Linking Getting to School With Going to School

Abstract

Researchers, policymakers, and practitioners have recently aligned efforts to reduce school absenteeism, particularly during kindergarten when excessive absences are highest out of all elementary grades. Little is known, however, about whether the way in which students get to school might influence if they go to school. To address this gap, this study was the first to address the role of school bus-taking on reducing school absences. Using a national large-scale dataset of children (the Early Childhood Longitudinal Study–Kindergarten Class of 2010–2011), the findings suggest that children who took the school bus to kindergarten had fewer absent days over the school year and were less likely to be chronically absent compared with children who commuted to school in any other way. Given that many districts are considering cutting or restricting bus services, this study brings to question whether doing so might limit the resources upon which families rely to ensure their children attend school each day. Implications are discussed.

Keywords

school bus absenteeism chronic absence descriptive analysis kindergarten

Districts across the country are trying to balance tight budgets with providing students with the supports needed for learning, including determining ways to reduce school absenteeism. An often overlooked but potentially important discussion is the relationship between the way children get to school and whether they go to school. Recent policy dialogue suggests that transportation to school is an increasing challenge for both students and school districts, thereby raising the worry that reducing access to and utilization of district transportation could have implications for student absenteeism and, by consequence, other important student outcomes. Using nationally representative data, this study explored the potential relationship between taking a school bus to school and absenteeism of kindergarten students, a group most dependent on adults for their transportation to school and most prone to absenteeism out of all elementary school years (Balfanz & Byrnes, 2012).

Specifically, this study asked whether kindergartners who took the school bus to school had fewer total absences and lower chances of being chronically absent compared with those who went to school by other means, and whether any observed association differed by individual and family characteristics. No other known study has examined whether taking the school bus is associated with fewer absences or reduced chronic absenteeism. Thus, even though school bus-taking has a long-standing history in our nation’s school systems, this study contributed new knowledge in an under-researched area regarding whether taking the bus can serve as one way to address absenteeism issues for those just starting out in school.

The Status of the School Bus

The question as to whether or not schools provide families with bus transportation has recently moved into the spotlight. Although school bus transportation has been part of U.S. schooling since the early 20th century (National Association of State Directors of Public Transportation Services, 2000), the resurgence of the issue is making its way into present-day policy debates given that managing and providing bus transportation is costly and that these costs are on the rise. The National Center for Education Statistics (NCES; U.S. Department of Education, NCES, 2016) reported that the national average per-pupil expenditure on public bus transportation was US$961 in 2011–2012; in constant dollars, that expense was only US$531 in 1980–1981. While the number of public school students riding the school bus has only increased by about 4 million between 1980–1981 and 2011–2012, total national costs have risen by more than US$12 billion (in constant dollars) over the same time period. In other words, school bus services are not inexpensive. Given the rising costs of bus transportation in addition to key macroeconomic events over the past 10 years such as higher fuel prices, weakened national and local economies, and slashed school budgets, one way districts have relieved recent financial pressure has been to cut bus service (Safe Routes to School, n.d.).

While some schools and districts like Chicago Public Schools are halting bus service entirely (Lulay, 2015), other districts have reduced access and availability, such as lowering the number of routes, cutting bus stops, or increasing the walk radius (i.e., the proximity to school within which students are ineligible for service). Reduced capacity in this way has minimized options by which families in our school systems can rely to send their children to school. To compound this, in many districts and states, courts have ruled that charter schools are not required to offer bus service (Smith, 2015), and private schools are also restricting their services (Oglesby, 2015; Redmond, 2016). Therefore, as bus service is completely halted or as access is constricted throughout the nation, families have become more vocal about expressing growing concerns about being able to get their children to school (Safe Routes to School, n.d.). These parental concerns range from new impositions on parents’ work schedules (for pick up/drop off) to fear over neighborhood safety (crime, hazards) for children who now have to walk, bike, or ride public transit to school instead of taking the bus (Chandler, 2015; Safe Routes to School, n.d.).

In sum, on one side of the issue, schools and districts are considering (or acting upon) constricting bus service as a way to reduce costs. On the other side, families are raising concerns that cutting access to this school service might impede their ability to get their children to school. And yet, given this debate that is making news headlines across the country, surprisingly little is known about whether taking the bus to school is actually a successful means of routinely getting children to school. This is a significant oversight given that at its fundamental core, the purpose of the school bus is to bring children to school on a daily basis.

This oversight, however, may not be without cause. For instance, on the website for the national American School Bus Council (n.d.), of the five key “Issues” that are presented, only one of these discusses how bus service helps children get to school; within this prong, getting to school (i.e., mitigating absenteeism) is nested within a larger “Access to Learning” category. The other “Issues” pertain to safety standards, bus drivers, the environment, and community development. Therefore, there is a void in both research and practice surrounding school transportation; little attention has traditionally focused on how taking the bus might be an effective means to get children to school each day and hence reduce absenteeism.

The Status of Student Absenteeism

If taking the school bus does indeed link to helping children show up at school, this has the potential to be significant. Currently, the United States is facing a school absence epidemic, and, much like school busing, absenteeism is also moving into the policy spotlight. According to current estimates, approximately 5 to 7.5 million children in K–12 schools across the United States are missing 1 month or more of school each year, which translates to 150 to 225 million instructional days lost per year in the United States (Chang & Davis, 2015). There are financial ramifications of absences on our school systems. For instance, many school districts receive funding based on average daily student attendance, and thus when students are absent from school, the district receives less funding from the state. Data from the Office of the California Attorney General (2015) found that in the 2014–2015 school year alone, school absences cost California school districts US$1 billion in forgone revenue (i.e., hypothetical revenue at 100% attendance minus revenue based on actual attendance). From 2011 to 2015, California school districts have lost a total of US$4.5 billion in potential revenue due to absenteeism.

Perhaps as, if not more, concerning than the financial consequences of absenteeism is that not showing up to school has ramifications for individual growth and development. Simply stated, absenteeism is harmful. Although much of the work in the area of absences is correlational (for more causal studies, see Gershenson, Jacknowitz, & Brannegan, 2016; Goodman, 2014; Gottfried, 2010, 2011b; Marcotte & Hemelt, 2008), research has unequivocally found that children who miss school tend to have lower standardized testing outcomes, higher chances of grade retention and dropout, more difficulty with social development and greater feelings of alienation, increased drug and alcohol use once in young adulthood, and worsened employment prospects (Alexander, Entwisle, & Horsey, 1997; Broadhurst, Paton, & May-Chahal, 2005; Chen & Stevenson, 1995; Connell, Spencer, & Aber, 1994; Ekstrom, Goertz, Pollack, & Rock, 1986; Finn, 1993; Gershenson et al., 2016; Goodman, 2014; Gottfried, 2009, 2010, 2011b, 2014; Hallfors et al., 2002; Kane, 2006; Morrissey, Hutchison, & Winsler, 2014; Newmann, 1981). More so, it has been established that a spillover effect exists in regard to absenteeism; when classmates miss more school, all students in the classroom tend to have lower test scores (Gottfried, 2011a, in press).

Of great concern among educators and policymakers is that many of the consequences of missing school emerge in kindergarten. In fact, out of all years of elementary school, absenteeism is highest in kindergarten (Balfanz & Byrnes, 2012). Estimates suggest that about 25% of all kindergartners are missing almost 1 month or more of school (Romero & Lee, 2007). This jumps to 55% in many low-income or urban schools (Chang & Romero, 2008). This is alarming because while many of the previously mentioned consequences of absenteeism of course may not apply at such a young age such as drug use and unemployment, certainly educational and developmental ramifications can emerge. For instance, Chang and Romero (2008) found that kindergartners with more absences also had lower scores on first-grade exams. Connolly and Olson (2012) linked high rates of absenteeism in kindergarten to lower achievement levels and also to grade retention and future absenteeism (at which point many of the other above-mentioned consequences might arise). Ready (2010) found that kindergartners with more absences had lower literacy gains in first grade. In addition to negative consequences on achievement, Gottfried (2014) found that highly absent kindergartners exhibited lower frequencies of favorable socioemotional behaviors. Romero and Lee (2007) found that current absences predict future absences—most of those children in first grade who had high rates of absences also had high rates of absences in kindergarten.

It is not entirely understood why absences are highest for kindergartners compared with any other elementary school year. But, one prominent explanation is grounded in the notion of school transitions. Ollendick and Mayer (1984) found that peak school refusal (i.e., absenteeism) behavior occurs at ages 5 and 6 and again at 10 and 11—the first being the transition into kindergarten and the second being the transition into middle school. As for the former, kindergarten is an extremely formative educational transition for both children and their families (Gill, Winters, & Friedman, 2006; Ladd & Price, 1987; McIntyre, Eckert, Fiese, DiGennario, & Wildeneger, 2007; Pianta & Cox, 2002; Pianta & Kraft-Sayre, 1999). For children, this transition entails adaptation to academic challenges as well as socialization (Bogart, Jones, & Jason, 1980; Holland, Kaplan, & Davis, 1974). In recent years, as this transition has become intensified and as kindergarten has become more aligned with first grade than with prekindergarten (Bassok, Latham, & Rorem, 2016; Little, Cohen-Vogel, & Curran, 2016), any failure to master the challenges in kindergarten has the potential to increase frustration and anxiety (Allan, Hume, Allan, Farrington, & Lonigan, 2014; Blair, 2002; Ponitz, McClelland, Matthews, & Morrison, 2009). Greater anxiety and difficulty adapting can spur negative attitudes about going to school (Shore, 1998), which is linked to more absences (Gottfried, 2015).

But families must also adapt to the transition into kindergarten, namely, through establishing school-going practices (Chang & Romero, 2008; Gottfried, 2015). Even in families when children left the house to attend preschool, many of these same families still experience dramatic changes and challenges between preschool and kindergarten (Quas, Murowchick, Bensadoun, & Boyce, 2002). Consequently, the transition into kindergarten is characterized by needing to address school-going logistics for that child, like determining transportation options and adjusting work schedules, as well as needing to establish daily routines, such as packing lunches and leaving the house at the same time each morning: The failure to address these can inhibit good school attendance (Chang & Romero, 2008).

Child health may also seem at face value to be a key factor in absenteeism for kindergarten-aged children. However, Kerr et al. (2012) conducted an intervention of school nurses, in which school nurses called or visited parents of P–3 schoolchildren to determine actual reasons for absence as well as test for illness. First, the study found that the majority of students who reported illness as reason for absence were not actually sick. Second, when asking parents reason for absence, out of all possible reasons that parents could select for a child’s absence (health or family-related), the second highest reason for any absence whatsoever was transportation issues.

Linking School Buses and Student Attendance

The school bus has the potential to serve as a key way to improve school attendance. Although there is no known research connecting bus-taking to absenteeism, the link between school transportation and student absences could be surmised vis-à-vis features of the kindergarten transition mentioned above. In addition, family characteristics may play a role in how bus-taking might be linked to absenteeism. Each is discussed.

Routines

In kindergarten, it is the family who, potentially for the first time, must ensure that children get to school every day. As a direct link between bus-taking and absenteeism, setting family routines becomes particularly important in this first year, and having a regular bus pickup may help to solidify new daily family activities when it comes to getting young children out of the house and into school each morning. There are also indirect ways that might link bus-taking and absenteeism. As mentioned above, the kindergarten transition is a potentially high period of stress and anxiety for both children and families (Fiese et al., 2002; Wildeneger, McIntyre, Fiese, & Eckert, 2008). It has been shown that the establishment of routines reduces family stress during transitions (Wolin & Bennett, 1984). If having access to bus services helps to establish daily routines, this might lower household stress, increased positive attitudes about going to school, and ultimately reduce absences.

Logistics

Utilizing the school bus may also directly aid with developing getting-to-school logistics, again possibly for the first time for the family: For instance, access to bus services may be particularly helpful in families where work schedules cannot be adjusted (Gottfried, 2015). Much like for routines, there may be also indirect effects associated with bus-taking and school-going logistics. For instance, having access to bus services might reduce family stress because of lowered financial and time burdens associated with getting children to school each day (Rhoulac, 2005).

Family Characteristics

Prior research has suggested that family characteristics might be linked to absenteeism as well. For instance, access to a school bus pickup may be helpful for parents in larger households with multiple siblings, especially if kindergarten is not located at the same school where other siblings may go (Rhoulac, 2005). Even in multi-child families when a kindergartner has an older sibling in the same school, having access to a bus may still be helpful for attendance; it has been speculated that in families with more than one sibling, there is often parental resource/attention diffusion (Downey & Condron, 2004). Thus, in this scenario, a bus pickup has the potential to help parents ensure that one or more children in the family can arrive at the same school each day. Moreover, family socioeconomic status (SES) may play a role. In lower SES families, there might be concerned about neighborhood safety for young children and their parents who are “commuting” to school for the first time. Therefore, having access to bus services may assuage anxiety around safety concerns regarding parents who must take young children to school by foot or by public transit to school (Martin & Carlson, 2005), all of which may increase families’ safety risks in lower SES neighborhoods. Or, low-SES families might be concerned about new costs that families had previously not had to address, whether they be monetary (e.g., paying for extra gas for a new driving commute, buying metro cards, etc.) or time (e.g., commute time to/from school for parents, potentially twice a day).

As with routines and logistics, there may be an indirect effect here as well. Access to the bus may increase family-level positivity, and thus increasing positivity toward school increases child-level positive feelings about school (Giallo, Treyvaud, Matthews, & Kienhuis, 2010). Thus, if the kindergarten transition can become less stressful through access to bus transit and/or reduction of potential costs of getting their kindergartner to school, then perhaps not only will absences actually directly decline from using the bus but also parents’ and students’ anxieties about school (e.g., school refusal behavior) may dissipate and absences subsequently decline in this way as well.

Hence, there appears to be potential for bus-taking to be linked to a reduction in school absences. But, as described by both Jacobson (2008) and Kerr et al. (2012), studying early absenteeism is even overshadowed by studying truancy at later grades. This study fills a critical void by asking two key questions:

Do children who take the school bus to kindergarten have lower frequencies of absences?

Are there differences in this association based on key child and family characteristics?

This is the first time in the literature that these questions have been asked. Therefore, this descriptive study fills a critical gap by addressing how children get to school might be linked to going to school. To do so, this study evaluated the most current national dataset of kindergarten children in the United States. As absenteeism is highest for children in this first year of schooling out of all elementary school years (Balfanz & Byrnes, 2012), understanding what practices are best suited for reducing this damaging behavior has the potential to set early trajectories. Gill et al. (2006) specifically cited familiarity with the bus (as well as with the kindergarten and teacher) as a key way to improve the transition into kindergarten, thereby highlighting the potential importance of this transportation mode. Focusing on school practices like bus transportation becomes increasingly important such that we can identify specific, school resources that may be effective at getting children into school.

Method

Source of Data

This study relied on a rich source of data compiled by the NCES at the U.S. Department of Education. The Early Childhood Longitudinal Study–Kindergarten Class of 2010–2011 (ECLS-K:2011) contains data on a nationally representative cohort of children in kindergarten in the 2010–2011 school year. NCES collected information on children and their families and schools through direct assessments of children as well as interviews and surveys of parents, teachers, and school administrators. To ensure that the data were nationally representative, NCES employed a three-stage stratified sampling design. The first sampling unit was geographic region followed by public and private school as the second sampling unit followed by children stratified by race/ethnicity as the third sampling unit. These data represent the most contemporaneously available set of data representing a diversity of school types, socioeconomic levels, and racial and ethnic backgrounds from across the United States.

During the 2010–2011 kindergarten school year, data were collected in two waves: in the fall of the school year as well as in the spring of that same year. This study utilized data from both waves. Missing data ranged from 0% to approximately 25% for the measures in this study. Therefore, chained multiple imputation was employed (Royston, 2004). Data for which there were missing values were imputed back to the sample for which there were nonzero weights. Given that missing data were up to 25% on some measures, 20 sets of plausible values were imputed to resemble the distributions of the observed variables (Graham, Olchowski, & Gilreath, 2007). Sample weights provided by NCES for the ECLS-K:2011 dataset were employed both in the imputation and in the analysis. After imputation, this sample consisted of approximately N = 14,370 child observations. Sample sizes are rounded to the nearest tens digit, per NCES rules of using the restricted version of these data.

Absence Measures

Table 1 presents all outcomes and independent variables analyzed in this study, broken out by those who took the bus and those who did not. Statistical tests of mean differences between are presented in the far-most right column. The outcomes consisted of two measures of child absences. In the spring survey, a child’s teacher reported on the number of absences that the child had up to that point in the year. The teacher selected from a discrete set of answer choices: 0, 1 to 4, 5 to 7, 8 to 10, 11 to 19, and 20 or more. Note that these categories in the survey align with similar category brackets established in prior research (Chang & Romero, 2008; Morrissey et al., 2014). The greatest percentage of children in the dataset missed between 1 and 7 days of school; note that only 4% of the sample had zero absences, and similarly, 4% of the sample had 20 or more absences.

Table 1

Descriptive Statistics (N = 14,370)

	Bus		No bus
	M	SD	M	SD
Absence measures
Number of days absent	5.72	4.60	5.98	4.72
Chronic absenteeism	0.12	0.33	0.14	0.33
Child demographic characteristics
Male	0.53	0.50	0.51	0.50**
Black	0.16	0.35	0.13	0.30***
Hispanic	0.17	0.39	0.28	0.44***
Asian	0.03	0.24	0.04	0.29***
Other	0.06	0.24	0.05	0.24
English language learner	0.13	0.37	0.17	0.40***
Kindergarten entry age (months)	66.44	4.51	66.01	4.65***
Poorer health rating	0.14	0.35	0.14	0.33
Disability	0.24	0.42	0.20	0.39***
Measured skills at kindergarten entry
Reading	33.96	11.55	34.61	12.12***
Math	29.07	10.94	28.92	10.91***
Self-control	3.07	0.63	3.07	0.62
Interpersonal skills	2.96	0.64	2.98	0.63***
Approaches to learning	2.90	0.68	2.95	0.67***
Internalizing problem behaviors	1.46	0.49	1.46	0.49
Externalizing problem behaviors	1.60	0.63	1.61	0.61
Eager to attend school	0.85	0.35	0.84	0.35
Kindergarten and prekindergarten experiences
Full-day kindergarten	0.79	0.42	0.84	0.38***
Public kindergarten	0.97	0.21	0.87	0.38***
Distance from school (miles)	5.71	4.87	4.72	3.56***
Average minutes to school	17.26	8.83	10.64	5.84***
Before/after school center care in kindergarten year	0.14	0.36	0.21	0.42***
Hours of before/after school care in kindergarten year	5.92	10.52	5.93	9.61
Center-based prekindergarten care	0.65	0.47	0.21	0.46***
Hours of center-based prekindergarten care	14.60	13.69	15.76	14.23***
Out-of-home care prior to prekindergarten	0.39	0.49	0.41	0.50**
Household characteristics
Two-adult household	0.77	0.42	0.76	0.41
Number of siblings	1.60	1.17	1.46	1.07***
Older sibling attends same school	0.56	0.50	0.51	0.50***
Age of mother at first birth	23.92	5.80	24.24	5.82***
Number of children’s books at home	84.47	142.05	85.00	134.74
Mother-reported depression	0.20	0.40	0.20	0.39
Mother’s education
Less than high school	0.13	0.34	0.13	0.33*
Some college	0.33	0.46	0.33	0.46
College graduate or beyond	0.29	0.46	0.33	0.48***
Father’s education
Less than high school	0.10	0.31	0.10	0.30
Some college	0.23	0.41	0.20	0.40
College graduate or beyond	0.23	0.43	0.26	0.46***
Household income	62,003.01	53,118.11	63,612.27	54,469.24***
Received government food assistance	0.43	0.48	0.36	0.46***
Urban	0.17	0.38	0.37	0.48***
Rural	0.33	0.47	0.20	0.38***
Problems with safety on school route	0.02	0.14	0.02	0.15
Safe to play outside	0.75	0.44	0.68	0.47***
Maternal employment
Full-time	0.33	0.47	0.37	0.48***
Part-time	0.28	0.45	0.26	0.44
Paternal employment
Full-time	0.65	0.48	0.62	0.47
Part-time	0.04	0.20	0.05	0.22
Regular bedtime	0.92	0.27	0.91	0.28
Number of breakfasts at regular time	5.34	1.77	5.41	1.65**
Number of breakfasts together as family	3.72	2.46	3.84	2.45***
Number of dinners at regular time	5.53	1.83	5.48	1.82
Number of dinners together as family	5.87	1.65	5.79	1.67**
n	3,490		10,880

Note. Tests of mean differences between groups at *p < .05, **p < .01, and ***p < .001.

In the most recent literature on absences, student absenteeism is typically measured in two ways. Like Gottfried, Egalite, and Kirksey (2016), this study examined both. The first way is the total number of days absent from school which corresponds to past literature, including those who have used ECLS-K data (e.g., Gershenson et al., 2016; Goodman, 2014; Gottfried, 2009, 2011a, 2011b, 2014; Gottfried et al., 2016; Morrissey et al., 2014; Ready, 2010). As consistent with prior research using categorical teacher response survey measures in ECLS-K (e.g., Gottfried et al., 2016; Guarino, Dieterle, Bargagliotti, & Mason, 2013), categorical responses were recoded as category midpoints except for the end categories which took the mode value of 0 or 20.

The second outcome was a binary indicator for whether the child was chronically absent as consistent with recent absence literature (Chang & Romero, 2008; Connolly & Olson, 2012; Gottfried, 2014, 2015; Gottfried et al., 2016; Morrissey et al., 2014). No absolute definition of chronic absenteeism exists (Gottfried, 2014). Prior research has deemed a child as being chronically absent after missing more than 2 weeks of the school year (Allensworth & Easton, 2007; Gottfried, 2014, 2015). This was the definition applied in this study; a binary indicator that identified students who were chronically absent defined as missing more than 10 days of kindergarten.

Note that all models also include a control measure for the date of the teacher spring survey. Controlling for survey administration date was necessary given that the date when the survey was administered to teachers in the spring varied. This would influence the number of reported absences.

School Bus

In the spring, NCES provided parents with a discrete set of choices to describe how their children most typically got to school each morning during that year (parents could only select one answer). Based on the response options for this survey question, a binary measure was created to indicate whether parents selected the response that their children took a “school bus” to school each morning. Other options were parent drives, carpool, walk, rides a bike/scooter, other. Note that it is in “other” where parents could designate public transportation options. Therefore, the measure of school bus-taking in this study pertained specifically to buses provided by a school system.

Approximately 24% of the sample commuted to school by school bus. In this study, the comparison was made between those who took the school bus each morning and those who did not take the school bus. The most common mode of getting to school in this non-bus-taking comparison group was being driven by parents (approximately 70% of the nonbus sample). Those who were driven in any car (parents, relatives, other) comprised approximately 86% of the nonbus sample. Therefore, the binary measure of taking the school bus was essentially comparing students who took the bus versus those who were driven in cars. Note that the largest category outside of bus and car was walking (approximately 8% of the sample). However, removing this 8% of observations from the control group did not change the results to follow. In addition, when running a model with walking as its own separate measure, walking was never statistically significant.

Control Variables

Table 1 also presents all control variables utilized in this study. ECLS-K:2011 provides a wide array of measures pertaining to the individual child as well as his or her family.

Child Demographic Characteristics

First, a common set of demographic characteristics were employed. These included the following variables: gender (female as the reference category), race (White as the reference category), English language learner (ELL) status based on home primary language, kindergarten entry age in months, and parent-rated health. For this latter measure, parents reported on child health in the fall of kindergarten on a scale of 1 to 5 with 1 being the highest. For this study, any child with a rating of 3, 4, or 5 was deemed to be rated as having poorer health. In addition, an indicator measure was included for whether a child had a disability.

Measured Skills at Kindergarten Entry

There were several ways to account for a child’s skills at kindergarten entry. First, in the fall of kindergarten, children were given direct assessments of their academic ability in both reading and math. Children’s item response theory–scaled scores on fall reading and math assessments were included as control measures in this study.

Next, teachers rated a child’s social-emotional skills across five measures. The details of these social-emotional skills are provided in the publicly available user’s manual for ECLS-K:2011. Here, a brief summary of these scales is provided. NCES based the teacher-rated social-emotional scales on the Social Skills Rating System (SSRS; Gresham & Elliott, 1990). Like the original scales, all scales in the ECLS-K:2011 study were continuous, with higher scores indicating more frequent behavior. All scales had high internal consistency, with the alpha reliability coefficients ranging from .79 to .91, as noted in the user’s manual (Tourangeau et al., 2013).

The four-item self-control scale measured the extent that the child was able to control his or her temper, respect others’ property, accept his or her peers’ ideas, and handle peer pressure. The five-item interpersonal skills scale measured the frequency by which a child was able to get along with others, form and maintain friendships, help other children, show sensitivity to the feelings of others, and express feelings, ideas, and opinions in positive ways. The seven-item approaches to learning scale measured the frequency that the child was able to keep his or her belongings organized, show eagerness to learn new things, adapt to change, persist in completing tasks, pay attention, and follow classroom rules. The five-item externalizing behaviors scale measured the frequency with which a child argued, fought, got angry, acted impulsively, and disturbed ongoing activities. The four-item internalizing behaviors scale measured the extent that the child exhibited anxiety, loneliness, low self-esteem, and sadness.

Finally, in the fall survey, parents rated the frequency with which their child expressed eagerness to attend school on a 3-point scale: never, once per week or less,more than once per week. A binary measure was created, indicating whether expressing eagerness occurred more than once per week.

Kindergarten and Prekindergarten Experiences

Several measures were included to account for kindergarten and prekindergarten experiences. The first four measures in this subsection of Table 1 pertain to the kindergarten experience. Binary measures indicated whether a child was in full-day kindergarten as well as whether this kindergarten was public.

This study included the distance in miles to kindergarten as well as the time it takes for the child to get to kindergarten. Information about distance to school was collected in nine categories: less 1/8 mile, 1/8 to 1/4 mile, more 1/4 mile to less 1/2 mile, 1/2 mile to less 1 mile, 1 to 2.5 miles, 2.6 to 5 miles, 5.1 to 7.5 miles, 7.6 to 10 miles, and 10.1 miles or more. The most common mileage from school was between 1 and 5 miles, which was approximately 45% of the sample. Next, minutes to school was coded as the three category midpoints of response choices: 7.5, 22.5, and 30 minutes.

Finally, parents were surveyed regarding the care that their child received outside of school during kindergarten, in the year before kindergarten (i.e., prekindergarten), and in the years prior to prekindergarten. Binary measures indicated whether a child attended a center-based care setting before/after school during the kindergarten year, center-based care (including Head Start) during prekindergarten, and center-based care in any year prior to prekindergarten. In addition, the number of hours of center-based care during kindergarten and prekindergarten was included in the analysis. These measures were included because prior research has explored both center-based kindergarten and prekindergarten care and its link to absenteeism (Gottfried, 2015).

Household Characteristics

The final section of Table 1 presents household, family, and parent characteristics. The set of household control measures can be loosely grouped into structure, environment, and SES. First, the measures of structure included an indicator for whether a child lived in a two-adult household (spouse or partner), a continuous number of siblings, an indicator for whether the child had an older sibling in the same school, and the age of the child’s mother when she first gave birth to any of her children. Second were measures of environment. These included the number of books in the home as well as reported maternal depression as recent research has suggested that maternal depression links to early absenteeism for children (Claessens, Engel, & Curran, 2015). Maternal depression was self-reported on how one felt over the past week with options ranging from never, some of the time, a moderate amount of the time, and most of the time.

Finally were measures of SES. Maternal and paternal education were included, with the category “high school graduate” serving as the omitted reference category. Household income was included, as was an indicator for whether a family received governmental meal assistance over the past 12 months. Indictors for whether the child attended a suburban or rural school were included, with attending school in an urban setting serving as the omitted reference category. In addition, the parent reported whether there were problems with safety when going to school, which has been included as a binary measure. A second neighborhood safety binary measure included here was based on parents reporting whether it was safe to play outside during the daytime. Maternal and paternal employment status was included, where not working was the omitted reference category. Last were measures of family routines, including an indicator for whether the parents reported their child going to bed at the same time each night as well as the number of breakfasts and dinners eaten at a regular time each week and the number of breakfasts and dinners eaten together by the family.

Analytic Approach

Baseline Model

A first model was employed to examine whether taking the bus to school was associated with differences in absenteeism for all students in the sample:

A_{i k} = β_{0} + β_{1} S B_{i} + β_{2} D_{i} + β_{3} E_{i} + β_{4} K P_{i} + β_{5} H_{i} + ε_{i k},

where A represents an absence outcome for child i in kindergarten k. The key measure in this study was SB_i, which represents the binary indicator for having taken the school bus to kindergarten. The other terms represent the control variables described in the preceding section: D_i (demographics), E_i (entry skills), KP_i (kindergarten and prekindergarten experiences), and H_i (household characteristics). The error term was adjusted for the fact that ECLS-K:2011 sampled children within the same school (White, 1980; Wooldridge, 2002). Recall that all models going forward also include spring survey assessment date, as mentioned previously.

Note that when the model presented in Equation 1 pertains to the outcome as the total number of absences, a standard linear regression was run. In this case, the coefficient on taking the bus to school could be interpreted as the number of days missed. When the model was run for chronic absenteeism, the outcome was binary and so the equation above represents a linear probability model.

District Variation

In addition to the unusually rich set of control variables included in the above model, there may still have been other factors—namely, at the level of the district—influencing the link between taking the school bus and absenteeism. One way to address this issue was by using school district fixed effects for the sample of children in public schools:

A_{i k} = β_{0} + β_{1} S B_{i} + β_{2} D_{i} + β_{3} E_{i} + β_{4} K P_{i} + β_{5} H_{i} + δ_{d} + ε_{i k},

In this model, $δ_{d}$ represents district fixed effects—that is, a set of indicator variables notating that a child i attended a public school in district d. District fixed effects may be particularly important in a study on bus-taking, as decisions about transportation can stem from district offices. These decisions entail not only if, but also the extent to which, a district offers school buses (i.e., route feasibility, distance cutoffs, vendor contracts). Therefore, district fixed effects help to control for district-to-district variation that exists in school bus availability and usage. Moreover, if one district had administrators who desired expansive bus policies as well as increased efforts to reduce school absences, then the prior baseline association between bus-taking and absenteeism might have been overestimated.

The importance of relying on the public school sample in this district fixed effects models was that public kindergartens may face a different set of transportation issues than private kindergartners, even within the same city. Therefore, all children in private kindergartens were removed. Also, while there is significant differentiation in private schools, public schools are supported as presenting a more uniform sample of students in national data (Wang, 2015).¹

Additional Robustness Checks

A model of school fixed effects was employed on the sample of public kindergartens. School fixed effects allowed the analysis to test for whether school-specific differences were biasing the association between bus-taking and absenteeism, if it were the case that the decision to offer bus services at a particular school was left up to the discretion of school-level administrators.

Next, returning to the full sample in a final test of robustness, propensity score matching was employed. Thus far, more precise associations between bus-taking and absenteeism were derived by relying on within-district or within-school samples. Nonetheless, within these samples, all children taking the bus were still being compared with all children who did not take the bus. A more restricted control group based on the propensity to having taken the bus might make for a more robust estimation. To run this propensity-matched design, the analysis occurred in two stages. In the first stage, the probability of taking the bus (i.e., the propensity score) was calculated for all children in the sample. The propensity score was estimated based on the set of control variables described in Table 1. Note that one required assumption when running propensity score matching on this sample was that the control variables were fairly time-invariant between being measured at kindergarten entry and the time period just before kindergarten when the decision about using school bus transportation was being made. Using logistic modeling (Rosenbaum & Rubin, 1983), each child was then assigned a propensity score for bus-taking.

In the second stage, children who did and did not take the bus were matched based on the propensity score assigned in the first stage. The matching method employed was one-to-one nearest-neighbor matching without replacement. The analytic sample was approximately n = 11,100. This reduction in sample size is consistent with prior research using propensity score matching in the kindergarten wave of the ECLS-K:2011 dataset (Gottfried, 2015). With this smaller-yet-matched sample, the standardized mean differences between groups on all variables to generate propensity matching analysis were reduced to |.10| or less, which is also as consistent with prior work in ECLS-K:2011 (Gottfried, 2015). In this way, the absence outcomes of children who took the bus were compared with those children who did not take the bus, but nonetheless displayed a similar propensity to have done so.

Results

Baseline Model

Table 2 presents the baseline results. The first column presents the model for the number of days absent. The second column presents the model for the probability of being chronically absent. In both models, standard errors clustered at the school are presented in parentheses.

Table 2

Bus Transportation and Absenteeism: Baseline Model

	Number of days absent	Chronic absenteeism
Bus to school	−0.39 (0.15)**	−0.03 (0.01)***
Child demographic characteristics
Male	−0.29 (0.09)**	−0.01 (0.01)
Black	−0.66 (0.19)***	−0.01 (0.01)
Hispanic	−0.18 (0.15)	−0.01 (0.01)
Asian	−0.08 (0.29)	0.02 (0.02)
Other	0.42 (0.21)*	0.04 (0.02)*
English language learner	−1.16 (0.21)***	−0.06 (0.01)***
Kindergarten entry age (months)	−0.02 (0.01)	0.00 (0.00)
Poorer health rating	0.90 (0.17)***	0.05 (0.01)***
Disability	0.34 (0.14)*	0.02 (0.01)*
Measured skills at kindergarten entry
Reading	−0.02 (0.01)*	0.00 (0.00)*
Math	−0.01 (0.01)	0.00 (0.00)
Self-control	0.13 (0.17)	0.01 (0.01)
Interpersonal skills	0.12 (0.15)	0.00 (0.01)
Approaches to learning	−0.62 (0.13)***	−0.03 (0.01)***
Internalizing problem behaviors	0.42 (0.11)***	0.03 (0.01)**
Externalizing problem behaviors	−0.40 (0.15)**	−0.03 (0.01)*
Eager to attend school	−0.20 (0.14)	0.00 (0.01)
Kindergarten and prekindergarten experiences
Full-day kindergarten	0.42 (0.17)*	0.02 (0.01)
Public kindergarten	−0.05 (0.23)	−0.01 (0.01)
Distance from school	0.03 (0.02)*	0.00 (0.00)*
Average minutes to school	0.01 (0.01)	0.00 (0.00)
Before/after school center care in kindergarten year	−0.60 (0.14)***	−0.04 (0.01)***
Hours of before/after school care in kindergarten year	0.01 (0.01)	0.00 (0.00)
Center-based prekindergarten care	−0.34 (0.12)**	−0.02 (0.01)*
Hours of center-based prekindergarten care	−0.01 (0.00)*	0.00 (0.00)
Out-of-home care prior to prekindergarten	0.11 (0.11)	0.00 (0.01)
Household characteristics
Two-adult household	−0.04 (0.21)	0.01 (0.02)
Number of siblings	−0.12 (0.05)*	−0.01 (0.00)*
Older sibling attends same school	−0.56 (0.13)***	−0.03 (0.01)**
Age of mother at first birth	−0.07 (0.01)***	0.00 (0.00)***
Number of children’s books at home	0.00 (0.00)	0.00 (0.00)
Mother-reported depression	0.35 (0.14)*	0.02 (0.01)*
Mother’s education
Less than high school	−0.04 (0.20)	0.01 (0.01)
Some college	−0.14 (0.14)	−0.01 (0.01)
College graduate or beyond	−0.27 (0.17)	−0.02 (0.01)
Father’s education
Less than high school	−0.15 (0.20)	−0.01 (0.01)
Some college	−0.19 (0.14)	−0.01 (0.01)
College graduate or beyond	−0.01 (0.17)	0.00 (0.01)
Household income	0.00 (0.00)	0.00 (0.00)
Received government food assistance	0.04 (0.13)	0.00 (0.01)
Urban	−0.02 (0.18)	−0.01 (0.01)
Rural	0.18 (0.16)	0.00 (0.01)
Problems with safety on school route	0.75 (0.42)	0.06 (0.03)
Safe to play outside	0.06 (0.13)	0.00 (0.01)
Maternal employment
Full-time	−1.08 (0.15)***	−0.06 (0.01)***
Part-time	−0.74 (0.13)***	−0.04 (0.01)***
Paternal employment
Full-time	−0.52 (0.17)**	−0.03 (0.01)*
Part-time	−0.27 (0.30)	−0.01 (0.02)
Regular bedtime	−0.64 (0.22)**	−0.04 (0.02)**
Number of breakfasts at regular time	0.04 (0.04)	0.00 (0.00)
Number of breakfasts together as family	0.00 (0.02)	0.00 (0.00)
Number of dinners at regular time	−0.03 (0.03)	0.00 (0.00)
Number of dinners together as family	0.02 (0.04)	0.00 (0.00)
N	14,370	14,370
R ²	.08	.05

Note. Standard errors adjusted for clustering in parentheses.

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

Beginning with the first column, children who took the school bus had fewer days absent compared with other children, controlling for all other measures. The coefficient on bus-taking translates to an approximate effect size (using standardized beta coefficients) of −0.10σ. In the second column, children who took the school bus were three percentage points less likely to be chronically absent. Regardless of how absenteeism was operationalized, the results link bus-taking to fewer absences. Note that the size of this effect for number of days absent is similar in size to intervention studies with absences as an outcome (see, for example, Avvisati, Gurgand, Guyon, & Maurin, 2013). The size of the coefficient in the linear probability model was also consistent with intervention work with absences as an outcome (see, for example, Kraft & Rogers, 2015). Finally, the size of these effects for chronic absence was also similar to other secondary data analysis examining school programs’ influence on chronic absenteeism (see, for example, Gottfried, 2015).

As a robustness check, the days absent model was rerun as an ordered logistic regression, and the chronic absenteeism model was run as a logistic regression model. The odds ratio for the former was 0.87 (p < .01). The odds ratio for the latter was .80 (p < .01). The interpretation for both was similar to the linear models: Children who took the bus to school were less absent than those who did not take the bus.

District Fixed Effects

Table 3 presents the results based on using district fixed effects models on the sample of public school students. The presentation in this table is analogous to that in Table 2. The first grouping presents findings for the measure of number of absent days. The second grouping presents findings for the probability of being chronically absent. Clustered errors are presented in parentheses.

Table 3

Bus Transportation and Absenteeism: School District Fixed Effects

	Number of days absent	Chronic absenteeism
Bus to school	−0.28 (0.10)**	−0.02 (0.01)*
Child demographic characteristics
Male	−0.28 (0.16)	−0.01 (0.01)
Black	−0.55 (0.20)**	0.00 (0.01)
Hispanic	−0.22 (0.16)	−0.01 (0.01)
Asian	0.03 (0.26)	0.04 (0.02)*
Other	0.19 (0.23)	0.02 (0.02)
English language learner	−1.10 (0.21)***	−0.06 (0.02)**
Kindergarten entry age (months)	−0.01 (0.01)	0.00 (0.00)
Poorer health rating	1.02 (0.18)***	0.05 (0.01)***
Disability	0.30 (0.14)*	0.02 (0.01)*
Measured skills at kindergarten entry
Reading	−0.01 (0.01)	0.00 (0.00)
Math	−0.01 (0.01)	0.00 (0.00)
Self-control	0.08 (0.17)	0.01 (0.01)
Interpersonal skills	−0.03 (0.16)	−0.01 (0.01)
Approaches to learning	−0.60 (0.13)***	−0.03 (0.01)**
Internalizing problem behaviors	0.37 (0.12)**	0.02 (0.01)*
Externalizing problem behaviors	−0.48 (0.14)***	−0.03 (0.01)*
Eager to attend school	−0.21 (0.15)	−0.01 (0.01)
Kindergarten and prekindergarten experiences
Full-day kindergarten	−0.05 (0.29)	0.00 (0.02)
Distance from school	0.03 (0.02)	0.00 (0.00)
Average minutes to school	0.01 (0.01)	0.00 (0.00)
Before/after school center care in kindergarten year	−0.55 (0.14)***	−0.03 (0.01)**
Hours of before/after school care in kindergarten year	0.01 (0.01)	0.00 (0.00)
Center-based prekindergarten care	−0.31 (0.13)*	−0.02 (0.01)
Hours of center-based prekindergarten care	−0.01 (0.00)	0.00 (0.00)
Out-of-home care prior to prekindergarten	0.11 (0.12)	0.00 (0.01)
Household characteristics
Two-adult household	−0.06 (0.22)	0.00 (0.02)
Number of siblings	−0.11 (0.06)	−0.01 (0.00)
Older sibling attends same school	−0.55 (0.14)***	−0.03 (0.01)*
Age of mother at first birth	−0.06 (0.01)***	0.00 (0.00)***
Number of children’s books at home	0.00 (0.00)	0.00 (0.00)
Mother-reported depression	0.25 (0.14)	0.01 (0.01)
Mother’s education
Less than high school	0.06 (0.21)	0.02 (0.01)
Some college	−0.15 (0.14)	−0.01 (0.01)
College graduate or beyond	−0.30 (0.18)	−0.02 (0.01)
Father’s education
Less than high school	−0.08 (0.20)	0.00 (0.01)
Some college	−0.11 (0.15)	−0.01 (0.01)
College graduate or beyond	0.05 (0.18)	0.01 (0.01)
Household income	0.00 (0.00)	0.00 (0.00)
Received government food assistance	0.13 (0.13)	0.01 (0.01)
Urban	−0.49 (0.28)	−0.02 (0.02)
Rural	0.26 (0.26)	0.01 (0.02)
Problems with safety on school route	0.78 (0.45)	0.06 (0.03)
Safe to play outside	−0.01 (0.14)	0.00 (0.01)
Maternal employment
Full-time	−1.07 (0.16)***	−0.06 (0.01)***
Part-time	−0.74 (0.14)***	−0.04 (0.01)***
Paternal employment
Full-time	−0.44 (0.18)*	−0.02 (0.01)
Part-time	−0.12 (0.31)	0.00 (0.02)
Regular bedtime	−0.59 (0.22)**	−0.04 (0.02)*
Number of breakfasts at regular time	0.03 (0.04)	0.00 (0.00)
Number of breakfasts together as family	−0.02 (0.02)	0.00 (0.00)
Number of dinners at regular time	−0.05 (0.03)	0.00 (0.00)
Number of dinners together as family	0.02 (0.04)	0.00 (0.00)
n	12,390	12,390
R ²	.21	.15

Note. Standard errors adjusted for clustering in parentheses.

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

Overall, there was a great deal of consistency in the interpretation of the findings between the baseline and district fixed effects models. However, once controlling for between-district differences and including only children in public kindergarten, the magnitudes of the coefficients have been tempered. Children who took the bus to kindergarten had −0.28 fewer absences than children who did not take the bus to school with a slightly lower standardized beta of −0.06σ. In addition, children who took the bus to school now had a two percentage point lower likelihood of being chronically absent.

That said, when comparing Tables 2 and 3, similar conclusions emerged. Children who took the bus to kindergarten had fewer total missed days of school and lower likelihood of being chronically absent. This remained the case even after controlling for a large span of covariates, including a clustered error, and now accounting for differences across districts. However, the tempered coefficients in the district fixed effects models for public school students suggested that between-district differences or including the sample of private school students might have caused some overestimation in the previous findings in the baseline model. Although controlling away for these between-district differences does provide a more robust estimate of bus-taking, it is important to keep in mind that there appears to be factors and processes at the district level that matter for how bus-taking links to absenteeism that may merit future investigation.

Even when using a within-district analysis, it could have been possible that the no-bus group within a single district was comprised of students who lived extremely close to school and had no access to a bus or lived far and opted out of taking the bus. Although specific walk-zone cutoffs were not provided in the dataset, one specification check was conducted where those students in the top and bottom 10% deciles of the distance-to-school variable were removed from each district. When removing either the top decile, bottom decile, or both, the results presented in Table 3 were robust to any of these trimmed samples.

Additional Robustness Checks

Several additional tests were run to confirm the robustness of these findings. As mentioned, these tests included school fixed effects and propensity score matching results.

In Table 4, the coefficient on taking the bus to school is presented for all approaches in this study: baseline (relying on within-sample variation), district fixed effects (relying on within-district variation for kindergartners in public schools), school fixed effects (relying on within-school variation for kindergartners in public schools), and propensity score matching (for all children in the sample). Note that each cell represents the bus-taking coefficient from unique regression that includes all control variables as well as cluster-adjusted standard errors.

Table 4

Additional Tests of Robustness

Outcome: Number of days absent
Baseline sample (full sample)	−0.39 (0.15)**
District fixed effects (public schools)	−0.28 (0.10)**
School fixed effects (public schools)	−0.32 (0.13)**
Propensity matched sample (full sample)	−0.26 (0.10)**
Outcome: Chronic absenteeism
Baseline sample (full sample)	−0.03 (0.01)***
District fixed effects (public schools)	−0.02 (0.01)*
School fixed effects (public schools)	−0.03 (0.01)**
Propensity matched sample (full sample)	−0.02 (0.01)**

Note. Standard errors adjusted for clustering in parentheses.

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

Across all models, children who took the bus to school had fewer absences than did children who did not go to school by bus as well as a lower likelihood of being chronically absent. The sizes of the days absent coefficients were quite similar between the district and school fixed effects models, though the coefficient on days absent was slightly lower in the propensity score matching model. Nonetheless, the coefficients from all tests of robustness were similar in size and lower than what the baseline model had suggested. The coefficient from the chronic absence model was similar across all tests of robustness as well and was lower than the baseline model. Therefore, it appears that any degree of overestimation in the association between bus-taking and child absences occurred when not having accounted for these alternative sources of bias.

Group Differences

A final analysis examined differences in the size of the coefficient on taking the bus based on key differences. Prior research as well as the findings for the control measures in this study suggest that differences in absenteeism may arise due to health, center-based care experiences, SES, or travel distances (Allen, 2003; Gottfried, 2010, 2015; Nauer, Mader, Robinson, & Jacobs, 2014). Therefore, this final approach evaluated whether there were distinctions in the nature of the previously established associations based on differences in these key factors (see Table 5).

Table 5

Differences by Key Characteristics

	Number of days absent	Chronic absenteeism
Bus to school × Health
Poorer health rating	0.14 (0.36)	−0.01 (0.03)
Disability	−0.18 (0.25)	−0.01 (0.02)
Bus to School × Commute and Care
Distance to school	0.00 (0.04)	0.00 (0.00)
Minutes to school	0.00 (0.02)	0.00 (0.00)
Attended before/after center-based care in kindergarten	0.20 (0.31)	0.01 (0.02)
Attended before/after center-based care in prekindergarten	0.11 (0.25)	0.02 (0.02)
Bus to School × Household Characteristics
Older sibling attends same school	0.05 (0.22)	0.00 (0.02)
Received government food assistance	0.11 (0.23)	0.01 (0.02)
Urban	0.25 (0.29)	0.02 (0.02)
Rural	−0.62 (0.28)*	−0.04 (0.02)*
Problems with safety on school route	−0.70 (0.92)	−0.03 (0.07)
Safe to play outside	0.13 (0.25)	0.00 (0.02)
Mother works full-time	0.18 (0.24)	0.01 (0.02)

Note. Standard errors adjusted for clustering in parentheses. Each cell represents the estimate of bus-taking from a unique regression.

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

To test for these effects, partially interacted models were run in which an interaction between bus-taking and the individual characteristic was included in the model. In Table 5, each cell is a unique regression. The estimate is the coefficient on the interaction of taking the school bus with the characteristic in the row, which was derived from a district fixed effects model with cluster-adjusted errors. All models included all covariates from prior models. All models also include the coefficients from the main effects of the interaction (i.e., bus and the row term). For the sake of clarity, these were not presented in the table, although they are available upon request.

There was almost no differentiation. Therefore, the association between bus-taking and absenteeism was experienced more widespread by all students in the sample as opposed to by specific groups of students or students within certain types of families. One difference that potentially has emerged was for rural students. Children in rural areas were more influenced from taking the school bus. This is evidenced in Table 5 by the statistically significant coefficients for the interaction between bus-taking and the indicator for rural.

Interestingly, several interactions that did not emerge as statistically significant were noteworthy. First, the findings on absenteeism were not differentiated by SES when measured by poverty, as consistent with prior research on absences in early childhood (Gottfried, 2015). Second, distance to school did not differentiate the influence that taking the school bus had on absence outcomes. The distance interaction model was rerun where distance in miles was replaced by an indicator for an above-average distance from school. However, this interaction remained nonsignificant. In this study, there is no evidence that students who lived particularly far from school were especially helped by taking the bus. Note that minutes to school was also tested in this way, and it remained nonsignificant. The evidence here was descriptive and based on categorical measures of both distance and time; therefore, further work with more detailed data and a quasi-experimental design (e.g., distance cutoffs for transportation eligibility) may provide additional insight.

Discussion

Taking a school bus to school is a ubiquitous practice in the United States and has been engrained in our nation’s school-going history over the past century. Since the establishment of school bus programs in the early 20th century, there has been much dialogue surrounding school bus safety and transportation standards (National Association of State Directors of Public Transportation Services, 2000). Surprisingly, less dialogue surrounds how taking the school bus actually serves the fundamental purpose of getting children to school. As mentioned in the introduction, of the five key school bus “Issues” featured on the website for the American School Bus Council (n.d.), only one of them (“Access to Learning”) raises the point that taking the school bus might be an effective way to help children not miss school. With this purpose buried deep within a rhetoric engulfed by talk of safety standards and commuting traffic, it is unsurprising that little attention has been paid to the student-level benefits of taking the bus. However, this is problematic: Little acknowledgment of these benefits of bus-taking may have spurred districts’ and schools’ actions to cut or reduce bus capacity (or consider doing so) as a way to address budget constraints. Parents have expressed concern and outrage as a result: They claim to be losing a key support helping to ensure their children can get to school.

Simultaneously, a call for reducing school absenteeism continues to gain momentum in policy and practice. Thus, identifying key supports that may help children get to school has become even more relevant in research. No known prior study has examined the extent to which taking the bus might be an effective strategy of helping children get to school. Therefore, for the first time in the field, this study inquired into whether taking the school bus was associated with reduced absence behavior.

All findings suggested that children who took the school bus had fewer absences and lower likelihoods of being chronically absent. Prior research in absenteeism has concluded that students with more absences have fewer opportunities to learn in school and perform more poorly on exams as a consequence of missing school (Chen & Stevenson, 1995; Nichols, 2003). Hence, if taking the bus lowers absenteeism, then access to this resource may benefit students in ways that have implications for individual learning and, ultimately, academic success. There is also the potential for aggregate effects. Districts may benefit by mitigating the need for remedial activities, as fewer absent students implies fewer missed opportunities to learn at school. In this way, bus-taking may indirectly benefit aggregate school performance.

Given these findings and potential benefits, the positive link between bus-taking and fewer absences raises several key implications. First, bus-taking appears to matter, and this has been a relatively unexplored school resource as it relates to absenteeism. This is surprising given the number of students utilizing the school bus system nationally. Because bus-taking may potentially benefit students and districts as mentioned above, this study urges for districts to consider how the usage of buses might add value as it relates to reducing absenteeism, rather than simply considering bus transit as an expendable component of the budget. This study also urges school bus organizations to reconsider how access to learning and mitigating absenteeism can better stand out as a priority. Burying this deep in rhetoric and marketing may inadvertently send the wrong message about the lack of educational importance of school bus services.

Second, at this point, little is known about families and bus usage, and thus districts might make a more concerted effort to examine who is taking the bus and if providing the service justifies the expense. In this vein, this study also urges for further research to explore bus-attendance issues through a cost–benefit analysis, considering both the costs of providing transportation and having the benefits of fewer student absences, versus the benefits that the district gains from cutting bus service and having to bear the costs of higher student absenteeism. Districts must assess whether cutting bus services is an effective way to manage budgets, and, vice versa, whether bus services is indeed a cost-effective way to improve attendance, learning, and test performance.

Finally, related to the issue of which families do and do not have access to buses is the concern about which districts in particular are cutting these services. Prior research has found that students in lower income districts have disproportionately higher absence rates (Chang & Romero, 2008; Nauer et al., 2014). It is in lower income districts where services are being cut, and it is also in lower income districts where there are fewer school-based transition practices being employed in kindergarten (Little et al., 2016). Therefore, if resource-constrained districts have greater absence problems and fewer transitions practices put in place and are, at the same time, cutting the very services like buses that may help to address this issue, then the students who require the greatest supports are losing them. Given the absence–achievement link, cutting off bus services in high-needs districts may ultimately widen achievement gaps between lower and higher income districts. Another gap-widening concern is grounded in the fact that many courts have ruled that charter schools do not need to offer bus services. In the expansion of school choice programs in high-needs districts, if students and families choose schools that eventually cut bus services, then this adds an additional school-going challenge that these families must then face.

Limitations and Further Research

In addition to revealing important issues worth probing more deeply, this study also highlights methodological concerns for future analyses. First, because the measure of absences was derived from the total number of missed school days reported by the teacher, there was no detail on the timing or types of absences that each child incurred. Therefore, it was not possible to determine whether bus-taking was linked to reducing excused versus unexcused absences or a reduction in absence spells (i.e., many absences in a row). Using district-level data merged with bus-taking data would allow for these details to be examined, including whether bus-taking affects fall absences more than spring absences in kindergarten. It was also not possible to determine a student’s full year’s sum of absences, as the teacher surveys were given in the spring and not necessarily at the end of the school year.

Similarly, there were also no data on student tardies, and this study could not analyze how bus-taking (or lack thereof) might be linked to differences in arriving to school on time. Having these additional details would certainly provide further insights into the mechanisms between bus-taking and missing school. Other datasets that contain detailed absence or tardy information often lack contextual data, such as administrative data; however, there may be some added benefit to addressing the link between bus-taking and differentiated forms of absenteeism even if contextual data and a wide span of control variables are not available.

Second, although this was the first to examine the effect of bus-taking on student outcomes, the approach was nonetheless descriptive. Although the methodological approaches taken in this study do suggest a degree of robustness in the findings, they only adjusted for measured variables and may have missed unmeasured common causes that might have biased the estimates. Thus, it remains important for future research in this area to focus on an analytic strategy where causality can be further explored through the evaluation of specific bus policies. For instance, with an appropriate district dataset that includes information on walk zones as well as precise distance from home, a future study could employ a regression discontinuity design, which would be particularly valuable when there have been changes to busing policies within a district. In this way, it would be possible to compare students who live equally as close to school with one subsection of the sample living just beyond the walking-distance cutoff requirement, thereby making the comparison between the bus-taking group and no-bus group even more robust. More so, by examining different busing policies as a result of policy changes within a single district or by comparing policies in different districts, future analyses can generate conclusions about which bus policy structure (e.g., walk-zone distances, direct vs. circulator routes) might be most effective for reducing absenteeism.

Finally, this issue of causality could also be pushed further if the estimates of bus-taking were to be derived from an experimental design. Future research might consider a randomized trial of transportation to school or an intervention based on increasing family logistics as it pertains to absenteeism. Although many previously studied factors of absenteeism cannot be randomly assigned such as SES, offering various school transportation services is a factor that might lend itself to experimentation. The findings of this study, then, could be corroborated with future experimental research to test the generalizability of these findings. Much like bus-taking is a potential factor of absences that is malleable for policymakers and practitioners, there is also potential here for it to be malleable for researchers. Hence, the potential is great to further develop how getting to school can reduce going to school for our rising cohorts of U.S. schoolchildren.

Footnotes

Declaration of Conflicting Interests

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

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by a grant from the Stuart Foundation.

1.

Note that counties tend to be diverse and can contain multiple school districts, each of which could potentially be enacting its own transportation policies. When it comes to bus-taking availability, it would be surmised that district fixed effects may be a more precise way to test model robustness. As an ancillary test, county fixed effects were employed. The results were almost identical to the baseline model, thereby suggesting that controlling for counties may have little influence on the results given the potential degree of within-county variation. This provides further motivation for relying on district, rather than county, fixed effects.

Author

Michael A. Gottfried, PhD, is an associate professor in the Gevirtz Graduate School of Education at University of California, Santa Barbara, Santa Barbara, CA, USA. His research interests pertain to the economics of education and education policy, in areas around truancy, early school settings, STEM (science, technology, engineering, and mathematics), and students with disabilities.

References

Alexander

K. L.

Entwisle

D. R.

Horsey

C. S.

(1997). From first grade forward: Early foundations of high school dropout. Sociology of Education, 70, 87–107.

Allan

N. P.

Hume

L. E.

Allan

D. M.

Farrington

A. L.

Lonigan

C. J.

(2014). Relations between inhibitory control and the development of academic skills in preschool and kindergarten: A meta-analysis. Developmental Psychology, 50, 2368–2379.

Allen

(2003). The impact of elementary school nurses on student attendance. The Journal of School Nursing, 19, 225–231. doi:10.1177/10598405030190040801

Allensworth

E. M.

Easton

J. Q.

(2007). What matters for staying on-track and graduating in Chicago public high schools. Chicago, IL: University of Chicago Consortium on Chicago School Research.

American School Bus Council. (n.d.). Available from http://www.americanschoolbuscouncil.org/

Avvisati

Gurgand

Guyon

Maurin

(2013). Getting parents involved: A field experiment in deprived schools. Review of Economic Studies, 81, 57–83.

Balfanz

Byrnes

(2012). The importance of being there: A report on absenteeism in the nation’s public schools. Baltimore, MD: Johns Hopkins University Center for Social Organization of Schools.

Bassok

Latham

Rorem

(2016). Is kindergarten the new first grade? AERA Open, 2(1). doi:10.1177/2332858415616358

Blair

(2002). School readiness: Integrating cognition and emotion in a neurobiological conceptualization of children’s functioning at school entry. American Psychologist, 57, 111–127.

10.

Bogart

G. A.

Jones

J. W.

Jason

L. A.

(1980). School transitions: Preventive intervention following an elementary school closing. Journal of Community Psychology, 8, 343–352.

11.

Broadhurst

Paton

May-Chahal

(2005). Children missing from school systems: Exploring divergent patterns of disengagement in the narrative accounts of parents, carers, children and young people. British Journal of Sociology of Education, 26, 105–119.

12.

Chandler

M. A.

(2015, August 21). District children lose the buses that took them to high-performing schools. The Washington Post. Retrieved from https://www.washingtonpost.com/local/education/district-families-lose-buses-to-higher-performing-schools/2015/08/21/7903f11e-476d-11e5-8e7d-9c033e6745d8_story.html

13.

Chang

H. N.

Davis

(2015). Mapping the early attendance gap. Retrieved from http://www.attendanceworks.org/wordpress/wp-content/uploads/2015/07/Mapping-the-Early-Attendance-Gap-Final-4.pdf

14.

Chang

H. N.

Romero

(2008). Present, engaged, and accounted for. The critical importance of addressing chronic absence in the early grades. New York, NY: National Center for Children in Poverty, Mailman School of Public Health at Columbia University.

15.

Chen

Stevenson

H. W.

(1995). Motivation and mathematics achievement: A comparative study of Asian-American, Caucasian-American, and East Asian high school students. Child Development, 66, 1215–1234.

16.

Claessens

Engel

Curran

F. C.

(2015). The effects of maternal depression on child outcomes during the first years of formal schooling. Early Childhood Research Quarterly, 32, 80–93.

17.

Connell

J. P.

Spencer

M. B.

Aber

J. L.

(1994). Educational risk and resilience in African-American youth: Context, self, action, and outcomes in school. Child Development, 65, 493–506.

18.

Connolly

Olson

L. S.

(2012). Early elementary performance and attendance in Baltimore City schools’ pre-kindergarten and kindergarten. Baltimore, MD: Baltimore Education Research Consortium.

19.

Downey

D. B.

Condron

D. J.

(2004). Playing well with others in kindergarten: The benefit of siblings at home. Journal of Marriage and Family, 66, 333–350.

20.

Ekstrom

R. B.

Goertz

M. E.

Pollack

J. M.

Rock

D. A.

(1986). Who drops out of high school and why? Findings from a national study. Teachers College Record, 87, 356–373.

21.

Fiese

B. H.

Tomcho

T. J.

Douglas

Josephs

Poltrock

Baker

(2002). A review of 50 years of research on naturally occurring family routines and rituals: Cause for celebration? Journal of Family Psychology, 16, 381–390.

22.

Finn

J. D.

(1993). School engagement and students at risk. Washington, DC: U.S. Department of Education.

23.

Gershenson

Jacknowitz

Brannegan

(2016). Are student absences worth the worry in U.S. primary schools? Education Finance and Policy. Retrieved from http://www.mitpressjournals.org/doi/abs/10.1162/EDFP_a_00207#.WL8CIfkrJdg

24.

Giallo

Treyvaud

Matthews

Kienhuis

(2010). Making the transition to primary school: An evaluation of a transition program for parents. Australian Journal of Educational & Developmental Psychology, 10, 1–17.

25.

Gill

Winters

Friedman

(2006). Educators’ views of pre-kindergarten and kindergarten readiness and transitions practices. Contemporary Issues in Early Childhood, 7, 213–227.

26.

Goodman

(2014, June). Flaking out: Student absences and snow days as disruptions of instructional time (Working Paper No. 20221). Cambridge, MA: National Bureau of Economic Research.

27.

Gottfried

M. A.

(2009). Excused versus unexcused: How student absences in elementary school affect academic achievement. Educational Evaluation and Policy Analysis, 31, 392–415. doi:10.3102/0162373709342467

28.

Gottfried

M. A.

(2010). Evaluating the relationship between student attendance and achievement in urban elementary and middle schools: An instrumental variables approach. American Educational Research Journal, 47, 434–465. doi:10.3102/0002831209350494

29.

Gottfried

M. A.

(2011a). Absent peers in elementary years: The negative classroom effects of unexcused absences on standardized testing outcomes. Teachers College Record, 113, 1597–1632.

30.

Gottfried

M. A.

(2011b). The detrimental effects of missing school: Evidence from urban siblings. American Journal of Education, 117, 147–182.

31.

Gottfried

M. A.

(2014). Chronic absenteeism and its effects on students’ academic and socioemotional outcomes. Journal of Education for Students Placed at Risk, 19, 53–75. doi:10.1080/10824669.2014.962696

32.

Gottfried

M. A.

(2015). Can center-based childcare reduce the odds of early chronic absenteeism? Early Childhood Research Quarterly, 32, 160–173.

33.

Gottfried

M. A.

(in press). Chronic absenteeism in the classroom context: Effects on achievement. Urban Education.

34.

Gottfried

M. A.

Egalite

Kirksey

J. J.

(2016). Does the presence of a classmate with emotional/behavioral disabilities link to other students’ absences in kindergarten? Early Childhood Research Quarterly, 36, 506–520.

35.

Graham

J. W.

Olchowski

A. E.

Gilreath

T. D.

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

36.

Gresham

F. M.

Elliot

S. M.

(1990). Social skills rating system. Circle Pines, MN: American Guidance Service.

37.

Guarino

Dieterle

S. G.

Bargagliotti

A. E.

Mason

W. M.

(2013). What can we learn about effective early mathematics teaching? A framework for estimating causal effects using longitudinal survey data. Journal of Research on Educational Effectiveness, 6, 164–198. doi:10.1080/19345747.2012.706695

38.

Hallfors

Vevea

J. L.

Iritani

Cho

Khatapoush

Saxe

(2002). Truancy, grade point average, and sexual activity: A meta-analysis of risk indicators for youth substance use. The Journal of School Health, 72(5), 205–211.

39.

Holland

J. V.

Kaplan

D. M.

Davis

S. D.

(1974). Interschool transfers: A mental health challenge. Journal of School Health, 44(2), 74–79.

40.

Jacobson

(2008). Absences in early grades tied to learning lags. Education Week, 28, 1–2.

41.

Kane

(2006). School exclusions and masculine, working-class identities. Gender and Education, 18, 673–685.

42.

Kerr

Price

Kotch

Willis

Fisher

Silva

(2012). Does contact by a family nurse practitioner decrease early school absence? Journal of School Nursing, 28, 38–46.

43.

Kraft

Rogers

(2015). The underutilized potential of teacher-to-parent communication: Evidence from a field experiment. Economics of Education Review, 47, 49–63.

44.

Ladd

G. W.

Price

J. M.

(1987). Predicting children’s social and school adjustment following the transition from preschool to kindergarten. Child Development, 58, 1168–1189.

45.

Little

M. H.

Cohen-Vogel

Curran

F. C.

(2016). Facilitating the transition to kindergarten: What ECLS-K data tell us about school practices then and now. AERA Open, 2(3), 1–18.

46.

Lulay

(2015, July 24). CPS cutting 270 bus stops, forcing some kids to walk 1.5 miles to get ride. DNA Info. Retrieved from https://www.dnainfo.com/chicago/20150724/near-west-side/cps-cut-bus-stops-by-more-than-half-save-money-this-year

47.

Marcotte

D. E.

Hemelt

S. W.

(2008). Unscheduled closures and student performance. Education Finance and Policy, 3, 316–338.

48.

Martin

Carlson

(2005). Barriers to children walking to or from school—United States, 2004. Journal of the American Medical Association, 54, 949–952.

49.

McIntyre

L. L.

Eckert

T. L.

Fiese

B. H.

DiGennario

F. D.

Wildeneger

L. K.

(2007). The transition to kindergarten: Family experiences and involvement. Early Childhood Education Journal, 35, 83–88.

50.

Morrissey

T. W.

Hutchison

Winsler

(2014). Family income, school attendance, and academic achievement in elementary school. Developmental Psychology, 50, 741–753.

51.

National Association of State Directors of Pupil Transportation Services. (2000). Transporting the Nation’s School Children [Position paper]. Retrieved from Retrieved from http://www.nasdpts.org/Documents/Paper-Transit.pdf

52.

Nauer

Mader

Robinson

Jacobs

(2014). A better picture of poverty: What chronic absenteeism and risk load reveal about NYC’s lowest-income elementary schools. New York: Center for New York City Affairs.

53.

Newmann

F. M.

(1981). Reducing student alienation in high schools: Implications of theory. Harvard Educational Review, 51, 546–564.

54.

Nichols

J. D.

(2003). Prediction indicators for students failing the state of Indiana high school graduation exam. Preventing School Failure: Alternative Education for Children and Youth, 47, 112–120. doi:10.1080/10459880309604439

55.

Office of the California Attorney General. (2015). In school + on track: Attorney General’s 2015 report on California’s elementary school truancy & absenteeism crisis. Sacramento, CA: Author.

56.

Oglesby

(2015, August 4). Brick parents panicked over private school busing cuts. Asbury Park Press. Retrieved from http://www.app.com/story/news/education/in-our-schools/2015/08/04/brick-private-school-busing-cuts/31130685/

57.

Ollendick

T. H.

Mayer

J. A.

(1984). School phobia. In Turner

S. M.

(Ed.), Behavioral theories and treatment of anxiety (pp. 367–411). Boston, MA: Springer.

58.

Pianta

R. C.

Cox

M. J.

(2002). Transition to kindergarten. Early Childhood Research and Policy Briefs, 2(2). https://www.gpo.gov/fdsys/pkg/ERIC-ED463897/pdf/ERIC-ED463897.pdf

59.

Pianta

R. C.

Kraft-Sayre

(1999). Parents’ observations about their children’s transitions to kindergarten. Young Children, 54(3), 47–52.

60.

Ponitz

C. C.

McClelland

M. M.

Matthews

J. S.

Morrison

F. J.

(2009). A structured observation of behavioral self-regulation and its contribution to kindergarten outcomes. Developmental Psychology, 45, 605–619.

61.

Quas

Murowchick

Bensadoun

Boyce

(2002). Predictors of children’s cortisol activation during the transition to kindergarten. Developmental and Behavioral Pediatrics, 23, 304–313.

62.

Ready

D. D.

(2010). Socioeconomic disadvantage, school attendance, and early cognitive development: The differential effects of school exposure. Sociology of Education, 83, 271–286. doi:10.1177/0038040710383520

63.

Redmond

(2016, July 22). Cuts to Ramapo school busing anger parents. Lohud: The Journal News. Retrieved from http://www.lohud.com/story/news/local/rockland/ramapo/2016/07/21/ramapo-central-busing-anger-parents/87349742/

64.

Rhoulac

(2005). Bus or car? The classic choice in school transportation. Transportation Research Record: Journal of the Transportation Research Board, 1922, 98–104.

65.

Romero

Lee

Y.-S.

(2007). A national portrait of chronic absenteeism in the early grades. New York, NY: National Center for Children in Poverty.

66.

Rosenbaum

P. R.

Rubin

D. B.

(1983). The central role of the propensity score in observational studies for causal effects. Biometrika, 70(1), 41–55. doi:10.2307/2335942

67.

Royston

(2004). Multiple Imputation of Missing Values. The Stata Journal, 4, 227-241.

68.

Safe Routes to School. (n.d.). School bus funding: Cuts and hazard/courtesy busing. Retrieved from http://www.saferoutespartnership.org/state/bestpractices/schoolbuscuts

69.

Shore

(1998). Ready schools. Washington, DC: National Education Goals Panel.

70.

Smith

(2015). Florida Judge Rules Charter Schools don’t have to provide bus service. Retrieved from http://www.americanbussales.net/florida-judge-rules-charter-schools-dont-provide-bus-service/

71.

Tourangeau

Nord

Sorongon

A. G.

Hagerdorn

M. C.

Daly

Najarian

(2013). User’s manual for the ECLS-K:2011 kindergarten data file and electronic codebook (NCES 2013-061). Washington, DC: National Center for Education Statistics.

72.

U.S. Department of Education, National Center for Education Statistics. (2016). Digest of education statistics, 2014 (NCES 2016-006). Washington, DC: Author.

73.

Wang

(2015). Pathway to a baccalaureate in STEM fields: Are community colleges a viable route and does early STEM momentum matter? Educational Evaluation and Policy Analysis, 37, 376–393.

74.

White

(1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica, 48, 817–838.

75.

Wildeneger

L. K.

McIntyre

L. L.

Fiese

B. H.

Eckert

T. L.

(2008). Children’s daily routines during Kindergarten transition. Early Childhood Education Journal, 36, 69–74.

76.

Wolin

S. J.

Bennett

L. A.

(1984). Family rituals. Family Process, 23, 401–420.

77.

Wooldridge

J. M.

(2002). Econometric analysis of cross section and panel data. Cambridge, MA: The MIT Press.