U.S. School District Dropout Prevention and Recovery Practices Linked to Graduation Rate Performance Daniel Princiotta Renee Ryberg Johns Hopkins University School of Education Child Trends Bethesda Policy Research, LLC Presented at the Annual Meeting of the American Educational Research Association Philadelphia, PA | April 4, 2014
Introduction • One in five eighth graders drops out during high school 5 • 600,000 public high school students drop out each year 3 • About 76 percent of U.S. public high school students graduated on- time in 2008-09 11 • Costs of the dropout problem are profound • On average, each dropout costs the public sector $209,100 over a lifetime 7 • Dropouts cost the United States more than $300 billion per year 1
Introduction • Little national research on district factors tied to graduation rates 9 • Graduation rates vary substantially by district. In 2007-08: 10 • Los Angeles Unified School District: 49 percent graduation rate • Montgomery County Public Schools in Maryland: 87 percent graduation rate • As of 2012, districts are reporting graduation rates the same way and being held accountable for graduation rate improvement 12 • But what are districts doing to help solve the dropout problem, and is there any evidence of their effectiveness?
Purpose • Provide a snapshot of the prevalence of certain district-level dropout prevention practices in the United States • Dropout Early Warning Systems • Informational intervention strategies • Dropout recovery strategies • Compare dropout prevention practices of districts that are over- and under-performing with respect to graduation rates • An exercise in “benchmarking” 6 • Identify whether particular practices are tied to better-than-expected district graduation rate performance
Data and sample • Fast Response Survey System (FRSS) Dropout Prevention Services and Programs Survey (2010-11) • Covers a wide array of district dropout prevention services and programs 2 • FRSS Surveyed 1,090 school districts • When weighted appropriately, represent 13,400 districts nationwide • Common Core of Data (CCD 2008-09) • American Community Survey (ACS) data (2006-2010) • Small Area Income and Poverty Estimates (SAIPE 2003-06)
Methodology • Step 1: Generate an Averaged Freshman Graduation Rate (AFGR) for all eligible school districts in the country (10,520 districts, 1,090 FRSS districts) • Step 2: Model AFGR as a function of school district and community predictor variables, as well as state (9,210 districts, 800 FRSS districts) • Step 3: Generate a predicted graduation rate for each district based on this model • Step 4: Determine the difference between each district’s observed and predicted graduation rates • Step 5: Classify districts as over- or under-performing (top/bottom 20 percent) • Step 6 : Compare dropout prevention practices among over- and under- performing districts using FRSS sample (130 and 160 districts, respectively, representing 1,820 and 1,850 districts nationwide)
Methodology: Modeling AFGR
Predictor variables • CCD (2008-09) • American Community Survey (2006-10) • Gender (% male) • Locale (rural, city, suburb, town) • Educational Attainment • % less than a high school credential • Race/ethnicity • % w/ a bachelor's degree or higher • % American Indian/Alaskan Native • Household type (% w/ 2 parents) • % Black • % Hispanic • Mobility (% moved in past year) • SAIPE (2003-06) • Poverty Rate
District dropout prevention practices investigated • Dropout Early Warning Systems (prevalence & indicators used) • Informational intervention strategies. Provide information on: • Financial and employment implications of dropping out • Other educational or training options (alternative schools, GED, job training) • Dropout recovery strategies • Try to determine status of students who did not return to school in the fall • Follow up with school-year dropouts to encourage them to return before the next school year
Results
Figure 1. Percentage of districts with a Dropout Early Warning System: 2010–11 41 43 47 53 57 59 43 57 Over-Peforming Under-Peforming All Districts Districts Districts Have dropout early warning system Do not have a dropout early warning system
Figure 2A. Percentage of DEWS districts that reported using various factors to identify at-risk students to a moderate or large extent: 2010–11 Factors used to identify at-risk students 97 Academic failure (grades/credits/retention) 100 97 93 Truancy or excessive absences 97 95 88 Behaviors that warrant suspension or expulsion 95 86 81 Involvement with criminal justice system 87 80 78 Failure on state standardized tests 76 76 0 20 40 60 80 100 Percent All districts Over-performing districts Under-performing districts
Figure 2B. Percentage of DEWS districts that reported using various factors to identify at-risk students to a moderate or large extent: 2010–11 Factors used to identify at-risk students 77 Behaviors that warrant other disciplinary action 80 75 76 Substance abuse 89 76 74 Change in student attitude or life conditions 78 69 74 * Homelessness or frequent address change 86 68 67 Pregnancy/teen parenthood 69 63 0 20 40 60 80 100 Percent All districts Over-performing districts Under-performing districts
Figure 2C. Percentage of DEWS districts that reported using various factors to identify at-risk students to a moderate or large extent: 2010–11 Factors used to identify at-risk students 65 Learning disability as indicated in an IEP 63 72 64 Involvement with social services or foster care 62 62 61 Mental health problems 66 60 54 Limited English proficiency 49 51 43 Migrant status 39 33 0 20 40 60 80 100 Percent All districts Over-performing districts Under-performing districts
Figure 3. Percentage of districts that provide information about the employment or financial consequences of dropping out to likely dropouts: 2010–11 Percent 100 * 80 71 57 60 53 38 40 29 * 24 18 20 5 6 0 All students Some students No students All districts Overperforming districts Underperforming districts
Figures 4A-D. Percentage of districts that provide information about schooling options to likely dropouts: 2010–11 Percent Percent Alternative Schools or Programs Combined Job Training + GED Programs 100 100 79 80 80 70 * 65 63 60 60 49 43 39 40 40 32 26 25 * 22 19 18 * 14 20 20 13 12 7 3 0 0 All students Some students No students All students Some students No students All districts Overperforming districts Underperforming districts All districts Overperforming districts Underperforming districts Percent GED or Adult Ed Programs Percent Job Training Programs 100 100 80 80 68 58 60 60 * 51 49 42 40 35 40 40 33 32 31 29 26 24 25 23 20 20 20 8 7 0 0 All students Some students No students All students Some students No students All districts Overperforming districts Underperforming districts All districts Overperforming districts Underperforming districts
Figure 5. Percentage of districts that try to determine the status of students who did not return to school in the fall as expected: 2010–11 Percent 100 78 77 80 73 60 40 19 14 20 13 12 9 4 0 All students Some students No students All districts Overperforming districts Underperforming districts
Figure 6. Percentage of districts that follow up with school-year dropouts to encourage them to return before the next school year : 2010–11 Percent 100 80 * 55 60 52 37 36 * 34 40 31 30 14 20 11 0 All students Some students No students All districts Overperforming districts Underperforming districts
Conclusion: Implications • Six in 10 districts lack DEWS, leaving substantial room for improvement in targeting dropout prevention efforts efficiently • Investigate address changes and homelessness as DEWS indicators, particularly in districts with high student mobility/homelessness • Substantial overlap exists between over-performing and under- performing district practices • It’s not just about what practices are in place, but how well they are implemented • It is critical to systematically apply dropout prevention and recovery strategies to all relevant students
Conclusions: Future research directions to address current limitations • Running analysis using new 2010-11 cohort graduation rates • Use updated predictor variables aligned with new base year • Strengthen over/under-performing metric by • Generating school-level predictions and then aggregating to district level • Incorporating student achievement data from EDFacts into model • Impute missing district-level data • Investigate additional district dropout prevention practices • Cross-check district-reported survey responses with outside data • Perform a multivariate investigation into dropout prevention practices
Recommend
More recommend
