Residential Mobility Research Brief Page 1 of 12 Perfect Strangers: Residential Mobility and Educational and Behavioral Outcomes in Metro Nashville Public Schools Adam Voight, Maury Nation, and Beth Shinn “Home is where one starts from.” – T.S. Eliot The focus of this research brief is the residential mobility of students in Metro-Nashville Public Schools (MNPS) middle schools. The purpose is to provide a summary of key findings from survey research conducted collaboratively by Metro-Nashville Public Schools and Peabody College of Vanderbilt University. The research demonstrates the deleterious effect that residential mobility has on the educational and behavioral outcomes of students who experience the instability as well as the negative effect that aggregate mobility at the school level has on students in the school. Housing issues have taken on greater significance in light of the recent economic recession and foreclosure crisis. In the Nashville Metropolitan area, the past five years have seen higher home vacancy and mobility rates, alongside larger average household sizes1. Housing instability has been connected to a variety of negative outcomes for homeowners and communities, but it may have particularly harmful effects on children and their educational experience, especially when it is associated with a change of schools. The adjustment that is required in moving homes during childhood and adolescence can create a stressful home environment that negatively impacts children’s education and early life outcomes. Children who change schools may miss key educational material, hampering their school performance. Further, children and parents who relocate must often develop new social networks and thus may not be privy to important information related to school success. Also, children attending a new school may feel socially marginalized, potentially leading to problem behavior and disengagement from school. Residential mobility is not frequently considered among the range of issues with which school must cope, but there are steps that schools can take to engage this risk factor. This report uses data from 11 MNPS middle schools and summarizes key findings related to residential mobility and student educational outcomes. Middle school is a period during which youth may be particularly vulnerable to residential instability due to their rapid social development. This report first describes the scope of residential mobility among middle school students; it then examines the relationship between mobility and educational outcomes; finally, it suggests possible avenues for schools to address problems associated with mobility. 2. The Landscape of Residential Mobility in MNPS Middle Schools2 According to the U.S. Census, 18.2% of the population in the Nashville metropolitan area reported in 2008 that they had changed their primary residence during the preceding year. Compared to the national mobility rate of 15.0%, Nashville has a relatively high incidence of residential mobility1. This citywide trend of high mobility is reflected in the mobility rates of MNPS middle school students3. As Figure 2.1 shows, less than 25% of MNPS middle school students live at the same address continuously over a six year period. Almost 70% of middle school students move between one and three times during the same period. A smaller number, approximately 10%, are highly mobile, moving residences four or more times in six years. Moving homes is not an uncommon experience for students in MNPS middle schools. 20 15 10 5 0 Percent of students 25 Residential Mobility Research Brief Page 2 of 12 0 1 2 3 4 5 6 Residential moves over six years (2003-2009) Figure 2.1 – Number of residential moves over six years (2003-2009) for MNPS middle school students Overall, on average MNPS middle school students moved almost two times in six years. This number varies according to students’ backgrounds. Is this experience of relocation equal for all students, or is it associated more strongly with students of different backgrounds? Students who are eligible for the free and reduced lunch program (FRPL)—which is often considered a proxy for socioeconomic status (SES)—have moved almost twice as frequently compared to students who are not eligible, as shown in Table 2.1. Further, Black students are the most frequent movers of any racial group, followed by Latino/a students and White students, respectively. These patterns are consistent with national trends4. Table 2.1 – Average number of residential moves over six years (2003-2009), by race and socioeconomic status Student characteristics Average number of moves Eligible for free/reduced lunch Not eligible for free/reduced lunch 1.89 1.09 Black Latino/a White 2.00 1.70 1.36 Total 1.71 Below, Figure 2.2 depicts the interaction of race and socioeconomic status, showing that low-SES Black students moved over twice as often as high-SES White students. MNPS middle school students of all backgrounds experience residential mobility to some degree, but Black, Latino/a, and low-SES students appear to be more frequent movers. 2 1 High SES te W hi o/ a La t in ac k Bl ia n As te W hi o/ a La t in Bl As ia n ac k 0 Average number of moves 3 Residential Mobility Research Brief Page 3 of 12 Low SES Figure 2.2 – Average number of residential moves over six years (2003-2009), by race and socioeconomic status Figure 2.3 offers an alternative way of understanding differences in mobility by race. Breaking down the frequency of mobility into three different categories—no moves, one to three moves, and more than three moves—shows that 80% of Black middle school students in MNPS move at least once over six years, with 40% experiencing extreme mobility, over three moves. By comparison, a third of White students stay in the same home during the six-year period, while fewer than a quarter are extremely mobile. Asian Black 12% 16% 25% 40% 44% 63% Latino/a 27% 20% White 21% 33% 53% No moves More than three moves 46% One to three moves Figure 2.3 – Residential mobility over six years (2003-2009), by race Residential Mobility Research Brief Page 4 of 12 3. The Relationship between Residential Mobility and Educational Success Residential mobility is important because it has been associated with a variety of relevant outcomes for cities, neighborhoods, homeowners, and children. Families move for many reasons, and the data examined here do not distinguish a residential move made for positive reasons versus one made under duress. However, a substantial body of research has demonstrated that residential moves, made for any reason, may have a negative effect on students’ achievement, aptitude, and retention5. Table 3.1 shows that MNPS middle school students with moderate and extreme levels of mobility score lower on achievement tests of math and reading and have more absences and disciplinary referrals. Table 3.1 – Average educational and behavioral outcomes by residential mobility category Number of moves 0 1-3 >3 Math TCAP Reading TCAP Absences Disciplinary referrals 48 42 38 47 42 37 6.4 6.8 9.4 1.5 1.9 2.8 Notes: (1) TCAP scores are reported in Normal Curve Equivalent scores; (2) absences are combined excused and unexcused absences 30 40 50 There is an average ten-point difference in terms of Normal Curve Equivalent6 (NCE) scores in the Math and Reading TCAP between students who have not moved at all during a six-year period compared to those who are highly mobile. Figures 3.1 and 3.2 graphically depict the changes in math and reading scores, respectively, by race and socioeconomic status as residential mobility increases. 0 1 2 3+ Residential moves over six years (2003-2009) All students Black Low SES Latino/a Figure 3.1 – Math TCAP scores (in NCEs) and number of residential moves over six years (2003-2009), by race and socioeconomic status Residential Mobility Research Brief Page 5 of 12 30 40 50 Regression analyses confirm that the decrease in Math and Reading TCAP scores associated with more residential moves, shown in Figures 3.1 and 3.2, is statistically significant, even when race and socioeconomic status are taken into consideration7. These analyses suggest that for the entire population of middle school students in MNPS, race and socioeconomic status being equal, one could expect highly mobile students to score almost three NCE points lower on their Reading and Math TCAP compared to residentially stable students, with roughly two and half additional absences during the academic year and almost one additional disciplinary referral. Highly mobile students demonstrate poorer educational outcomes. 0 1 2 3+ Residential moves over six years (2003-2009) All students Black Low SES Latino/a Figure 3.2 – Reading TCAP scores (in NCEs) and number of residential moves over six years (2003-2009), by race and socioeconomic status 4. Schools with Highly Mobile Students There are certain consequences for students who change residences, but there may be additional consequences for students who attend schools with high overall rates of mobility. That is, even if a student does not move herself, her educational outcomes may be affected by a school peer group that is highly mobile. A school with a large number of highly mobile students deals with significant academic and environmental challenges. A high turnover of students often leads to disruption of class curricula as teachers backtrack to accommodate the content knowledge of new students. Further, social relationships among students and between students and teachers are complicated by a changing enrollment8. Perhaps most fundamentally, serial residential mobility in childhood can limit the social capital to which a student has access, as her relationships with the community and school are often interrupted9. Highly mobile students may lack social support and suffer from lowered selfconcepts10. The average number of moves that students experience varies across schools, and these averages are shown below in Table 4.1 for all 11 of the middle schools included in this report. The overall student mobility average for all of the schools in the data is 1.74 moves. The average number of moves ranges from a low of 1.24 in West End Middle School to a high of 2.12 in W.A. Bass Middle School. What are the implications of student body mobility on the educational outcomes of students in these schools? Table 4.1 – Residential moves per student over six years (2003-2009), by school School Average moves per student Apollo M.S. Bailey M.S. Bellevue M.S. Cameron M.S. Donelson M.S. Ewing Park M.S. Jere Baxter M.S. McMurray M.S. Neely’s Bend M.S. W.A. Bass M.S. West End M.S. 1.64 2.00 1.38 1.65 1.73 1.87 1.90 1.76 1.81 2.12 1.24 40 50 60 Figure 4.1, below, is similar to the figures in the above section that depict the relationship between residential mobility and academic achievement by race and socioeconomic status. The difference here is that instead of modeling individual student mobility on the horizontal axis, school-average mobility is shown. That is, moving from left to right on the horizontal axis indicates an increase in average student mobility at the school level. Using the school average from Table 4.1, for example, West End Middle School would be located at the far left end of the horizontal axis (with an average student mobility of 1.24) and W.A. Bass Middle School would be at the far right end (with an average student mobility of 2.12). 30 Math TCAP score (2009) Residential Mobility Research Brief Page 6 of 12 1.2 1.4 1.6 1.8 2.0 Average number of moves of students in school All students Black Low SES Latino/a Figure 4.1 – School-average reading TCAP scores (in NCEs) and school-average residential moves over six years (2003-2009), by race and socioeconomic status Residential Mobility Research Brief Page 7 of 12 6 5 4 Absences (2009) 7 Figure 4.1 shows that schools with more mobile students have lower average Math TCAP scores. Multilevel regression analyses confirm that this relationship is statistically significant, even when controlling for school demographics and quality11. Regression analyses also show that school-average residential mobility is significantly associated with a greater number of absences and a decreased overall perception of school climate. The relationship between school-average residential mobility and absence is depicted in Figure 4.2. Students in schools with very low average levels of mobility— such as West End Middle School—miss an average of four days per academic year. Students in schools with high average mobility—such as W.A Bass Middle School—are absent an average of seven days per year. 1.2 1.4 1.6 1.8 2.0 Average number of moves of students in school Figure 4.2 – School-average absences and school-average residential moves over six years (2003-2009), by race and socioeconomic status The relationship between school-average residential mobility and school climate is worth particular mention. School climate is often measured by averaging individuals’ perceptions of some characteristic of the school—in this case, the relationships among students. In Section 3, above, there was no significant relationship found between a student’s personal experience of mobility and that student’s perceptions of school climate. That is to say, moving frequently does not impact one’s feelings about the school. However, in the regression models used to examine mobility at the school level, a significant relationship was found between school-average mobility and school-average perceptions of school climate. In other words, a school with a large number of highly mobile students is more likely to have low climate, but these low climate perceptions are not being driven by the students who move frequently. The results suggest that it is the more stable students in the school whose perceptions of school climate are damaged by a large proportion of highly mobile peers. The relationship between school-average residential mobility and school climate is shown graphically below in Figure 4.3. The difference in school climate between a school with extremely low average mobility and a school with extremely high average mobility is .25. School climate was measured using a scale of 1 to 5 that asked students to agree or disagreement with statements about the nature Residential Mobility Research Brief Page 8 of 12 3.4 of relationships among students at their school. The standard deviation between schools is .14, thus a difference of .25 is very powerful. School-average residential mobility is strongly related to school climate. Cameron Apollo 3.2 West End Bellevue 3.1 Neeley's Bend Bass 3 Donelson Ewing Park Bailey 2.9 School climate 3.3 McMurray Jere Baxter 1.2 1.4 1.6 1.8 2 2.2 Average residential moves of students in school Figure 4.3 – School-average residential mobility and school climate in 11 MNPS middle schools The empirical findings from Sections 3 and 4 provide compelling evidence that residential mobility is an important factor in the educational experience of MNPS middle school students. Individual students who move frequently are subject to lower academic achievement, more absences, and more disciplinary problems. Additionally, schools with many highly mobile students suffer from poorer overall achievement, lower attendance, and inferior school climate. 5. Program and Policy Implications for Schools Urban school districts are already greatly overburdened with the task of helping their students to be successful. School decision-makers may look at residential mobility as a phenomenon that is outside of their school walls and thus out of their control. To a degree, this sentiment is valid. However, there are real steps that schools can take to address issues related to residential mobility among their students. The following recommendations are drawn from the research literature on residential mobility and offer concrete strategies for schools12. Strategies for individual schools: • Provide educational supports for mobile students, such as tutoring, afterschool programs, summer programs, and transportation within high mobility zones. • Provide supplementary non-academic resources for mobile students, such as individual and group counseling and other student services, nutrition programs, and recreational opportunities. • Reduce risk and stress by making efforts to reduce student and teacher turnover, teaching stress management skills, and providing crisis services, such as transition planning. • Support mobile students’ parents in developing parenting skills, fostering bonds with other competent and caring adults, and fostering school engagement. Residential Mobility Research Brief Page 9 of 12 Strategies for the school district: • Improve accessibility and transferability of records to ease the transition of students from one school to another. • Increase the stability of key aspects of academic curricula across schools in the district. Most simply, the knowledge that residential mobility is a risk factor for lower achievement, lower attendance, and higher problem behavior can help to understand the needs of the student body. This report concludes that residential mobility is an issue that should be taken seriously in MNPS middle schools. By identifying these students and targeting them for additional resources, schools may have the ability to overcome the challenges associated with residential mobility. Who are the truly at-risk students? Table 3.1, above, showed that students who are highly mobile—that is, students who have more than three residential moves over a seven-year period—score ten NCE points lower on the Math and Reading TCAP than students experience no moves, and they have more absences and disciplinary referrals. These students should be a priority for schools when they determine who to target for the aforementioned interventions. However, over 16% of the MNPS middle school population has moved more than three times, and given the finite resources with which schools work, it may be difficult (if not unrealistic) to follow the abovementioned strategies for such a large proportion of the student body. A profile for students who are extraordinarily at-risk due to residential mobility can be illustrated by analyzing data only for students who are highly mobile. Among the highly mobile group of students, significant gaps in outcomes are evident for the following groups: Table 5.1 – Group differences in educational and behavioral outcomes among highly mobile students Groups Math TCAP Reading TCAP Absences Black v. non-Black FRPL v. non-FRPL ELL v. non-ELL SPED v. non-SPED -9.5 -13.2 -15.8 -15.6 -8.3 -12.7 -22.2 -12.2 -1.4 3.6 -2.8 5.2 Disciplinary referrals 1.9 1.1* -0.4* 2.2 Notes: (1) TCAP scores are reported in Normal Curve Equivalent scores; (2) absences are combined excused and unexcused absences; (3) * = non-significant differences, statistically; (4) differences in gender and grade level did not predict significant outcome gaps Table 5.1 shows that among the highly mobile population of MNPS students being Black and eligible for the free/reduced priced lunch program, English language learned services, and special education services all serve as “double” risk factors for highly mobile students. Students eligible for special education services have more absences and disciplinary referrals than their peers, as well, among this highly mobile subgroup. These differences in educational and behavioral outcomes by group are indeed evident in the full population of students, not just the highly mobile subgroup. Nevertheless, looking at factors that compound the potentially harmful effects of residential mobility may help direct scarce resources to those most in need. For example, only 2% of students are both highly mobile and eligible for special education services. This is a much more manageable Residential Mobility Research Brief Page 10 of 12 proportion of students to target for supplemental services than the 16% of the student population that is highly mobile. Table 5.2 – Proportion of MNPS middle school students Student characteristics Highly mobile Highly mobile and Black Highly mobile and FRPL Highly mobile and ELL Highly mobile and SPED Percentage of students 16.1 9.2 14.8 1.9 1.8 Table 5.2 shows the percentage of middle school students in MNPS who are highly mobile and members of a second at-risk group. For schools looking for a place to start addressing the issue of residential mobility among their students, students who are eligible for English language learner and special education services may be a manageable subpopulation with which to begin. Schools with a broader approach to targeting at-risk students with supplemental services may consider including all highly mobile students among their service groups. U.S. Census Bureau. (2008). Selected social characteristics in the United States: 2008. 2008 American Community Survey. Retrieved July 25, 2010 from http://factfinder.census.gov. 1 The data used for this report is drawn from a sample of approximately 5,000 students in 11 middle schools in MNPS. These 11 schools were selected due to their high incidences of “indirect assaults” (i.e., behavior that can be characterized as disrespectful, noncompliant, disruptive, threatening/intimidating, use of profanity, displaying gang signs, and repeated rules violation). The descriptive and inferential statistics presented herein are based on these 11 schools and may not be wholly representative of all 37 middle schools in the district. 2 A residential mobility index was created to make mobility between students comparable. MNPS records indicate, on an annual basis, whether or not a student’s home address remains the same from the previous year. If it does not, the student is said to have had a change of address from the previous year, but there is no indication of whether the student experienced multiple changes of address within any given year. There are six years of data collection and therefore six is the maximum number of moves that any one student may have in the data. Because of this, the mobility index artificially deflates the true overall rate of student mobility. Further, because some students have been in the sample for the full six years of data collection while others (for example: fifth-graders or students who transfer into MNPS from an outside district) have not, the total residential moves that the student has experienced while in MNPS is divided by the number of years that they have been a student in MNPS and then multiplied by six: 3 Mobility Index = (Total moves while in MNPS / Number of years in MNPS) x 6 This index acts as an estimate for the number of residential moves that the student would have on record had they been in MNPS for the full six years of data collection, 2003 to 2009. Burkam, D., Lee, V., & Dwyer, J. (2009, June 29). School mobility in the early elementary grades: Frequency and impact from nationally representative data. Presented at the Workshop on the Impact of Mobility and Change on the Lives of Young Children, Schools, and Neighborhoods, Washington, D.C. 4 Residential Mobility Research Brief Page 11 of 12 Astone, N. M., & McLanahan, S. S. (1994). Family structure, residential mobility, and school dropout: A research note. Demography, 31(4), 575–584.; Burkam, D., Lee, V., & Dwyer, J. (2009, June 29). School mobility in the early elementary grades: Frequency and impact from nationally representative data. Presented at the Workshop on the Impact of Mobility and Change on the Lives of Young Children, Schools, and Neighborhoods, Washington, D.C.; Reynolds, A. J., Chen, C., & Herbers, J. E. (2009, June 29). School mobility and educational success: A research synthesis and evidence on prevention. Presented at the Workshop on the Impact of Mobility and Change on the Lives of Young Children, Schools, and Neighborhoods, Washington, D.C.; Temple, J. A., & Reynolds, A. J. (1999). School Mobility and Achievement: Longitudinal Findings From an Urban Cohort. Journal of School Psychology, 37(4), 355–377. 5 Normal Curve Equivalent transformations were introduced by the U.S. Department of Education to allow for comparison of test scores over time and across sites. According to the Tennessee Department of Education website, “The Normal Curve Equivalent shows your student’s scores using a scale that ranges from 1–99. Scores from 1–34 are Below Average, 35–65 are Average, and 66–99 are Above Average” (available online at http://www.state.tn.us/education/assessment/doc/Form_R_Parent.pdf). 6 The statistical model employs a multilevel design to account for the grouping of students by school and the subsequent lack of independence between students in the same school. Residential mobility is included in the model as a set of binary variables (yes = 1, no = 0) to indicate the number of moves a student has made over the six year period from 2003-2009. Binary variables for students who are Black, Latino/a, and eligible for the free and reduced lunch program are included in the model alongside residential mobility as covariates. The output for the regressions with (a) Math TCAP, (b) Reading TCAP, (c) attendance, (d) disciplinary referrals, and (e) perception of school relational climate (SD=0.67) is given below: 7 Outcome variable One residential move over six years (2003-2009) Two residential moves Three or more residential moves Free/reduced lunch program eligible Black Latino/a N (students) * p<.10 ** p<.05 Math TCAP Reading TCAP Absences -1.290 (1.31) -1.538 (1.13) -2.714 ** (1.14) -6.664 *** (1.65) -7.281 *** (1.13) -1.166 (1.36) 4700 *** p<.01 -2.423 (1.54) -1.851 (1.04) -2.734 (1.14) -7.040 (1.53) -6.142 (1.55) -2.870 (2.30) 4670 0.283 (0.37) 0.644 (0.36) 2.408 (0.31) 2.858 (0.49) -2.858 (0.55) -3.837 (0.59) 4906 * ** *** *** * *** *** *** *** Disciplinary referrals Perceptions of school climate 0.293 (0.28) 0.186 (0.22) 0.697 (0.24) 0.865 (0.24) 1.303 (0.20) -0.535 (0.25) 4289 -0.002 (0.03) -0.006 (0.03) 0.017 (0.03) 0.045 (0.04) -0.060 (0.04) 0.089 (0.05) * 4483 *** *** *** ** National Research Council and Institute of Medicine. (2010). Student Mobility: Exploring the Impact of Frequent Moves on Achievement: Summary of a Workshop. A. Beatty, Rapporteur. Committee on the Impact of Mobility and Change on the Lives of Young Children, Schools, and Neighborhoods. Board on Children, Youth, and Families, Division of Behavioral and Social Sciences and Education. Washington, DC: The National Academies Press. 8 Petit, B., & McLanahan, S. (2003). Residential mobility and children’s social capital: Evidence from an experiments. Social Science Quarterly, 84(3), 632-649. 9 Hendershott, A. B. (1989). Residential mobility, social support, and adolescent self-concept. Adolescence, 24(93), 217232. 10 The statistical model employs a multilevel design to account for the grouping of students by school and the subsequent lack of independence between students in the same school. Only school-level variables are included in the model. Average student residential mobility over the six-year period from 2003 to 2009 is included in the model as a continuous variable. The proportion of students in the school who are eligible for the free or reduced lunch program 11 Residential Mobility Research Brief Page 12 of 12 (FRLP) and the average years of experience among teachers in the school are the two covariates included in the model. The output for the regressions with (a) Math TCAP, (b) Reading TCAP, (c) attendance, (d) disciplinary referrals, and (e) perception of school relational climate (SD=0.67) is given below: Outcome variable Average student residential mobility in school Proportion of students in school eligible for FRLP Average years of experience among teachers in school N (schools) * p<.10 ** p<.05 Math TCAP -13.550 ** (6.60) -20.626 *** (6.23) -0.127 (0.32) 11 *** p<.01 Reading TCAP Absences -7.040 (7.58) -25.647 *** (7.17) 0.087 (0.37) 11 4.378 * (2.25) -3.823 * (2.12) -0.120 (0.11) 11 Disciplinary referrals -3.189 ** (1.62) 2.575 * (1.53) -0.229 *** (0.08) 11 Perceptions of school climate -0.654 ** (0.29) 0.472 * (0.27) -0.003 (0.01) 11 Note: The variables “Average student residential mobility in school” and “Proportion of students in school eligible for FRLP” are highly collinear, and this multicollinearity appears to be distorting the parameter estimates in the model with disciplinary referrals as an outcome. In a model with the FRPL proportion variable held out, the residential mobility variable has a significant effect in a negative direction. General Accounting Office. (1994). Elementary school children: Many change schools frequently, harming their education. Washington, D.C.: General Accounting Office.; National Research Council and Institute of Medicine. (2010). Student mobility: Exploring the impact of frequent moves on achievement: Summary of a workshop. A. Beatty, Rapporteur. Committee on the impact of mobility and change on the lives of young children, schools, and neighborhoods. Board on Children, Youth, and Families, Division of Behavioral and Social Sciences and Education. Washington, DC: The National Academies Press. 12