Read and select out the salient points and forward it to the class on the discussion board using a power-point of 8-10 slides.

Read and select out the salient points and forward it to the class on the discussion board using a power-point of 8-10 slides. If slides can’t be provided, just give facts about it and I will revert it to a powerpoint.

1
Keeping Promises: Single Mothers, Race, and Elementary Educational Engagement
Yiwan Ye
Larissa Saco
University of California, Davis
September 2019
WP19-11-FF
*Paper prepared for presentation at the American Sociological Association’s annual
meeting, August 2019. Direct all correspondence to Yiwan Ye, Office 286, Department
of Sociology, Social Sciences and Humanities Building, UC Davis, 1 Shields Avenue,
Davis, CA, 95616 (ywye@ucdavis.edu). This paper is supported by the Department of
Sociology’s Travel Grant and Graduate Student Association Travel Award of UC Davis
to the first author.
Keeping Promises: Single Mothers, Race, and Elementary Educational Engagement
2
The present study explores how household arrangement influences parental
engagement in children’s elementary education among mothers in U.S. urban settings.
Using two waves of panel data from the Fragile Families and Child Wellbeing Study (N =
2,982), the present paper compares the difference in educational engagement between
coupled (married or cohabiting) and single mothers. Logistic regression models are
utilized to examine the impacts of household arrangement on the possibility of enrolling
children in tutoring, initiating a conversation with teachers, and frequent book reading
with children. After controlling for household structure, financial factors, and mother and
child characteristics, results suggest that, compared with mothers who live with partners,
single mothers who consistently live alone at waves 4 and 5 are just as likely to hire a tutor
but less likely to initiate conversations with teachers. The results also suggest no
differences in after-school tutoring enrollment and frequent book reading across
household arrangements. This paper also discusses some racial disparities found for
parental engagement outcomes. Black mothers are more likely to hire tutors and, like
single mothers in general, are less likely to initiate discussions with teachers compared to
their White counterparts. Hispanic mothers read with their children less frequently than
non-Hispanic mothers, which could possibly be explained by the lesser availability of
children’s literature written in the Spanish language compared with the English language.
The findings of this paper have important implications for understanding both the
engagement strategies employed and obstacles faced by single mothers in urban areas,
and suggest new hypotheses for future study of racial gaps in parental engagement in
children’s education.
Key Words: cultural capital, educational engagement, Fragile Families and Child
Wellbeing Study, living arrangement, race, single motherhood, tutoring
3
BACKGROUND
Numerous studies have provided empirical evidence suggesting that family
background is a structural force for children’s educational achievement (Quinn, 2015;
Raudenbush & Eschmann, 2015; Reardon, 2011). Among family background
characteristics, household and relational factors such as marital status, residential
arrangements, and partnership status are strongly associated with the social and cognitive
wellbeing of mothers and children (McLanahan, 2009; McLanahan et al., 2010; Teitler &
Reichman, 2008). Recent research featuring Fragile Families data examines the
relationship between marital status and child outcomes, suggesting that a mother’s change
in marital status or partners has a statistically significant negative impact on their children’s
wellbeing such as school readiness (Cooper et al., 2011; McLanahan, Tach, & Schneider,
2014).
Studies have also found that there are no substantial differences in various family
outcomes among certain living arrangements, such as cohabitation and living with
children’s grandparents.
1 After controlling for socioeconomic factors, studies report no
significant variation in children’s or the mothers’ educational success or mental wellbeing
between unmarried and married couples (Brown et. al., 2015; McLanahan et al., 2010;
Shapiro & Keyes, 2007). Some studies do suggest that single-parent families have
substantial positive impacts on children’s prospects for success (Edin & Kefalas, 2005;
Musick & Meier, 2010). However, especially among single-parent mothers who
experience family disruptions – divorce, separation, and widowing – the single-parent
1 Hence, the present paper does not compare married and unmarried couples.
4
family structure has also been found to be associated with a negative effect on children’s
educational achievement, while other family transitions – marriage and living with
romantic partners – have been found to yield a positive effect (Amato & Keith, 1991;
Wagmiller et. al., 2010).
The present study focuses on children between 5 and 9 years old (4th and 5th
waves) – an age range where many children are enrolled in elementary school. We did not
include the 6th wave when children reached the age of 15 because elementary children are
arguably more susceptible to family structural factors and changes in household
arrangement than middle school children who become more independent in their social
development. The effect of household arrangements on education for elementary schoolers
and middle schoolers are likely to be nonlinear and can be examined independently,
because the biological, cognitive, and social development for these two age and education
groups are qualitatively different.
According to social science literature, the educational disadvantages of children
from single-parent families can in large part be attributed to low financial and childcare
resources available. Empirical studies have provided evidence suggesting that a two-parent
family structure is beneficial for child development, household resources, and overall
mental health (McLanahan, 2009; McLanahan et. al., 2010). Some research has shown that
when single mothers become married, this family change can promote children’s
educational success due to a boost in financial security from the respective partner (Amato
et al., 2007). However, prior studies examining children’s educational outcomes neglect
some direct mechanisms that link parental relationship status and children’s academic
achievement – in other words, parental engagement in children’s educational affairs.
5
Therefore, it is important to investigate these specific parental processes that may influence
children’s academic achievement.
Early childhood educational development plays an increasingly important role in
preparing children for future academic achievement, improving the likelihood of social
mobility, and closing racial achievement gaps, especially among socioeconomically
disadvantaged children (Apple, 2015; Condron, 2009; Downey et al., 2004; Heckman,
2006). Living arrangements outside of two-parent households have been reported to be
associated with young, low socioeconomic status, and minority adults (Pew Research
Center 2014). Single motherhood is one of the weakest risk factors of being poor compared
to risk factors such as low education, unemployment, and young headship (Brady,
Finnigan, & Hubgen, 2017). Given that children from single-parent households have been
reported to face some educational disadvantages due to financial constraints, this study
investigates the question of whether single-parent families are more or less likely to be
involved and invest in their children’s education compared with two-parent households,
after controlling for relevant factors.2 The Fragile Families and Child Wellbeing Study is
a suitable dataset for this study because it focuses on unmarried couples who are at risk of
separating and whose children are vulnerable to living in poverty.
Hypotheses
The study explores both the time-deficit framework and cultural factors to
hypothesize maternal engagement differences between the two living arrangements of
single- and two-parent households. The time deficit framework hypothesizes that single
mothers are less likely to engage or invest in children’s educational affairs because they
2 The variables reflecting relevant factors are based on Cooper and colleagues (2011) and
McLanahan and colleagues (2010).
6
may have less time and financial resources available to allocate towards this end compared
with mothers who share some of the time and financial resources with their partners. This
framework is derived from time-spending literatures which suggest that single parent
households that are also characterized as working poor have inadequate discretionary time
to engage in household production such as child care and education (Kalenkoski et al.,
2011; Vickery, 1997; Zacharias, 2011). Time-deficit literatures argue that disadvantaged
families experience scarce finances and time for extensive child care activities due to
prolonged working hours and barriers from low standards of living (Kalenkoski et. al.,
2011; Mullainathan & Shafir, 2013). This framework may also apply to single mothers in
urban settings who are in precarious financial and time-constrained situations.
Alternatively, cultural factors illuminate the possibility that single mothers have
perspectives and make decisions concerning their children’s education that are different
from those of mothers who have a partner. According to Usdansky & McLanahan (2003),
being a single and college-educated mother is associated with holding more independent
views about marriage, education, and work. In Promises I Can Keep (2005), Edin and
Kefalas find that poor single mothers who desire marriage show commitment to their
children’s education and prioritize children’s wellbeing above their marriage. Resting on
these empirical findings, we hypothesize that mothers who consistently live alone may
experience motivation from self-reliant and hardworking ethics to be active in parental
engagement in order for their children to have educational opportunities comparable to
children of non-single mothers who may benefit from more time and financial resources.
Hypotheses
7
In order to test the efficacy of the time-deficit framework and empirical
explanations of cultural factors, we test the following hypotheses to explore the potential
impact of living arrangement on parental engagement:
1. The likelihood of single mothers letting their children participate in after-school tutoring
and math lab is different from that of mothers with partners.
2. The likelihood of single mothers initiating discussion with the teacher to talk about
children’ academic and behavioral problems is different from that of mothers who have
partners.
3. The likelihood of single mothers reading books to their children more than once a week
is different from that of mothers who have partners.
For our study, we control for mothers’ socioeconomic status, educational
attainment, demographics, as well as children’s prior educational achievement and basic
demographics. We control for child’s gender because a recent study on residential
transition and children’s academic performance suggests a significant educational disparity
between children of different genders (Cooper et al., 2011). As mentioned above, some
research has suggested that a residential transition can have significant impact on children’s
educational outcomes, whether that be to support or impede children’s school readiness.
Controlling for other material conditions, we test whether single mothers invest additional
effort to improve their children’s academic achievement, potentially motivated by a drive
to overcome material disadvantages associated with single parenthood.
DATA
8
All data are derived from the Fragile Families and Child Wellbeing Study (FFCWS)
with an original cohort of 4,898 families. This study randomly selected 16 large U.S. cities
from 20 total and surveyed both non-marital births and marital births, with an intentional
oversampling of “fragile families” or unmarried parents and their children. In the first
wave, mothers were surveyed from 1998 to 2000 shortly after giving birth. The follow-up
surveys were conducted in 1, 3, 5 and 9 years after this first birth. The latest of these waves
ends in 2009, when most children turned age 9 to 10, and about 3,515 couples remain in
the sample after 5 waves of surveys. Since FFCWS concentrates on “at-risk” parents in
urban settings who cohabited or were separated when giving birth to their first child,
FFCWS purposely under-samples traditional households (24% of the families in the
original sample) characterized by mothers with post-conception in marital unions or those
who were married to their children’s fathers when entering the survey.3
Analytic Sample
Our analysis focuses on the in-home surveys of mothers and the focal child in the
first five waves of FFCWS. In year 9, FFCWS collects the child’s academic, behavioral,
and school information in the in-home surveys, which also include information on
parenting and parents’ educational engagement. Our sampling frame applies to mothers
who participated in the in-home study at wave 5, and thus it excludes 943 mothers who did
not participate in the in-home study and 667 mothers who exited the FFCWS before wave
5. In addition, among the 3,288 mothers who participated in the in-home study at wave 5,
we excluded 231 mothers who skipped important questions that were included in our
measurements of interest. A total of 2,982 mothers are in the final analytic sample.
3 Even though the under-sampling substantially limits external validity for generalizing findings, this issue
may be resolved after applying constructed national sampling weight (Reichman et al., 2001).
9
MEASUREMENT
Dependent Variables
The maternal educational engagement variables measure how active the mother is
in engaging with and investing in the child’s academic activities that were included in the
survey. Our dependent variables measure possibilities of the mother engaging in activities
that facilitate children’s school readiness and academic achievement at wave 5, such as (1)
children’s participation in after-school academic activities, (2) initiating discussion with
teachers about children’s academic and behavioral performance, and (3) reading books to
children more than once a week. These three binary variables measure three domains of
maternal educational engagement: third-party educational investment, parent-teacher
interaction, and parent-child interaction.
The first dependent variable – tutoring – measures the possibility that children have
participated in individual tutoring or math lab (p5i2e). Although the survey asked both
parents about who made the decisions for their child to participate in these activities, the
study focuses on mothers based on reports that mothers on average exhibit greater
educational involvement and child attachment than fathers (Coyl-Shepherd & Newland,
2013). The second variable – initiation – measures the possibility that the mother takes the
initiative to discuss behavioral and academic problems with the child’s teachers (p5l18).4
Initiation is included to test whether mothers are proactive in communicating with their
4 In the variable initiation, “1” denotes mothers starting discussions with teachers, “0” denotes other family
members or teachers starting discussions on the child’s problems at school.
10
children’s instructors.5 The third dependent variable – reading – measures whether mothers
read books or talked about books with their children more than once per week in the past
month (p5ile). Mothers who read books to their children several times or more per week
are assigned a value of “1”, and mothers who read books to their children less than twice
per week are assigned a value of “0.”
Independent Variables
Living arrangement can encompass the mother’s residential status with the focal
child’s biological father, the number of the focal child’s siblings, and the presence of the
child’s grandparents in the household. Therefore, our independent variables include the
following to account for mothers’ family structures: (1) the mother living alone versus
living with a partner (husband or unmarried biological father), (2) having any partnership
or marital transition, (3) the focal child’s siblings, and (4) living with the child’s
grandparent(s). Given the study’s focus on parental engagement, the mother’s residential
status with the biological father is of primary interest in the present paper. With this
variable, we look at whether mothers consistently lived alone (without any romantic
partner) from wave 4 to wave 5 (year 5 to year 9).6 Since this variable captures at least
four years of living alone, we can isolate the effect of single motherhood based on the
assumption that partner separation took place before the survey of mother’s educational
engagement, and we can expect measurable observance of any effect that living alone may
5 FFCWS includes additional variables that measure whether the mother simply discusses academic
problems or behavioral problems with teachers. However, these two variables do not measure who
proactively initiates the discussion and therefore we do not include these variables in our analysis.
6 Mothers who only report living alone at wave 5 are not statistically different from mothers who live with
partners in any of the models.
11
have on engagement given the length of separation. In our analysis sample, approximately
46% or exactly 1,428 mothers lived without a romantic partner from year 5 to year 9.
The relationship transition variable measures whether there is any change in marital
and residential status between the mother and the child’s biological father. In other words,
the transition variable assigns a “1” if the mother got married, divorced, became widowed,
changed romantic partners, started cohabiting, or separated from the child’s biological
fathers since the baseline year. In our sample, 46% of mothers experience at least one type
of partnership or relationship transition in the past 9 years. The siblings variable measures
whether the focal child has a maternal sibling; in other words, if mothers have more than
one biological child since the baseline year. About 62% of mothers have more than one
biological child after the baseline year, and 32% of mothers have exactly 2 children in the
past 9 years. The grandparents variable measures whether at least one of the child’s
grandparents (grandmother, grandfather, or both) live with mothers since the baseline year.
Approximately 40% of mothers live with at least one of the child’s grandparents in the past
9 years, and most of them are living with grandmothers (36%).
Control Variables
Our first set of control variables accounts for mother’s characteristics that are
hypothesized to be correlated with mother’s educational engagement, such as mother’s
race, educational attainment, income, employment status, and experience of material
hardship.7 The mother’s race and ethnicity are grouped into four categories – White nonHispanic, Black non-Hispanic, Hispanic, and other races. The other race category includes
Asian, American Islander, and other races, which comprises 4% of the analytic sample.
7 Mother’s age at wave 5 is automatically omitted in all logistic models due to no association with the
dependent variables, and thus we dropped this variable from the analysis.
12
The White, Black, and Hispanic races comprise approximately 21%, 48%, and 27% of the
sample respectively. The racial disparity in maternal education engagement remains
prominent even after controlling for class and other socioeconomic status variables.
Therefore, we included in the analysis additional socioeconomic and cultural variables to
reduce any omitted variable bias and conduct sensitivity tests on the significance of racial
disparity.
We included mother’s educational attainment level at year 9, which constitutes
“less than high school” (34%), “high school” (26.6%), “some college/technical school”
(25.6%), and “college or graduate school” (13.8%). The income variable is a constructed
continuous variable in FFCWS that measures mother’s household income in thousands of
dollars in year 9, with a median income of $34,000. Some of the missing income values
are imputed by FFCWS. Employment status is a binary variable that measures whether
mothers have been consistently employed from year 5 to year 9, and about 43% of mothers
were consistently employed in those five years. Similar to the living alone variable, the
duration of employment variable should be long enough so that it could exert potential
influence on maternal engagement. The material hardship variable measures whether
mothers experience any material hardship since the baseline year. This hardship variable
and the income variable are not redundant, because the former reflects the actual financial
condition and standard of living, while the income variable does not illustrate the cost of
living. Material hardship is a binary variable constructed from a series of questions such
as “being hungry because couldn’t afford enough food,” “did not pay the full amount of
rent/mortgage,” “gas/electric service turned off,” and “did not see a doctor because of
13
cost.” Approximately 90% of mothers experienced one of these material hardships since
the birth of the focal child.
The second set of control variables account for the child’s characteristics that might
be confounding the relationship between household structure and maternal educational
engagement, such as children’s age, gender, repeating grades, and Woodcock Johnson
vocabulary test scores. Children’s age is measured in years and months, with a median age
of 9.2, and about 90% of children are around 9 years-old. The percent of female children
is about 47%. Repeating grade is a binary variable that indicates whether children repeat
any grades from kindergarten up until wave 5. The Woodcock Johnson test score is an
interval ratio variable that measures children’s standardized vocabulary capacity and
reading comprehension skills, ranging from 0 to 14 with a median score of 9.9.
METHOD
To test our hypotheses, we used four nested logistic regression models with robust
standard errors, which include a simple bivariate logistic regression model, incrementally
followed by a model with household characteristics, a model controlling for mother’s
characteristics, and a full model that controls for children’s characteristics. We ran four
nested models for each dependent variable. Our main findings will only include the full
models, and the less constrained models are included in the appendix. We conducted the
regression analysis for the three dependent variables separately. In addition, we used robust
regression models and reported the odds ratios and t-statistics for each model, using the
reference group of mothers who live with partners.
Logistic Regression Models
14
Our models examine the effect of mother’s living arrangement, especially single
motherhood, on each of the maternal education engagement measures – tutoring, talking
to teachers, and reading books to children. The following is the formula for the full model:
Where i: individual mother in the sample; p(engage): probability of mother
engaging in child’s educational affairs; b: slope coefficient of a variable; b0: slope
coefficient of the reference group; w: slope coefficients of a set of variables; Alone:
binary variable indicating single motherhood for the last 4 years; Mother_Demo:
mother’s demographic variables – age, race and education; Mother_Econ:
mother’s/household economic variables – income, employment status, and material
hardship; Child_Demo: child’s demographic variable – gender, age, repeating grade,
vocabulary test scores; e: a robust error term.
RESULTS8
Table 2 reports the odds ratios and standard errors for key predictors and control
variables that show significant association with maternal engagement outcomes. The table
omitted mother’s age, income, employment status, material hardship, child’s gender and
age, repeating grade, and reference group. The reference group is mothers who are White,
with an educational attainment level of less than high school completion, and consistently
married or cohabiting with no grandparents in the house. In Model 1, the results show no
difference between the proportions of single mothers and married/cohabiting mothers for
enrolling their children in “academic activities like tutoring or math lab” after controlling
for everything else. In other words, children from single mother households are just as
8 Table 1 for descriptive statistics will be included in a future version of this paper.
15
likely to participate in after-school tutoring or math lab as children from two-parent
families.
Model 1 also shows that mothers with more than some college education are more
likely to have children enrolled in tutoring or math lab than mothers with an educational
attainment level less than high school (1.35 times for mothers with some college education
and 1.39 times for college or graduate school educated mothers). This result is in line with
cultural capital studies and theory (Brown, 1974), suggesting college-educated mothers
pass on knowledge and skills useful for navigating the education system to their children.
We also found that children of Black mothers are on average 1.87 times more likely to
enroll in after-school academic activities compared with their White counterparts holding
all else constant (p-value < 0.001). This racial disparity requires future study, and the
potential mechanism for this racial gap will be discussed. The Woodcock Johnson test of
vocabulary has a negative significant association with the possibility of tutoring/math lab
participation. The model suggests that for every 10-point increase in the vocabulary test,
the probability of enrolling child in tutoring/math lab is reduced by 9% holding all else
constant, suggesting that children who are academically “gifted” or prepared are in less
need of tutoring services.
Model 2 suggests that single mothers are approximately 22% less likely to make
contact with a teacher to discuss the child’s academic or behavioral issues compared with
mothers living with partners holding all else constant (p-value < 0.01). This result may
support the time-deficit hypothesis that single mothers who are consistently employed may
have less time and energy to initiate discussion with the teacher. As for the control
variables, mothers with more than one child are also 14% less likely to start a conversation
16
with teachers compared with mothers with only one child after controlling for everything
else. This finding is a further evidence for the time-deficit hypothesis where mothers are
less likely to initiate conversation if they have to distribute child-caring time among
multiple children. In addition, college-educated mothers are 1.8 times more likely to initiate
conversation about the child’s academic affairs with teachers than mothers with less than
a high-school education.
Model 3 shows that the probability of reading books to children more than once a
week for single mothers is the same as that of married/cohabiting couples holding all else
constant. The model also indicates no statistical difference in the probability of reading
books across educational level and number of children in the household. However,
Hispanic mothers are approximately 32% less likely than non-Hispanic White mothers to
read books to their children more than once a week. One mechanism that may help explain
this variation is the unavailability of children’s books written in the Spanish language
compared with those written in the English language. After controlling for whether mothers
used Spanish language questionnaires, the statistical significance of the coefficient for
Hispanic mothers disappears.
DISCUSSION
In general, the results from the full models suggest that single mothers who live
alone are just as likely to personally engage in their child’s education as cohabiting and
married mothers. The study also reaffirms that there is no significant difference in parental
education engagement between cohabiting mothers and married mothers. Cultural capital
theory may help explain what discourages Black mothers from engagement with teachers
17
while encouraging educational investments in after-school activities. School-initiated
stigma around race and single motherhood may intersect to further discourage parentinitiated interactions with teachers. The racial disparity in child tutoring between Black and
White mothers could be explained by the following hypotheses that warrant future study.
Black mothers: are more likely to enroll their children in schools that offer after-school
tutoring or math lab; are systematically recruited by tutoring firms and school programs;
utilize tutoring services to help their children have more equal educational opportunities to
those of White children; have families or communities with more activated social capital
or higher levels of attachment between the mother and children; and are clustered in certain
occupations that require long work hours and therefore child enrollment in after-school
activities.
The race effect on mother-initiated communication with children’s teachers is
hypothesized to reflect some influence from demographic mismatch between teachers and
parents. A potential explanation for the disparity in parent-initiated discussion with
teachers between single mothers and mothers with partners is that school-initiated
stigmatization surrounding single motherhood and a lack of dominant cultural capital
discourage single mothers from initiating conversation with the child’s teacher. Therefore,
Black mothers and single mothers may prefer after-school program enrollment and reading
at home as their engagement strategies rather than teacher communication. The similar
levels of engagement overall between single mothers and mothers in two-parent
households demonstrate that single mothers are just as active in children’s education as
their partnered counterparts.
18
Another potential piece of the puzzle for explaining similar engagement levels is
targeted-selection, which hypothesizes that single mothers and other disadvantaged
families are specifically targeted by the government, schools, and private programs that
aim to assist socioeconomically vulnerable families. Because of the resources that become
available from these institutions, families are more likely to take advantage of services such
as children’s tutoring or after-school math lab. Unfortunately, this level of information
about tutoring or after-school programs is not identified in the FFCWS. This framework
takes an optimistic stand on the effectiveness of funded programs for increasing child
enrollment in academic enrichment activities, and compliments the time-deficit framework
wherein single mothers have less time to allocate to parental educational engagement due
to work obligations and thereby choose to enroll their children in private tutoring or math
lab because they have the resources to do so.
The present study has several limitations. Some of the variables may have internal
validity issues. For example, the variable for after-school tutoring and math lab does not
specify who made the decision for the child to participate in the activity. The models also
do not account for interaction effects of living arrangement and race. Single motherhood
may have a multiplicative negative effect on Black mothers compared to the effect of Black
married mothers or the effect of non-Black single mothers. The small sample size for the
reference group reduces coefficient significance for the main effects when running
interaction analysis. The small sample size may diminish the potential moderating effect
of mother’s race in the analysis of living arrangement’s impact on maternal engagement.
A tremendous research and policy spotlight is currently on how schools influence
children, but the connection between household structure and children’s educational
19
success deserves more attention from researchers and policymakers. Therefore, a study like
this is one endeavor to increase understanding about the evidence and policy implications
for after-school and family factors that shape children’s academic progress. For instance,
some general policy implications include the possibility of schools removing barriers and
implementing evidence-based strategies to encourage inclusive teacher-parent interactions,
schools and local libraries could include more books in multiple languages such as Spanish,
and educational institutions could provide free English language programs for Spanishspeaking mothers.
20
REFERENCES
Apple, M. W. (2015). “Reframing the Question of Whether Education Can Change
Society.” Educational Theory. 65(3): 299-315.
Amato, P., Booth, A., Johnson, D. R., & Rogers, S. J. (2007). Alone Together: How
Marriage in America is Changing. Harvard University Press. MA: Cambridge.
Amato, P.R., & B. Keith. (1991). “Parental Divorce and Adult Well-being: A Meta-analysis.”
Journal of Marriage and Family. 53(1): 43-58.
Brady, D., Finnigan, R.M., and S. Hubgen. (2017). “Rethinking the Risks of Poverty: A
Framework for Analyzing Prevalences and Penalties.” American Journal of Sociology.
123(3): 740-786.
Brown, S. L., Manning, W. D., & Payne, K. K. (2015). Relationship Quality Among Cohabiting
Versus Married Couples. Journal of Family Issues. doi:10.1177/0192513×15622236
Brown, R. (1974). “Knowledge, Education and Cultural Change: Papers in Sociology of
Education.” Social Science Paperbacks.
Condron, D.J. (2009). “Social Class, School, and Non-school Environments, and Black/White
Inequalities in Children’s Learning.” American Sociological Review. 74(5): 685-708.
Cooper, C.E., Osborne, C.A., Beck, A.N., and McLanahan, S.S. 2011. “Partnership Instability,
School Readiness, and Gender Disparities.” Sociology of Education. 84(3): 246-259.
Coyl-Shepherd, D. D., & Newland, L. A. (2013). Mothers’ and fathers’ couple and family
contextual influences, parent involvement, and school-age child attachment. Early Child
Development and Care, 183(3-4), 553-569. doi:10.1080/03004430.2012.711599
Downey, D., von Hippel, P.T., & B.A. Broh. (2004). “Are Schools the Great Equalizer? Cognitive
Inequality during the Summer Months and the School Year.” American Sociological
21
Review. 69(5): 613-635.
Edin, K. & Kefalas, M. (2005). “Promises I Can Keep: Why Poor Women Put Motherhood
before Marriage.” University of California Press, Berkeley, California.
Gibson-Davis, C.M. 2009. “Money, Marriage, and Children: Testing the Financial
Expectations and Family Formation Theory.” Journal of Marriage and Family.
71(1): 146-160.
Heckman, J.J. 2006. “Skill Formation and the Economics of Investing in Disadvantaged
Children.” Science. 312: 1900-1902.
Kalenkoski, C. Hamrick, K., and Andrews, M. (2011). Time Poverty Thresholds and Rates
for the US Population. Social Indicators Research. Vol. 104. Issue 1, pp. 129-155.
Quinn, D.M. 2015. “Kindergarten Black-White Test Score Gaps: Re-examining the Roles
of Socioeconomic Status and School Quality with New Data.”
Pew Research Center. (2014). “Appendix C: Young Adult Living Arrangements and
Household Incomes”. Web.
<http://www.pewsocialtrends.org/2014/02/11/appendix-c-young-adult-livingarrangements-and-household-incomes/>
Raudenbush, S.W., & R.D. Eschmann. 2015. “Does Schooling Increase or Reduce Social
Inequality.” Annual Review of Sociology. 41:443-470.
Reichman N.E., Teitler, J.O., Garfinkel, I., & S.S. McLanahan 2001. “Fragile Families:
sample and design.” Children and Youth Services Review. 23(4-5): 303-326.
Shapiro, A., & Keyes, C. L. M. (2007). Marital Status and Social Well-Being: Are the
Married Always Better Off? Social Indicators Research, 88(2), 329-346.
doi:10.1007/s11205-007-9194-3
22
Teitler, J., & Reichman, N. (2008). “Mental Illness as a Barrier to Marriage Among
Unmarried Mothers.” Journal of Marriage and Family. 70: 772-782.
McLanahan, S. (2009). “Fragile Families and the Reproduction of Poverty.” The Annals of
the American Academy of Political and Social Science. 621(1): 111-131.
McLanahan, S., Garfinkel, I., Mincy, R., & Donahue, E. (2010). “Introducing the Issue.”
The Future of Children. 20 (2).
McLanahan, S., Tach, L., & D. Schneider. 2014. “The Causal Effects of Father Absence.”
Annual Review of Sociology. 39: 399-427.
Mullainathan, Sendhil. and Shafir, Eldar. (2013). Scarcity: Why Having Too Little Means
So Much. Times Books.
Musick, K., & A. Meier. (2010). “Are Both Parents Always Better Than One? Parental
Conflict And Young Adult Well-Being.” Social Science Research. 39(5): 814-830.
Reardon, S.F. (2011). “The Widening Income Achievement Gap.” Faces of Poverty. 70(8):
10-16
Usdansky, M. L. & McLanahan, S. (2003). “Looking for Murphy Brown: Are CollegeEducated, Single Mothers Unique?” Center for Research on Child Wellbeing.
Working Paper # 03-05-FF.
Vickery, Clair. (1977). The Time-Poor: A New Look At Poverty. The Journal of Human
Resources. XII. 1.
Wagmiller, R. L. Jr., Gershoff, E., Veliz, P., & Clements, M. (2010). “Does Children’s
Academic Achievement Improve when Single Mothers Marry?” Sociology of
Education. 82(2), 201-226.
23
Zacharias, Ajit. (2011). The Measurement of Time and Income Poverty. Working Paper
No.690. Levy Economics Institute of Bard College. New York.
24
Table 1 [please email the first author for descriptive statistics]
Table 2: Odds Ratio of Logistic Regression Models Predicting Maternal Education
Engagement using FFCWS 2005-2009 (N = 2,982)
(1) Tutor or
Math Lab
(2) Initiation
Discussion
(3) Read books >
than Once a Week
Living alone since year 5 1.130
(1.41)
0.779**
(-2.96)
1.064
(0.70)
Residential transition 1.054
(0.64)
0.973
(-0.35)
1.037
(0.43)
Grandparents in the
house1
0.965
(-0.42)
0.974
(-0.31)
0.940
(-0.70)
Have more than one
child 1.058
(0.64)
0.756**
(-3.13)
0.845
(-1.82)
Black 1.868***
(5.40)
0.759*
(-2.40)
1.060
(0.50)
Hispanic 1.192
(1.38)
0.836
(-1.42)
0.680**
(-3.10)
Other Race 1.194
(0.74)
0.676
(-1.64)
0.736
(-1.29)
High school 1.028
(0.27)
1.091
(0.88)
1.000
(-0.00)
Some college 1.349**
(2.77)
1.279*
(2.33)
1.209
(1.72)
College or Grad 1.393*
(2.14)
1.833***
(3.79)
1.344
(1.82)
Woodcock Johnson Test 0.910***
(-3.58)
1.031
(1.20)
0.983
(-0.60)
BIC 3989.9 4035.3 3827.0
t statistics in parentheses
* p<0.05 ** p<0.01 *** p<0.001″
Variables omitted in this table are mother’s age, income (2), employment status (3),
material hardship (4), child’s gender and age, (5) repeating grade, and reference
group. The Reference group is White, less than high school educated, consistently
married/cohabiting, mothers with more than one child and no grandparents in the
house.
1. At least one grandparent lived with mother in any wave since baseline.
2. Mother’s household income is imputed and in thousand of dollars.
3. Stay employed since year 5.
4. Mother experienced any material hardship since baseline.
5. Children repeat any grade since baseline.
25
Table 3: Odds Ratio Results of Logistic Regression Models Predicting Children’s
Enrollment in Tutoring and Math Lab using FFCWS 2005-2009 (N = 2,982)
Binary
Household
Structure
Mother’s
Characteristics Full Model
Living alone since year 5 1.233**
(2.74)
1.242**
(2.79)
1.116
(1.30)
1.130
(1.42)
Marital/residential
transition
1.071
(0.89)
1.066
(0.80)
1.054
(0.65)
Grandparents in the house 1.016
(0.19)
0.971
(-0.35) 0.965 (-0.42)
Have more than one child 1.085
(1.02)
1.077
(0.84)
1.058
(0.64)
Mother’s Characteristics
Age at year 9 0.989
(-1.39)
0.990
(-1.29)
Black 1.961***
(5.87)
1.799***
(5.15)
Hispanic 1.218
(1.54)
1.192
(1.36)
Other Race 1.200
(0.76)
1.194
(0.74)
High school 1.021
(0.21)
1.028
(0.27)
Some College 1.279*
(2.31)
1.349**
(2.78)
College or Grad 1.305
(1.74)
1.393*
(2.14)
Household
income2
1.001
(1.22)
1.002
(1.92)
Employed3 0.873
(-1.67)
0.905
(-1.21)
Material
hardship4 1.060
(0.43)
1.056
(0.39)
Children’s
Characteristics
Female 1.105
(1.27)
Age in months 0.973*
(-1.96)
Children repeat
grades5
1.139
(1.20)
Woodcock
Johnson Test 9
0.910***
(-3.63)
BIC 3937.8 3960.0 3980.3 3989.9
t statistics in parentheses
26
* p<0.05 ** p<0.01 *** p<0.001″
27
Table 4: Odds Ratio Results of Logistic Regression Models Predicting Mother-Initiated Discussion with School
Teachers using FFCWS 2005-2009 (N = 2,982)
Binary Household Structure
Mother’s
Characteristics Full Model
Living alone since year 5
0.690***
(-4.90)
0.686***
(-4.87)
0.781**
(-2.96)
0.779**
(-2.94)
Any partner/residential
transition
0.892
(-1.49)
0.978
(-0.28)
0.973
(-0.34)
Grandparents in the
house1
0.874
(-1.69)
0.974
(-0.32)
0.974
(-0.31)
Have more than one child 0.706*** 0.754** 0.756**
(-4.31) (-3.18) (-3.15)
Mother’s Characteristics
Age at year 9 1.008
(1.04)
1.008
(1.00)
Black 0.746** 0.759*
(-2.58) (-2.40)
Hispanic 0.828 0.836
(-1.51) (-1.43)
Other Race 0.672 0.676
(-1.69) (-1.65)
High school 1.094 1.091
(0.91) (0.87)
Some College 1.289* 1.279*
(2.41) (2.32)
College or Grad 1.859*** 1.833***
(3.82) (3.72)
Household
income2
1.000
(0.42)
1.000
(0.29)
Employed3 1.008 0.997
(0.09) (-0.03)
Material
hardship4
0.766
(-1.85)
0.761
(-1.89)
Children’s Characteristics
Female 1.094
(1.16)
Age in months 1.004
(0.31)
Children repeat
grades5
1.048
(0.44)
Woodcock Johnson Test 9
1.031
(1.19)
t statistics in parentheses
* p<0.05 ** p<0.01 *** p<0.001
28
Table 5 [please email the first author for the odds ratios results for the initiating
conversation w/ teachers].
Figures: they will added in future revisions or in publication.