in Response to Act 1329 Dr. Gary Ritter Kaitlin Anderson Office - PowerPoint PPT Presentation

School Discipline and Student Achievement: An Overview of Results in Response to Act 1329 Dr. Gary Ritter Kaitlin Anderson Office for Education Policy University of Arkansas Presentation for the Arkansas State Board of Education August 13, 2015 1

• AR Education Reports • Policy Briefs • Report Cards • Newsletters • Data Resources www.officeforeducationpolicy.org/ 2

Agenda 1. Introduction and Motivation 2. Four Questions and Results 3. Conclusion and Next Steps 3

Introduction and Motivation 4

Background on School Discipline - US • US Department of Education’s Office for Civil Rights Database (2014) – African-American students without disabilities are more than three times as likely as their white peers without disabilities to be expelled or suspended – Over 50% of students involved in school-related arrests or referred to law enforcement are Hispanic or African-American 5

Introduction to Arkansas Act 1329 • OEP presented on this topic in July, 2014 in response to Act 1329: An Act to Evaluate the Impact of School Discipline on Student Achievement; And For Other Purposes – Annual report to include • District enrollment, subgroup enrollment, disciplinary rates, achievement, and disciplinary disparity between subgroups • Possible disciplinary strategies and resources Arkansas school districts can access 6

Challenges of Interpreting Data • Unclear how to interpret any potential disparities – Act 1329: “Disparity in discipline rates does not necessarily indicate discrimination ; it can result from an ineffective school climate or from cultural strategies that are not successful in engaging the academic efforts of all students.” • This year, we can improve upon prior presentations with new data … • Next year, we will examine even further 7

Questions and Results 8

Questions to Answer Today I. How often do students get cited for behavior & which types of schools write up students most often? II. Which types of schools give stricter (more days) punishments, for the same infractions? III. Which types of students (on average) receive stricter punishments for the same infractions? IV. Finally, within school , do specific types of students receive stricter punishments for the same infractions? 9

Q1: Which types of schools write up students most often? • Infractions per 100 students per year by type of school • Grouped by severity of infraction: “Severe 6” infractions lead to ISS, OSS, or Expulsion in at least 90% of the cases: – Fighting, gang-related activity, drugs, alcohol, knives, guns • Overview of Results: – SW and SE regions had the most infractions – Surprisingly, the smallest and largest schools had the most infractions – Jr. High Schools and HS had highest rates – The more African-American students, the more infractions per student – The poorer the school (by % FRL), the more infractions per student – Lower performing schools had more infractions per student 10

Q1: SW and SE are Trouble Spots Number of Infractions Per 100 Students Per Year (2010- 11 to 2012-13) 63.4 63.1 Total Infractions 44.5 42.4 44.2 6.8 4.0 "Severe 6" Infractions 5.4 4.6 3.0 - 10.0 20.0 30.0 40.0 50.0 60.0 70.0 Southeast Southwest Central Northeast Northwest 11

Q1: Junior High Schools, High Schools Cite Students More Frequently Number of Infractions Per 100 Students Per Year by School Type (2010-11 to 2012-13) 42.9 55.7 Total Infractions 89.6 63.9 41.1 38.3 48.3 Other Infractions 82.3 59.1 37.8 4.6 7.4 "Severe 6" Infractions 7.3 4.8 3.4 0.0 20.0 40.0 60.0 80.0 100.0 Elementary Middle Junior High Senior High Comprehensive K-12/Other 12

Q1: Smallest and Largest Schools Cite Students More Frequently 13

Q1: Predominantly African-American Schools Cite Students More Frequently Number of Infractions Per 100 Students Per Year by % African-American (10-11 to 12-13) 80.0 74.0 74.1 70.9 70.0 69.1 67.0 69.8 62.4 61.0 60.0 50.0 46.9 More than 3x 43.6 40.0 mostly white 32.2 30.0 28.4 23.2 26.5 22.0 schools 29.2 26.1 23.6 20.0 20.7 19.3 7.1 10.0 6.7 9.9 4.2 3.0 2.6 2.8 3.2 2.7 2.3 - Decile 1 Decile 2 Decile 3 Decile 4 Decile 5 Decile 6 Decile 7 Decile 8 Decile 9 Decile 10 (Lowest % (Highest % AA) AA) "Severe 6" Infractions Other Infractions Total Infractions 14

Q1: Low-Income Schools Cite Students More Frequently Number of Infractions Per 100 Students Per Year by % FRL-Eligible (10-11 to 12-13) 80.0 74.3 70.0 66.5 59.4 60.0 54.5 45.8 50.0 40.1 More than 2x 40.0 41.8 richest 31.2 36.5 30.0 schools 28.0 20.0 10.0 4.9 4.0 3.6 7.8 3.2 - Quintile 1 (Least Quintile 2 Quintile 3 Quintile 4 Quintile 5 (Most FRL) FRL) "Severe 6" Infractions Other Infractions Total Infractions 15

Q1: Results By Benchmark Scores Number of Infractions Per 100 Students Per Year by Benchmark "GPA" (10-11 to 12-13) 120.0 Almost 8x difference between 104.5 100.0 highest performing and lowest performing schools 80.0 93.3 60.0 45.2 45.9 40.0 40.0 41.8 24.0 13.7 20.0 21.4 11.3 12.7 4.1 5.2 2.7 1.0 - Quintile 1 (Low Quintile 2 Quintile 3 Quintile 4 Quintile 5 (High Performing) Performing) "Severe 6" Infractions Other Infractions Total Infractions And 11x difference in rate of “Severe 6 16

Yes, certain types of schools are writing students up for misbehavior more often … So … how about the consequences? 17

Q2: Which types of schools give stricter (more days) punishments, for the same infractions? Overview of Analytic Strategy: • For every infraction, we observe punishment, and compute an average punishment (number of days) for each infraction • “Strictness” = punishment longer or shorter than average • The “residual” for each infraction is the number of days punished above (positive) or below (negative) average • Then, we can see what types of schools tend to assign longer (more strict) punishments based on school averages 18

Q2: Which types of schools give stricter (more days) punishments, for the same infractions? Step 1: Using multivariate regression analysis, create residuals (measure of severity of punishment above or below average) at infraction level: 𝑒𝑏𝑧𝑡_𝑞𝑣𝑜 𝑗𝑢 = 𝛾 0 + 𝛾 1 𝑕𝑠𝑏𝑒𝑓 𝑗𝑢 + 𝛾 2 𝑗𝑜𝑔𝑠𝑏𝑑𝑢𝑗𝑝𝑜_𝑢𝑧𝑞𝑓 𝑗𝑢 + 𝛾 3 𝑗𝑜𝑔𝑠𝑏𝑑𝑢𝑗𝑝𝑜_𝑝𝑠𝑒𝑓𝑠 𝑗𝑢 + 𝛾 4 𝑡𝑗𝑢𝑓_𝑣𝑡𝑓 𝑗𝑢 +𝑣 𝑗𝑢 Our statistical model controls for: 𝑕𝑠𝑏𝑒𝑓 𝑗𝑢 = a vector of grade dummies (with 8 th grade as baseline) 𝑗𝑜𝑔𝑠𝑏𝑑𝑢𝑗𝑝𝑜_𝑢𝑧𝑞𝑓 𝑗𝑢 = a vector of infraction dummies (disorderly conduct as baseline) 𝑗𝑜𝑔𝑠𝑏𝑑𝑢𝑗𝑝𝑜_𝑝𝑠𝑒𝑓𝑠 𝑗𝑢 = a vector of dummies for first, second, third, fourth, fifth, sixth, seven or more infraction (student by school by year) plus last infraction (student by school by year) 𝑡𝑗𝑢𝑓_𝑣𝑡𝑓 𝑗𝑢 = a vector of dummies for school site-use (Pre-K/Kindergarten, Elementary, Middle, Junior High, High school, Comprehensive K-12) 19

Q2: Which types of schools give stricter (more days) punishments, for the same infractions? Step 2: Using multivariate regression analysis, predict average residuals at school level using school characteristics 𝑏𝑤𝑓𝑠𝑏𝑕𝑓_𝑠𝑓𝑡𝑗𝑒𝑣𝑏𝑚 𝑡𝑢 = 𝛾 0 + 𝛾 1 ln 𝑓𝑜𝑠𝑝𝑚𝑚𝑛𝑓𝑜𝑢 𝑡𝑢 + 𝛾 2 𝑡𝑗𝑢𝑓_𝑣𝑡𝑓 𝑡𝑢 + 𝛾 3 𝑞𝑓𝑠𝑑_𝐺𝑆𝑀 𝑡𝑢 + 𝛾 4 𝑞𝑓𝑠𝑑_𝐵𝐵 𝑡𝑢 + 𝛾 5 𝑝𝑤𝑓𝑠𝑏𝑚𝑚_𝑢𝑓𝑡𝑢_𝐻𝑄𝐵 𝑡𝑢 + 𝛾 6 𝑠𝑓𝑕𝑗𝑝𝑜 𝑡𝑢+ 𝑣 𝑗𝑢 Our statistical model controls for: ln 𝑓𝑜𝑠𝑝𝑚𝑚𝑛𝑓𝑜𝑢 𝑡𝑢 =natural log of school enrollment 𝑡𝑗𝑢𝑓_𝑣𝑡𝑓 𝑡𝑢 = a vector of dummies for school site-use (Elementary, Middle/Junior High, High school, Comprehensive K-12, with Elementary as the baseline) 𝑞𝑓𝑠𝑑_𝐺𝑆𝑀 𝑡𝑢 = school % free-and reduced-lunch eligible 𝑞𝑓𝑠𝑑_𝐵𝐵 𝑡𝑢 = school % African American 𝑝𝑤𝑓𝑠𝑏𝑚𝑚_𝑢𝑓𝑡𝑢_𝐻𝑄𝐵 𝑡𝑢 = average “GPA” of benchmark and EOC scores where 1 = Below Basic and 4 = Advanced 𝑠𝑓𝑕𝑗𝑝𝑜 𝑡𝑢 = Geographical Region (baseline = Northwest) 20

Q2: Which Schools are Stricter? School Average Residual Ln(enrollment) 0.111 (0.0769) Using a statistical strategy to % FRL-Eligible -0.123 consider all of these variables at (0.228) once, we tend to see … % African American 1.567 *** (0.189) Longer Punishments Overall Test Score GPA -0.219 • (0.183) Schools with more African- Northeast Region -0.246 *** American students (0.0908) Central Region -0.242 ** Shorter Punishments (0.101) • Northeast and Central (relative to Southwest Region -0.135 Northwest, Southwest, Southeast) (0.103) Southeast Region 0.388 (0.271) Constant -0.0960 (1.012) Observations 2,780 R-squared 0.064 Robust standard errors in parentheses 21 *** p<0.01, ** p<0.05, * p<0.1