1.08k likes | 1.36k Views
Trial Slides . Michael Podgursky Department of Economics University of Missouri – Columbia. Analysis of Student Level 2006 MAP data. Inequality of student achievement. The vast majority of inequality in student achievement as measured by MAP is within rather than between school districts.
E N D
Trial Slides Michael Podgursky Department of Economics University of Missouri – Columbia
Inequality of student achievement • The vast majority of inequality in student achievement as measured by MAP is within rather than between school districts. • If all differences in average student achievement between Missouri school districts were eliminated, 85-90 percent of the inequality of student achievement would remain.
Student Achievement InequalityTotal Variation in State MAP Achievement = W + B W= variation within districts B= differences in average scores between districts
The vast majority of the variation in student achievement is within rather than between schools • If all differences in average student achievement between Missouri schools were eliminated, 80-95 percent of the inequality of student achievement would remain.
Frequency distributions of student test scores • Plaintiff and non-plaintiff districts • High and low spending districts • Knowing whether the student attended a high or low spending district tells us almost nothing about his test score
Distribution of 2006 8th Grade Math MAP Scores in Plaintiff And Non-Plaintiff Districts
Distribution of 2006 8th Grade Communication Arts MAP Scores in Plaintiff And Non-Plaintiff Districts
Distribution of 2006 8th Grade Math MAP Scores in Top 20 And Bottom 20 Percent of Districts by Current Spending per Student
Distribution of 2006 8th Grade Communication Arts MAP Scores in Top 20 And Bottom 20 Percent of Districts by Current Spending per Student
Scattergrams of individual student test scores against district spending per student • No detectable positive relationship between student achievement and spending per student.
2006 Grade 8 Student Math MAP Scores and Current Spending Per Student
2006 Grade 8 Student Communication Arts MAP Scores and Current Spending Per Student
2006 Grade 8 Student Math MAP Scores and Current Spending Per Student: African-American Students Who Are FRL-Eligible
2006 Grade 8 Student Com. Arts MAP Scores and Current Spending Per Student: African-American Students Who Are FRL-Eligible
2006 Grade 8 Student Math MAP Scores and Current Spending Per Student: White Students Who Are FRL-Eligible
2006 Grade 8 Student Com. Arts MAP Scores and Current Spending Per Student: White Students Who Are FRL-Eligible
Sample Selection • 2006 MAP Scores by district and grade • Grades 3-8, 10 Math • Grades 3-8, 11 Communication Arts • Sample restricted to districts/grades with at least 10 test-takers in 2006 • Current spending per student in 2006
2006 Grade 3 Communication Arts MAP Index Score and Per Pupil Current Spending
2006 Grade 3 Math Index Score and Per Pupil Current Spending
2006 Grade 4 Communication Arts MAP Index Score and Per Pupil Current Spending
2006 Grade 4 Math MAP Index Score and Per Pupil Current Spending
2006 Grade 5 Communication Arts MAP Index Score and Per Pupil Current Spending
2006 Grade 5 Math Index Score and Per Pupil Current Spending
2006 Grade 6 Communication Arts MAP Index Score and Per Pupil Current Spending
2006 Grade 6 Math MAP Index Score and Per Pupil Current Spending
2006 Grade 7 Communication Arts MAP Index Score and Per Pupil Current Spending
2006 Grade 7 Math MAP Index Score and Per Pupil Current Spending
2006 Grade 8 Communication Arts MAP Index Score and Per Pupil Current Spending
2006 Grade 8 Communication Arts MAP Index Score and Per Pupil Current Spending (p = plaintiff district)
2006 Grade 8 Math MAP Index Score and Per Pupil Current Spending
2006 Grade 8 Math MAP Index Score and Per Pupil Current Spending
2006 Grade 11 Communication Arts MAP Index Score and Per Pupil Current Spending
2006 Grade 10 Math MAP Index Score and Per Pupil Current Spending
Multivariate Regression Models • A(t) = b0 + b1 S(t) • A(t) = b0 + b1 S(t) + b2 %FRL(t) • A(t) = b0 + b1 S(t) + b2 %FRL(t) + b3 A(t-k) • A(t) = b0 + b1 S(t) + b2 %FRL(t) + b3 A(t-k) + b4 %MIN(t) A(t), A(t-k) = MAP achievement in the district in year t, t-k S(t) = Current spending per student in the district in year t in thousands of dollars FRL(t) = Percent of district students FRL-eligible in year t MIN(t) = Percent of district students minority (unweighted OLS estimates)
Example of Value-Added Model A(g8, 2006) = b0 + b1 S(2006) + b2 %FRL(2006) + b3 A(g3, 2001) Communication Arts Achievement, Grade 8, 2006 Communication Arts, Grade 3, 2001 (same cohort of students, earliest year MAP available)
Percent Proficient or Advanced: 2006 Grade 8 and High School
4. Value-added + %Min 2. %FRL only 3. Value-added % FRL only 1. No controls Value-added No controls
Of four pairs of value-added estimates of the marginal effect of spending only one pair is statistically significant. • In all cases, the point estimates of the marginal effects are small, and sometimes negative. • All point estimates were less than one percentage point gain in %proficient or advanced per $1000 of additional spending • Roughly 60 percent of students scored less than proficient in 2006
What does “proficient” mean? • Variable and subjective measure • Early and current MAP standards differ • High school math MAP has challenging questions • Many students who score less than proficient on HS MAP succeed as freshman in Missouri colleges.
How MAP Performance-Level Cut Scores Were Set(early MAP) • Expert Panels (~40 members per panel) • Educators, educated lay public, school board members • “Bookmark” method • Math (1997) • Communication Arts (1998)