The Policy Choices of Effective Principals

1. The Policy Choicesof Effective Principals David Figlio, Northwestern U/NBER Tim Sass, Florida State U July 2010

2. Introduction Considerable research energy on measuring teacher (and now, principal) value added Increasing evidence of large variation in principal effectiveness; unsurprisingly, easily measured characteristics aren’t particularly explanatory Very little knowledge of what effective principals actually do, and whether these actions might be transferable

3. This Project Merge measures of principal value added with rich survey data on instructional policies and practices in Florida Survey data collected on three occasions for 80% of all Florida public schools; today’s first pass: 2001-02 and 2003-04 surveys of elementary school principals Panel nature of survey allows us to observe policy changes that come about with principal changes at a school

4. Research Questions Do school policies change when the identity of the principal changes? What are the policy choices of effective principals? Do high value added principals “bring with them” different policy/practice choices from their prior schools? Are the policy/practice choices of new principals in a school different for principals with high value added?

5. Defining Principal Value Added Estimate “value-added” model of student achievement, including principal-by-year fixed effects Time period = 2000/01-2004/05 school years Estimate four models which vary by: Extent of persistence in prior inputs Complete persistence (dependent variable is achievement gain) vs. partial persistence (dependent variable is current achievement level, lagged score as an explanatory variable) Inclusion vs. exclusion of controls for teacher characteristics If exclude teacher characteristics, principal value-added measure includes average teacher quality

6. Correlation of Principal Value-Added Measures

7. School Policies and Practices Survey asked 65 questions about instructional policies and practices Because of a school budget constraint, and the fact that these are often variations of a theme, we combine these questions into domains Domains are weighted averages of individual policy responses, weighted by the variation in the question response

8. Policy and Practice Domains Policies to improve low-performing students Lengthening instructional time Reduced class size for specific subjects Narrowing of curriculum Systems of scheduling and class organization Policies to improve low-performing teachers Teacher resources Teacher incentives Teacher autonomy

9. Example: Scheduling Systems • Component questions: • Block scheduling • Common prep periods • Subject matter specialist teachers • Organizing teachers into teams • Looping • Multi-age structure • Other scheduling systems

10. Example: Policies to Improve Low-Performing Students • Component questions: • Require grade retention • Require summer school • Require before/after school tutoring • Require in-school supplemental instruction • Require tutoring • Require Saturday school • Require other policy

11. Do New Principals Affect School Policy? For each of nine policy domains (Z), regress absolute value of difference in policy in 2003/04 and 2001/02 at school k on a constant and an indicator for change in identity of principal (DP):

12. Estimated Effect of Principal Change on Normalized Absolute Value of Policy Change

13. A Simple Model of School Policy Assume current policy in school k with principal j is related to policy at school k in prior period, whether a change in school leadership has occurred (DP), and policy preferences of the new principal (DP x C):

14. Do Principals Bring their old Policies to New Schools? • Dependent variable: school policy domain in 2003/04 • Right hand size variables: • School policy domain in 2001/02 • New principal since 2001/02 • Interaction between new principal and new principal’s 2001/02 policy domain in their prior school

15. Do Principals Bring their Old Policies to New Schools?

16. Does New Principal Value Added Relate to School Policy Adoption? • Dependent variable: school policy domain in 2003/04 • Right hand size variables: • School policy domain in 2001/02 • New principal since 2001/02 • Interaction between new principal and new principal’s 2001/02 value added • Graph shows value-added from model with partial persistence and controls for teacher characteristics

17. Does New Principal Value Added Relate to School Policy Adoption?

18. Summary • Too early to draw any firm conclusions • Early evidence suggests that: • New principals do have an impact on school policies even in the short run (2 years or less) • In most cases principals do not simply “port” their policies from one school to another • High value added principals implement different new policies when they arrive at a new school • Tend to favor policies directed toward improving low-performing students and low-performing teachers