Analyzing the Effects of Distance on Student Attendance Rates ...
ANALYZING THE EFFECTS OF DISTANCE ON STUDENT ATTENDANCE RATES USING A GIS‐ BASED METHOD Taylor Krieger, MPP Office of Accountability School District of Philadelphia
Attendance in Philadelphia schools is one of the lowest nationwide • The average high school student attends class only 84% of the time. – This result in nearly 29 days worth of absence. • State test scores are 15‐20% less for chronically absent students
• Factors affecting attendance – Parents views on education – Performance – Distance??
Philadelphia School Action Plans • Plan of best practices • Used to address school level problems of crime and safety, academic performance, and attendance. • Created by principals and staff of every school • Presented to central administrators for comments
Does distance affect Average Daily Attendance (ADA)? • Anecdotal evidence “My students miss class because they live far away.”
• Previously done studies – Focus on general attendance in short‐term public health programs – Do not focus on students – Do not consider school attendance
Distance did not affect attendance to these programs
Current attendance‐improvement policies target trans pass students • Subsidized public transportation – Substitute for a district level bus system – ≥1.5 mile eligibility
• Public service announcements (TV, radio, billboards) – Substantial presence of PSAs in public transportation – High costs in marketing funds and administrative hours • Could this be put to better use?
How effective are these policies? Are new policies necessary?
USING GIS… IS THERE A SPATIAL RELATIONSHIP BETWEEN DISTANCE AND ATTENDANCE?
Methods • Buffer definition Radial distance of .5 miles ≥10.5 miles defined as one buffer
• Students flagged based on intersection with buffer • Additional flags given for each buffer intersected
Flagging Students by Buffer Intersection • Distance group |Sum of flags – 20|
• Students grouped by distance • Each group has ADA weighted by total days enrolled
One‐Half Mile Buffer Intervals by School
Example Script
Philadelphia Public High Schools • Neighborhood High Schools – General Admission by catchment area – Default school
• Citywide Admissions – Some admissions Criteria – Often Career and Technically Oriented – Lottery Based
• Magnet Schools – Special Admissions – Strict Criteria (students removed if chronically absent)
Enrollment by Distance and School Type Percent of Whole All Students
School Type Neighborhood Special Admissions Citywide Admission
Enrollment 33,656 9,557 7,158
Enrollment by Distance and School Type Count of All Students
Neighborhood Student Counts ≤ 1.5 mles
> 1.5 miles
≤ 3 miles
> 3 miles
23288
10134
29460
3962
ADA by Distance and School Type For all Students
School Type Neighborhood Citywide Magnet
ADA Coefficient 0.026 0.020 0.022
ADA Significance 0.000 0.089 0.035
Cases 33422 7158 9612
Looking at Other Cohorts Do atrisk groups show a relationship between distance and attendance?
Looking at Other Cohorts 6year Cohort Dropout Rates by Ethnicity and Gender 60% 51% 50% 44%
43%
39%
40%
30%
43%
36%
36%
32%
30%
37%
36%
GIRLS
30%
29%
BOYS
25%
ALL
20%
20%
10%
0% African American
Asian
Latino
White
*Source: The African American and Latino Male Dropout Task Force Report, September 2 2010
Other
Enrollment by Distance and School Type Percent of Whole Black Males Only
School Type Neighborhood Special Admissions Citywide Admission
Enrollment 11059 1707 2455
Enrollment by Distance and School Type Percent of Whole Black Males Only
Neighborhood Student Counts ≤ 1.5 mles
> 1.5 miles
≤ 3 miles
> 3 miles
7530
3529
9621
1438
ADA by Distance and School Type Percent of Whole Black Males Only
School Type Neighborhood Citywide Magnet
ADA Coefficient ADA Significance .033 .000 .005 .800 .054 .026
Cases 11059 2455 1707
Enrollment by Distance and School Type Percent of Whole Latino Males Only
School Type Neighborhood Special Admissions Citywide Admission
Enrollment 3182 500 302
ADA by Distance and School Type Percent of Whole Latino Males Only
School Type Neighborhood Citywide Magnet
ADA Coefficient ADA Significance .029 .108 .003 .951 .104 .072
Cases 3182 500 302
Summary • Distance appears to have no relationship with attendance rates – Zero effect for all cohorts
• Attendance for Citywide schools higher overall despite distance • Is increased distance illustrative of increased preference?
2009 Transit Strike Absence comparisons
• Remember: Students beyond 1.5 miles are eligible for subsidized public transportation • November 2009 Septa strike – Three day disruption in all public transportation
• Also occurred 2005 – Five days disruption
2009 Transit Strike Absence comparisons REGION 26‐Oct (Monday)
Out of Buffer Students Absences PERCENT Neighborhood All Other
In Buffer Student Absences PERCENT Neighborhood All Other
28.9%
19.5%
33.9%
25.3%
27‐Oct
33.6%
18.8%
35.3%
24.1%
28‐Oct
33.6%
19.1%
36.3%
25.0%
29‐Oct
31.2%
17.9%
32.1%
23.1%
30‐Oct
33.7%
18.7%
37.6%
26.2%
29.8%
17.0%
31.7%
20.0%
31.8%
18.5%
34.5%
24.0%
4‐Nov
42.4%
30.2%
30.2%
19.5%
5‐Nov
38.1%
27.5%
29.8%
19.3%
6‐Nov
35.7%
22.4%
27.1%
14.0%
2‐Nov (Monday) Pre‐Strike Average
Conclusions • Distance should not be a determinant in matching students to schools – Goodness of fit will likely have more of an effect
• Human and financial resources should be directed towards other means of truancy reduction – Increased support for 8th grade counselors would help students find a school best suited for their needs. – Virtually all Latino males are enrolled in neighborhood schools, enrollment in a “better fit” could increase attendance and performance
• GIS not just about showing spatial relationships, but also showing where they do not exist despite popular thought
Future Work Distance, no… …time, yes?
ANY QUESTIONS?
Attendance in Philadelphia schools is one of the lowest nationwide • The average high school student attends class only 84% of the time. – This result in nearly 29 days worth of absence. • State test scores are 15‐20% less for chronically absent students
• Factors affecting attendance – Parents views on education – Performance – Distance??
Philadelphia School Action Plans • Plan of best practices • Used to address school level problems of crime and safety, academic performance, and attendance. • Created by principals and staff of every school • Presented to central administrators for comments
Does distance affect Average Daily Attendance (ADA)? • Anecdotal evidence “My students miss class because they live far away.”
• Previously done studies – Focus on general attendance in short‐term public health programs – Do not focus on students – Do not consider school attendance
Distance did not affect attendance to these programs
Current attendance‐improvement policies target trans pass students • Subsidized public transportation – Substitute for a district level bus system – ≥1.5 mile eligibility
• Public service announcements (TV, radio, billboards) – Substantial presence of PSAs in public transportation – High costs in marketing funds and administrative hours • Could this be put to better use?
How effective are these policies? Are new policies necessary?
USING GIS… IS THERE A SPATIAL RELATIONSHIP BETWEEN DISTANCE AND ATTENDANCE?
Methods • Buffer definition Radial distance of .5 miles ≥10.5 miles defined as one buffer
• Students flagged based on intersection with buffer • Additional flags given for each buffer intersected
Flagging Students by Buffer Intersection • Distance group |Sum of flags – 20|
• Students grouped by distance • Each group has ADA weighted by total days enrolled
One‐Half Mile Buffer Intervals by School
Example Script
Philadelphia Public High Schools • Neighborhood High Schools – General Admission by catchment area – Default school
• Citywide Admissions – Some admissions Criteria – Often Career and Technically Oriented – Lottery Based
• Magnet Schools – Special Admissions – Strict Criteria (students removed if chronically absent)
Enrollment by Distance and School Type Percent of Whole All Students
School Type Neighborhood Special Admissions Citywide Admission
Enrollment 33,656 9,557 7,158
Enrollment by Distance and School Type Count of All Students
Neighborhood Student Counts ≤ 1.5 mles
> 1.5 miles
≤ 3 miles
> 3 miles
23288
10134
29460
3962
ADA by Distance and School Type For all Students
School Type Neighborhood Citywide Magnet
ADA Coefficient 0.026 0.020 0.022
ADA Significance 0.000 0.089 0.035
Cases 33422 7158 9612
Looking at Other Cohorts Do atrisk groups show a relationship between distance and attendance?
Looking at Other Cohorts 6year Cohort Dropout Rates by Ethnicity and Gender 60% 51% 50% 44%
43%
39%
40%
30%
43%
36%
36%
32%
30%
37%
36%
GIRLS
30%
29%
BOYS
25%
ALL
20%
20%
10%
0% African American
Asian
Latino
White
*Source: The African American and Latino Male Dropout Task Force Report, September 2 2010
Other
Enrollment by Distance and School Type Percent of Whole Black Males Only
School Type Neighborhood Special Admissions Citywide Admission
Enrollment 11059 1707 2455
Enrollment by Distance and School Type Percent of Whole Black Males Only
Neighborhood Student Counts ≤ 1.5 mles
> 1.5 miles
≤ 3 miles
> 3 miles
7530
3529
9621
1438
ADA by Distance and School Type Percent of Whole Black Males Only
School Type Neighborhood Citywide Magnet
ADA Coefficient ADA Significance .033 .000 .005 .800 .054 .026
Cases 11059 2455 1707
Enrollment by Distance and School Type Percent of Whole Latino Males Only
School Type Neighborhood Special Admissions Citywide Admission
Enrollment 3182 500 302
ADA by Distance and School Type Percent of Whole Latino Males Only
School Type Neighborhood Citywide Magnet
ADA Coefficient ADA Significance .029 .108 .003 .951 .104 .072
Cases 3182 500 302
Summary • Distance appears to have no relationship with attendance rates – Zero effect for all cohorts
• Attendance for Citywide schools higher overall despite distance • Is increased distance illustrative of increased preference?
2009 Transit Strike Absence comparisons
• Remember: Students beyond 1.5 miles are eligible for subsidized public transportation • November 2009 Septa strike – Three day disruption in all public transportation
• Also occurred 2005 – Five days disruption
2009 Transit Strike Absence comparisons REGION 26‐Oct (Monday)
Out of Buffer Students Absences PERCENT Neighborhood All Other
In Buffer Student Absences PERCENT Neighborhood All Other
28.9%
19.5%
33.9%
25.3%
27‐Oct
33.6%
18.8%
35.3%
24.1%
28‐Oct
33.6%
19.1%
36.3%
25.0%
29‐Oct
31.2%
17.9%
32.1%
23.1%
30‐Oct
33.7%
18.7%
37.6%
26.2%
29.8%
17.0%
31.7%
20.0%
31.8%
18.5%
34.5%
24.0%
4‐Nov
42.4%
30.2%
30.2%
19.5%
5‐Nov
38.1%
27.5%
29.8%
19.3%
6‐Nov
35.7%
22.4%
27.1%
14.0%
2‐Nov (Monday) Pre‐Strike Average
Conclusions • Distance should not be a determinant in matching students to schools – Goodness of fit will likely have more of an effect
• Human and financial resources should be directed towards other means of truancy reduction – Increased support for 8th grade counselors would help students find a school best suited for their needs. – Virtually all Latino males are enrolled in neighborhood schools, enrollment in a “better fit” could increase attendance and performance
• GIS not just about showing spatial relationships, but also showing where they do not exist despite popular thought
Future Work Distance, no… …time, yes?
ANY QUESTIONS?
