Page 1 of 18
Advances in Social Sciences Research Journal – Vol. 9, No. 9
Publication Date: September 25, 2022
DOI:10.14738/assrj.99.13030. Zheng, H., Elliott, W., & O’Brien, M. (2022). The Importance of Parent/Child Communications About Children’s Savings Accounts for
Developing a College-Bound Identity. Advances in Social Sciences Research Journal, 9(9). 173-190.
Services for Science and Education – United Kingdom
The Importance of Parent/Child Communications About
Children’s Savings Accounts for Developing a College-Bound
Identity
Haotian Zheng
University of Michigan School of Social Work
1080 S University Ave, Ann Arbor, MI 48109 USA
William Elliott
University of Michigan School of Social Work
1080 S University Ave, Ann Arbor, MI 48109 USA
Megan O’Brien
University of Michigan School of Social Work
1080 S University Ave, Ann Arbor, MI 48109 USA
ABSTRACT
Children’s Savings Accounts (CSAs) not only can help children and families to pay
for expenses related to postsecondary education but also produce positive social
and psychological effects, such as children’s college-bound identity. In this sense,
CSAs provide a rich context for parent and child to communicate about college and
ultimately for children to co-construct a college-bound identity with parents. This
study examines the relationship between parent/child communication about
participation in CSA and children’s college-bound identity. Using data from 506
parent/child dyads participating in the Promise Scholars CSA, we examine if
parent/child communications about their CSA are associated with a stronger
college-bound identity, an important predictor of future academic success.
Propensity score weighting is performed to minimize the impact of selection bias.
Multiple imputations are performed to handle missing data. We find evidence of
higher college-bound identity scores among communicator dyads compared to non- communicator dyads and the comparison group. We also find that the difference in
college-bound identity scores is not statistically significant between non- communicator dyads and the comparison group. Given this finding, CSA programs
may consider interventions that may facilitate parent/child communications about
their CSA as a way of encouraging children to form a college-bound identity. Further,
we posit that parent/child communications about CSA are another important way
for programs to understand engagement as most programs aim to increase college
attendance. Until recently, CSA programs have often narrowly understood
engagement as whether families make contributions or not.
Key words: Promise Scholars, Children’s Savings Accounts, Assets, College-Bound
Identity, Parent/Child Communication
Page 2 of 18
174
Advances in Social Sciences Research Journal (ASSRJ) Vol. 9, Issue 9, September-2022
Services for Science and Education – United Kingdom
Highlights:
• Students of the communicator group have stronger college-bound identity than students
of the comparison group as well as the non-communicator group.
• Students of the non-communicator group did not have statistically different college- bound identity scores than children in the comparison group.
Highlights:
• The college-bound identity of communicator group student is higher than the non- communicator group and the comparison group.
• The college-bound identity of students in the non-communicator group is not different
than children in the comparison group.
INTRODUCTION
In their simplest form, Children’s Savings Accounts (CSAs), sometimes called Child
Development Accounts (CDAs) are asset-building accounts for children. In their current state,
these accounts usually have the goal of helping children pay for expenses related to
postsecondary education, but they can be used for broader asset building activities (e.g., saving
for a home, retirement, or a business). In addition to providing families with an account for
making their own contributions, they augment families’ efforts by providing them with an
initial deposit, incentives, and matching funds. According to Prosperity Now [1], by the end of
2020, there were 109 active CSA programs serving about 922,000 children in more than 36
states. This was a 30% increase from the end of 2019. Typically, CSA programs start at birth or
when children enter kindergarten. While family contributions are important within CSA
programs, it is not the only, or even main mechanism that CSAs provide for building assets.
When a child has an account, the scaffolding is put in place to augment the saving efforts of low- income families through government, philanthropy, and other sources. Some programs are
experimenting with targeted ongoing deposits as a means of building assets, some universities
are matching family savings up to a specified amount, while some foundations are converting
traditional scholarships into early award scholarships.
By augmenting low-income families’ contributions with initial deposits, matches, incentives, as
well as the opportunity for third party contributions, CSAs mimic early educational transfers
wealthy families provide to their children. These early investments in a child’s educational
trajectory are crucial for their outcomes as adults. Research shows that predicted household
income and net worth are higher for adults who received parental financial support for college
as children than for those receiving no such support [2]. Because CSAs are often designed
specifically with the goal in mind of providing low-income children financial support for college,
we posit, like early educational transfers of wealthy families, CSAs have the potential for
increasing income and wealth in adulthood among children born into low-income households.
When families make contributions to their child’s CSA, they are engaging with its wealth
building capacity. However, as Sherraden [3] notes, CSAs also can create social and
psychological effects. When Sherraden [3] discusses CSAs’ ability to create social and
psychological effects, he primarily talks about this from an institutional perspective; how
institutions shape individuals and society. But like with building assets, the institution not only
provides the structure needed for asset accumulation (e.g., initial deposits, matches, the
account itself), it also provides the opportunity for families to engage in building assets through
Page 3 of 18
175
Zheng, H., Elliott, W., & O’Brien, M. (2022). The Importance of Parent/Child Communications About Children’s Savings Accounts for Developing a
College-Bound Identity. Advances in Social Sciences Research Journal, 9(9). 173-190.
URL: http://dx.doi.org/10.14738/assrj.99.13030
their own individual contributions. This is what most human behavior looks like, institutions
augment what people can do (or provide the bounds on what they can do). At the same time,
you cannot fully understand the impact of institutions on behavior without accounting for
differing levels of effort and ability by individuals. And so, in as much as CSAs create social and
psychological effects, it also means at the same time they provide individuals with the
opportunity to engage in producing social and psychological effects above what the institution
can create on its own.
Acknowledging that both institutions and individuals play a key role in determining individual
outcomes, in this study, we investigate how might families engage with CSAs’ capacity as an
institution for producing social and psychological effects? We suggest one way is through
parent/child communications. From this perspective, CSAs not only facilitate greater asset
accumulation among low-income families, but they also facilitate communication between
parent and child about college and the need to begin planning for it now, not later. We use
identity-based motivation theory developed and tested by Daphna Oyserman and her
colleagues to make specific hypotheses [4-6].
While it is not the focus of this paper, we think it is still important to recognize individual
outcomes are not solely determined by institutions or individuals’ effort, and ability; history
also plays a role. In America, there is a long history of inequitable wealth distribution [7]. This
history impacts the return low-income and Black children can receive from education alone.
For example, research shows low-income children as adults with a four-year degree earn about
one-third less than higher-income families with a four-year degree [8]. Regarding wealth,
research shows Black families who have a head of household who graduated from college have
about 33% less wealth than White families who have a head of household who dropped out of
high school [9]. And so, if the goal is to ultimately reduce wealth inequality, access to small- dollar CSAs (initial deposits of $5 to $1,000) will not be enough, nor will individuals own use of
effort and ability. Substantial targeted ongoing deposits are needed to overcome existing
wealth gaps that derive from a legacy of wealth inequality in America.
REVIEW OF RESEARCH
Research on CSAs has for some time theorized that it matters if children know whether they
have an account or not (e.g., [10]). Early on this was articulated as children having an account
in their own name with the assumption being, if they have an account but are not aware of it, it
will not produce the kinds of social and psychological effects that have facilitated the growth of
CSAs. The focus on whether the account was in the child’s name was driven by available data at
the time. Early studies, before primary data was available, largely relied on secondary data from
the Panel Study of Income Dynamics (PSID). The Child Development Supplement (CDS) of the
PSID asks children first if they have a bank account in their name and then whether they have
money set aside in that account for future schooling such as college. What became clearer as
program data became available, was not so much that the account is in the child’s name, but
that the child has knowledge that money was being put aside for them in an account to attend
college.
This paper builds on these nascent ideas within the CSA field, that children who know that
money is being put aside for them in college savings accounts are more likely to form an identity
as being college-bound than those who are not aware, despite being in a CSA program. One way
Page 4 of 18
176
Advances in Social Sciences Research Journal (ASSRJ) Vol. 9, Issue 9, September-2022
Services for Science and Education – United Kingdom
to measure that knowledge is by asking parents if they are communicating with their children
about participating in the CSA program. Why communication is important is because programs
can have a direct impact on parental communications. More specifically, if parent/child
communications are found to be an important predictor of a child’s formation of a college- bound identity, interventions can be easily developed that help remind parents of the
importance of talking to their children about their CSA.
Educational Expectations, used as a Proxy for College-Bound Identity
There is a well-established body of research in the education field that indicates parental and
children’s college expectations are strong predictors of children’s educational success (e.g., [10-
12]). Moreover, research shows CSAs are associated with children having more positive
expectations about their ability to attend college (e.g.,[10]). Drawing on research done by social
psychologist Daphne Oyserman and her team (e.g.,[4-6]), researchers have suggested that CSAs
help children form and act on their college-bound identity [13-15]. Children’s college
expectations have been used by researchers as a proxy for college-bound identity (e.g., [13]).
The idea that children form a college-bound identity is based on Identity-Based Motivation
theory (IBM) which has three key components: salience (images of future selves are on the
mind now), group congruence (people like me also go to college), and difficulty as normal [16].
These core components have been shown to be important predictors of children’s school
behaviors [16] and may help researchers understand more about how CSAs influence children’s
behaviors (e.g., [14-15]). In a recent paper, Elliott et al. [14] use a preliminary instrument to
measure children’s college-bound identity. They find that participating in a CSA program is
associated with children (fourth through ninth grades) forming a college-bound identity.
Parental Educational Communication
Building on the early thinking about CSA awareness and resulting influence on the way children
think about their future, this paper investigates the role of parental communications within CSA
programs. Parental communications generally fall under the study of parenting practices.
Parenting practices have been shown to be an important factor for understanding child
functioning [17]. According to Darling and Steinberg [18], parental practice consists of parental
attitudes and patterned behaviors (such as parental involvement to include communication,
structure, discipline, psychological control, and other behaviors) which make up the parent- child relationship. While parenting practices are influenced by several different domains, the
domain most relevant to the relationship between CSAs and parenting practices is family
context [19]. A part of the family context domain is socioeconomic status (e.g., income, wealth,
employment) (for a complete list of domains see [20]).
In a statewide randomized control trial conducted by the Center for Social Development (CSD)
at Washington University, Huang et al. [21] found that CSAs are causally related to a reduction
in punitive-parenting practices, even a tool for positive parent-child interactions. Researchers
from CSD also found that CSAs reduce maternal depression [22] and result in more positive
parental educational expectations for their children [23]. Qualitative research suggests that
parents in CSA programs may be more likely to talk to their children about college [24].
However, research on CSAs has not examined whether these communications may increase the
likelihood that children develop a positive college-bound identity. Parental communications
might be a way that CSAs influence children’s development of a college-bound identity, and
Page 5 of 18
177
Zheng, H., Elliott, W., & O’Brien, M. (2022). The Importance of Parent/Child Communications About Children’s Savings Accounts for Developing a
College-Bound Identity. Advances in Social Sciences Research Journal, 9(9). 173-190.
URL: http://dx.doi.org/10.14738/assrj.99.13030
importantly, point to a way for programs to strengthen CSA effects through interventions that
encourage families to communicate with their children about their CSAs.
THEORY
In this paper we posit that parental communication about CSAs is associated with children
forming a college-bound identity. We use Identity-Based Motivation (IBM) theory to
understand how parent-child communications about CSAs is related to children’s formation of
a college-bound identity. IBM is a social psychological theory that explains how people form
identities (possible identity or possible self) and under which situations people are likely to act
on an identity [6, 25-26]. A possible identity is a child’s perception of their potential to become
a certain type of person in the future within a particular social and cultural context [27-29].
From this perspective, a college-bound identity is a child’s perception of their potential for
becoming a college student in the future within their particular social and cultural context. A
college-bound identity would come with a set of personal traits and characteristics about what
it means to be college-bound [30]. People form many possible [16] that are positive or negative
[30]. People do not act on each identity they form (for a review, [31]). To be actionable, possible
identities must be linked to strategies to reach them and important for the current context [32].
However, this study is focused on the formation of identities not on whether children act on
them.
Drawing from IBM, we hypothesize a way college-bound identity develops is through parent- child conversations regarding college and career within the context of being a CSA participant.
CSAs provide an opportunity for families to engage in conversations with their children about
college. For example, most CSAs have as one of their primary goals, children being able to pay
for college. They are often administered through a state 529 plan. State 529 plans are tax- advantaged savings plans designed specifically for the purpose of helping families pay for
college. As such, CSAs by their very nature provide families with a context for talking about
college with their children. It is not only the structure of CSAs that provides the context for
families to talk to their children about college, but it is also the program materials and quarterly
financial statements that families receive. For example, in a qualitative study of SEED OK
participants, researchers find, “Many respondents say that the SEED OK program materials and
quarterly SEED OK 529 statements remind them that their child has an account for
postsecondary education and prompt them to think about saving.” So, these material and
statements cue to parents that their child has a strategy for going to college, bringing it to the
forefront of their minds and providing them with another opportunity to talk to their children
about their account and what it is for. In this way, CSAs provide parents and children the
opportunity to talk about their CSA and going to college.
Importantly, evidence suggests parent-child communicationns are key to children’s formation
of possible-identities [33-35, 28]. In essence, the parent and child co-construct the child’s
college bound identity [35]. Marshall et al.’s [35] findings that parent and child co-construct the
child’s possible identities is supported by research that finds parent and child often share
similar possible identities for the child [36]. Therefore, we posit parent and child co-construct
the child’s college-bound identity. Further, we posit that CSAs provide a rich context for parent
and child to communicate about college and ultimately for children to co-construct their
college-bound identity with their parent. It is important to note, however, the child does not
blindly accept parental views about what their college-bound identity should be. Instead,
Page 6 of 18
178
Advances in Social Sciences Research Journal (ASSRJ) Vol. 9, Issue 9, September-2022
Services for Science and Education – United Kingdom
research shows that children bring into the conversation information from other contexts
which allows them to alter parent views to fit their own views [37]. And so, individuals and
their experience also matter and will help explain whether children form a college-bound
identity.
Promise Scholars
In this paper, we use data from Promise Scholars. Promise Scholars is a CSA program started
by the Community Foundation of Wabash County (CFWC) for children in Wabash County,
Indiana. The program serves students in grades 4-8 by linking a CSA with additional
opportunities to regularly earn scholarships and match by satisfying objective criteria. Earned
scholarship awards captured in their CSA are later granted (up to the student’s age of 23) for
qualified educational expenses once a student enrolls in a qualifying post-secondary institution.
For a full description of the Promise Scholars program and the activities and amounts of the
scholarships provided see [14].
METHODS
The population of this study included a total of 1,579 fifth, sixth, seventh, and ninth grade
students attending Wabash County schools during the 2019-2020 academic year. The
population was made up of 413 fifth graders, 393 sixth graders, 404 seventh graders, and 369
ninth graders. Surveys were separately conducted for parents and students. The Promise
Scholars program has partnered with the University of Southern California and Summitlab
Corporation to conduct a randomized controlled trial (RCT) of Pathways to Success. Pathways is
a 12-session classroom-based social-psychological intervention that aims to help eighth grade
students transition to high school. It is implemented twice a week during the first six weeks of
the school year. Therefore, the current study did not include eighth grade students in the study
population. We plan to interview eighth grade students for future studies.
Data
Parent Survey
Parents of 1,579 fifth, sixth, seventh, and ninth grade students were distributed surveys
through email, with follow-up reminders via mail and at school-based events. CFWC provided
the mailing list, to which links to an online survey were mailed with periodic reminders before
the closure of the survey. Eighty-seven percent of the parents had access to emails and received
online invitations to complete the survey. The response rate to the online survey was 44%.
Along with the parent survey, a comprehensive consent form was presented to explain, and
seek consent for, the child survey. For families that had multiple children in the eligible sample,
a separate email with a unique survey link was sent for each child. Respondents received a $20
pre-paid debit card for each survey completed.
There were about 100 items in the parent survey, which covered a broad range of topics
including: parent engagement in students’ academic activities, knowledge of savings programs
and communication with child about savings early childhood education and childcare, and
discussion of post-secondary goals; parental educational expectations, planning/saving for
post-secondary education; parent and child health history, academic performance, material
hardships; select demographics about primary and secondary caregiver including
race/ethnicity, education, employment, household income, and net worth.
Page 7 of 18
179
Zheng, H., Elliott, W., & O’Brien, M. (2022). The Importance of Parent/Child Communications About Children’s Savings Accounts for Developing a
College-Bound Identity. Advances in Social Sciences Research Journal, 9(9). 173-190.
URL: http://dx.doi.org/10.14738/assrj.99.13030
Student Survey
Each school provided a place for students whose parents consented to participate in the child
study to complete the survey. Information about the student survey was emailed to eligible
students one-day before the survey date. In groups by grade, a CFWC or school representative
read the consent information out and began the survey. To indicate completion of the survey,
the student showed the facilitator the final screen that indicated the survey has been completed.
Each student received $10 cash for taking part in the survey. CFWC followed up with students
absent on the date of the survey to ensure they had the opportunity to participate in the study.
Child Assessment Tool- Electronic (CHAT-E), developed by Terri Sabol and her team at
Northwestern University, was used to collect child data. Preliminary findings from
kindergartners indicated that CHAT-E had adequate construct validity [38]. CHAT-E consisted
of 36 statement items and two open-ended questions on children’s perceptions of college,
school, and experience with money and savings. By design, CHAT-E was a developmentally
informed tool that assessed how young children made meaning of school and college. CHAT-E
for young children was administered on a tablet application. Informed by the Berkeley puppet
interview, CHAT-E contained puppet-like exchange on a tablet showing two identical animated
figures who made different statements on same topics such as school, college, and money (To
learn more about the Berkeley puppet interview see
https://dslab.uoregon.edu/about/berkeley-puppet-interview/). Students were prompted to
select puppets whose statement they mostly agreed with.
Given the population with older students for the current study, we collaborated with CHAT-E
developers to modify animated items for online and paper-and-pencil modalities. Still, as with
the original, each student was presented with a pair of statements and asked to select the one
that was most like them. Despite the age differences, all items except for one, retained the
original wording. One item about parents buy toys was changed to “nice things”. Among 637
complete parent surveys, 93% (n = 594) consented for child to take part in the student survey.
Sample
There were 506 parents and students in the final analytic sample. Forty-six students were not
included in the sample because they had an Indiana CollegeChoice 529 Direct Savings Plan but
were not enrolled in the Community Foundation Promise Scholarship program. Seventy-eight
complete student surveys without a matched parent survey were removed. Therefore, the
analytic sample for children’s college-bound identity consist of 506 matched parent and student
pairs.
In the analytic sample of the current study, half of the students were female, 24% were fifth
graders, 27 % were sixth graders, 28 % were seventh graders, and 21 % were ninth graders.
The composition of surveyed students from different grades was comparable to the
composition of the population. Among parents, 34% received a bachelor’s degree or above,
41% earned less than $55,000 in 2019, 73% were married, and 95% were white. According to
Census Bureau [39], 18% of Wabash County residents received a bachelor’s degree or above,
the median household income was $50,637 in 2018, 52% residents were married, and 97%
residents identified as white. Therefore, the parent sample were made up of more college- educated, married, and higher-income respondents than typical residents of Wabash County.
Page 8 of 18
180
Advances in Social Sciences Research Journal (ASSRJ) Vol. 9, Issue 9, September-2022
Services for Science and Education – United Kingdom
Measures
The variable of interest in this study was a three-level variable constructed based on students’
Promise Scholar status (1 = Promise Scholars; 0 = Comparison) and communication between
parents and students about Promise Scholars (1 = Communicated; 0 = No communication).
Regarding communications, respondents were specifically asked, “In the past 12 months, did
you or another adult in your household talk with [child] specifically about [child]’s Promise 529
account or the Community Foundation Promise Early Award Scholarship Fund?” Students in the
comparison group (not in Promise Scholars) were coded 0, parents of Promise Scholars who
did not communicate about the program were coded 1, and parents of Promise Scholars who
communicated about the program was coded 2.
Dependent variables
College-bound identity was measured with three subscales (i.e., salience, difficulty as normal,
and group congruence). The three subscales consist of 27 items altogether: identity salience
with 7 items, group congruence with 10 items, and interpretation of difficulty as normal with
10 items. Five items were removed after unidimensionality and reliability tests. The total score
of college-bound identity ranges from 0 to 22, with a higher score indicating a more positive
college-bound identity. The composite reliability and the Alpha of the scale were respectively
0.72 and 0.67 [14], which indicate acceptable reliability.
Control variables
There were seven control variables used in this study including child’s gender (1 = Female; 0 =
Male), child’s overall academic performance perceived by one parent (1 = Above average; 0 =
Average or worse), child’s health status rated by one parent (1 = Excellent; 0 = Very good or
worse), child’s grade (1 = fifth grade, 2 = sixth grade, 3 = seventh grade, and 4 = ninth grade),
parent marital status (1 = Married; 0 = Not married), household net worth, and household
income.
Analytic strategy
The following hypotheses were tested in this study:
H1: Promise Scholars whose parents communicated about the program with them (i.e.
communicators) have higher college-bound identity than do the comparison group.
H2: Promise Scholars whose parents communicate about the program with them (i.e.
communicators) have higher college-bound identity than do Promise Scholars whose parents
do not communicate about the program with them (i.e. non-communicators).
Three steps of analysis were conducted in this study. The first step was to analyze the pattern
of missing data to determine whether the default listwise deletion by STATA was appropriate
to handle missing information. Missing information within a data frame may compromise
inferences from obtained results. In the present study, results of Little’s test suggest that
variable of interests were not (p < .05) missing completely at random (MCAR), with a range of
0.5 percent to 15 percent. Therefore, multiple imputation by chained equations (MICE) was
employed for all variables included in the analytic models because MICE does not assume a joint
multivariate normal distribution for categorical variables [40]. Ten cycles of imputation were
carried out.
Page 9 of 18
181
Zheng, H., Elliott, W., & O’Brien, M. (2022). The Importance of Parent/Child Communications About Children’s Savings Accounts for Developing a
College-Bound Identity. Advances in Social Sciences Research Journal, 9(9). 173-190.
URL: http://dx.doi.org/10.14738/assrj.99.13030
The second step addressed the potential for selection bias that could arise when only
respondents who were Promise Scholars were included. To address potential selection bias, an
Inverse Probability Weight (IPW) for survey participants who were Promise Scholars and those
who were not was constructed. Briefly, we used information available for all survey
respondents to predict the probability of signing up for Promise Scholars and used the inverse
of the probabilities as weights in the analysis so that the results would be representative for the
initial populations of the sample [41]. Inverse probability weights were estimated with a
multinomial logistic regression model that used Promise Scholars account ownership as the
dependent variable and all available covariates as independent variables [42].
All covariates described above were included in the weight estimation, as they were thought to
be potentially correlated with signing up for Promise Scholars. Balance checks were conducted
to determine the ability of IPW to balance relevant covariates. Multinomial logistic models that
used Promise Scholar account ownership as the dependent variable and each covariate as the
independent variable were estimated. P-values of the coefficients from each model before and
after applying inverse probability weights are displayed. In the unweighted sample, many of
the covariates showed significant group differences between respondents who signed up for
Promise Scholars and those who did not. Once the inverse probability weight was applied,
group differences on all covariates were no longer significant (see Table 1).
Table 1. Data Balance Before & After Propensity Score Inverse-Probability Weight Adjustmenta
Promise Scholars vs.
Comparison
Promise Indiana
Participants vs.
Comparison Covariates Unweighted
P-Value
Weighted
P-Value
Unweighted
P-Value
Weighted
P-Value
Household net worth: Negative .124 .374 .847 .982
Household net worth: $0-$55,000 .050 .929 .708 .289
Household net worth: $55,001-$150,000 .696 .206 .9.3 .724
Household net worth: > $150,001 .002 .454 .324 .211
Household income: < $55,000 <.001 .973 .608 .905
Household income: $55,001-$95,000 .552 .922 .452 .198
Household income: > $95,001 <.001 .953 .058 .259
Marital status <0.001 0.976 0.176 0.551
Child gender 0.769 0.883 0.707 0.971
Child’s grade: 5th grade .477 .560 .198 .821
Child’s grade: 6th grade .971 .426 .334 .316
Child’s grade: 7th grade .510 .917 .917 .660
Child’s grade: 9th grade .900 .776 .520 .800
Child’s health status 0.115 0.967 0.695 0.413
Overall academic performance <0.001 0.618 0.451 0.936
a P-values of the coefficient of Promise Scholars and Promise Indiana participants are
displayed
The third step was to analyze children’s college-bound identity using ordinary least squares
regression (OLS). Data analysis steps were conducted using STATA [43]. Ordinary least squares
regression was used to predict scores of children’s college bound identity. The mi estimate: reg
commands in STATA were used to perform OLS regression across the ten imputed data sets.
Page 10 of 18
182
Advances in Social Sciences Research Journal (ASSRJ) Vol. 9, Issue 9, September-2022
Services for Science and Education – United Kingdom
Joint tests of transformed coefficients were conducted after multiple imputations estimates
were obtained for selected groups [44]. Findings at a significance level of p < .05 were noted in
the tables.
RESULTS
Descriptive results
Table 2 displayed descriptive statistics by communication status. There were more high-net
worth and high-income households in the communicator group than in the non-communicator
group and the comparison group. More households in the communicator group were married
than in the other two groups. More children in the communicator group were rated excellent
health and above average academic performance by their parents than the other two groups.
Regarding the variables of interest, children in the communicator group had the highest
college-bound identity mean scores (19.69, standard deviation 2.93), while those in the
comparison group had the lowest mean scores (17.58, standard deviation 4.17).
Page 11 of 18
183
Zheng, H., Elliott, W., & O’Brien, M. (2022). The Importance of Parent/Child Communications About Children’s Savings Accounts for Developing a
College-Bound Identity. Advances in Social Sciences Research Journal, 9(9). 173-190.
URL: http://dx.doi.org/10.14738/assrj.99.13030
Table 2. Descriptive Statistics by the Communication Statusa (N = 506)
Communicator
(%)
Non-Communicator
(%)
Comparison
Group(%)
Categorical Variables
Net worth
Negative net worth 20 (4) 23 (5) 16 (3)
0 - $55,000 80 (16) 81 (16) 50 (10)
$55,001 - $150,000 66 (13) 36 (7) 23 (5)
$150,001 or more 72 (14) 24 (5) 9 (2)
Income
Less than $55,000 75 (14) 86 (16) 61 (12)
$55,001-$95,000 102 (19) 57 (11) 36 (7)
$95,001 or above 70 (13) 35 (7) 8 (2)
Marital status
Not married 45 (8) 60 (11) 45 (8)
Married 213 (39) 125 (23) 64 (12)
Child Gender
Male 130 (22) 100 (17) 55 (10)
Female 131 (23) 105 (18) 60 (10)
Child Grade
5th grade 72 (13) 54 (10) 31 (6)
6th grade 73 (13) 50 (9) 27 (5)
7th grade 62 (11) 58 (11) 23 (4)
9th grade 48 (9) 37 (7) 18 (3)
Child Health
Very good or worse 103 (18) 79 (14) 55 (10)
Excellent 159 (28) 113 (20) 59 (10)
Academic Performance
Average or below 68 (12) 69 (12) 66 (12)
Above average 194 (35) 120 (21) 45 (8)
Variables of Interest
Educational expectations
Less than four-year college 71 (12) 72 (13) 65 (11)
Four-year college or above 190 (33) 126 (22) 49 (9)
Communicator Non-Communicator Comparison Group
Continuous Variable Mean (S.D.) Mean (S.D.) Mean (S.D.)
College-Bound Identity 19.69 (2.93) 18.48 (3.79) 17.58 (4.17)
Note. Missing information are not included for the calculation of descriptive statistics
OLS regression results
We examined the effect of communication about Promise Scholars on students’ college-bound
identity using weighted OLS regression. Students in the communicators group had stronger
college-bound identity than their counterparts in the non-communicators group and the
comparison group. On average, students in the communicators group had 1.24 points higher
college-bound identity scores than those in the comparison group, holding other variables
constant (β = 1.24, p < .05). Students in the non-communicators group did not have statistically
Page 12 of 18
184
Advances in Social Sciences Research Journal (ASSRJ) Vol. 9, Issue 9, September-2022
Services for Science and Education – United Kingdom
different college-bound identity scores than children in the comparison group (p > .05). Among
covariates, the college-bound identity scores for students whose parents perceived them
having above average academic performance were on average 1.20 points higher than that of
students whose parents perceived them having average or below academic performance (β =
1.20, p < .05). On average, sixth graders had 1.58 points higher college-bound identity scores
than fifth graders (β = 1.58, p < .05), while the college-bound identity of students in other grades
were not significantly from fifth graders.
Parent/child Communication
Given the insignificant coefficient of non-communicators on college-bound identity, the
difference between communicators and non-communicators might not be statistically
significant. Therefore, this study examined the difference between communicators and non- communicators in terms of their coefficients. The results of the difference test indicated that
children in the communicators group had stronger college-bound identity than students in the
non-communicators group. On average, the college-bound identity scores for the
communicators group were 0.81 points higher than for the non-communicators group (F1, 391 =
4.21, p < .05).
Page 13 of 18
185
Zheng, H., Elliott, W., & O’Brien, M. (2022). The Importance of Parent/Child Communications About Children’s Savings Accounts for Developing a
College-Bound Identity. Advances in Social Sciences Research Journal, 9(9). 173-190.
URL: http://dx.doi.org/10.14738/assrj.99.13030
Table 3. The Estimated OLS Model of Children’s College-Bound Identity (N = 506)
Coef. Std. Err. [95% Conf. Interval]
Constant 16.08*** 0.81 14.48 17.68*
Net worth
Negative net worth - - - -
0 - $55,000 -0.30 0.57 -1.42 0.83
$55,001 - $150,000 -0.81 0.72 -2.23 0.61
$150,001 or more 0.47 0.76 -1.03 1.96
Income
Less than $55,000 - - - -
$55,001-$95,000 0.50 0.69 -0.86 1.85
$95,001 or above -0.01 0.86 -1.71 1.69
Marital status
Not married - - - -
Married -0.15 0.62 -1.38 1.07
Child Gender
Male - - - -
Female 0.21 0.49 -0.75 1.16
Child Grade
5th grade - - - -
6th grade 1.58* 0.56 0.47 2.69
7th grade 0.74 0.80 -0.83 2.31
9th grade 0.36 0.70 -1.02 1.75
Child Health
Very good or worse - - - -
Excellent 0.90 0.54 -0.17 1.96
Academic Performance
Average or below - - - -
Above average 1.20* 0.56 0.11 2.30
Educational expectations
Less than four-year college - - - -
Four-year college or above 0.40 0.68 -0.95 1.74
Communication Status
Comparison group - - - -
Non-Communicator 0.43 0.53 -0.62 1.47
Communicator 1.24* 0.50 0.24 2.23
* p < .05, ** p < .01, *** p < .001, two-tailed test
DISCUSSION
This paper investigated the relationship between parental communications about CSA
programs with their children and children’s formation of a college-bound identity. This line of
investigation came out of early thinking on CSAs that suggests in order to produce the types of
social and psychological effects Sherraden [3] originally hypothesized, children must know that
they have an account. This question about parental communications has practical implications
for the CSA field. If parental communications are associated with improved outcomes with CSA
programs, then developing interventions that help foster this communication may be important
Page 14 of 18
186
Advances in Social Sciences Research Journal (ASSRJ) Vol. 9, Issue 9, September-2022
Services for Science and Education – United Kingdom
for strengthening CSA effects. Moreover, this is the type of intervention that programs can
relatively easily implement without incurring high costs, making it all the more important to
understand.
While no existing research specifically tests the relationship between parental communications
about CSA programs with their children and children’s college-bound identities, there is
evidence to suggest that CSAs can influence parenting practices more generally [21, 19, 45].
Parenting communications fall under the broad research on parenting practices. Findings from
this study indicate that parental communications about the program are associated with
children’s formation of a college-bound identity. This is important because college expectations,
a proxy for college-bound identity, have been shown to be important factors for predicting
children’s education outcomes [10-12].
Furthermore, findings from this study align with findings on the relationship between CSAs and
children’s college expectations. Prior research has found that children who have savings set
aside for college, a proxy for CSAs, is positively associated with children’s college expectations
(e.g., [10, 46-47]). In this study, however, instead of using a single survey question (“do you
expect to attend college”) as a proxy for children’s formation of a college-bound identity, the
researchers used a new measure of college-bound identity which consists of three subscales
with 27 items. The three subscales are meant to capture the three core components of a college- bound identity: salience (college is on the forefront of the mind), difficulty as normal
(schoolwork is supposed to be hard and I have strategies to succeed), and group congruence
(people like me go to college).
Findings from this study also align with studies conducted by identity-based motivation
researchers. Specifically, we find that parent-child communications about CSAs is associated
with children developing a college-bound identity. This aligns with research conducted by IBM
researchers that finds parent-child communications are associated with children forming
possible identities related to career and schooling [33-35, 28]. Findings from this study are also
consistent with IBM research that indicates that possible identities are co-constructed [35].
That is, both parent and child can play a role in the formation of a child’s college-bound identity.
Implications
The primary implication drawn from this study is that CSA programs should consider
implementing strategies that encourage families to talk to their children about their CSA if a
goal of the program is developing a college-bound identity among their participants. There are
many ways that programs may encourage families to communicate with their children about
their CSA and college. For example, programs could send out text messages to families
reminding them of the importance of talking to their children about their CSA. They could send
out more frequent financial statements and encourage parents to sit down and talk to their
children about the statement. Another thing that programs can do is encourage families to make
depositing funds a parent-child event. Certainly, there are additional ways that programs could
encourage parents to talk to their children about their CSA, but these are a few.
Another implication of this study is the idea that there are ways that families can engage with
the CSA program that go beyond making deposits. Often CSA programs too narrowly define
engagement as being whether families make a deposit or simply end up prioritizing making