabilities of participating in education also infrequently participated in education (figure available from the first author). These results suggest that the baseline characteristics used in the principal-score models have predictive validity.</p> <hd id="AN0126281570-27">Descriptive Subgroup Differences</hd> <p>Means and standard deviations for the matched groups of compliers, always takers, and never takers on covariates and parenting outcomes are presented in Table 1. The observed differences in covariate means across groups suggest these are distinct groups of mothers. For instance, the always takers are more advantaged than the never takers and compliers in many ways, including having higher baseline education, lower depression, better self-reported health, and higher literacy scores, and they report more beneficial parenting overall. Although not directly measuring mothers' motivation, these descriptive differences do fit with the expectation that the always takers may be particularly motivated toward education and supporting their children's learning to begin with. Similarly, the expectation that the never takers might have had prior negative educational experiences or already have stable jobs was supported because these mothers had much lower levels of education and were more likely to work full time at baseline compared to the other two groups of mothers.</p> <hd id="AN0126281570-28">Associations Between Participation in Maternal Education and Parenting</hd> <p>Principal-score matched regression analyses were conducted to explore the relationship between random assignment to Head Start and parenting in the matched sample of mothers from the Head Start treatment and control groups who we predicted would have participated in education only when assigned to Head Start (compliers). Because only mothers in the Head Start group participated in education, these results give the effect of maternal educational participation. As shown in Table 3, in year one, after adjusting for covariates, mothers' participation in education when children were assigned to Head Start was significantly associated with fewer hours of TV watching, having more types of printed media in the home (at trend-level), more strongly endorsing the importance of early learning, more frequent participation in cultural activities, and more involvement in the care setting. In year two, the associations were maintained for endorsing the importance of early learning and participating in cultural activities.[<reflink idref="bib8" id="ref108">8</reflink>]</p> <p>Table 3. Regression coefficients for associations between participating in maternal education and parenting in the complier groups.</p> <p> <ephtml> <table><thead><tr><td /><td>Year one</td><td>Year two</td></tr><tr><td>Outcomes</td><td><italic>B</italic></td><td><italic>SE</italic></td><td><italic>B</italic></td><td><italic>SE</italic></td></tr></thead><tbody><tr><td>Frequency of cognitively stimulating activities</td><td char=".">−0.08</td><td char=".">0.21</td><td char=".">−0.27</td><td char=".">0.19</td></tr><tr><td>Hours of TV watched</td><td char=".">−1.21</td><td char=".">0.31<xref ref-type="fn" rid="t3fn0004" /></td><td char=".">−0.40</td><td char=".">0.26</td></tr><tr><td>Types of printed media in the home</td><td char=".">0.08</td><td char=".">0.04<xref ref-type="fn" rid="t3fn0001" /></td><td char=".">0.05</td><td char=".">0.03</td></tr><tr><td>Beliefs about the importance of early learning</td><td char=".">0.29</td><td char=".">0.09<xref ref-type="fn" rid="t3fn0003" /></td><td char=".">0.27</td><td char=".">0.10<xref ref-type="fn" rid="t3fn0003" /></td></tr><tr><td>Frequency of participation in cultural activities</td><td char=".">0.10</td><td char=".">0.04<xref ref-type="fn" rid="t3fn0002" /></td><td char=".">0.08</td><td char=".">0.04<xref ref-type="fn" rid="t3fn0002" /></td></tr><tr><td>Involvement in care setting</td><td char=".">0.76</td><td char=".">0.25<xref ref-type="fn" rid="t3fn0003" /></td><td char=".">0.04</td><td char=".">0.19</td></tr><tr><td>Perceived support</td><td char=".">0.14</td><td char=".">0.14</td><td char=".">0.10</td><td char=".">0.13</td></tr></tbody></table> </ephtml> </p> <ulist> <item>9 <emph>Note</emph>. These models control for all covariates.</item> <item>10 <emph>N</emph> = 161 (73 control, 88 treatment).</item> <item>30001 <emph>p</emph> <.10.</item> <item>30002 <emph>p</emph> <.05.</item> <item>30003 <emph>p</emph> <.01.</item> <item>30004 <emph>p</emph> <.001.</item> </ulist> <p>Regression analyses exploring the effects of random assignment to Head Start on parenting were then conducted within each of the strata of mothers predicted to be always and never takers. These analyses help to explore the extent to which the associations of participation in maternal education and parenting may be because of other mechanisms related to Head Start because these mothers have the same educational participation as each other, but differ in random assignment status, whereas the compliers differ in both educational participation and random assignment status. As shown in Table 4, for mothers who participated in education and were predicted to be always takers, random assignment to Head Start was only significantly associated with more frequent engagement in cognitively stimulating activities (at trend), participation in cultural activities (at trend), and greater involvement in the care setting in year one. For mothers who did not participate in education and were predicted to be never takers, random assignment to Head Start was associated with greater frequency of cognitively stimulating activities, fewer hours of TV watched, greater frequency of participation in cultural activities, greater involvement in the care setting, and higher perceived support. None of these effects lasted into year two. Cohen's <emph>d</emph> effect sizes for random assignment to Head Start were compared for mothers who were predicted to be compliers, always takers, and never takers. As shown in Table 5, for most outcomes, the effect sizes were substantially larger for the compliers, and <emph>H<subs>T</subs></emph> tests indicated these were statistically significantly larger in some cases, suggesting the importance of maternal educational participation for parenting.</p> <p>Table 4. Regression coefficients for associations between random assignment to Head Start and parenting in the always takers and never takers groups.</p> <p> <ephtml> <table><thead><tr><td /><td>Year one</td><td>Year two</td></tr><tr><td /><td>Always takers</td><td>Never takers</td><td>Always takers</td><td>Never takers</td></tr><tr><td>Outcomes</td><td><italic>B</italic></td><td><italic>SE</italic></td><td><italic>B</italic></td><td><italic>SE</italic></td><td><italic>B</italic></td><td><italic>SE</italic></td><td><italic>B</italic></td><td><italic>SE</italic></td></tr></thead><tbody><tr><td>Frequency of cognitively stimulating activities</td><td char=".">0.28</td><td char=".">0.15<xref ref-type="fn" rid="t4fn0001" /></td><td char=".">0.30</td><td char=".">0.06<xref ref-type="fn" rid="t4fn0004" /></td><td char=".">−0.10</td><td char=".">0.13</td><td char=".">0.08</td><td char=".">0.06</td></tr><tr><td>Hours of TV watched</td><td char=".">−0.19</td><td char=".">0.16</td><td char=".">−0.24</td><td char=".">0.09<xref ref-type="fn" rid="t4fn0003" /></td><td char=".">−0.08</td><td char=".">0.17</td><td char=".">−0.09</td><td char=".">0.08</td></tr><tr><td>Types of printed media in the home</td><td char=".">0.01</td><td char=".">0.03</td><td char=".">0.04</td><td char=".">0.01<xref ref-type="fn" rid="t4fn0004" /></td><td char=".">−0.01</td><td char=".">0.03</td><td char=".">0.01</td><td char=".">0.01</td></tr><tr><td>Beliefs about the importance of early learning</td><td char=".">−0.03</td><td char=".">0.06</td><td char=".">0.02</td><td char=".">0.03</td><td char=".">0.01</td><td char=".">0.07</td><td char=".">0.03</td><td char=".">0.03</td></tr><tr><td>Frequency of participation in cultural activities</td><td char=".">0.05</td><td char=".">0.03<xref ref-type="fn" rid="t4fn0001" /></td><td char=".">0.04</td><td char=".">0.01<xref ref-type="fn" rid="t4fn0004" /></td><td char=".">0.03</td><td char=".">0.03</td><td char=".">−0.01</td><td char=".">0.01</td></tr><tr><td>Involvement in care setting</td><td char=".">0.40</td><td char=".">0.15<xref ref-type="fn" rid="t4fn0003" /></td><td char=".">0.30</td><td char=".">0.06<xref ref-type="fn" rid="t4fn0004" /></td><td char=".">−0.11</td><td char=".">0.14</td><td char=".">−0.03</td><td char=".">0.05</td></tr><tr><td>Perceived support</td><td char=".">0.10</td><td char=".">0.08</td><td char=".">0.16</td><td char=".">0.04<xref ref-type="fn" rid="t4fn0004" /></td><td char=".">−0.10</td><td char=".">0.09</td><td char=".">0.04</td><td char=".">0.04</td></tr></tbody></table> </ephtml> </p> <ulist> <item>11 <emph>Note</emph>. These models control for all covariates.</item> <item>12 Always takers <emph>N</emph> = 321 (119 control, 202 treatment); Never takers <emph>N</emph> = 1,452 (536 control, 916 treatment).</item> <item>40001 <emph>p</emph> <.10.</item> <item>40002 <emph>p</emph> <.05.</item> <item>40003 <emph>p</emph> <.01.</item> <item>40004 <emph>p</emph> <.001.</item> </ulist> <p>Table 5. Cohen's d effect sizes and HT tests for associations between random assignment to Head Start and parenting, in the compliers, always takers, and never takers groups.</p> <p> <ephtml> <table><thead><tr><td /><td>Year one</td><td>Year two</td></tr><tr><td>Outcomes</td><td>C</td><td>AT</td><td>NT</td><td>χ<sup>2</sup> difference test</td><td>C</td><td>AT</td><td>NT</td><td>χ<sup>2</sup> difference test</td></tr></thead><tbody><tr><td>Frequency of cognitively stimulating activities</td><td char=".">−0.07</td><td char=".">0.24</td><td char=".">−0.28</td><td char=".">− 4.70<xref ref-type="fn" rid="t5fn0001" /></td><td char=".">−0.27</td><td char=".">−0.09</td><td char=".">−0.07</td><td char=".">5.25<xref ref-type="fn" rid="t5fn0001" /></td></tr><tr><td>Hours of TV watched</td><td char=".">−0.67<sup>a,b</sup></td><td char=".">−0.14<sup>a</sup><xref ref-type="fn" rid="t5fn0002" /></td><td char=".">−0.14<sup>b</sup></td><td char=".">−10.35<xref ref-type="fn" rid="t5fn0003" /></td><td char=".">−0.27<xref ref-type="fn" rid="t5fn0002" /></td><td char=".">−0.06<xref ref-type="fn" rid="t5fn0002" /></td><td char=".">−0.06<xref ref-type="fn" rid="t5fn0002" /></td><td char=".">1.77</td></tr><tr><td>Types of printed media in the home</td><td char=".">−0.33</td><td char=".">−0.04<xref ref-type="fn" rid="t5fn0002" /></td><td char=".">−0.20</td><td char=".">− 1.77</td><td char=".">−0.25<xref ref-type="fn" rid="t5fn0002" /></td><td char=".">−0.04<xref ref-type="fn" rid="t5fn0002" /></td><td char=".">−0.05<xref ref-type="fn" rid="t5fn0002" /></td><td char=".">2.21</td></tr><tr><td>Beliefs about the importance of early learning</td><td char=".">−0.61<sup>c,d</sup></td><td char=".">−0.06<sup>c</sup><xref ref-type="fn" rid="t5fn0002" /></td><td char=".">−0.04<sup>d</sup><xref ref-type="fn" rid="t5fn0002" /></td><td char=".">−13.47<xref ref-type="fn" rid="t5fn0004" /></td><td char=".">−0.50<sup>e,f</sup></td><td char=".">−0.02<sup>e</sup><xref ref-type="fn" rid="t5fn0002" /></td><td char=".">−0.06<sup>ff</sup></td><td char=".">7.89<xref ref-type="fn" rid="t5fn0002" /></td></tr><tr><td>Frequency of participation in cultural activities</td><td char=".">−0.46</td><td char=".">−0.23</td><td char=".">−0.19</td><td char=".">− 2.64</td><td char=".">−0.36<sup>g</sup></td><td char=".">−0.12<xref ref-type="fn" rid="t5fn0002" /></td><td char=".">−0.03<sup>g</sup><xref ref-type="fn" rid="t5fn0002" /></td><td char=".">6.48<xref ref-type="fn" rid="t5fn0002" /></td></tr><tr><td>Involvement in care setting</td><td char=".">−0.75</td><td char=".">−0.41</td><td char=".">−0.34</td><td char=".">− 3.32</td><td char=".">−0.04<xref ref-type="fn" rid="t5fn0002" /></td><td char=".">−0.12</td><td char=".">−0.04<xref ref-type="fn" rid="t5fn0002" /></td><td char=".">0.56</td></tr><tr><td>Perceived support</td><td char=".">−0.19</td><td char=".">−0.14</td><td char=".">−0.23</td><td char=".">− 0.44</td><td char=".">−0.13<xref ref-type="fn" rid="t5fn0002" /></td><td char=".">−0.13<xref ref-type="fn" rid="t5fn0002" /></td><td char=".">−0.05<xref ref-type="fn" rid="t5fn0002" /></td><td char=".">2.11</td></tr></tbody></table> </ephtml> </p> <ulist> <item>13 <emph>Notes</emph>. Coefficients with the same superscript are significantly different from one another at <emph>p</emph> <.05. These models control for all covariates.</item> <item>14 C = compliers, AT = always takers, NT = never takers.</item> <item>50001 <emph>p</emph> <.10.</item> <item>50002 <emph>p</emph> <.05.</item> <item>50003 <emph>p</emph> <.01.</item> <item>50004 <emph>p</emph> <.001.</item> </ulist> <hd id="AN0126281570-29">Sensitivity Analyses</hd> <p></p> <hd id="AN0126281570-30">Combined Across Multiple Imputations</hd> <p>The primary results just presented and all the other results presented in this article were conducted in a single imputed data set selected at random because we wanted to be able to compare Cohen's <emph>d</emph> effect sizes for the results across groups of mothers in different predicted strata given the substantial differences in the size of the groups of compliers, always takers, and never takers. To test whether these results were consistent across the four other imputed data sets, the entire principal-score matching procedure was conducted in each data set and estimates were combined using Rubin's rules to account for uncertainty in the data sets (Rubin, [<reflink idref="bib62" id="ref109">62</reflink>]). As shown in Appendix C, the results for the compliers were relatively consistent with the primary results, but the coefficients for involvement in the care setting and for stronger endorsement of beliefs about the importance of early learning were smaller and no longer statistically significant.[<reflink idref="bib9" id="ref110">9</reflink>]</p> <hd id="AN0126281570-31">Benjamini-Hochberg Procedure to Adjust for Multiple Comparisons</hd> <p>We were interested in testing a wider set of parenting outcomes than has been explored in prior research and so we explored a number of parenting outcomes in the year of random assignment and one year later. We applied a Benjamini-Hochberg ([<reflink idref="bib5" id="ref111">5</reflink>]) procedure with a false discovery rate of.10 to adjust our primary results for multiple comparisons. After applying this correction, we found that the result for having more types of printed media in the home was no longer statistically significant.</p> <hd id="AN0126281570-32">Traditional Mediation Analysis Approach</hd> <p>The analytic approach and results for the traditional mediation analyses are presented in Appendix D. Overall, results were consistent with the results of the primary analyses. In year one, participation in maternal education was found to mediate the effects of Head Start on hours of TV watching, types of printed media in the home, and participation in cultural activities, but it was not found to be a statistically significant mediator of the effects of Head Start on the frequency of cognitively stimulating activities and perceived support. In contrast to the primary principal-score analyses, in the mediation analyses participation in maternal education was not significantly associated with involvement in the care setting in year one. Moreover, because there were no significant effects of Head Start on any parenting outcomes in year two (or with beliefs in year one), mediation could not be established, even though educational participation was associated with a number of these parenting practices.</p> <hd id="AN0126281570-33">Discussion</hd> <p>The aim of this study was to use rigorous quasi-experimental analyses to explore whether maternal participation in education influenced a broad set of parenting outcomes that are likely associated with children's later academic development. In the primary principal-scoring analyses, maternal educational participation when children were assigned to Head Start was concurrently associated with children watching fewer hours of TV per day, greater involvement in the care setting, and having more different types of printed media, such as children's books and dictionaries, available in the home. Both concurrently and one year after participation in education, educational participation was related to greater endorsement of beliefs about the importance of early learning and greater frequency of participation in cultural activities, such as visiting museums. We conducted three sets of sensitivity analyses—combining estimates from multiply imputed data to account for uncertainty in the estimates to address the fact that our primary analyses were conducted in a single imputed data set, applying a Benjamini-Hochberg correction to adjust for multiple comparisons, and conducting traditional mediation analyses—and we focus on results that were statistically significant at least at trend level in two or more of the sensitivity analyses—watching fewer hours of TV, having more types of printed media in the home, and more frequently participating in cultural activities in both years.</p> <p>The approach used in the article leveraged random assignment to identify comparable groups of mothers and compare the parenting of mothers in the Head Start treatment group who we assumed participated in education because their children were offered a spot in Head Start to mothers in the control group who we assumed would have participated in education if their children were offered a spot in Head Start (compliers). This means that the influence of maternal educational participation is inseparable from being assigned to Head Start. However, one strength of this method is that it is possible to explore the extent to which other factors related to random assignment to Head Start may be driving the association between maternal educational participation and parenting by conducting analyses across groups of mothers who had the same educational participation as one another but differed in random assignment to Head Start. These analyses suggested that for mothers who we predicted would have participated in education regardless of random assignment (always takers), being assigned to Head Start was not significantly associated with most of the parenting outcomes examined. This may be because these mothers were more advantaged and already engaged in high levels of many of the parenting practices at baseline, so there was little room for these mothers to show improvements in parenting. Alternatively, control always takers may have sought out other high-quality early childhood education programs that offer similar parenting supports to Head Start while they participated in education, meaning the treatment contrast may be smaller for these mothers than for the other mothers. Overall, this seemingly distinct group of mothers engaged in positive parenting behaviors regardless of whether or not they were assigned to Head Start. In contrast, for mothers who we predicted would not participate in education regardless of random assignment (never takers), being offered a spot in Head Start was associated with a number of parenting outcomes in year one. Specifically, assignment to Head Start was related to less TV watching, more types of printed media in the home, greater frequency of participation in cultural activities, more involvement in the care setting, greater frequency of cognitively stimulating activities, and higher perceptions of support. These results are consistent with Head Start's provision of family services to support parents.</p> <p>These results may raise some concern that the observed relations between participation in education and parenting might be because of other mechanisms related to Head Start. Although the primary results do illustrate the influence of participation in education when families were assigned to Head Start, there are a number of indications that maternal educational participation has additional influence over and above Head Start. First, the effects of Head Start likely vary for mothers in different educational participation strata because these mothers are theoretically and descriptively different from one another. Accordingly, the treatment effects of Head Start on parenting that are observed for the never takers would not necessarily be observed for the compliers if they did not participate in education, just as they are not observed for the always takers. Second, for almost all of the parenting outcomes the effect sizes were larger for the compliers than for the always and never takers, and it may be that the smaller sample size of compliers explains the lack of statistical significance of some of the outcomes. Third, the only result that was maintained in year two was for mothers who participated in maternal education, perhaps because the improvements in parenting related to random assignment to Head Start fade once fewer treatment group families are participating in Head Start.[<reflink idref="bib10" id="ref112">10</reflink>] Finally, mediation analyses that included Head Start treatment and participation in maternal education in the same analytic model suggested that a portion of the treatment effect of Head Start on parenting was statistically mediated by educational participation.</p> <p>If we accept that participation in maternal education is associated with parenting, this raises the question of why participation in education is associated with some parenting outcomes and not others. Maternal participation in education was associated with over an hour less TV watching per day. Although some of this decrease may be explained by children spending more time in child care settings while mothers are participating in education, mothers may also place more limits on their children's television viewing as they gain more education. Research suggests there are differences in children's TV environments according to maternal education such that the children of more highly educated mothers are less likely to have a TV in their bedroom and are more likely to be in households that have rules prohibiting eating while watching TV (Hesketh, Ball, Crawford, Campbell, & Salmon, [<reflink idref="bib26" id="ref113">26</reflink>]). Watching fewer hours of TV has been associated with positive cognitive, behavioral, and health outcomes for children (Ozmert, Toyran, & Yurdakok, [<reflink idref="bib55" id="ref114">55</reflink>]; Schmidt & Anderson, [<reflink idref="bib65" id="ref115">65</reflink>]). Participation in maternal education was also found to be associated with having more different types of printed media in the home. Although this may be partly explained by children receiving learning materials in Head Start, this may also be because mothers who are participating in education have materials such as novels, biographies, and nonfiction books in the home to use for their own study. In qualitative research, mothers reported their children's interest in, and imitation of, their studying behaviors, and this role modeling could have later benefits for children's education (Sommer et al., [<reflink idref="bib69" id="ref116">69</reflink>]). Educational participation was also associated with greater frequency of participation in cultural activities, such as visiting museums and the zoo, in years one and two. Mothers who participate in education gain more exposure to cultural activities, which can reinforce their preferences for engaging in such activities with their children (Harding et al., [<reflink idref="bib25" id="ref117">25</reflink>]). Notably, participation in high-culture activities has been associated with teacher-reported academic outcomes for children over and above numerous covariates, including socioeconomic status and ability (Dumais, [<reflink idref="bib16" id="ref118">16</reflink>]; Wildhagen, [<reflink idref="bib77" id="ref119">77</reflink>]). There was some suggestion that maternal education was significantly associated with mothers' beliefs about the importance of early learning and involvement in the care setting, but these results were not consistent across the sensitivity analyses, so future research is needed to further test the relationship between maternal education and these parenting outcomes. These findings lend some support to the idea that participation in maternal education increases mothers' participation in parenting activities related to their human and cultural capital (Harding et al., [<reflink idref="bib25" id="ref120">25</reflink>]), but importantly a number of other parenting outcomes were not consistently related to maternal educational participation.</p> <p>Contrary to expectations, participation in education was not related to greater frequency of engagement in cognitively stimulating activities. This could be because the measure of cognitively stimulating activities asked how often mothers participated in these activities between the hours of 8 a.m. and 6 p.m. Monday through Friday and mothers who participate in education may have educational commitments during the day. Participation in education was also not significantly related to perceived support from friends, professionals, and social groups. Education may provide opportunities to connect with more highly educated others, but may not affect general perceptions of social support. Finally, a number of the effects on parenting observed in year one are no longer statistically significant by year two, although they remain moderately sized and in the same direction. This is consistent with the only other study we are aware of that looked at the associations of maternal educational participation with later parenting, which also found that the coefficients for the longer term outcomes were of similar magnitude to the concurrent effects, but that the effects were no longer statistically significant (Domina & Roksa, [<reflink idref="bib15" id="ref121">15</reflink>]). More research is needed to understand this issue because theories of capital would predict that the effects of maternal educational participation and attainment are long lasting.</p> <hd id="AN0126281570-34">Limitations and Future Research</hd> <p>This research predicted mothers' membership in educational participation strata using principal scores based on an extensive set of covariates in an attempt to explore more causal relations between participation in maternal education and parenting. As described earlier, this method invokes a number of assumptions that need to be weighed. One of the primary assumptions is ignorability—that participation in maternal education is not related to confounds that also predict parenting. This is an untestable assumption and unobserved characteristics of mothers could indeed affect their participation in education and parenting. For example, mothers may change their beliefs about the value of education, which could relate to their decision to participate in education as well as to thinking it is important to have different types of printed materials in the home. Extensive baseline controls were included in analyses to address this issue, to the extent possible. Moreover, the principal-score approach provides some advantages in addressing this issue because we explore parenting for mothers who were predicted to have participated in education because they were randomly assigned to Head Start rather than for those who chose to participate. Because we leverage the randomization of families to Head Start to create groups, we can assume that mothers who are similar to mothers who were induced into education exist (Hill, Waldfogel, & Brooks-Gunn, [<reflink idref="bib29" id="ref122">29</reflink>]). This does require accepting a second assumption—that appropriate comparison mothers have been identified. Although the prediction models had descriptive validity in predicting mothers' educational participation strata membership, uncertainty in the model used to predict participation in education was not taken into account. Thus, although this research extends prior research, replication of these findings in other studies that also employ methods of causal inference would strengthen the conclusions presented here.</p> <p>Another limitation is the measurement of educational participation. The question used to define participation asks whether mothers enrolled or attended any classes at a school, college, or university, so it is unclear what intensity of services mothers participated in. It is possible that some mothers reported attending parenting workshops held at a school as educational participation whereas other mothers may have participated in credit-bearing courses. In addition, increases in mothers' educational credentials (e.g., moving from a high school diploma to gaining an associate's degree) likely have stronger associations with parenting than educational participation does given that gaining credentials requires prolonged participation and can bring benefits through improving employment prospects and income. Unfortunately, the group of mothers who reported increases in maternal education in year one was too small to explore whether gaining educational credentials affects parenting, but future research could explore how increases in mothers' credentials shape parenting.</p> <p>Next, although this study extends prior research by exploring a broader set of parenting outcomes than is typically examined, data limitations precluded a full examination of all theoretically relevant outcomes, such as the educational attainment of mothers' social connections. Moreover, all of the parenting outcomes that are present in the data are self-reported. It is possible that mothers who participate in education receive more encouragement to engage in certain parenting practices and report doing so because of social desirability. However, one might suspect that expectancy bias would affect all of the parenting outcomes and not just some of them, providing at least some indication that this may not be the entire explanation for the observed results. Nonetheless, future research using observed measures of parenting would contribute to understanding more about how mothers' participation in education is related to parenting.</p> <p>Finally, the generalizability of the sample is limited in a number of ways. First, results are from a sample of low-income families who applied for Head Start. Although the low-income nature of the sample is appropriate because this is the group for whom increases in maternal education are more likely and because research suggests increases in maternal education matter most at the low end of the educational distribution (Magnuson, [<reflink idref="bib47" id="ref123">47</reflink>]), the specific population should be considered when interpreting the results. Generalizability is further limited to mothers who reported information on the key predictor: maternal educational participation. Although these mothers were slightly more advantaged than the full sample, we did not impute the key predictor because this would have meant the same variables were used to impute the predictor and then match mothers (Harding, [<reflink idref="bib24" id="ref124">24</reflink>]). Finally, our primary analyses focus on mothers who are expected to participate in education because of access to Head Start. Again, this analytic sample is relevant given that such mothers are likely targets of two-generation programs that aim to use child care centers as a platform for recruiting and engaging parents in education.</p> <p>Despite these limitations, this research uses rigorous techniques to explore the relationship between participation in maternal education and a broad range of parenting outcomes that likely influence children's later development. Methodologically, the techniques employed in this study offer a possible way to gain more knowledge from randomized controlled trials through helping to get inside the "black box" of mechanisms by which randomized treatments influence outcomes. Practically, the demonstrated relationship between participation in maternal education and some parenting outcomes suggests that two-generation programs that aim to increase educational participation could positively influence family functioning.</p> <hd id="AN0126281570-35">Funding</hd> <p>The research reported here was supported by the American Psychological Foundation Elizabeth Munsterberg Koppitz Graduate Student Fellowship and the New York University Predoctoral Interdisciplinary Research Training Fellowship. The opinions expressed are those of the authors.</p> <ref id="AN0126281570-36"> <title> Footnotes </title> <blist> <bibl id="bib1" idref="ref46" type="bt">1</bibl> <bibtext> The small proportion of control families who received Head Start at the center they applied to (5.6%) and the treatment families who did not receive Head Start at the center they applied to (7.4%) were included within their originally assigned groups to retain the intent-to-treat component of the analysis.</bibtext> </blist> <blist> <bibl id="bib2" idref="ref9" type="bt">2</bibl> <bibtext> Analyses were also conducted in the sample with complete information available (<emph>N</emph> = 1,282). Although some of the coefficients were of a similar magnitude to the primary results, most of the results were not statistically significant (see Appendix A in the online supplemental material). This is likely because the sample of predicted compliers was 66% of the size of the imputed sample (<emph>N</emph> = 96–104).</bibtext> </blist> <blist> <bibl id="bib3" idref="ref78" type="bt">3</bibl> <bibtext> This analysis assumes there are no "defiers"—mothers who would not take up education when assigned to Head Start, but who would take up education when assigned to the control group. It is difficult to think of a situation where random assignment would cause defiance systematically, although there are plausible individual cases where this could occur.</bibtext> </blist> <blist> <bibl id="bib4" idref="ref20" type="bt">4</bibl> <bibtext> Although language from instrumental variables (IV) analysis was used to describe the different groups of mothers, IV was not used because this case does not meet the exclusion restriction, which assumes that random assignment to Head Start only affects parenting through maternal education (Moulton, Peck, & Bell, [52]). There are a number of other ways that Head Start could affect parenting given that many Head Start centers offer parents' workshops and opportunities to be involved. Moreover, the bias introduced by not meeting the exclusion restriction is larger when there are low odds that someone is a complier, which is a concern in the present case because the effect of random assignment on participation in maternal education is small (i.e., the first-stage <emph>F</emph> statistic was less than 10). IV analyses with site-by-treatment interactions have been used to identify multiple mediators when the exclusion restriction is implausible, but these models introduce additional strong assumptions (Reardon & Raudenbush, [58]).</bibtext> </blist> <blist> <bibl id="bib5" idref="ref94" type="bt">5</bibl> <bibtext> One criticism of principal-score approaches is that the model that generates the principal scores may provide a better fit among the group on which it was modeled (Hill, Weiss, & Zhai, 2011; Moulton et al., 2014; Peck, 2003). To address this concern, a similar method to the current approach, the Analysis of Symmetrically Predicted Endogenous Subgroups method uses a modeling sample or a cross-validation approach to predict subgroup membership (Moulton et al., 2014; Peck, 2003). We do not use these techniques in the current application because of the complexity of trying to identify multiple subgroups. As such, we use observed educational stratum membership when it is available; we compare observed control always takers to predicted treatment always takers and observed treatment never takers to predicted control never takers. However, membership in our key comparison group—compliers—was predicted for both the treatment and control groups, lessening concerns about mismatch in the accuracy of subgroup prediction.</bibtext> </blist> <blist> <bibl id="bib6" idref="ref24" type="bt">6</bibl> <bibtext> The matching methods in Step 1c (caliper matching, without replacement) and Step 2c (one-to-one matching, with replacement) were selected in order to create proportionally sized groups across the Head Start treatment and control conditions. Other matching and adjustment techniques were used, including inverse probability of treatment weighting in Step 2c, and results were consistent with the primary analyses (results available from the first author). See Stuart (2010) for a description of matching methods.</bibtext> </blist> <blist> <bibl id="bib7" idref="ref51" type="bt">7</bibl> <bibtext> Interestingly, mothers in the Head Start treatment group who had low probabilities of educational participation were more likely to participate in education than similar mothers in the control group. It may be that these are the mothers who are being induced into education because of the opportunities provided by random assignment to Head Start. Alternatively, the prediction model may provide a better fit for the control group given that it was modeled on this group (Peck, [56]). Unfortunately, we are unable to adjudicate between these possibilities, but it seems plausible that mothers with a lower predicted probability of educational participation are those who most need the additional supports provided by Head Start to participate in education.</bibtext> </blist> <blist> <bibl id="bib8" idref="ref40" type="bt">8</bibl> <bibtext> The primary results do not control for baseline parenting because baseline interviews occurred between 1–5 months after random assignment to Head Start and random assignment was significantly associated with many of the "baseline" parenting variables. Additional principal-score matched regression analyses were conducted controlling for the parenting measures available at baseline (frequency of cognitively stimulating activities, types of printed media in the home, frequency of participation in cultural activities, positive parenting, and perceived support). Results for the compliers showed the same pattern of significant results as the primary analyses, except that the result for types of printed media was no longer statistically significant (results available from the first author).</bibtext> </blist> <blist> <bibl id="bib9" idref="ref1" type="bt">9</bibl> <bibtext> The multiply imputed results for the never and always takers groups were almost exactly the same as the single imputation estimates presented in Table 4, so they are not presented (results available from the first author).</bibtext> </blist> <blist> <bibtext> Although treatment families could still participate in Head Start in year two, a far smaller proportion of the sample did participate (66%), and this proportion was more similar to the proportion of control group families who attended Head Start who could also experience the potential benefits to parenting (49%).</bibtext> </blist> <blist> <bibtext> Jessica F. Harding is now at Mathematica Policy Research.</bibtext> </blist> <blist> <bibtext> Supplemental data for this article can be accessed at https://doi.org/10.1080/19345747.2016.1266536.</bibtext> </blist> </ref> <ref id="AN0126281570-37"> <title> References </title> <blist> <bibtext> Angrist, J. D., Imbens, G., & Rubin, D. B. (1996). Identification of causal effects using instrumental variables. Journal of the American Statistical Association, 91(434), 444–455.</bibtext> </blist> <blist> <bibtext> Augustine, J. M. (2014). Maternal education and the unequal significance of family structure for children's early achievement. Social Forces, 93(2), 687–718.</bibtext> </blist> <blist> <bibtext> Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychology research: Conceptual, strategic, and statistical consideration. Journal of Personality and Social Psychology, 51(6), 1173–1182.</bibtext> </blist> <blist> <bibtext> Becker, G. S. (2002). Human capital. In D. R. Henderson (Ed.), The concise encyclopedia of economics. Indianapolis, IN: Liberty Fund. Retrieved from <ulink href="http://www.econlib.org/library/Enc/HumanCapital.html">http://www.econlib.org/library/Enc/HumanCapital.html</ulink></bibtext> </blist> <blist> <bibtext> Benjamini, Y., & Hochberg, Y. (1995). Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society, Series B (Methodological), 57, 289–300.</bibtext> </blist> <blist> <bibtext> Bourdieu, P. (1986). The forms of capital. In J. Richardson (Ed.), Handbook of theory and research for the sociology of education (pp. 241–258). New York, NY: Greenwood.</bibtext> </blist> <blist> <bibtext> Brown, T. (2006). Confirmatory factor analysis for applied research. New York, NY: Guilford.</bibtext> </blist> <blist> <bibtext> Chase-Lansdale, P. L., & Brooks-Gunn, J. (2014). Two-generation programs in the twenty-first century. The Future of Children, 24, 14–39.</bibtext> </blist> <blist> <bibtext> Choi, K. H., Raley, R. K., Muller, C., & Riegle-Crumb, C. (2008). A longitudinal analysis of contextual exposure to peers with college educated parents and students' college enrollment. Social Science Quarterly, 89(4), 846–866. doi:10.1111/j.1540-6237.2008.00587.x</bibtext> </blist> <blist> <bibtext> Cochran, M. M., & Brassard, J. A. (1979). Child development and personal social networks. Child Development, 50, 601–616.</bibtext> </blist> <blist> <bibtext> Coleman, J. S. (1988). Social capital in the creation of human capital. American Journal of Sociology, 94, S95–S120.</bibtext> </blist> <blist> <bibtext> Crosnoe, R., & Kalil, A. (2010). Educational progress and parenting among Mexican immigrant mothers of young children. Journal of Marriage and Family, 72(4), 976–990. doi:10.1111/j.1741-3737.2010.00743.x</bibtext> </blist> <blist> <bibtext> DiStefano, C., Zhu, M., & Mindrila, D. (2009). Understanding and using factor scores: Considerations for the applied researcher. Practical Assessment, Research & Evaluation, 14(20), 1–11.</bibtext> </blist> <blist> <bibtext> Dollaghan, C. A., Campbell, T. F., Paradise, J. L., Feldman, H. M., Jonosky, J. E., Pitcairn, D. N., & Kurs-Lasky, M. (1999). Maternal education and measures of early speech and language. Journal of Speech, Language, and Hearing Research, 42, 1432–1443.</bibtext> </blist> <blist> <bibtext> Domina, T., & Roksa, J. (2012). Should Mom go back to school? Post-natal educational attainment and parenting practices. Social Science Research, 41(3), 695–708. doi:10.1016/j.ssresearch.2011.12.002</bibtext> </blist> <blist> <bibtext> Dumais, S. A. (2006). Early childhood cultural capital, parental habitus, and teachers' perceptions. Poetics, 34(2), 83–107. doi:10.1016/j.poetic.2005.09.003</bibtext> </blist> <blist> <bibtext> Dunn, L. M., & Dunn, L. M. (1997). Peabody Picture Vocabulary Test–Third Edition. Circle Pines, MN: American Guidance Service.</bibtext> </blist> <blist> <bibtext> Estabrook, R., & Neale, M. (2013). A comparison of factor score estimation methods in the presence of missing data: Reliability and an application to nicotine dependence. Mulitvariate Behavioral Research, 48(1), 1–27.</bibtext> </blist> <blist> <bibtext> Fan, X., & Chen, M. (2001). Parental involvement and students' academic achievement: A meta-analysis. Educational Psychology Review, 13(1), 1–22.</bibtext> </blist> <blist> <bibtext> Friedman-Krauss, A. H., Connors, M. C., & Morris, P. A. (2016). Unpacking the treatment contrast in the Head Start Impact Study: To what extent does assignment to treatment affect quality of care? Journal of Research on Educational Effectiveness, 10, 68–95. doi:10.1080/19345747.2016.1147627</bibtext> </blist> <blist> <bibtext> Gelman, A., Hill, J., Su, Y. S., Yajima, M., & Pittau, M. G. (2014). Missing data imputation and model checking: R Software package.</bibtext> </blist> <blist> <bibtext> Gennetian, L. A., Magnuson, K., & Morris, P. A. (2008). From statistical associations to causation: What developmentalists can learn from instrumental variables techniques coupled with experimental data. Developmental Psychology, 44(2), 381–394.</bibtext> </blist> <blist> <bibtext> Greenberg, D., Meyer, R. H., & Wiseman, M. (1993). Prying the lid from the black box: Plotting evaluation strategy for welfare employment and training programs (Discussion Paper No. 999-93). Madison, WI: University of Wisconsin–Madison, Institute for Research on Poverty.</bibtext> </blist> <blist> <bibtext> Harding, J. F. (2015). Increases in maternal education and children's cognitive and behavioral outcomes. Developmental Psychology, 51(5), 583–599.</bibtext> </blist> <blist> <bibtext> Harding, J. F., Morris, P. A., & Hughes, D. (2015). The relationship between maternal education and children's academic outcomes: A theoretical framework. Journal of Marriage and Family, 77, 60–76. doi:10.1111/jomf.12156</bibtext> </blist> <blist> <bibtext> Hesketh, K., Ball, K., Crawford, D., Campbell, K., & Salmon, J. (2007). Mediators of the relationship between maternal education and children's TV viewing. American Journal of Preventive Medicine, 33(1), 41–47.</bibtext> </blist> <blist> <bibtext> Hicks, R., & Tingley, D. (2012). MEDIATION: Stata module for causal mediation analysis and sensitivity [Computer software]. Retrieved from <ulink href="http://econpapers.repec.org/software/bocbocode/s457294.htm">http://econpapers.repec.org/software/bocbocode/s457294.htm</ulink></bibtext> </blist> <blist> <bibtext> Hill, J., Brooks-Gunn, J., & Waldfogel, J. (2003). Sustained effects of high participation in an early intervention for low-birth-weight premature infants. Developmental Psychology, 39(4), 730–744.</bibtext> </blist> <blist> <bibtext> Hill, J., Waldfogel, J., & Brooks-Gunn, J. (2002). Differential effects of high-quality child care. Journal of Policy Analysis and Management, 21(4), 601–627. doi:10.1002/pam.10077</bibtext> </blist> <blist> <bibtext> Hill, J., Waldfogel, J., Brooks-Gunn, J., & Han, W. J. (2005). Maternal employment and child development: A fresh look using newer methods. Developmental Psychology, 41, 833–850.</bibtext> </blist> <blist> <bibtext> Hill, J., Weiss, C., & Zhai, F. (2011). Challenges with propensity score strategies in a high-dimensional setting and a potential alternative. Multivariate Behavioral Research, 46(3), 477–513. doi:10.1080/00273171.2011.570161</bibtext> </blist> <blist> <bibtext> Holmlund, H., Lindahl, M., & Plug, E. (2011). The causal effect of parents' schooling on children's schooling: A comparison of estimation methods. Journal of Economic Literature, 49(3), 615–651.</bibtext> </blist> <blist> <bibtext> Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6, 1–55.</bibtext> </blist> <blist> <bibtext> Imai, K., Keele, L., & Tingley, D. (2010). A general approach to causal mediation analysis. Psychological Methods, 15(4), 309–334. doi:10.1037/a0020761</bibtext> </blist> <blist> <bibtext> Imai, K., Keele, L., Tingley, D., & Yamamoto, T. (2011). Unpacking the black box of causality: Learning about causal mechanisms from experimental and observational studies. American Political Science Review, 105(4), 765–789. doi:10.1017/s0003055411000414</bibtext> </blist> <blist> <bibtext> Jo, B. (2008). Causal inference in randomized experiments with mediational processes. Psychological Methods, 13(4), 314–336.</bibtext> </blist> <blist> <bibtext> Jo, B., & Stuart, E. A. (2009). On the use of propensity scores in principal causal effect estimation. Statistics in Medicine, 28(23), 2857–2875.</bibtext> </blist> <blist> <bibtext> Jo, B., Stuart, E. A., MacKinnon, D. P., & Vinokur, A. D. (2011). The use of propensity scores in mediation analysis. Multivariate Behavioral Research, 46(3), 425–452. doi:10.1080/00273171.2011.576624</bibtext> </blist> <blist> <bibtext> Kalil, A., Ryan, R., & Corey, M. (2012). Diverging destinies: Maternal education and the developmental gradient in time with children. Demography, 49(4), 1361–1383.</bibtext> </blist> <blist> <bibtext> Kaufman, A. S. (1994). Kaufman Functional Academic Skills Test (K-FAST). Circle Pines, MN: American Guidance Service.</bibtext> </blist> <blist> <bibtext> Kemple, J. J., Snipes, J. C., & Bloom, H. (2001). A regression-based strategy for defining subgroups in a social experiment. New York, NY: MDRC.</bibtext> </blist> <blist> <bibtext> Kingston, P. W., Hubbard, R., Lapp, B., Schroeder, P., & Wilson, J. (2003). Why education matters. Sociology of Education, 76(1), 53–70.</bibtext> </blist> <blist> <bibtext> Lareau, A. (2011). Unequal childhoods: Class race, and family life. Berkeley: University of California Press.</bibtext> </blist> <blist> <bibtext> LeVine, R. A., LeVine, S., & Schnell, B. (2001). "Improve the Women": Mass schooling, female literacy, and worldwide social change. Harvard Educational Review, 71(1), 1–51.</bibtext> </blist> <blist> <bibtext> LeVine, R. A., LeVine, S., Schnell-Anzola, B., Rowe, M. L., & Dexter, E. (2012). Literacy and mothering: How women's schooling changes the lives of the world's children. Oxford, England: Oxford University Press.</bibtext> </blist> <blist> <bibtext> Lin, N. (2000). Inequality in social capital. Contemporary Sociology, 29(6), 785–795.</bibtext> </blist> <blist> <bibtext> Magnuson, K. (2007). Maternal education and children's academic achievement during middle childhood. Developmental Psychology, 43(6), 1497–1512. doi:10.1037/0012-1649.43.6.1497</bibtext> </blist> <blist> <bibtext> Magnuson, K., Sexton, H. R., Davis-Kean, P. E., & Huston, A. C. (2009). Increases in maternal education and young children's language skills. Merrill-Palmer Quarterly, 55(3), 319–350.</bibtext> </blist> <blist> <bibtext> McCormick, M. P., Cappella, E., O'Connor, E., Hill, J., & McClowry, S. G. (2016). Do effects of social-emotional learning programs vary by level of parent participation? Evidence from the randomized trial of INSIGHTS. Journal of Research on Educational Effectiveness, 9, 364–394. doi:10.1080/19345747.2015.1105892</bibtext> </blist> <blist> <bibtext> McPherson, M., Smith-Lovin, L., & Cook, J. M. (2001). Birds of a feather: Homophily in social networks. Annual Review of Sociology, 27, 415–444.</bibtext> </blist> <blist> <bibtext> Mirowsky, J., & Ross, C. E. (2003). Education, social status and health. New York, NY: Aldine de Gruyter.</bibtext> </blist> <blist> <bibtext> Moulton, S., Peck, L. R., & Bell, S. H. (2014). Social Policy Impact Pathfinder (SPI-Path) analytic suite: SPI-Path|Individual user guide version 2. Bethesda, MD: Abt Associates Inc.</bibtext> </blist> <blist> <bibtext> Muthen, L. K., & Muthen, B. O. (1998–2011). MPlus user's guide (6th ed.). Los Angeles, CA: Author.</bibtext> </blist> <blist> <bibtext> Office of Head Start. (n.d.). Using the Head Start parent, family and community engagement framework in your program: Markers of progress. Retrieved from <ulink href="http://eclkc.ohs.acf.hhs.gov/hslc/tta-system/family/docs/ncpfce-markers-of-progress.pdf">http://eclkc.ohs.acf.hhs.gov/hslc/tta-system/family/docs/ncpfce-markers-of-progress.pdf</ulink></bibtext> </blist> <blist> <bibtext> Ozmert, E., Toyran, M., & Yurdakok, K. (2002). Behavioral correlates of television viewing in primary school children evaluated by the Child Behavior Checklist. Archives of Pediatric & Adolescent Medicine, 156, 910–914.</bibtext> </blist> <blist> <bibtext> Peck, L. R. (2003). Subgroup analysis in social experiments: Measuring program impacts based on post-treatment choice. The American Journal of Evaluation, 24(2), 157–187. doi:10.1016/s1098-2140(03)00031-6</bibtext> </blist> <blist> <bibtext> Perna, L. W., Fester, R., & Walsh, E. (2010). Exploring the college enrollment of parents: A descriptive analysis. Journal of Student Financial Aid, 40(1), 6–16.</bibtext> </blist> <blist> <bibtext> Reardon, S. F., & Raudenbush, S. W. (2013). Under what assumptions do site-by-treatment instruments identify average causal effects? Sociological Methods and Research, 42, 143–163.</bibtext> </blist> <blist> <bibtext> Rhemtulla, M., Brosseau-Liard, P. E., & Savalei, V. (2012). When can categorical variables be treated as continuous? A comparison of robust continuous and categorical SEM estimation methods under suboptimal conditions. Psychological Methods, 17(3), 354–373.</bibtext> </blist> <blist> <bibtext> Roid, G., & Miller, L. (1997). Leiter-R AM battery. Wood Dale, IL: Stoelting.</bibtext> </blist> <blist> <bibtext> Rubin, D. B. (1974). Estimating causal effects of treatments in randomized and non-randomized studies. Journal of Educational Psychology, 66, 688–701.</bibtext> </blist> <blist> <bibtext> Rubin, D. B. (1987). Multiple imputation for nonresponse in surveys. New York, NY: John Wiley & Sons.</bibtext> </blist> <blist> <bibtext> Rubin, D. B., & Thomas, N. (2000). Combining propensity score matching with additional adjustments for prognostic covariates. Journal of the American Statistical Association, 95(450), 573–585.</bibtext> </blist> <blist> <bibtext> Sabol, T. J., & Chase-Lansdale, P. L. (2015). The influence of low-income children's participation in Head Start on their parents' education and employment. Journal of Policy and Management, 34(1), 136–161.</bibtext> </blist> <blist> <bibtext> Schmidt, M. E., & Anderson, D. R. (2007). The impact of television on cognitive development and educational achievement. In N. Pecora, J. P. Murray & E. A. Wartella (Eds.), Children and television: Fifty years of research (pp. 65–84). Mahwah, NJ: Lawrence Erlbaum Associates.</bibtext> </blist> <blist> <bibtext> Schochet, P. Z., & Burghardt, J. (2007). Using propensity scoring to estimate program-related subgroup impacts in experimental program evaluations. Evaluation Review, 31(2), 95–120.</bibtext> </blist> <blist> <bibtext> Seligman, M. E. P. (1993). What you can change ... and what you can't. New York, NY: Ballantine Books.</bibtext> </blist> <blist> <bibtext> Sirin, S. R. (2005). Socioeconomic status and academic achievement: A meta-analytic review of research. Review of Educational Research, 75(3), 417–453. doi:10.3102/00346543075003417</bibtext> </blist> <blist> <bibtext> Sommer, T. E., Chase-Lansdale, P. L., Brooks-Gunn, J., Gardner, M., Rauner, D. M., & Freel, K. (2012). Early childhood education centers and mothers' postsecondary attainment: A new conceptual framework for a dual-generation education intervention. Teachers College Record, 114, 1–40.</bibtext> </blist> <blist> <bibtext> St. Pierre, R. G., Ricciuti, A. E., & Rimdzius, T. A. (2005). Effects of a family literacy program on low-literate children and their parents: Findings from an evaluation of the Even Start family literacy program. Developmental Psychology, 41(6), 953–970. doi:10.1037/0012-1649.41.6.953</bibtext> </blist> <blist> <bibtext> Stuart, E. A. (2010). Matching methods for causal inference: A review and a look forward. Statistical Science, 25(1), 1–21.</bibtext> </blist> <blist> <bibtext> Stuart, E. A., & Jo, B. (2011). Assessing the sensitivity of methods for estimating principal causal effects. Statistical Methods in Medical Research, 24(6), 657–674.</bibtext> </blist> <blist> <bibtext> Suizzo, M., & Stapleton, L. M. (2007). Home-based parental involvement in young children's education: Examining the effects of maternal education across U.S. ethnic groups. Educational Psychology, 27(4), 533–556. doi:10.1080/01443410601159936</bibtext> </blist> <blist> <bibtext> Torr, J. (2004). Talking about picture books: The influence of maternal education on four-year-old children's talk with mothers and pre-school teachers. Journal of Early Childhood Literacy, 4(2), 181–210. doi:10.1177/1468798404044515</bibtext> </blist> <blist> <bibtext> U.S. Department of Health and Human Services. (2010a). Head Start Impact Study: Final report. Washington, DC: U.S. Department of Health and Human Services, Administration for Children and Families.</bibtext> </blist> <blist> <bibtext> U.S. Department of Health and Human Services. (2010b). Head Start Impact Study: Technical report. Washington, DC: U.S. Department of Health and Human Services, Administration for Children and Families.</bibtext> </blist> <blist> <bibtext> Wildhagen, T. (2009). Why does cultural capital matter for high school academic performance? An empirical assessment of teacher-selection and self-selection mechanisms as explanations of the cultural capital effect. The Sociological Quarterly, 50, 173–200.</bibtext> </blist> <blist> <bibtext> Woodcock, R. W., McGrew, K., & Mather, N. (2001). Woodcock-Johnson tests of cognitive abilities and tests of achievement (3rd ed.). Rolling Meadows, IL: Riverside.</bibtext> </blist> </ref> <aug> <p>By Jessica F. Harding; Pamela A. Morris and Jennifer Hill</p> <p>Reported by Author; Author; Author</p> </aug> <nolink nlid="nl1" bibid="bib14" firstref="ref2"></nolink> <nolink nlid="nl2" bibid="bib68" firstref="ref3"></nolink> <nolink nlid="nl3" bibid="bib22" firstref="ref4"></nolink> <nolink nlid="nl4" bibid="bib25" firstref="ref5"></nolink> <nolink nlid="nl5" bibid="bib47" firstref="ref6"></nolink> <nolink nlid="nl6" bibid="bib48" firstref="ref7"></nolink> <nolink nlid="nl7" bibid="bib32" firstref="ref8"></nolink> <nolink nlid="nl8" bibid="bib73" firstref="ref11"></nolink> <nolink nlid="nl9" bibid="bib39" firstref="ref12"></nolink> <nolink nlid="nl10" bibid="bib57" firstref="ref13"></nolink> <nolink nlid="nl11" bibid="bib51" firstref="ref14"></nolink> <nolink nlid="nl12" bibid="bib75" firstref="ref15"></nolink> <nolink nlid="nl13" bibid="bib64" firstref="ref16"></nolink> <nolink nlid="nl14" bibid="bib42" firstref="ref18"></nolink> <nolink nlid="nl15" bibid="bib44" firstref="ref19"></nolink> <nolink nlid="nl16" bibid="bib11" firstref="ref21"></nolink> <nolink nlid="nl17" bibid="bib74" firstref="ref23"></nolink> <nolink nlid="nl18" bibid="bib45" firstref="ref25"></nolink> <nolink nlid="nl19" bibid="bib43" firstref="ref26"></nolink> <nolink nlid="nl20" bibid="bib46" firstref="ref28"></nolink> <nolink nlid="nl21" bibid="bib50" firstref="ref29"></nolink> <nolink nlid="nl22" bibid="bib10" firstref="ref30"></nolink> <nolink nlid="nl23" bibid="bib12" firstref="ref33"></nolink> <nolink nlid="nl24" bibid="bib15" firstref="ref34"></nolink> <nolink nlid="nl25" bibid="bib70" firstref="ref35"></nolink> <nolink nlid="nl26" bibid="bib69" firstref="ref36"></nolink> <nolink nlid="nl27" bibid="bib54" firstref="ref37"></nolink> <nolink nlid="nl28" bibid="bib20" firstref="ref38"></nolink> <nolink nlid="nl29" bibid="bib28" firstref="ref41"></nolink> <nolink nlid="nl30" bibid="bib37" firstref="ref42"></nolink> <nolink nlid="nl31" bibid="bib38" firstref="ref43"></nolink> <nolink nlid="nl32" bibid="bib72" firstref="ref44"></nolink> <nolink nlid="nl33" bibid="bib76" firstref="ref47"></nolink> <nolink nlid="nl34" bibid="bib24" firstref="ref48"></nolink> <nolink nlid="nl35" bibid="bib30" firstref="ref49"></nolink> <nolink nlid="nl36" bibid="bib53" firstref="ref50"></nolink> <nolink nlid="nl37" bibid="bib59" firstref="ref52"></nolink> <nolink nlid="nl38" bibid="bib18" firstref="ref53"></nolink> <nolink nlid="nl39" bibid="bib209" firstref="ref55"></nolink> <nolink nlid="nl40" bibid="bib33" firstref="ref56"></nolink> <nolink nlid="nl41" bibid="bib13" firstref="ref57"></nolink> <nolink nlid="nl42" bibid="bib40" firstref="ref59"></nolink> <nolink nlid="nl43" bibid="bib67" firstref="ref60"></nolink> <nolink nlid="nl44" bibid="bib60" firstref="ref61"></nolink> <nolink nlid="nl45" bibid="bib78" firstref="ref62"></nolink> <nolink nlid="nl46" bibid="bib17" firstref="ref63"></nolink> <nolink nlid="nl47" bibid="bib21" firstref="ref64"></nolink> <nolink nlid="nl48" bibid="bib62" firstref="ref66"></nolink> <nolink nlid="nl49" bibid="bib61" firstref="ref67"></nolink> <nolink nlid="nl50" bibid="bib36" firstref="ref86"></nolink> <nolink nlid="nl51" bibid="bib52" firstref="ref88"></nolink> <nolink nlid="nl52" bibid="bib66" firstref="ref89"></nolink> <nolink nlid="nl53" bibid="bib63" firstref="ref96"></nolink> <nolink nlid="nl54" bibid="bib23" firstref="ref98"></nolink> <nolink nlid="nl55" bibid="bib27" firstref="ref101"></nolink> <nolink nlid="nl56" bibid="bib34" firstref="ref102"></nolink> <nolink nlid="nl57" bibid="bib35" firstref="ref103"></nolink> <nolink nlid="nl58" bibid="bib49" firstref="ref105"></nolink> <nolink nlid="nl59" bibid="bib41" firstref="ref106"></nolink> <nolink nlid="nl60" bibid="bib26" firstref="ref113"></nolink> <nolink nlid="nl61" bibid="bib55" firstref="ref114"></nolink> <nolink nlid="nl62" bibid="bib65" firstref="ref115"></nolink> <nolink nlid="nl63" bibid="bib16" firstref="ref118"></nolink> <nolink nlid="nl64" bibid="bib77" firstref="ref119"></nolink> <nolink nlid="nl65" bibid="bib29" firstref="ref122"></nolink>
Header DbId: eric
DbLabel: ERIC
An: EJ1159786
AccessLevel: 3
PubType: Academic Journal
PubTypeId: academicJournal
PreciseRelevancyScore: 0
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: Understanding Associations between Low-Income Mothers' Participation in Education and Parenting
– Name: Language
  Label: Language
  Group: Lang
  Data: English
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AR" term="%22Harding%2C+Jessica+F%2E%22">Harding, Jessica F.</searchLink><br /><searchLink fieldCode="AR" term="%22Morris%2C+Pamela+A%2E%22">Morris, Pamela A.</searchLink><br /><searchLink fieldCode="AR" term="%22Hill%2C+Jennifer%22">Hill, Jennifer</searchLink>
– Name: TitleSource
  Label: Source
  Group: Src
  Data: <searchLink fieldCode="SO" term="%22Journal+of+Research+on+Educational+Effectiveness%22"><i>Journal of Research on Educational Effectiveness</i></searchLink>. 2017 10(4):704-731.
– Name: Avail
  Label: Availability
  Group: Avail
  Data: Routledge. Available from: Taylor & Francis, Ltd. 530 Walnut Street Suite 850, Philadelphia, PA 19106. Tel: 800-354-1420; Tel: 215-625-8900; Fax: 215-207-0050; Web site: http://www.tandf.co.uk/journals
– Name: PeerReviewed
  Label: Peer Reviewed
  Group: SrcInfo
  Data: Y
– Name: Pages
  Label: Page Count
  Group: Src
  Data: 28
– Name: DatePubCY
  Label: Publication Date
  Group: Date
  Data: 2017
– Name: TypeDocument
  Label: Document Type
  Group: TypDoc
  Data: Journal Articles<br />Reports - Research
– Name: Audience
  Label: Education Level
  Group: Audnce
  Data: <searchLink fieldCode="EL" term="%22Preschool+Education%22">Preschool Education</searchLink>
– Name: Subject
  Label: Descriptors
  Group: Su
  Data: <searchLink fieldCode="DE" term="%22Correlation%22">Correlation</searchLink><br /><searchLink fieldCode="DE" term="%22Low+Income%22">Low Income</searchLink><br /><searchLink fieldCode="DE" term="%22Mothers%22">Mothers</searchLink><br /><searchLink fieldCode="DE" term="%22Scores%22">Scores</searchLink><br /><searchLink fieldCode="DE" term="%22Parenting+Styles%22">Parenting Styles</searchLink><br /><searchLink fieldCode="DE" term="%22Educational+Attainment%22">Educational Attainment</searchLink><br /><searchLink fieldCode="DE" term="%22Mother+Attitudes%22">Mother Attitudes</searchLink><br /><searchLink fieldCode="DE" term="%22Prediction%22">Prediction</searchLink><br /><searchLink fieldCode="DE" term="%22Disadvantaged+Youth%22">Disadvantaged Youth</searchLink><br /><searchLink fieldCode="DE" term="%22Preschool+Education%22">Preschool Education</searchLink><br /><searchLink fieldCode="DE" term="%22Preschool+Children%22">Preschool Children</searchLink><br /><searchLink fieldCode="DE" term="%22Child+Development%22">Child Development</searchLink><br /><searchLink fieldCode="DE" term="%22Statistical+Analysis%22">Statistical Analysis</searchLink><br /><searchLink fieldCode="DE" term="%22Family+Characteristics%22">Family Characteristics</searchLink><br /><searchLink fieldCode="DE" term="%22Cultural+Activities%22">Cultural Activities</searchLink><br /><searchLink fieldCode="DE" term="%22Reading+Materials%22">Reading Materials</searchLink><br /><searchLink fieldCode="DE" term="%22Family+Environment%22">Family Environment</searchLink><br /><searchLink fieldCode="DE" term="%22Randomized+Controlled+Trials%22">Randomized Controlled Trials</searchLink><br /><searchLink fieldCode="DE" term="%22Regression+%28Statistics%29%22">Regression (Statistics)</searchLink>
– Name: DOI
  Label: DOI
  Group: ID
  Data: 10.1080/19345747.2016.1266536
– Name: ISSN
  Label: ISSN
  Group: ISSN
  Data: 1934-5747
– Name: Abstract
  Label: Abstract
  Group: Ab
  Data: Maternal education is one of the strongest predictors of children's academic outcomes. One possible explanation for this is that more highly educated mothers more frequently engage in parenting practices that may promote children's later cognitive development; however, most of this evidence is correlational. This study uses Head Start Impact Study data (N = 1,953) to explore whether low-income mothers' participation in education affects their parenting practices and beliefs. Principal scores, which predict maternal educational participation based on covariates, were used for analysis. Principal score matching was used to identify mothers who we predicted participated in education because their children were randomly assigned to Head Start. We compared these mothers' outcomes to those of mothers we predicted would have participated in education if they were assigned to Head Start. For these mothers, participation in maternal education was associated with children watching fewer hours of TV, having more types of printed media at home, and more frequent participation in cultural activities, but it was not associated with a host of other parenting outcomes. Changing parenting is one potential pathway by which maternal educational participation may influence children's later academic outcomes.
– Name: AbstractInfo
  Label: Abstractor
  Group: Ab
  Data: As Provided
– Name: Ref
  Label: Number of References
  Group: RefInfo
  Data: 86
– Name: DateEntry
  Label: Entry Date
  Group: Date
  Data: 2017
– Name: AN
  Label: Accession Number
  Group: ID
  Data: EJ1159786
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=eric&AN=EJ1159786
RecordInfo BibRecord:
  BibEntity:
    Identifiers:
      – Type: doi
        Value: 10.1080/19345747.2016.1266536
    Languages:
      – Text: English
    PhysicalDescription:
      Pagination:
        PageCount: 28
        StartPage: 704
    Subjects:
      – SubjectFull: Correlation
        Type: general
      – SubjectFull: Low Income
        Type: general
      – SubjectFull: Mothers
        Type: general
      – SubjectFull: Scores
        Type: general
      – SubjectFull: Parenting Styles
        Type: general
      – SubjectFull: Educational Attainment
        Type: general
      – SubjectFull: Mother Attitudes
        Type: general
      – SubjectFull: Prediction
        Type: general
      – SubjectFull: Disadvantaged Youth
        Type: general
      – SubjectFull: Preschool Education
        Type: general
      – SubjectFull: Preschool Children
        Type: general
      – SubjectFull: Child Development
        Type: general
      – SubjectFull: Statistical Analysis
        Type: general
      – SubjectFull: Family Characteristics
        Type: general
      – SubjectFull: Cultural Activities
        Type: general
      – SubjectFull: Reading Materials
        Type: general
      – SubjectFull: Family Environment
        Type: general
      – SubjectFull: Randomized Controlled Trials
        Type: general
      – SubjectFull: Regression (Statistics)
        Type: general
    Titles:
      – TitleFull: Understanding Associations between Low-Income Mothers' Participation in Education and Parenting
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Harding, Jessica F.
      – PersonEntity:
          Name:
            NameFull: Morris, Pamela A.
      – PersonEntity:
          Name:
            NameFull: Hill, Jennifer
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 01
              M: 01
              Type: published
              Y: 2017
          Identifiers:
            – Type: issn-print
              Value: 1934-5747
          Numbering:
            – Type: volume
              Value: 10
            – Type: issue
              Value: 4
          Titles:
            – TitleFull: Journal of Research on Educational Effectiveness
              Type: main
ResultId 1