Beruflich Dokumente
Kultur Dokumente
The purpose of this study is to develop early indicators for the Montgomery Public Schools
system to identify students who are at-risk of not graduating. Similar studies have been
conducted around the country to assist policy-makers and administrators in developing strategies
to address the drop-out problem.
This report is modeled on earlier work that the Public Affairs Research Council of Alabama
(PARCA) performed with the Mobile Area Education Foundation and the Mobile County Public
Schools, with the assistance of Dr. Bob Balfanz of Johns Hopkins University. PARCA’s work is
supported by the Montgomery Education Foundation, with technical assistance from the
Montgomery Public Schools, Q3 Solutions, of Prospect, Kentucky, and the Alabama State
Department of Education.
1
Data and Methodology
The findings of this study rely on the analysis of a specific cohort. It is based on a statistical
sample drawn randomly to test any particular hypotheses about influences of particular factors.
It is a benchmarking endeavor specific to the schools which these students attended. Thus, the
purpose is not a generalization about education in any manner, but about the functioning of a
particular school system during a particular point in time with this specific cohort.
Montgomery Public Schools provided STI databases for the school years 2002-2003 through
2009-2010. These data sets were mounted on a server and updated to SQL Server 2008 format
for analysis. Q3 Solutions provided an extraction of 130,836 records for 2,937 students in the
cohort enrolled in 6th Grade during the 2002-2003, hereafter 2003, school year. A complete
listing of the record fields is provided in Appendix I.
The methodology of this study relies on the extraction of data for a cohort of students who were
in 6th Grade in 2003, which are the earliest available data to us. Relevant indicators were
selected for each student for each year. These included socio-economic status, ethnicity, grades,
attendance, and disciplinary records for each student. This study compares their performance at
the 6th Grade and follows each student through the year 2010, one year beyond their normal
graduation year of 2009.
One of the features of a cohort analysis is that there is a certain mortality associated with the
cohort. Some of the original members disappear from the data set as a result of moving away, or
transferring to another school, etc. PARCA has reviewed the coding for withdrawals and
developed the following rules for keeping a student in the cohort for the purpose of calculating
risk for dropping out or not graduating.
Appendix II provides a detailed table on the transfer codes that are associated with students who
were removed from the analysis as their presence in the schools terminated for reasons that
appear to be valid educational alternatives. Codes related to withdrawals referring to the home
were not excluded as they tend to occur well past the point at which an alternative educational
experience culminating in a valid diploma seems to be a likely or plausible outcome.
Some withdrawal codes indicate that the student transferred to home school or a church/home,
we have no evidence that such a code results in a successful educational outcome, and suspect
that many of the students coded as such are merely dropping out and going home.
2
A number of students whose records included withdraw codes continued to have records in the
system after the year in which a withdrawal was coded, including 29 students who have a
withdrawal code indicating that they had transferred out, but also appeared in the data with a
valid graduation code. Those students were not eliminated from the cohort analysis.
2,937 Students
– 549 Transferred
– 82 Incomplete Records
2,306 Adjusted Base Cohort
The cohort was reduced by 549 to account for transfers out of the system, and by another 82
graduates for whom incomplete records prevented comparisons relevant indicators. That
resulted in a remaining 2,306 students for whom our tracking applies in the subsequent analysis.
Tables 1 and 2 provide a breakdown of the students who were removed from the analysis due to
transfers to schools outside the system.
TABLE 1
TABLE 2
3
IDENTIFICATION OF GRADUATES
The Alabama State Department of Education (ALSDE) website listed 1,219 graduates for 2009.
The cohort from the STI database initially showed a count of 1,556 students with graduation
codes. This count was reviewed again by MPS and ALSDE. A second list of students from the
Montgomery Public Schools (MPS) for 2009 or 2010 yielded 903 matches. A second match was
performed by the ALSDE that yielded a total of 1,111 graduates. Complete records for the
duration of the period (2006-2010) were not available for 82 of those graduates, resulting in
1,029 graduates to compare to non-graduates.
To summarize, this study is based on total of 2,306 students in the 2003 6th Grade cohort that
included a total of 1,029 graduates in 2009 and 2010.
Indicators
While there are many factors cited in the literature on dropouts that have been found to influence
behaviors related to dropping out of school, the intent of this study is to develop indicators, using
existing administrative data to identify students who are not likely to graduate. The intention is
to provide early indicators that can be detected in 6th grade so that school staff can provide
services targeting those students who are at risk of not graduating.
In collaboration with the Montgomery Education Foundation, PARCA examined three major
factors. In order of significance they are:
1. Unexcused Absences
2. Suspensions
3. Course Failures
Chart 1 provides an overview of outcomes and indicators for the cohort. Individual tables and
charts for each factor follow, with the percentage of those students with and without that
indicator, and the percentages of each who graduated. Some students withdraw and return to the
system.
Unless a student transferred out of the system without any further records or graduated, but for
whom records were incomplete (records terminate in years prior to graduation), we have
included that student in the subsequent analysis. See Appendix III for other limitations in the
data. Chart 1 on page 6 displays the percentages for the students remaining in the analysis of the
cohort.
4
2003 6th Grade Cohort Results with Indicators*
One or more
unexcused absences
270
One or more failed courses 12%
61
3% One or more suspensions
8 Failures/Suspensions
0% 2
0%
No Indicators
587
25%
All Three
Indicators
114
Overlapping Indicators 5%
Suspensions/Absences
5
351
15% 159
7%
Graduates Failures/Absences
CHART1: 2003 5th Grade Cohort Results with Indictors
1029 76
45% 3%
TABLE 3
The graduation rate for poverty students is nine percentage points below that of non-poverty
students in this cohort. White students graduated at lower rates than any other group, which is
typically not the case in the general population, and male students were less likely than female
students to graduate. The lowest graduation rate is for Hispanic male students.
TABLE 4
Race, Sex and Graduation
Graduate? Percent
Race No Yes Total Graduating
ALL Students 1277 1029 2306 45%
ALL Females 550 591 1,141 52%
ALL Males 727 438 1,165 38%
ASIAN 15 14 29 48%
Female 7 8 15 53%
Male 8 6 14 43%
BLACK 943 842 1785 47%
Female 401 501 902 56%
Male 542 341 883 39%
HISPANIC 13 2 15 13%
Female 4 1 5 20%
Male 9 1 10 10%
NO RESPONSE 3 1 4 25%
Female 3 3 0%
Male 1 1 100%
WHITE 303 170 473 36%
Female 135 81 216 38%
Male 168 89 257 35%
6
THE IMPACT OF UNEXCUSED ABSENCES IN 6TH GRADE
TABLE 5
FIGURE 1
2,306
6th GradeCohort
Students
One or More
Absences No absences
938 1,368
TABLE 6
7
THE IMPACT OF SUSPENSIONS IN 6TH GRADE
TABLE 7
FIGURE 2
2,306
6th GradeCohort
Students
One or More
Suspensions No Suspensons
394 1,912
TABLE 8
8
THE IMPACT OF FAILING A COURSE IN 6TH GRADE
TABLE 9
FIGURE 3
2,306
6th GradeCohort
Students
TABLE 10
9
THE IMPACT OF GRADES in 6TH GRADE
TABLE 11
FIGURE 4
6th Grade
Students*
Avg. GPA = 2.38
TABLE 12
10
FIGURE 5: OVERLAPPING INDICATORS AND NO INDICATORS
TABLE 13
11
OTHER INDICATORS ARE NEEDED
TABLE 14
Of the 1,277 students who did not graduate, 587 did not have an indicator as recorded by MPS
data. Nevertheless, the overwhelming majority of students who do not graduate, and do not have
an indicator, simply disappear without any recorded explanation.
Many have records that indicate the date of departure, but nothing else. Some codes indicate that
a reason was given, but not specified in the records received, and others were simply blank (98),
while 263 could not be found at all in a search by ALSDE.
We do know something about their grades in 6th Grade. Of the 587 students, grade records were
not available for 138. The remainder of students who did not graduate had an average GPA of
2.47, which is above average (2.38) for all students.
TABLE 15
12
Study indicators identified only 54 percent of the students who did not graduate. As a result, 46
percent of the students who did not graduate could not have been identified by using these
indicators. Given that almost half of the students who do not graduate exhibited none of the
indicators examined, what is not known is almost as important as is known.
Most disturbing are more than 500 students who did not graduate and disappeared from the
records with no explanation. Of the students who disappear without an indicator, more than 40
percent disappeared from the records received before the age of 16. Of those, 239 students, or 98
percent of those who disappear before age 16, had no explanation for their disappearance from
the records.
TABLE 16
Obviously, more information is needed to identify students who did not have one of these three
major indicators related to graduation outcomes.
13
SUMMARY AND CONCLUSION
The most critical single indicator for graduation is a failing grade in 6th Grade, with only 14
percent graduating. Only 295 students failed a class. They accounted for 23 percent of all
students who did not graduate.
The next most critical indicator is a suspension, with 28 percent graduating. Only 22 percent of
those who did not graduate had a suspension.
Only 34 percent of students with unexcused absences graduated. They accounted for almost
half, 48 percent, of those who do not graduate.
Having all three indicators was nearly fatal to graduation, with those students having only a 10
percent chance of graduating.
Despite these strong indicators, 54 percent of students with no indicators at all did not
graduate.
14
APPENDINX I
Database Fields:
studentnumber
SchoolYear
SCHOOLNUMBER
RACEDESCR
SEX
DOB
GRADE
SPECED
LUNCHDESCR
pupil_days
ExcusedAbsence
UnexcusedAbsence
enrollment_code
withdrawal_code*
Icode
Infraction
Dcode
Disposition
InfractionTeacherName
NbrOfDays
CourseName
CourseTeacherName
FinalScore
Gradcode
Diploma
Failures
Absences
Disciplinary Actions
Suspensions
Corporal Punishment
*Additional Fields from an extract by the Alabama State Department of Education included
description of withdrawal codes.
For the purposes of this study, graduates are defined as having an advanced diploma or a regular
diploma. Other codes included alternative adult diploma, graduation certification –special
education (137), and occupational diplomas.
15
APPENDIX II
16
APPENDIX III
1. Codes are not consistent or not consistently entered over time (e.g. Disciplinary,
Withdraw Codes, etc)
2. Codes that do not fit the expected range of values (e.g. 12/35/97, or nonexistent WA9 for
withdrawal codes)
3. Conflicts in coding for the same observation/case.
a. There may be multiple withdrawal codes for the same student in the same year
(this may be true, but no dates are specified and re-entry dates are not available).
b. There may be conflicts in the demographic data for the same student:
i. Age
ii. Date of birth
iii. Race/Ethnicity
4. Logical conflicts occur regarding withdrawal and graduates, including multiple
withdrawals and completion codes for the same student: For example a student may have
a graduation code in 2009, but also reported attendance in 2010.
5. Graduates appear in the data without any attendance in the prior year and there are
graduates who do have not have enough attendance or not enough classes to graduate.
6. In some cases, attendance and /or Out-of-School suspensions exceed total days per year.
7. Years and grades regress, or go backwards without explanation (e.g. in 10th grade in
2008, but in 9th in 2009).
8. Some data entries were simply incomplete, for example, 6/21/ for a date.
9. Graduation and diploma codes do not match.
10. Student ID numbers were incomplete, missing digits, or had two additional digits.
17