MPS High School Evaluation
Council of the Great City Schools Annual Fall Conference
October, 2010
Deb Lindsey, Milwaukee Public SchoolsBradley Carl, Wisconsin Center for Education Research
High School Evaluation: Purpose, Design, Methodology• Evaluation designed as comparisons
between high school types (Small vs. Large, Charter vs. Non-Charter, etc.)– Not designed as a comparison/ranking of
individual high schools– Not all schools included in some
comparisons– Small/Large cutoff: 400 students– “Selectivity” defined somewhat narrowly:
admissions requirement (4 schools)
High School Evaluation: Purpose, Design, Methodology• 2 context measures (enrollment &
demographics) + 11 outcome metrics (completion rate, test scores, attendance, suspensions, etc.)
• Time period generally 2004-05 through 2008-09, corresponding to start of HS Redesign process
• Descriptive data + inferential (regression) analysis to account for differences in students served
High School Evaluation: Key Findings
Enrollment: distinct and purposeful shift toward smaller high schools Corresponding increases in enrollment
(numerical and “market share”) for several subsets of small high schools: small charters, small newly-created charters, etc.
Demographics: no evidence of any school types consistently under-serving student subgroups of interest (SpEd, ELL, etc.)
High School Evaluation: Key Findings
• WKCE Test Performance:– Stagnant rates of non-cohort proficiency on
Grade 10 tests (% Proficient + Advanced); small increases in Grade 10 mean scale scores
– Small schools appear to serve lower-performing students overall
– Same-student gains (Grade 8-10; Fall 2005-Fall 2007 and Fall 2006-Fall 2008) for non-mobile students with matched tests:• Most remain in same proficiency level, but more
students drop 1+categories than increase• No school types produce consistently superior gains for
both Reading/Math for both growth cohorts; greater variation within school types than between
High School Evaluation: Key Findings
Mobility: higher rates of within-year mobility for Small sites, both descriptively (unadjusted) and inferentially (regression-adjusted): Based on month-to-month changes in school
of enrollment during 2005-06 and 2008-09 Regression controls for student/school
demographics + prior (grade 8) attendance & mobility
Variation in mobility again higher within school types than across school types
High School Evaluation: Key Findings
• Within-Year (September-May) Grade 9 Reading & Math Benchmark Gains:– Similar pattern to WKCE: no school type
produces consistently superior gains, either descriptively or regression-adjusted; lower prior achievement in Small sites
– Again, greater variation within than between types– Low participation rates may bias results
High School Evaluation: Key Findings
• Retention rates for first-time 9th graders: Small schools have higher rates descriptively (2005-06 through 2008-09), but lower regression-adjusted rate for 2008-09– Controls used in regression: student and
school-level demographics & prior retention history (past 5 years)
– Again, substantial within-type variance
High School Evaluation: Key Findings
• Attendance: 75-80% for grades 9-12 across school types; no significant change for most school types over past 5 years– Descriptive data: higher attendance rates
(grades 9-12) for Large sites– Regression-adjusted data: higher for first-
time 9th graders in Small sites after controlling for demographics + prior (grade 8) attendance history
– Again, large variance within school types
High School Evaluation: Key Findings
• GPA (overall and core subject): – Low GPA for all school types (1.5 - 1.8);
marginal (if any) improvement– GPA (both types) lower in Small sites;
likely reflects lower ability levels upon entering high school
High School Evaluation: Key Findings
• High School Completion Rate: – Insufficient data to make meaningful
comparisons (new schools/types + small student counts + high grade 9 retention rates)
– “Total Quality Credits” (during Year 1 of H.S. and overall) used as proxy for progression through high school; higher TQC attainment in Large sites, but (again) substantial within-type variance
Conclusions
• Conclusion 1: very limited evidence of systemic improvement in high school outcomes studied
• Conclusion 2: some evidence that Small high schools overall serve different student populations (lower prior achievement, etc.)
• Conclusion 3: some high school types fare better on some outcomes, but very limited evidence of consistently superior outcomes for any school type across all years/metrics– Outcome variance generally greater within school
types than across
Implications for HS Reform
• Change focus to a quality control/ performance management approach that encourages diversity of offerings + accountability for results – Select most valued/relevant outcomes,
establish expectations, monitor results–MPS already taking steps in this direction:
evaluating charters, EdStat, data warehouse reports, etc.