Between-Year and Within-Year School Mobility: Different Effects by Race/Ethnicity

Abstract

The current study investigated the effects of school mobility on the academic achievement of different racial/ethnic groups in four cohorts of students from a very large urban school district. In this study, I compared within-year and between-year mobility and, most importantly, account for all the schools students attended over the study period. Using a multiple membership model (MMM), the findings confirmed that, for all student groups, academic achievement was affected more by within-year school mobility than between-year school mobility. Black students had the highest mobility rates, both for between- and within-year mobility. Although Asian-American students achieved higher reading and math scores on average, they were more negatively impacted by within-year school mobility compared to other groups. The current study was able to pinpoint the students most at risk for negative outcomes following within-year mobility. The findings are discussed in the context of policy recommendations that can be adopted by school districts.

Keywords

school mobility between-year within-year race/ethnicity

Introduction

The United States is a country constantly on the move—almost one in six households change residence annually—a quarter of which are accompanied by school changes (Gasper et al., 2010). Scholars studying school mobility have paid special attention to mobility’s effects on student academic achievement. While a handful of scholars find no relationship between school mobility and academic outcomes (Alexander et al., 1996; Heinlein & Shinn, 2000), the preponderance of research suggests that there is a negative association between school mobility and student academic achievement (Burkam et al., 2009; Engec, 2006; Grigg, 2012; Hanushek et al., 2004; Kerbow, 1996). However, mobility might not affect all students in a negative way. Despite the generally adverse impacts of school mobility on academic achievement, students could achieve better academic outcomes if they transfer to schools that are higher quality or satisfy their needs (Hanushek et al., 2004; Holme and Richards, 2009; Schwartz et al., 2012; Xu et al., 2009).

Additionally, a distinction has been made between within-year (also called intra-year) school mobility and between-year (also called inter-year) mobility (Burkam et al., 2009; Grigg, 2012; Hanushek et al., 2004). Within-year mobility refers to changing schools during the academic year; between-year mobility refers to changing schools between academic years, often occurring during the summer. The timing of school mobility is often differentiated because the distinction matters to children’s school outcomes (Grigg, 2012; Hanushek et al., 2004). Specifically, school changes during the academic year tend to be “reactive moves.” Thus, within-year mobility is more likely to be disruptive to students’ studies than between-year mobility over the summer (Hanushek et al., 2004).

It is also important to note that not all racial/ethnic groups bear the burden of school mobility equally (Alexander et al., 1996; Hanushek et al., 2004). Racial minority students, especially African Americans, have particularly high school mobility rates compared to white students. School mobility for black students is often accompanied by parental divorce or financial problems (Rumberger, 2015). However, few studies have simultaneously considered the timing of school mobility and race/ethnicity heterogeneity when examining school mobility effects on students’ academic attainment (see Hanushek et al., 2004 for an exception). In spite of Hanushek et al.’s (2004) illuminating work, even they have neglected school transitions among Asian Americans, perhaps because many surveys fail to include Asian Americans in their sampling scope.

Using CISD¹ data, the current study aims to distinguish between the effects of within- and between-year mobility on students’ reading and math scores in a state-required test for grades 3-8. In addition to the timing distinction, I sought to examine how different racial/ethnic groups weather within- and between-year mobility differently. Thus, my study contributes to previous work on school mobility in several ways. First, I use multiple membership models to account for all schools attended by mobile students, rather than solely considering the most recent school. Second, I include Asian American students—a group often neglected by scholars—and estimate whether the effects of each type of school mobility vary between racial/ethnic groups. Third, I use data collected from one of the 20 largest school districts (out of more than 14,000 nationwide) with a relatively high student mobility rate. Despite the large size of the school district and the high mobility rate, mobility has rarely been studied in these types of contexts. Further, research to date has often focused on mostly white student school districts. Mobility analyses based upon this particular school district will provide unique insights because it is comprised of predominantly minority students and hence might exhibit different patterns of academic performance compared to school districts dominated by white students.

Literature

School Mobility

School mobility occurs when a student moves to a different school. School changes can be classified as promotional or non-promotional school moves (Fiel et al., 2013; Mao, 1997; Xu et al., 2009). Promotional school moves are school changes required for all students, such as the transition from elementary school to middle school (Xu et al., 2009). Non-promotional moves are school changes occurring via transfers, even though a student has not yet reached the highest grade level in that school. Students experience promotional school changes with peers in the same grade level because promotional school moves are compulsory for all students (Xu et al., 2009). Unlike promotional transfers, students often experience non-promotional moves alone (Xu et al., 2009).

Non-promotional school mobility is a common phenomenon throughout the United States (Gasper et al., 2010; Hanushek et al., 2004; Kerbow, 1996; Rumberger et al., 1999; Xu et al., 2009). For instance, in North Carolina, the student turnover rate reached 33% in 2004 (Xu et al., 2009). In Nashville, 8% to 15% of elementary and middle school students experienced non-promotional school mobility from 1998 to 2003 annually (Grigg, 2012). In New Orleans, for public school students on grades 1 to 7, only 60% of the students in 2004 and 67% in 2011 stayed in the same contiguous school (Maroulis et al., 2019). At the national level, 23% of middle and high school students changed schools between 1994 and 1996 (Metzger et al., 2015); around 80% of 12-year-old students experienced at least one non-promotional school transfer from 1997 through 2004 (Gasper et al., 2010).

Mobility Reasons and Classification

Students change schools for a variety of reasons (Burkam et al., 2009; Hanushek et al., 2004; Rumberger et al., 1999; Schwartz et al., 2012; Xu et al., 2009). One common reason is that students move due to factors related to schools (e.g., trouble with teachers or peers in prior schools, desire for schools with higher quality or special programs, and school closures; Burkam et al., 2009; Hanushek et al., 2004; Xu et al., 2009). Students also change schools because of family reasons (e.g., parental divorce, parental job changes, financial issues), which are often accompanied by residential moves (Burkam et al., 2009; Mao, 1997). Based on the various mobility reasons, scholars have distinguished between two types of non-promotional school mobility—strategic moves and reactive moves (Burkam et al., 2009; Hanushek et al., 2004; Rumberger et al., 1999; Schwartz et al., 2012). Strategic moves, also called “Tiebout type moves,” are planned school changes through which students seek to attend higher-quality or better-fitting schools (Hanushek et al., 2004; Schwartz et al., 2012). In contrast, reactive moves are often unplanned and occur due to unexpected family events or disciplinary actions (Grigg, 2012; Rumberger et al., 1999).

School mobility can also be classified by the timing of the mobility occurrence. Changing school while in session is considered to be within-year school mobility; school move that occur during the summer is considered to be between-year mobility. According to transfer policies in the school district under study, students can change schools either during the summer or during the school year. Even though both within-year mobility and between-year mobility can be reactive or strategic, moving during the summer is often planned ahead by parents and thus strategic, whereas moving during the academic year tends to be reactive (Hanushek et al., 2004).

Confounders of School Mobility and Academic Attainment

Students who change schools frequently are often different from those who do not change schools as often (Alexander et al., 1996; Burkam et al., 2009; Ingersoll et al., 1989; Kerbow, 1996; Rumberger et al., 1999). For example, mobile students are more likely to be racial minorities, homeless, in low socioeconomic status, in lower grade levels, in single-parent families, and to have limited English proficiency (Alexander et al., 1996; Ashby, 2010; Burkam et al., 2009; Fong et al., 2010; Hanushek et al., 2004; Kerbow, 1996; Reynolds et al., 2009). Sometimes advantaged students also change schools, but they are prone to do so to move to higher quality schools (Alexander et al., 1996; Xu et al., 2009).

School mobility is associated with several negative outcomes, such as achievement decline, grade retention, delinquency, drug use, and mental health problems (Alexander et al., 1996; Gasper et al., 2012; Xu et al., 2009). One crucial question surrounding school mobility is its effects on students’ school achievement. Because the aforementioned characteristics associated with high mobility rate are also causes of low academic performance more generally, many scholars have suggested that SES, race/ethnicity, or family structures could be the underlying joint causes of both school mobility and academic achievement (Alexander et al., 1996; Heinlein and Shinn, 2000). Therefore, the observed effects of school mobility on school outcomes may likely be explained by preexisting disparities.

The Independent Effects of School Mobility

Aside from the confounding factors on the link between mobility and academic attainment, a remaining question becomes whether mobility per se is harmful if background characteristics are held constant. A few scholars found a spurious relationship between school mobility and academic achievement after controlling for students’ demographic and background characteristics (Alexander et al., 1996; Heinlein and Shinn, 2000). However, most research finds that the negative effects of school mobility on students’ academic outcomes cannot be entirely explained by preexisting disparities in family background or other baseline factors (Engec, 2006; Grigg, 2012; Hanushek et al., 2004; Xu et al., 2009).

Several mediators have been proposed to account for the effects of school mobility on student outcomes, including social network characteristics, peer influences, psychological wellbeing, and school engagement (Coleman and Coleman, 1994; South et al., 2007). First, social capital is a collection of social ties that are embedded in families or communities (Coleman and Coleman, 1994). Switching schools might disrupt social networks between students and schools, making it difficult for students to fit into the new school environment, learn new course content, meet and communicate with new peers and teachers which may, eventually, result in a decline in school performance (Alexander et al., 1996; Astone and McLanahan, 1994; South et al., 2007). Second, peer effects try to quantify the influence of peers in schools for students’ outcomes. Mobile youths are more likely to befriend peers who exhibit delinquency and have worse academic performance (South et al., 2007). A third explanation for the negative association between school mobility and school outcomes is psychological wellbeing (South et al., 2007). Because children and adolescents are already transitioning their unique and important life stages, school mobility might cause extra stress and undermine their self-esteem (South et al., 2007). Another related explanation is school engagement, which suggests that school mobility might distract students from their study and make them less likely to aspire toward academic success (South et al., 2007).

Although many scholars agree that school mobility negatively impacts academic outcomes in general, others argue that switching to higher-quality or better-fitting schools might benefit students’ academic attainment in the long run (Alexander et al., 1996; Hanushek et al., 2004). In other words, it is the resources associated with the new school that contributes to better academic outcomes, rather than the move to a new school per se. It should be noted, though, the positive association only applies to students who transfer to a better school and stay in that school long enough to recover from any difficulties caused by the transition process (Hanushek et al., 2004). For frequent movers, even if they constantly transfer to better schools in quick succession, they do not seem to benefit much from the upward mobility to better schools.

The Timing of School Mobility

The timing of school mobility—either during or in between school years—adds more complexities to its effects (Hanushek et al., 2004; Mao, 1997). Hanushek et al. (2004) state that, “the negative effect of mid-year entry is at least twice as large as the effect of entrants at the beginning of the school year (p. 1742).” However, Grigg (2012) makes distinctions between mobility types that vary depending on whether they are compulsory and within/between school years. However, in his comparison, he only compared within- and between-year moves that were considered voluntary, leaving important groups excluded. One such group are those students who transfer schools over the summer because of disciplinary reasons. In the current study, I examine the entire spectrum of student mobility.

Heterogeneity across Racial/Ethnic Groups

In terms of school transition costs, researchers have considered heterogeneity in mobility effects among racial and ethnic groups. Alexander et al. (1996) suggest that school mobility is particularly harmful for racial minority and low SES students. Facing the same type of school mobility, black students experience a larger academic slip following mobility compared to Hispanic and white students in some studies (Hanushek et al., 2004). A study in North Carolina, however, suggests that Hispanic students experience more harmful academic consequences following mobility compared to black students (Xu et al., 2009). Therefore, researchers have not yet reached a consensus on which racial/ethnic group is most negatively impacted by school mobility.

Despite the heterogeneity discussion, little progress has been made to understand mobility consequences among Asian-American students. This oversight might be attributable to their relatively small population compared to other racial/ethnic groups. Even though Asian Americans only make up about 6% of the total U.S. population, they have been the fastest growing group in the United States, in terms of percentage increase (Lee and Zhou, 2015). Moreover, the academic success of Asian Americans compared to other racial/ethnic groups (including whites) results in a “model minority” image (Lee and Zhou, 2015). Hence, it is important to examine whether school mobility enlarges the achievement gaps among racial/ethnic groups, and whether Asian Americans’ academic advantage persists when they experience school mobility. Few studies have not done an adequate job of measuring heterogeneity in mobility effects, so racial/ethnic disparities require further investigation.

Methodological Approaches

Conventional multilevel models allow scholars to explore policy-driven research questions and produce relevant suggestions for policymaking. Nonetheless, traditional multilevel models cannot get around the first dilemma faced by fixed effects models, namely that students migrate from one second-level unit (i.e., school) to another second-level unit (school) within the same study (Burkam et al., 2009). Researchers often settle on only analyzing the last school that a student attended. School mobility studies that choose to ignore the previous schools’ effects likely underestimate the school effects, which may lead to inaccurate conclusions and incorrect inferences about the effects of school-level variables (Leckie, 2009).

A model that can take all the schools that a student has attended into consideration is the multiple membership model (MMM). The MMM data structure indicates a multiple membership relationship, that is, the lower-level units (i.e., students) can belong to more than one higher-level unit (i.e., schools; Leckie, 2013). MMMs were originally developed by Hill and Goldstein (1998) and then applied to various disciplines by other scholars (Browne et al., 2001; Chandola et al., 2005; Leckie, 2009). For example, MMMs has been used to study the illnesses and physical symptoms of patients taken care of by multiple nurses (Leckie, 2013). The biggest strength of this model is that it can take all the higher-level units into consideration, including the idea that two students can belong to the same school and that one of those two students could also belong to a second school (to which they moved).

Given the importance and complexity of school mobility, a re-estimation of within- and between-year school mobility effects on student outcomes is crucial. This study focuses on one of the biggest school districts in the United States, which has a relatively high school mobility rate. In particular, there are 13 elementary schools in this district with extremely high student turnover rates; around 40% of students change schools throughout the academic year. Moreover, the school district allows students to attend non-zoned schools (i.e., school choice districts). This greater availability of school choice increases school mobility (Fiel et al., 2013). As such, it is important to investigate the effects of school mobility on student academic achievement at this specific site. This study may have implications for other school districts, because the number of school choice districts has been increasing in recent years (Burkam et al., 2009). In the current study, I address two main research questions:

Do the effects of between- and within-year school mobility on student academic achievement differ? Specifically, do they differ when applying a different model (MMM) to data from a school choice district?

Do the effects of between- and within-year school mobility on student academic achievement vary by race/ethnicity? Specifically, how do they differ among Asian Americans compared to other racial/ethnic groups?

Data

I used data provided by the school district in this study. The school district is one of the twenty largest school districts in the United States. The data contain student-level demographic and background characteristics (e.g., race/ethnicity, gender, date of birth, and grade level), as well as school-level characteristics (e.g., proportion of students eligible for free/reduced price lunch, total student population). The data also include all the schools that students have attended, including the number of days they have been enrolled in each school, which enabled me to identify whether students have changed schools during the academic year or in the summer. Additionally, the data include students’ scores for a state-required test in the 2011 to 2012 and 2012 to 2013 academic year.

I dropped students who experienced a promotional school transfer, namely, transitioned from elementary school to middle school. This allowed me to compare students who underwent within- and between-year moves that were not normative transitions that students are expected to go through. I focused on four cohorts of students who have taken the tests in the 2011 to 2012 and 2012 to 2013 academic year. In the 2012 to 2013 school year, these cohorts were grades 4, 5, 7, and 8. With respect to the missing data, I examined the characteristics of missingness. There did not appear to be any systematic pattern of missingness, thus I assumed that there was likely no significant selection bias in the sample. The total student sample size was 34,299, nested within 202 elementary and middle schools.

Variables

The dependent variables were students’ reading and math scores in the 2012 to 2013 academic year. There are two score versions of the test—raw scores and scale scores. Raw scores only estimate how many questions a student has answered correctly; scale scores account for the difficulty level of each test and thus are comparable across different tests. Scale scores were used in the current analyses. I also used students’ reading and math scores in the 2011 to 2012 school year as independent variables, in order to control students’ prior academic performance. To better understand the extent of mobility effects, I standardized the scale scores for both 2011 to 2012 and 2012 to 2013 academic years.

Table 1 provides a descriptive summary of the original scale scores and scores after standardization in the 2012 to 2013 school year. The average math scale score is slightly higher than the average reading scale score. The math scores also have a larger range and standard deviation relative to the reading scores. For example, the maximum math scale score is around 14 standard deviations higher than the average math scale score.

Table 1.

Standardized Reading and Math Scale Scores in the 2012 to 2013 Academic Year.

Test	Mean	Standard deviation	Min	Max
Reading scale scores	1,571	134.6	811	2,141
Math scale scores	1,596	157.0	867	3,784
Standardized reading scores	0	1	−5.65	4.23
Standardized math scores	0	1	−4.64	13.93

My key independent variable is type of school mobility (1 = no school change, 2 = between-year school mobility, and 3 = within-year school mobility).² My other independent variables include race/ethnicity, gender, economic disadvantage status, grade level, limited English proficiency, gifted/talented program, prior test scores, and special education status.³ Aside from student-level factors, I also controlled for school-level variables, such as the percentage of students eligible for free/reduced price and the student-teacher ratio.⁴ In the data, the percentage of students receiving subsidized lunch was relatively high in most schools. For example, more than 75% of schools had at least 85% of students eligible for free/reduced price lunch.

Methodology

In this study, I chose a MMM for two reasons. First, each school may have a distinct impact on students, so we cannot assume that each student represents an independent observation, as OLS modeling does. Second, MMM is used in cross-classified data in which lower-level units can belong to more than one higher-level unit (Leckie, 2013). One crucial feature of the multiple membership data structure is that “the degree to which each lower level unit belongs to each higher level unit will often vary across those higher level units (Leckie, 2013: 3).” In addition, the degrees of variation often represent different weights of the higher-level units. Using different ways to assign weights and then fitting these weights into the multilevel models might result in completely different results. Particularly, simply assigning weights to just one higher-level unit overlooks other higher-level units and would likely result in misleading conclusions.

In this study, the multiple membership weight was defined as the proportion of days students spent in each school. This weight was chosen because the more time a student has spent in one school, the larger influences the school might have on that student. If a student has not changed schools, it means he/she attends one school for the entirety of the 2012 to 2013 academic year. Therefore, I assigned a value of 1 to that school for this student. If a student has enrolled in the first school (A) for 54 days and in the second school (B) for 126 days, the multiple membership weights would be 0.3 (54 days divided by the total enrollment days, which is 180 days) for school A and 0.7 (126 days divided by the total enrollment days) for school B. The multiple membership model can be written as:

\begin{array}{l} y_{i} = β_{0} + β_{1} x_{1 i} + β_{2} \sum_{j ϵ s c h o o l (i)} w_{j, i}^{(2)} x_{2 j}^{(2)} + \sum_{j ϵ s c h o o l (i)} w_{j, i}^{(2)} u_{j}^{(2)} + e_{i} \\ u_{j}^{(2)} ~ N (0, σ_{u (2)}^{2}) \\ e_{i} ~ N (0, σ_{e}^{2}) \end{array}

In this case, $i (i = 1, 2, \dots, 34299)$ represents the student population and $j (j = 1, 2, \dots, 202)$ measures the number of schools. Each student or school has its distinct ID number. Where $y_{i}$ is the reading/math score for student $i,$ $β_{0}$ is the average reading/math score for the entire population. In $β_{1} x_{1 i}$ , $x_{1 i}$ is the student-level variable with a slope coefficient $β_{1}$ . The fixed part $\sum_{j ϵ s c h o o l (i)} w_{j, i}^{(2)} u_{j}^{(2)},$ $w_{j, i}^{(2)}$ represents the degree student $i$ belongs to school $j$ with an associated effect of $u_{j}^{(2)}$ . $\sum_{j ϵ s c h o o l (i)} w_{j, i}^{(2)} x_{2 j}^{(2)}$ is the weighted sum of school-level variables, with a slope of $β_{2}$ . $e_{i}$ is the residual error, which is the difference between a student’s real score and the predicted score from using the full equation. In addition, $σ_{u (2)}^{2}$ is the school variance, which captures the differences between schools once the predictor variables are already controlled. This estimate can also be understood as the school-level residual. $σ_{e}^{2}$ is the student variance, measuring how students vary from each other within the same school.

For this specific analysis, three steps were taken to quantify the effects of mobility type and race/ethnicity on student achievement. First, to investigate the unadjusted achievement differences between groups, I ran OLS regressions by including types of school mobility as a single predictor. Second, I used multiple membership models to measure the adjusted achievement differences by controlling for student-level and school-level factors related to academic achievement. Third, I estimated the potential different effects of school changes by adding interactions between types of school mobility and race/ethnicity.

Results

Descriptive Findings

The descriptive statistics for each of the independent variables can be found in Table 2. The sample is dominated by Hispanic students (61.36%), followed by blacks (25.97%), whites (8.33%), Asians (3.41%), and others (0.93%). Four out of five students were classified as economically disadvantaged, either living in poverty or receiving a lunch subsidy. Approximately 3 in 10 had limited English proficiency, and 1 in 5 were in a gifted/talented program. As for student mobility rates, around 8.41% of students only changed schools during the summer before the 2012-2013 academic year, and 1.76% of students switched schools during this school year.

Table 2.

Descriptive Statistics (By Type of School Mobility).

Variables	Values	All youth (N = 34,299)	Between-year mobility (N = 2,883)	Within-year mobility (N = 604)
Ethnicity	White	8.33	3.82	3.15
	Black	25.97	35.73	44.37
	Hispanic	61.36	57.2	48.34
	Asian/PI	3.41	2.6	3.31
	Other	0.93	.66	0.83
Economically disadvantage	Non-disadvantaged	20.09	12.04	8.44
Economically disadvantage	Disadvantaged	79.91	87.96	91.56
Gender	Male	49.78	50.71	52.48
Gender	Female	50.22	49.29	47.52
Grade level	4	25.27	25.18	31.29
	5	29.34	27.61	28.18
	7	26.06	28.58	23.84
	8	19.33	18.63	16.72
Limited English	No	68.34	68.26	74.83
Limited English	Yes	31.66	31.74	25.17
Gifted/talented Program	No	79.59	87.96	93.38
Gifted/talented Program	Yes	20.41	12.04	6.62
Special education	No	96.25	96.36	95.7
Special education	Yes	3.75	3.64	4.3

I also examined how school mobility varied by students’ attributes. With regard to race/ethnicity, black students had the highest mobility rates, followed by Hispanics, whites, and Asians. For example, among the student who were mobile within a school year, 44.37% were blacks, which is much higher than their representation (25.97%) in the overall student population. Students who are economically disadvantaged were more likely to move, as were students not in gifted/talented programs. Furthermore, students with limited English proficiency (LEP) status did not display higher mobility rates relative to non-LEP students. Overall, the between-year mobility rates were higher compared to within-year mobility rates across various characteristics, reflecting that parents tend to make their children change schools following the end academic year rather than during it.

Unadjusted Achievement Differences

Before adding other control variables, I ran two regression models to estimate the effects of school mobility on standardized reading and math scores. As seen in Table 3, between-year school changers had a reading score 0.239 standard deviations lower than the reading score of non-mobile students (see Model 1). Within-year school changers’ reading scores were .545 standard deviations lower than students who did not change schools. The effects of school mobility on math followed the same pattern as reading scores, as shown in Model 4. Between-year school changers had a math score 0.234 standard deviations lower than non-mobile students. Within-year school changers’ math scores were 0.556 standard deviations lower than students who stay in previous schools. The decline of standard deviations in Model 4 relative to Model 1 reveals that mobility affects students’ math achievement more than reading. In sum, within-year changers had the lowest reading and math achievement relative to between-year changers and stable students.

Table 3.

Results from OLS Regression and Multiple Membership Models (On Standardized Reading and Math Scores).

Predictor	Reading			Math
	Model 1	Model 2	Model 3	Model 4	Model5	Model6
	Coef	Coef	Coef	Coef	Coef	Coef
Constant	0.034	0.247	0.234	0.034	0.044	0.023
Female (Hispanic)		0.021***	0.021***		−0.014*	−0.014*
White		0.055***	0.054***		0.066***	0.066***
Black		−0.05***	−0.048***		−0.036***	−0.039***
Asian		0.166***	0.17***		0.277***	0.284***
Other ethnic group		0.023	0.021		0.149***	0.153***
Limited English proficiency		−0.102***	−0.101***		−0.058***	−0.057***
Economically disadvantaged		−0.067***	−0.067***		−0.042***	−0.042***
Gifted/talented program		0.232***	0.232***		0.332***	0.332***
Special education		−0.179***	−0.179***		0.028	0.028
Scores in the prior year (Grade 4)		0.697***	0.697***		0.611***	0.61***
Grade 5		−0.106***	−0.106***		−0.011	−0.011
Grade 7		0.189***	0.187***		0.05**	0.046*
Grade 8 (no change)		0.216***	0.215***		0.32***	0.316***
Between-year change	−0.239***	−0.031**	−0.032**	−0.234***	−0.047***	−0.047***
Within-year change	−0.545***	−093***	−0.068**	−0.556***	−0.159***	−0.182***
Percentage of free/reduced lunch		−0.003***	−0.003***		−0.002***	−0.002***
Student-teacher ratio (Hispanic × within-year)		−0.003	−0.003		0.003	0.004
White × within-year			0.197			0.012
Black × within-year			−0.053			0.083
Asian × within-year			−0.299**			−0.396**
Other × within-year			−0.124			−0.242
Between school variance		0.011	0.011		0.036	0.035
Between individual variance		0.297	0.297		0.427	0.427

Note. *p < .1. **p < .05. ***p < .005.

Adjusted Achievement Differences

In addition, I ran two multiple membership models for reading and math separately. In Models 2 and 5, controlling for individual- and school-level factors, school mobility still negatively impacted student academic achievement. In Model 2, between-year changers had a reading score 0.031 standard deviations lower than students who did not change schools. Within-year changers’ reading scores were 0.093 standard deviations lower than students who did not change schools. In Model 5, the decreased magnitude of math scores was also larger for within-year school changers than for between-year changers, with all other variables controlled. Overall, changing schools during the academic year was more harmful to student academic performance than changing schools during the summer.

The effects of student-level predictors were largely consistent with the previous literature.

The percentage of students eligible for free/reduced price lunch was negatively associated with student academic achievement. If the proportion of students receiving subsidized lunch increased by 1%, both the reading scale score (by 0.003 SDs) and math scale score (by 0.002 SDs) decreased. Although this reduction seems small, this decrease could make a large difference when the discrepancy in lunch subsidies between two schools is large. The student-teacher ratio was not significant both in Model 2 and Model 4, which might be due to lack of variation in the ratios across schools.

Effects of School Mobility by Race/Ethnicity

To estimate whether the effects of school changes on student academic performance differed along racial/ethnic group, I added interaction terms between types of school mobility and racial/ethnic group into the MMMs. Following the rule of model parsimony (Aiken et al., 1991), I eliminated interaction terms that are not statistically significant. The remaining interaction terms include within-year school mobility and race/ethnicity, indicating that the harmful effects of between-year school mobility did not vary across race groups.

In Model 3, the interaction term between Asian and within-year change was negative and significant at the 0.05 level, indicating that within-year school mobility reduces the reading scores more for Asian students relative to Hispanic students. The interaction terms between white or black and within-year mobility are not significant, suggesting that within-year mobility affects Hispanics, blacks, and whites to similar degrees. Model 6 suggests that within-year change impacts Asians’ math performance more relative to Hispanic, black, and white students.

To reveal the disparities of academic performance among racial/ethnic groups more clearly, I plot the predicted scores based on the full models (from Model 3 for predicted reading scores and Model 6 for predicted math scores). Figure 1 presents the predicted reading scores by types of school mobility across race/ethnicity. Among student who did not move, Asian American students had the highest academic performance. If Asians experienced intra-year mobility, their academic achievement was the most impaired. Other racial/ethnic groups were not negatively affected by within-year mobility as much as Asian students. Similar patterns can be observed in Figure 2, which presents the predicted math scores across racial/ethnic groups.

Figure 1.

Predicted reading scores (by type of school mobility and race).

Figure 2.

Predicted math scores (by type of school mobility and race).

Discussion and Conclusion

Consistent with Hanushek et al.’s (2004) findings, both within- and between-year school mobility were harmful to students’ reading and math scores. Within-year mobility yielded more detrimental effects than between-year mobility. In addition, I found that, while the effects of within-year school mobility varied across racial/ethnic groups, the effects of between-year mobility did not vary across racial/ethnic groups. Even though black students had the highest mobility rate, within-year school mobility had similarly negative effects on academic attainment for Hispanic and black students.

Unexpectedly, Asian American students are set back most by within-year school mobility. To gain a sense of the mechanisms underlying this finding, I took a closer look at the attributes of Asian American students who changed schools during the academic year. There are two possible explanations for this phenomenon. First, Asian American students were most likely to be foreign-born among all within-year mobile students. After adding immigration status into the models as a control variable, the negative effects of within-year school change were slightly reduced for Asian Americans. I argue that perhaps immigration status exacerbated the negative effects of school mobility on student outcomes. Immigrant children likely have greater difficulty adjusting to new school environments due to language barriers and other cultural differences. Second, it is possible that mobile Asian students tend to be lower performing and from lower-SES backgrounds, because there is a significant difference in these characteristics among Asian subgroups (Lee and Zhou, 2015).

Reading and math scores were differentially compromised by school mobility. Specifically, students’ math performance was impacted more by school mobility than their reading performance. Learning mathematics is a cumulative process and thus may be more sensitive to disruption of course instruction; literacy instruction is relatively flexible (Kerbow, 1996; Swanson and Schneider, 1999). These differences might explain why math performance was more negatively affected.

Because the data only include information of students enrolled in the school district, I am not able to examine students not in this district before the 2012 to 2013 academic year and those who moved out of it during the 2012 to 2013 school year.⁵ In other words, this study only estimates intra-district school mobility and does not speak to between-district mobility. On the one hand, we can conclude that, even when changing schools within the same school district, mobile students have lower achievement on state-required tests. These findings echo Fitchen’s (1994) contention that “even a short-distance move may take a child into another school district with fundamentally different teaching approaches, methodology, and basal texts (p. 427).” On the other hand, we should be cautious in generalizing these findings to between-district or long-distance school transfers.

This study had several limitations, which also serve as suggestions for future studies on school mobility. First, this study only examined the short-term effects of school mobility on student academic achievement. Previous scholars have demonstrated that the negative effects from mobility might diminish with more acclimation to a new environment, and in the long term, students would probably recover from school changes (Hanushek et al., 2004; Mao, 1997). Even strategic moves appear costly if researchers only examine achievement right after the move (Kain & O’Brien, 1998). It is quite likely that the negative mobility effects would diminish over time for students. Therefore, we can only conclude that changing schools has immediate negative effects on test scores following the move; the longer-term effects of school mobility need further investigation. For the same reason, some non-mobile students in this sample might have changed schools the year earlier, which might even be reducing the discrepancy between mobile and non-mobile students.

Second, school changes are often accompanied by residential moves and the two types of mobility have different impacts on children (Alexander et al., 1996; Gasper et al., 2010; Swanson and Schneider, 1999). The two types of mobility involve different adjustments—one move involves adjustments to a new school; one move involves adjustments to a new neighborhood. Due to the difficulty of identifying whether students in the sample have changed residence, the distinctions between school mobility and residential mobility were beyond the scope of this study.

Finally, according to previous research, family factors are strong predictors of student academic achievement and school mobility (Gasper et al., 2010). For example, mobile children are more likely to change schools due to family reasons rather than due to school reasons (Burkam et al., 2009). Due to the lack of family information in the data, I was not able to control family factors in this study. Therefore, future educational research should consider incorporating other data sources that include family-related factors, such as parent occupation and family structure.

Overall, this study has practical implications for both parents and policy makers. Given the more severe consequences of within-year school mobility, the policies should specifically target students who change schools during the academic year. If a school move is avoidable, schools should put more effective strategies in place to address school mobility. For example, schools could attempt to retain their students by developing close social networks with their parents. It is important for parents to recognize the detrimental effects of student mobility, especially moves when school is in session. With closer relations and mutual trust between parents and schools, parents might be less likely to make their children transfer schools, even if when a residential move is inevitable.

However, school mobility is unavoidable oftentimes, for instance when a student’s family has to move due to eviction or occupational-related reasons. In these cases, there are still practices that schools could take to alleviate the negative outcomes of school mobility. For example, districts could establish standardized curriculum across all schools or provide proper adjustment programs for mobile students (Hanushek et al., 2004; Nelson et al., 1996). Adjustment programs could serve both academic and psychosocial purposes. On the one hand, schools provide mobile students with afterschool curricular programs so that they would be able to catch up with their non-mobile peers academically. On the other hand, counseling services targeted at mobile students are needed to help them form new friendship with peers, develop social support groups, and adjust to the new school environment.

Footnotes

Declaration of Conflicting Interests

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

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Jie Min

Notes

Author Biography

Jie Min received her Ph.D. in Sociology from Rice University. Her research interests include education, race and ethnic relations, immigration, and urban sociology. Jie Min is also affiliated with the Houston Education Research Consortium (HERC), a research-practice partnership between Rice University and eleven Houston area school districts.

References

Aiken

L. S.

West

S. G.

Reno

R. R.

(1991). Multiple regression: Testing and interpreting interactions. Sage.

Alexander

K. L.

Entwisle

D. R.

Dauber

S. L.

(1996). Children in motion: School transfers and elementary school performance. The Journal of Educational Research, 90(1), 3–12.

Ashby

C. M.

(2010). K-12 education: Many challenges arise in educating students who change schools frequently (Report to congressional requesters, GAO-11-40). US Government Accountability Office.

Astone

N. M.

McLanahan

S. S.

(1994). Family structure, residential mobility, and school dropout: A research note. Demography, 31(4), 575–584.

Browne

W. J.

Goldstein

Rasbash

(2001). Multiple membership multiple classification (MMMC) models. Statistical Modelling, 1(2), 103–124.

Burkam

D. T.

Lee

V. E.

Dwyer

(2009, June 29–30). School mobility in the early elementary grades: Frequency and impact from nationally-representative data [Paper presentation]. Workshop on the Impact of Mobility and Change on the Lives of Young Children, Schools, and Neighborhoods, Washington, DC.

Chandola

Clarke

Wiggins

R. D.

Bartley

(2005). Who you live with and where you live: Setting the context for health using multiple membership multilevel models. Journal of Epidemiology & Community Health, 59(2), 170–175.

Coleman

J. S.

Coleman

J. S.

(1994). Foundations of social theory. Harvard University Press.

Engec

(2006). Relationship between mobility and student performance and behavior. The Journal of Educational Research, 99(3), 167–178.

10.

Fiel

J. E.

Haskins

A. R.

Turley

R. N. L.

(2013). Reducing school mobility: A randomized trial of a relationship-building intervention. American Educational Research Journal, 50(6), 1188–1218.

11.

Fitchen

J. M.

(1994). Residential mobility among the rural poor. Rural Sociology, 59(3), 416–436.

12.

Fong

A. B.

Bae

Huang

(2010). Patterns of student mobility among English language learner students in Arizona Public Schools. Issues & Answers. REL 2010-No. 093 (Research reports). Regional Educational Laboratory West.

13.

Gasper

DeLuca

Estacion

(2010). Coming and going: Explaining the effects of residential and school mobility on adolescent delinquency. Social Science Research, 39(3), 459–476.

14.

Gasper

DeLuca

Estacion

(2012). Switching schools: Revisiting the relationship between school mobility and high school dropout. American Educational Research Journal, 49(3), 487–519.

15.

Grigg

(2012). School enrollment changes and student achievement growth: A case study in educational disruption and continuity. Sociology of Education, 85(4), 388–404.

16.

Hanushek

E. A.

Kain

J. F.

Rivkin

S. G.

(2004). Disruption versus Tiebout improvement: The costs and benefits of switching schools. Journal of Public Economics, 88(9–10), 1721–1746.

17.

Heinlein

L. M.

Shinn

(2000). School mobility and student achievement in an urban setting. Psychology in the Schools, 37(4), 349–357.

18.

Hill

P. W.

Goldstein

(1998). Multilevel modeling of educational data with cross-classification and missing identification for units. Journal of Educational and Behavioral Statistics, 23(2), 117–128.

19.

Holme

J. J.

Richards

M. P.

(2009). School choice and stratification in a regional context: Examining the role of inter-district choice. Peabody Journal of Education, 84(2), 150–171.

20.

Ingersoll

G. M.

Scamman

J. P.

Eckerling

W. D.

(1989). Geographic mobility and student achievement in an urban setting. Educational Evaluation and Policy Analysis, 11(2), 143–149.

21.

Kain

J. F.

O’Brien

D. M.

(1998). A longitudinal assessment of reading achievement: Evidence from the Harvard/UTD Texas Schools Project. University of Texas at Dallas.

22.

Kerbow

(1996). Patterns of urban student mobility and local school reform. Journal of Education for Students Placed at Risk, 1(2), 147–169.

23.

Leckie

(2009). The complexity of school and neighbourhood effects and movements of pupils on school differences in models of educational achievement. Journal of the Royal Statistical Society: Series A (Statistics in Society), 172(3), 537–554.

24.

Leckie

(2013). Multiple membership multilevel models. Centre for Multilevel Modelling.

25.

Lee

Zhou

(2015). The Asian American achievement paradox. Russell Sage Foundation.

26.

Mao

M. X.

(1997). Student mobility, academic performance, and school accountability. A Study of Student Mobility in Texas Public Schools, Statewide Texas Educational Progress Study Report (3), 349–355. https://files.eric.ed.gov/fulltext/ED409380.pdf

27.

Maroulis

Santillano

Jabbar

Harris

D. N.

(2019). The push and pull of school performance: Evidence from student mobility in New Orleans. American Journal of Education, 125(3), 345–380.

28.

Metzger

M. W.

Fowler

P. J.

Anderson

C. L.

Lindsay

C. A.

(2015). Residential mobility during adolescence: Do even “upward” moves predict dropout risk? Social Science Research, 53, 218–230.

29.

Nelson

P. S.

Simoni

J. M.

Adelman

H. S.

(1996). Mobility and school functioning in the early grades. The Journal of Educational Research, 89(6), 365–369.

30.

Reynolds

A. J.

Chen

C. C.

Herbers

J. E.

(2009, June 29–30). School mobility and educational success: A research synthesis and evidence on prevention [Paper presentation]. Workshop on the Impact of Mobility and Change on the Lives of Young Children, Schools, and Neighborhoods, Washington, DC.

31.

Rumberger

R. W.

(2015). Student mobility: Causes, consequences, and solutions. National Education Policy Center.

32.

Rumberger

R. W.

Larson

K. A.

Ream

R. K.

Palardy

G. J.

(1999). The educational consequences of mobility for California students and schools (Research reports). Policy Analysis for California Education.

33.

Schwartz

A. E.

Stiefel

Cordes

(2012, October 3). Moving matters: The causal effect of school mobility on student performance [Conference session]. Annual Conference of the Association for Education Finance and Policy, Institute for Education and Social Policy, Steinhardt and Wagner Schools, New York University.

34.

South

S. J.

Haynie

D. L.

Bose

(2007). Student mobility and school dropout. Social Science Research, 36(1), 68–94.

35.

Swanson

C. B.

Schneider

(1999). Students on the move: Residential and educational mobility in America’s schools. Sociology of Education, 72(1), 54–67.

36.

Hannaway

D’Souza

(2009). Student transience in North Carolina: The effect of school mobility on student outcomes using longitudinal data (Working Paper No. 22). National Center for Analysis of Longitudinal Data in Education Research.