You are on page 1of 25

The Chaos Factor:

A Study of Student Mobility in Indiana1


Claire Smither & Ben Clarke

Each year in Indiana, the Indiana Statewide Testing for Education-


al Progress (ISTEP) is administered to public school students as a
means of academic assessment. The many determinants of ISTEP
scores can be separated into two categories: (1) those characteristics
of the students and (2) those characteristics of the school. After some
investigatory research into these determinants, we became interested
in student educational mobility, the percentage of the student body
that switches schools for reasons other than promotion either within
(intra) or between (inter) the corporations during the school year.
A body of literature supports common intuition that frequently
changing schools lowers student achievement because it generates
chaos for both students and classrooms. High mobility within a cor-
poration can also jeopardize a corporation’s Annual Yearly Progress
(AYP) under the No Child Left Behind (NCLB) Act. Finally, unlike
personal characteristics and family situations of children, corpora-
tions can exercise some influence over the level of mobility with
their own policies locally (intra) and by working with other corpora-
tions (inter). Consequently, we analyzed mobility data for the 292
regular school corporations in Indiana with the goal of answering
the following question: can corporation pass rates on the Indiana
Statewide Testing for Educational Progress (ISTEP) tests be raised
by reducing student mobility? We hypothesized that schools with
elevated levels of intra- and inter-district mobility would have lower
ISTEP pass rates and tested the null hypothesis that mobility has no
effect on ISTEP pass rates. Our research shows that both intra- and

1
journal of undergraduate research

inter-district mobility have significant, negative impacts on corpora-


tions’ average ISTEP pass rates.

Causes of Student Mobility


The persistence of student mobility is a detriment to students and
schools. It disrupts the nature of education by “penetrating the es-
sential activity of schools – the interaction of teachers and students
around learning.”2 Student mobility, while mainly impacting urban
school districts, concerns districts nationwide. Student mobility at
the elementary level is considered the norm; according to data from
the National Assessment of Educational Progress (NAEP), approxi-
mately “34 percent of 4th graders…changed schools at least once in
the previous two years.”3 The causes of student mobility can be bro-
ken down into two categories: (1) purposeful, planned moves and
(2) incidental, impromptu moves. Purposeful educational moves are
generally seen as positive reasons for moving, including such sce-
narios as students and families moving residentially to escape a vio-
lent neighborhood, a mother and students escaping a violent father,
or students being accepted into a magnet program at a different
school.
The model for positive educational moves in economics is known
as the Tiebout Model.4 In this model, consumers move if they are
unhappy with the provided services; in this case, educational con-
sumers move if they are not satisfied with their school. This “voting
with your feet” model is held up as a means of choice for families. As
with most economic models, one of the assumptions of the Tiebout
model is the presence of adequate resources.5 Highly mobile stu-
dents, however, are more likely to be living in poverty6 and in situa-
tions with inadequate resources.7 Moreover, as is often the case with
highly mobile students, the cause of mobility is unrelated to the out-
come; that is, a residential move prompts an educational move even
if the family was happy with the educational product. In the Tie-
bout model, it is hard to disentangle the relevant cause for the move
and the desired outcome without a statement from the movers. An

2
Th e C h a o s F ac t o r : A S t u d y o f S t u de nt Mobil it y in Indiana

important policy nuance concerns corporation enrollment policy;


in corporations with a closed enrollment system, where school as-
signments are based on residential location within a corporation,
residential moves are almost guaranteed to cause school switching,
regardless of the family’s attitude towards the school. The Tiebout
Model would assume that a family switches residence based on their
dissatisfaction with their school, as opposed to the more likely sce-
nario of a family changing residence because of other factors, for
example, eviction, job loss, or divorce, regardless of their feelings
towards the school. Thus the reality of residential and educational
mobility is not always as clear as the Tiebout model presents.
In contrast to the choice models, there are negative reasons for
student mobility: (a) residential moves unrelated to educational con-
cerns, potentially leading to an increasingly unstable home life;8 (b)
parents’ fear of testing the student for special needs or for other such
school-related reasons; and (c) parental issues with school admin-
istration that result in student transfer. To use one example, in the
South Bend Community School Corporation (SBCSC) unrelated
residential moves seem to be the most prevalent cause of student
mobility.9 There are many possible combinations of residential and
educational mobility, all of which must be accounted for when de-
signing local enrollment policies.

The Consequences of Student Mobility


Just as there are many reasons for mobility, there are many conse-
quences and magnitudes felt in the entire educational community.
Many populations are directly impacted by student mobility, mov-
ing from micro to macro: mobile students, their classroom peers,
teachers, and the larger school community itself. Mobility impacts
the interactions between the different agents within a school as well
as the agents themselves.

3
journal of undergraduate research

The Impact of Mobility on Students


Existing studies find that students can suffer psychologically, social-
ly, and academically from mobility.10 Students face the challenge of
becoming acquainted with a new set of rules and regulations in their
new schools – a huge psychological pressure, especially for primary-
level students.11 New social situations are also stressful as children
are still developing social skills; mobility disrupts the trust needed
for the students to be in a place where they can learn and interact
meaningfully with other students. Finally, students suffer academi-
cally because of increased mobility. It is unclear whether the impact
on academic achievement is causational or corollary, since research
supports evidence on both sides. The causation argument (which
we make) states that it is the movement of the students and its con-
sequences that result in lowered academic performance. The corol-
lary story does not dispute the lowered academic performance, but
cites that students who are likely to be mobile are also more likely
to be poor, a member of a minority, and already at risk for poor
school performance. Therefore, the argument goes, student mobility
is correlated with low performance but does not cause it. This story
does not give any weight to the chaos created in the classrooms and
schools. Mobile students are more likely to be poor or minority stu-
dents – characteristics that are also associated with poor academic
achievement.12 Some past studies examining student mobility have
not controlled for family characteristics in their data. It is important
to account for these factors because student mobility is also a proxy
for family stability.13 Thus, when controlling for family characteris-
tics, the impact of mobility is greatly diminished.14
There are three specific negative consequences associated with
a high level of mobility: (1) a higher dropout rate in high school,15
(2) lower test scores,16 and (3) an increase in behavioral issues. Stu-
dents who switch schools are 35 percent more likely to have failed
a grade and they are more likely to lose interest in school and drop
out.17 Student mobility is strongly correlated with low attainment
on tests.18 This correlation could be causal if it is the disruption of
4
Th e C h a o s F ac t o r : A S t u d y o f S t u de nt Mobil it y in Indiana

moving that impedes learning, rather than a mere correlation with


student background characteristics and predicted academic achieve-
ment. Finally, behavioral issues can arise from an unstable home
life and a similarly unstable school life.19 These behavioral issues
thus far associated include a tendency to fight with the teacher and
other students, as well as lack of interest in current and future con-
sequences due to the possibility of frequently changing schools. 20
After controlling for socioeconomic differences, we found that 77
percent of school switchers are reported to have behavioral prob-
lems.21 Children will act out because they cannot develop a strong,
trusting bond with their teachers in a short period of time because
they do not feel as though they are members of the school commu-
nity.22 Mobile students take time and resources to build a trusting
relationship with their teacher, which has an impact on the students
who stay in a classroom.
The Impact on Students who Stay
The students who stay within a classroom are directly impacted by
other students’ mobility because of the introduction of the “chaos”
factor.23 Mobile students require more time from the teacher, de-
creasing teacher availability for the rest of the class. Instructional
routines are disrupted and the pace of instruction could slow to ac-
commodate the new students; the curriculum could become driven
by the needs of mobile students because they are behind.24 Without
adequate resources to address mobile students’ immediate needs,
the resources for non-mobile students could be crowded out in the
classroom.
The Impact on Teachers
Teachers feel the consequences of student mobility in many ways,
perhaps most as an increased workload.25 Often, because of the lack
of consistency in mobility policy, students’ records arrive after the
students do, and the teacher is presented with a pupil but not the
student’s pertinent educational background.26 The student may not

5
journal of undergraduate research

have studied what the rest of the class has, and the teacher must fa-
cilitate catch-up and integration, in addition to maintaining his or
her regular duties.
Teachers are also mainly responsible for the social well-being of
each student. The importance of a trusting bond between students
and teachers cannot be stressed enough, for without it, students are
less likely to learn. Teachers are the primary adults with whom stu-
dents interact; the necessity of trust for meaningful student learning
is paramount. Since the teacher spent the beginning of the school
year building a relationship with his students, it is disruptive to re-
peat the process for each new student throughout the year.
Schools As Communities
Looking at the larger picture, we see that schools lose social capi-
tal with an increasing rate of student mobility; social capital and
student mobility are inversely related.27 Social capital is composed
of social networks and norms, which give a sense of community
among a group of people – in this case, within a school.28 Schools
utilize social capital by creating a community of learners and greater
community members committed to the education of their students;
a sense of community is vital to a school’s success.29 Mobility de-
creases the connection between parents, students, schools, and the
greater community, and dissolves the sense of ownership and need
for participation traditionally present in school communities. Be-
cause mobile students and their schools create a tenuous bond, the
idea that a student would extend the school trust diminishes; mobile
students adapt to moving frequently by accepting their disconnec-
tion from the larger community. Thus, schools feel the impact of
mobility through decreased parental participation, community in-
volvement, and, perhaps most importantly, through decreased trust
between teachers and students. Mobility undermines the role of the
school and school authority in students’ lives because of the imper-
manence of interaction.30

6
Th e C h a o s F ac t o r : A S t u d y o f S t u de nt Mobil it y in Indiana

In addition to morale and community, the school’s academic


prominence could suffer. Schools are held accountable for their en-
rolled students, who may have been elsewhere for a significant por-
tion of the school year, at the time of the ISTEP tests.

Data
Our main data source was the Indiana Department of Education
website.31 When gathering data on student mobility, we found data
from several different sources, none of which ideally explained their
data gathering methods. To further our confusion, the different data
sources did not report the same mobility rates. As we shall explain,
in the end we used the mobility data from Indiana Annual School
Performance Reports (APR).32 When we initially compared these
data to the mobility information from each individual corporation’s
Student Migration data, we observed different rates from those pro-
vided by the APR.33 The Student Migration data shows the number
of students who transferred in and out of a given school for a school
year. For example, for the South Bend Community School Corpo-
ration (SBCSC, Corporation number 7205) the APR disclosed an
intra-district mobility rate of 9.1 percent, while the Student Migra-
tion data, after some tallying calculations, reported an intra-district
mobility rate of 19.27 percent.
We therefore investigated the Student Migration data set and
found reasons that the values of intra- and inter-district mobility
may be inaccurately reported. First, it is important to note that the
Student Migration data is raw, insomuch as it provides the number
of students moving into and out of a school corporation between
two school years, and thus calculations must be made to generate a
meaningful mobility rate. This not only leaves room for calculation
error, but also raises the questions of which definition of mobility
to use as well as how to create that measure. Second, it is also pos-
sible that Student Migration data is providing a count that is below
the actual mobility, since only schools that received or lost five or
more students within the corporation (or received or lost 10 or more

7
journal of undergraduate research

students to another corporation) were included in the student migra-


tion data. Even this definition can be misleading though, as students
were only counted for one move per school year, regardless of their
actual number of moves. Based on conversations with local educa-
tion professionals, it is our opinion that because of this practice of
single counting, any estimates based on these numbers are lower
bounds for any estimates of the impact of mobility; if there is any
bias to our estimates, it is that their impact is understated. With
these concerns, we turned to the APR data.
Since there was no clear explanation of the calculations or data
gathering methods for the APR data, it was difficult to verify the In-
diana Department of Education’s (DOE) work. In the end, however,
we chose to use the APR data because the mobility rate was already
calculated and available for each corporation, and because we have
no reason to believe that the Indiana DOE would be erroneous or
lax in its reporting.34

Methods
Theory
Since our null hypothesis analyzes the consequences of student mo-
bility on ISTEP pass rates, we regressed various independent vari-
ables on our dependent variable of ISTEP pass rates separately for
the math and English sections. This could also be helpful for corpo-
rations if they wish to look at the specific subject that holds the most
promise for improvement. Our final equation is below; we applied
the two dependent variables to the same independent variables:

PCTmath = β0 +β1INTRA +β2 INTER +β3 ELLpct +β4 ATTNpct +β5


STratio + β6 SPEDpct +β7 ENROLL +β8 ENROLLminPCT +β9 FREE-
LUNCHpct +β10 PPE + β12 town + β13 rural + β14 NoChoice + e

Our null hypothesis, which we will work to disprove, is that intra-


and inter-district mobility have no impact on a school corporation’s
ISTEP pass rates for the English and math sections.

8
Th e C h a o s F ac t o r : A S t u d y o f S t u de nt Mobil it y in Indiana

Dependent Variables
We included two different dependent variables, each a different cate-
gory of ISTEP passing rates measured by the average pass rate in the
corporation. These corporation averages are compiled from the aver-
age pass rate for the schools at each grade tested, grades 3 through
10. Our two dependent variables are: (1) PCTmath, the percentage
of students passing the math section, independent of their ISTEP
English score and (2) PCTeng, the percentage of students passing the
English section, independent of their math score. We chose these
two ISTEP pass rates because they are each a measure of AYP for a
given sub-group population under the federal government NCLB
Act,35 and thus important to school corporations. Also, math and
English each have different theoretical learning models, causing
mobility to impact each pass rate differently.
ISTEP tests are given to students in Indiana at the beginning of
the school year. Since there is little time for new curriculum before
students are tested, ISTEP tests are lagged one year. They are associ-
ated with the school year prior to the one in which they were given.
Therefore, while our independent variables are for the 2005-2006
school year, our ISTEP pass rates are from the 2006-2007 school
year.
The descriptive variables for each dependent variable are shown
in Table 1. The first column shows our two dependent variables. The
number of observations, means, standard deviations, minimums,
and maximums are shown for the two dependent variables in the
columns to the right, respectively.

Table 1. Summary of Dependent Variables


Variable Observations Mean Std.Dev. Min Max
PCTmath 292 75.58 7.47 40.53 94.74
PCTenglish 292 72.47 7.39 43.05 92.47

The mean pass rate for students passing math independent of English
is higher than the pass rate for students passing the English section

9
journal of undergraduate research

independent of math. The standard deviations of the two pass rates


are similar. The range for the average math pass rates is wider than
the range for the English average pass rates.
Independent Variables
Our goal was to construct important relationships between the data
that explain the effect of student mobility on 2006-2007 statewide
ISTEP pass rates. We began with a theoretical model based on the
idea that the following independent variables could significantly
explain variations in ISTEP pass rates: the percentage of students
who move from one school to another in the same school corpo-
ration (INTRA), and the percentage of students who move from
one school to another in a different school corporation (INTER). In
order to isolate the effects of mobility, we included numerous inde-
pendent variables to control for the many factors that influence stu-
dents’ academic achievement on standardized tests as measured by
corporation pass rates. In our regression, we wanted to control for
factors within two categories that may influence ISTEP pass rates:
(1) characteristics of schools and (2) characteristics of students.
For characteristics that affect schools, we first included in the
model the average attendance rate of each corporation (ATTNpct),
since students with a higher average attendance rate are more likely
to perform better on standardized tests due to more frequent ex-
posure to class work and in-school academic exercises. Second,
we included the average student-teacher ratio for the corporation
(STratio), since previous studies have shown that lower student-
teacher ratios allow teachers to spend more time and energy on each
individual student, thus more efficiently nourishing their academic
performance.36 Third, we included the per pupil expenditure (PPE)
because corporations that have more money to spend per student
can afford certified teachers, better facilities, and new and updated
textbooks. Fourth, a variable for the geographic setting of the cor-
poration is included (metro, town, or rural) because inner-city and
rural schools are typically underachieving in comparison to their

10
Th e C h a o s F ac t o r : A S t u d y o f S t u de nt Mobil it y in Indiana

suburban counterparts. Fifth, we included a dummy variable if a


corporation has no option for intra-district mobility (NoChoice) –
meaning just one elementary, one middle, and one high school in
the corporation. Finally, the number of students enrolled in the cor-
poration (ENROLL) is also included. By including these variables
in the model, the effects of mobility are isolated. In other words, by
including a variable for the average per pupil expenditure (PPE) for
each corporation, for example, the variations in ISTEP pass rates
can be more substantially attributed to the effects of student mobil-
ity because the average per pupil expenditure is being controlled for
by holding it constant in the regression.
For characteristics that affect students, we included all popu-
lations that are disaggregated by NCLB for their AYP sub-groups.
We first included the percentage of the student population coded as
English Language Learners (ELLpct). Students who are non-native
English speakers typically have a more difficult time grasping con-
cepts in the classroom and performing well on English portions of
standardized tests due to the language barrier. Second, we included
the percentage of the student population coded as special education
students (SPEDpct) since students with special learning needs are
likely to vary in academic performance as a result of those needs.
Third, we included the percentage of the student population who
are racial minorities (ENROLLminPCT) in order to control for the
positive and negative influences that a diverse student body may
have on academic performance. Fourth, we included the percentage
of the student population that qualifies for free or reduced lunch
prices (FREELUNCHpct). To qualify for free or reduced lunch, the
student’s family must be within certain levels of the federal poverty
line, usually between 180 and 130 percent of the federal poverty
line,37 so the FREELUNCHpct variable serves as a proxy for poverty
status. Issues relating to poverty can certainly impact a student’s aca-
demic performance – poverty is especially taxing on children who
are susceptible to malnutrition and lax homework help. Therefore,

11
journal of undergraduate research

we wanted to control for these student variables that would impact


academic performance.
A summary of the independent variables used is shown below
in Table 2. The first column shows the independent variables. The
associated summary statistics – the number of observations, mean,
standard deviation, minimum, and maximum – are shown in the
five columns to the right, respectively.

Table 2. Summary of Independent Variables


Variable Observations Mean Std. Dev. Min Max
INTRA 291 0.776 1.54 0 11.9
IINTER 291 7.46 2.66 1.1 18.9
ELLpct 292 3.54 6.662 0 51.38
ATTNpct 292 96.1 0.842 92.92 100
Stratio 292 17.5 2.07 8.03 23.82
SPEDpct 292 17.1 3.44 9.34 28.54
ENROLL 292 3515.18 4408.28 154 38142
ENROLLminPCT 292 10.3 15.00 0.21 99.34
FREELUNCHpct 292 23.6 12.55 0 81.3
PPE 291 9617.2 1179.91 7400 13400
Metro 292 (20)* 0.068 0.253 0 1
Town 292 (53)* 0.209 0.40 0 1
Rural 292 (120)* 0.410 0.493 0 1
NoChoice 292 (99)* 0.339 0.474 0 1
* Numbers shown in parenthesis are the number of observations for that geographic setting.

As we can see in Table 2, all variables had at least 291 observa-


tions. (For a complete definition of the independent variables, see
Appendix). Our measures of mobility are intra- (INTRA) and inter-
(INTER) district mobility. Since intra- and inter-district mobility
rates are measured on the same scale, the difference in their averages
is striking. While INTRA had a minimum of zero and a mean of
0.7756, INTER had a minimum of 1.1 and a mean of 7.4608. These
measures mean that if the average school corporation in Indiana had

12
Th e C h a o s F ac t o r : A S t u d y o f S t u de nt Mobil it y in Indiana

1000 students, then approximately 7 students would switch schools


within the corporation and 75 students would switch into the cor-
poration from a different corporation or state within a year after Oc-
tober 1st. Thus, on average, school corporations experience ten times
more inter-district mobility than intra-district mobility.
When looking at geographic settings – metro, town, rural, and
no choice – the number in parenthesis shows that more of Indiana’s
schools, 120 of 291, are in a rural setting than in any other geo-
graphic category. The fact that corporations experience more inter-
district mobility than intra-district mobility could be reflective of
the majority of school districts being coded as rural, as rural areas are
less likely to have multiple schools to which students can transfer.
Other notable variables include Per Pupil Expenditure (PPE), which
has a $6,000 range, from $7,400 to $13,400.

Results
The results of our econometric regressions are shown in Table 4.
The first column of the table lists the independent variables we used
in our regression. By including other explanatory variables, we at-
tempted to isolate the effect of intra- and inter-district mobility on
standardized test scores, independent of other characteristics. The
second and third columns show our math and English ISTEP pass
rate regressions, respectively. In these columns, the first number is
the coefficient assigned to each independent variable, and the num-
ber below in parenthesis is the standard error, used to calculate a
measure of significance. The coefficients with an asterisk are statisti-
cally significant from zero at the five percent significance level; co-
efficients with two asterisks are jointly statistically significant from
zero at the five percent significance level. The last row of the table
shows r2, the correlation coefficient. It shows the amount of the vari-
ation in the data for which we account: it is a goodness of fit for our
equation.

13
journal of undergraduate research

Table 4: OLS Estimates of the Effect of Mobility


on Student Performance 2006
Independent Dependent Variable
Variables (1) PCTmath (2) PCTeng
INTRA -1.081* (.437) -0.9095* (.381)

INTER -0.3793* (.150) -0.3558* (.152)

ELLpct -0.0130 (.046) -0.0720 (.047)

ATTNpct 1.2909* (.409) 1.2369* (.324)

Stratio 0.2898 (.178) 0.4783* (.174)

SPEDpct -0.1333 (.089) -0.0127 (.083)

ENROLL 0.0003 (.000) 0.0002 (.000)

ENROLLminPCT -0.1179* (.038) -0.0890* (.033)

FREELUNCHpct -0.2600* (.043) -0.3031* (.038)

PPE 0.0005* (.000) 0.0007* (.000)

rural 1.3696** (1.82) -0.6861** (1.82)

town 1.6008** (1.61) 0.8761** (1.52)

NoChoice -0.4002** (1.93) -1.6708** (1.93)

r2 .6918 .7181
*Statistically Significant at 5% level.
** Geographic settings are jointly significant at 5% level.

Intra- and Inter-District Mobility


The first regression (PCTmath) shows the effect of intra- and in-
ter-district mobility on the percentage of students per corporation
who passed the math section of the fall 2006 ISTEP, independent

14
Th e C h a o s F ac t o r : A S t u d y o f S t u de nt Mobil it y in Indiana

of their performance on the English section. The results indicate


that a one-point increase in the corporation’s intra-district mobility
rate decreases the percentage of students who passed the math sec-
tion by 1.08 percentage points. Meanwhile, a one-point increase in
the corporation’s inter-district mobility rate decreases the percent-
age of students who passed the math section by 0.38 percentage
points. Both of these coefficients are statistically significant. The
inverse of these statements is also true: a one-point decrease in the
corporation’s intra-district mobility rate increases the percentage of
students who passed the math section by 1.08 percentage points and
a one-point decrease in the corporation’s inter-district mobility rate
increases the percentage of students who passed the English section
by 0.38 percentage points.
The second regression (PCTeng) shows the effect of intra- and
inter-district mobility on the percentage of students per corporation
who passed the English section of the fall 2006 ISTEP, regardless of
their performance on the math section. The results indicate that a
one-point increase in the corporation’s intra-district mobility rate
decreases the percentage of students who passed the English section
by 0.91 percentage points; meanwhile, a one-point increase in the
corporation’s inter-district mobility rate decreases the percentage of
students who passed the English section by 0.36 percentage points.
Both of these coefficients are statistically significant.
In all of these regressions, the coefficients on intra- and inter-dis-
trict mobility were negative. Increasing intra- and inter-district mo-
bility rates had a negative effect on ISTEP scores as a whole and on
specific sections. These results make sense: the effects of an increased
mobility rate – e.g. more absences from class, differing curriculum,
the social and psychological influences of adjusting to new learning
environments – do appear to affect test scores negatively, all other
factors being held constant. Additionally, the coefficient of intra-
district mobility had the greatest or second greatest negative value
in both models. This means that of all the other factors that could

15
journal of undergraduate research

negatively affect test scores, the intra-district mobility rate affected


test scores by the greatest amount per unit change.
Remembering the averages for intra- and inter-district mobili-
ties, it is apparent that the magnitudes of the coefficients associated
with the variables do not appear to follow suit. Although on average
school corporations experience almost ten times more inter-district
mobility, the impact for increased inter-district moves is relatively
low compared to that for intra-district moves. In fact, reversing the
trend present in the averages, intra-district mobility impacts ISTEP
pass rates by more than double the magnitude of inter-district mo-
bility rates for both math and English. Nonetheless, it is important
for school corporations to be aware of the importance of intra-dis-
trict mobility as they create their own intra-district transfer and en-
rollment policies.
Attendance
In addition to mobility variables, there are other variables of impor-
tance in these regressions. The attendance rate was significant at the
1 percent level of significance in both the math and English regres-
sions; it had the greatest positive effect on standardized test scores.
This result is logical: students who attend class on a more regular
basis learn better the academic skills necessary to perform well on
tests. What seems more important is that attendance and mobility
are both significant in the same regressions; when including the at-
tendance rate in the regression, the coefficients on both intra- and
inter-district mobility are negative. This means that even when the
attendance rate is held constant, student mobility negatively affects
ISTEP pass rates. This notion is critical because it shows that the
process of switching schools – this chaos factor –negatively impacts
students by means other than just absences from class.
Geographic Settings
The three variables serving as proxies for corporation size and amount
of school choice – rural, town, and NoChoice – are jointly significant

16
Th e C h a o s F ac t o r : A S t u d y o f S t u de nt Mobil it y in Indiana

in both regressions. Since these variables are significant relative to


the omitted variable, the coefficients should be interpreted in com-
parison to metro. The coefficients in Table 4 show the comparison
between students in a metro school corporation and students in each
of the three other settings. The table also supports the estimation
that, in comparison with their metropolitan (urban) counterparts,
students in rural settings (both rural and NoChoice) perform worse
on average, and students in a town setting perform better on aver-
age. As we can see in Table 5, the means for all of the variables are
different for the subgroups of corporations based on the number of
schools they have.

Table 5. Variable Averages by Geographic Population


Geographic Setting
Metro Town Rural NoChoice
Observations 20 53 120 99
# of Schools ≥16 7 to 15 4 to 6 ≤3

Variable
PCTeng 65.54 72.56 73.44 72.66
PCTmath 67.56 74.74 76.84 76.13
INTRA 3.92 1.67 0.47 0.02
INTER 7.55 7.08 7.42 7.70
Stratio 17.07 17.66 17.99 17.03
SPEDpct 17.58 17.03 16.90 17.25
ENROLL 16,080 6,101 2,282 1,087
ENROLLminPCT 37.15 17.07 8.12 3.85
FREELUNCHpct 38.33 26.41 22.02 20.98
PPE $ 10,485.00 $ 9,896.23 $ 9,572.50 $ 9,343.88
ATTNpct 95.61 96.03 96.14 96.12
ELLpct 7.77 4.62 3.69 1.94

The range in their intra-district averages is dramatic, spanning


3.92 in metro areas down to .02 in corporations with no choice.
While the term “no choice” is meant to convey the lack of choice
in public schools in a corporation, the average most likely reflects

17
journal of undergraduate research

the movement from home-school to public school in these corpora-


tions. When looking at Table 5, we see that the regression result that
geographic setting has a statistically significant impact on ISTEP
pass rates is expected.

Limitations
We acknowledge that there is an omitted variable in our research: a
measure of student ability. Certain measurements, such as IQ scores,
indicate a person’s inherent intellectual capacity. Usually innate abil-
ity and academic performance illustrate a strong correlation – so
students who are naturally “smarter” would usually perform better
on standardized tests like the ISTEP. We could not, however, secure
any individual data to control for student ability in our regression.
Thus, when our regression model failed to pass the Ramsey Regres-
sion Equation Specification Error Test (RESET), we were not sur-
prised by this result. The RESET tests for the possibility of omitted
variables in an equation, though it does not test for what the variable
could be. For our equation, a measure of student ability is the likely
culprit. Additionally, given that our r2 was fairly high, we felt that
our regression was accurate for missing such an important predictor
of ISTEP pass rates.
A second problem may exist in our data that could potentially
skew the results. In an ideal study, we would like to include im-
portant characteristics of the population within a corporation to
more accurately control for population characteristics that impact
the school community. While we considered using data from the
Department of Education’s data extract site with census informa-
tion, the data was from the 2000 census and we judged it too dated
to be relevant. Thus, we lacked specific population characteristics
– such as the percentage of adults in the population with less than
a high school education – which could also influence the students’
academic environment, and thus their academic performance. Per-
haps future studies on mobility will be able to include such popula-
tion characteristics.

18
Th e C h a o s F ac t o r : A S t u d y o f S t u de nt Mobil it y in Indiana

Policy Implications
By examining the causes and consequences of student mobility, we
found that certain policy implications come to the forefront for
statewide Indiana school corporations. There are also implications
for mobility policy at the local level for school corporations such
as the South Bend Community School Corporation (SBCSC), and
such policies need to be carried out in a way that has a direct impact
on mobile students.38 In some ways local authorities may not be able
to address student mobility comprehensively, since it is a problem
that extends beyond their jurisdiction into further social realms such
as landlord-tenant relations.39 Policy changes could include modi-
fying the current Hardship Transfer Policy of the SBCSC to allow
students wishing to stay in their original school to do so without
the requirement of personal transportation. This issue is especially
pressing when compared to the transportation provided for students
who attend certain magnet schools, since transportation is provided
for these students regardless of where they live in the district. While
magnet school students usually perform well on ISTEP and other
such tests, the transportation entitlement is striking when compar-
ing “regular” schools and magnet schools, especially when consider-
ing the impact of mobility on students.
Both inter- and intra-district mobility are worth exploring as a
means of understanding the mobility picture. As Hanushek, Kain,
and Rivkin discover, the two types of mobility present different out-
comes for students and come about for different reasons.40 Local au-
thorities must unite for the cause of lowering inter-district mobility,
a goal with great obstacles, in order to limit the negative influence
of student mobility on academic achievement. Requirements would
include keeping accurate records for students in case they need to
transfer so there is no gap in school information; this way, one school
could pick up where another left off. It would also ensure that indi-
vidual student records were transferred between schools efficiently
and in a timely manner, so the student’s new school could appro-
priately gauge the student’s prior academic work and achievement.
19
journal of undergraduate research

Meanwhile, intra-district mobility is a problem with a locally-based


policy solution because a school corporation itself has control over
its enrollment policies.
In some school corporations, there is a lack of consistency at the
school level about mobility policy; i.e., there is no corporation-wide
system for welcoming families into the schools, transferring student
records, or incorporating students into classrooms and schools.41
With mobile students creating the “chaos factor” within a school,
the disorganization at the school and corporation level only adds
to the inconsistency, increasing the impact of mobility on students.
Without a formal policy, there is a real chance that students and
families fall through the cracks, increasing the negative impact of
mobility for students.

Conclusion
The results of our econometric analysis indicate, as previously de-
scribed, that increases in intra- and inter-district mobility rates have
a statistically significant negative effect on ISTEP pass rates. Based
on our results, then, school corporations’ academic performance can
be increased by decreasing student mobility. The specific estimates
for a given decrease in the mobility rate are shown below in Table 6.

Table 6. Changes in Mobility and Pass Rates


  MATH ENGLISH
Decrease of 1%  
INTRA + 1.1% + .91%
INTER + .38% + .36%
   
Decrease of 5%  
INTRA + 5.5% + 4.6%
INTER + 1.9% + 1.8%
   
Decrease of 10%  
INTRA + 10.8% + 9.1%
INTER + 3.8% + 4.6%

20
Th e C h a o s F ac t o r : A S t u d y o f S t u de nt Mobil it y in Indiana

Though one percent changes may seem small, when looking at a


decrease of 10 percentage points in the mobility rate, a corporation
could expect ISTEP pass rates to increase in an almost one-to-one
ratio, approximately 10.8 percentage points for math and 9.1 per-
centage points for English scores.
Students and school corporations could gain by increasing
academic performance through investment in policies designed
to reduce student mobility. Proposals for such policies are abun-
dant in academic papers; Rumberger has several pages of sugges-
tions.42 While the goal of this research was to determine the effect of
student mobility on academic performance, suggestions for policies
to reduce student mobility abound, it is to be hoped that schools
choose to employ them.

21
journal of undergraduate research

Appendix: Glossary of Variables


Dependent Variables
PCTeng = percentage of students passing only the English section of
the ISTEP.
PCTmath = percentage of students passing only the Math section of
the ISTEP.

Independent Variables
Corp = Corporation code number, unique for each corporation.
INTRA = percentage of students who move from one school to an-
other in the same school corporation; the APR data was taken from
the difference between the school roster on October 1st of two con-
secutive school years, only looking at students who transferred within
the school corporation.
INTER = percentage of students who move from one school to an-
other in a different school corporation; the APR data was taken from
the difference between the school roster on October 1st of two con-
secutive school years, only looking at students who transferred from
outside of the corporation.
ELLpct = percent of the corporation’s student population coded as
English Language Learners
ATTNpct = average percent of attendance per corporation.
STratio = ratio of the number of students enrolled to full-time equiva-
lent teachers per corporation.
SPEDpct = percent of corporation’s population that is coded as special
education students.
ENROLL = number of students enrolled per corporation.
ENROLLminpct = percentage of students enrolled who are of minor-
ity ethnicity, non-white.
FREELUNCHpct = percentage of students receiving free lunch per
corporation, having a family income below 130 percent of the pov-
erty line.
PPE = average per pupil expenditure in a corporation

22
Th e C h a o s F ac t o r : A S t u d y o f S t u de nt Mobil it y in Indiana

Metro = calculated proxy variable for school corporation geographic


setting. We designated a school corporation “metro” if it had 16 or
more schools.
Town = calculated proxy variable for school corporation geographic set-
ting. We designated a school corporation a “town” if it had between
7 and 15 schools, inclusive.
Rural = calculated proxy variable for school corporation geographic set-
ting. We designated a school corporation “rural” if it had between 4
and 6 schools, inclusive.
NoChoice = calculated proxy variable for school corporation geograph-
ic setting. We designated a school corporation “NoChoice” if it had
3 or fewer schools.

23
journal of undergraduate research

Endnotes
1
The authors wish to thank Professor Jennifer Warlick and the Fall 2007 class
of ECON 40520, The Economics of Education for their helpful revisions and
support. Claire wishes to thank Dean Stuart Greene for his support while she
was writing the literature review for this paper, originally created for his ESS
senior research seminar. Finally, part of the database used in this paper was
created for Claire’s senior thesis during summer 2007, funded by UROP. Both
authors are most appreciative of UROP’s support.
2
D. Kerbow, Patterns of Urban Student Mobility and Local School Reform.
Journal of Education for Students Paced at Risk, 1 (no.2, 1996): 1.
3
R.W. Rumberger, “The Causes and Consequences of Student Mobility,” Jour-
nal of Negro Education 72 (no.1, 2003): 6.
4
E.A. Hanushek, J.F. Kain, & S.G. Rivkin, “Disruption Versus Tiebout Im-
provements: The Costs and Benefits of Switching Schools,” National Bureau
of Economic Research, Working Paper 8479. http://www.nber.org/papers/
w8479 (Accessed 7 November 2007).
5
Ibid.
6
Rumberger, “The Causes and Consequences of Student Mobility,” 6-21.
7
G. M. Ingersol, J.P. Scamman, & W.D. Eckerling, “Geographic Mobility and
Student Achievement in an Urban Setting,” Educational Evaluation and Policy
Analysis 13 (no.2, 1989): 143-149.
8
C.B. Swanson & B. Schneider,“Students on the Move: Residential and Edu-
cational Mobility in America’s Schools,” Sociology of Education 72 (no.1,
1999): 54-67.
9
Angie Buysee, interview by the author, South Bend, Indiana (28 November
2007).
10
Rumberger, “The Causes and Consequences of Student Mobility,” 6-21.
11
Ibid.
12
Ibid.
13
C.J. Tucker, J. Marx, & L. Long, “Moving On:” Residential Mobility and
Children’s School Lives. Sociology of Education 71 (no. 2, 1998): 111-129.
14
Ibid.
15
R. Felner, J. Primavera & A. Cauce, “The Impact of School Transitions: A
Focus for Preventive Efforts.” American Journal of Community Psychology 9
(1981): 449-459.
16
Hanushek, “Disruption Versus Tiebout Improvements.”
17
Felner, “The Impact of School Transitions,” 449-459.
18
Hanushek, “Disruption Versus Tiebout Improvements.”
19
D. R. Sanderson, Veteran Teachers’ Perspectives on Student Mobility. Essays
in Education 4 (2003): 1-17.
20
Ibid.

24
Th e C h a o s F ac t o r : A S t u d y o f S t u de nt Mobil it y in Indiana

21
Tucker, “Moving On,” 111-129.
22
Jay Bankowski, interview by the author, South Bend, Indiana (28 November
2007).
23
A.A. Lash & S.L. Kirkpatrick, “A Classroom Perspective on Student Mobility,”
The Elementary School Journal 91 (no.2, 1990): 176-191.
24
Rumberger, “The Causes and Consequences of Student Mobility,” 6-21.
25
D.R. Sanderson, “Engaging Highly Transient Students,” Education 123 (no.3,
2003): 600-605.
26
D.B. Schuler, “Effects of family mobility on student achievement,” Education-
al Research Spectrum 8 (no.4, 1990): 17-24.
27
Bankowski, interview by the author.
28
R.D. Putnam, “Bowling Alone: America’s Declining Social Capital,” Journal
of Democracy 6 (no.1, 1995): 65-78.
29
V.E. Lee, “Catholic Lessons for Public Schools,” Chapter 6 in D. Ravitch and
J. Viteritti, New Schools for a New Century, 147-163, (New Haven, CT.: Yale
University Press, 1997).
30
Tucker, “Moving On,” 111-129.
31
http://www.doe.state.in.us/.
32
Data is also available on the website as School Snapshots http://mustang.doe.
state.in.us/SEARCH/search.cfm and http://www.doe.state.in.us/htmls/perfor-
mance.html.
33
CORP 7205, for example, http://mustang.doe.state.in.us/SEARsCH/feed-
corp.cfm?corp=7205.
34
When building our database, we used several means of gathering data: (1)
drawing data down straight from the IN Department of Education website,
(2) we found the data in hard copy and hand entered it, and (3) we extracted
the data like (1), but manipulated it to get the measure we wanted. The vari-
ables were entered into the database as follows. Method 1: ENROLL. Method
2: INTRA, INTER, PPE. Method 3: ELLpct, ATTNpct, STratio, SPEDpct,
ENROLLminPCT, FREELUNCHpct, metro, town, rural, nochoice.
35
US Department of Education, No Child Left Behind. http://www.ed.gov/
nclb/landing.jhtml.
36
36 F. Mosteller, “The Tennessee Study of Class Size in the Early School
Grades.” The Future of Children 5 (no.2, 1995): 113-127.
37
US Department of Agriculture, Food and Nutrition Services. http://www.fns.
usda.gov/cnd/governance/notices/iegs/iegs.htm.
38
Kerbow, “Patterns of Urban Student Mobility,” 147-169.
39
Schuler, “Effects of family mobility on student achievement,” 17-24.
40
Hanushek, “Disruption Versus Tiebout Improvements.”
41
U.S. General Accounting Office (1994). Elementary school children: Many
Change Schools Frequently, Harming Their Education. Washington, DC.
42
Rumberger, “The Causes and Consequences of Student Mobility,” 6-21.

25

You might also like