Assistance and Accountability in Externally Managed Schools: The Case of Edison Schools, Inc.

Abstract

Edison Schools, Inc., is the largest and most visible among a growing number of Education Management Organizations that have entered into contracts to manage public schools, including both conventional and charter schools. Edison's approach to managing schools is comprehensive, and it distinguishes itself from most other school improvement strategies by simultaneously addressing both the resources and assistance provided to schools and the accountability systems under which school staff operate. In this article we explore the ways in which the assistance and resources provided by Edison (including diverse professional development opportunities, materials, technology, and other tools), as well as accountability mechanisms (such as monitoring and rewards), have translated into principal and teacher actions, and the factors that facilitated or constrained educators' efforts to implement the Edison design and improve teaching and learning. Drawing on data gathered from extensive interviews, observations, and document reviews collected during a four-year comprehensive study of Edison schools, we demonstrate how Edison intends to promote not only educators' capacity but also their motivation and opportunity to deliver high-quality instruction. We examine variation that occurs across schools as teachers and principals respond to these system-level efforts. In addition, we identify several important predictors of variation in implementation, including the strength of instructional leadership provided by the principal and the presence or absence of district-imposed constraints such as union contract rules.

Portions of this article were drawn directly from a RAND report (see Gill et al., 2005).

Notes

¹ For further details, see Gill et al. (2005)

² For further details of Edison's history in Philadelphia and the financial ups and downs, see Gill et al. (2005)

³ Not all 23 schools participated in the study from the start. We initially selected 15 schools in 2001. By 2003, three of the original 15 case study schools were no longer under Edison operation and a fourth elected to drop out of the study (and soon thereafter terminated its relationship with Edison). We replaced these four schools with four schools that were new to Edison, permitting us to maintain our sample size and a sample that better represented Edison's current portfolio of schools. In the fall of 2004, the RAND study team concluded that it would be useful to conduct a few additional school site visits as our study neared completion. Rather than return to schools we had previously visited, we elected to add four new elementary schools selected to add more balance in terms of governance and number of years under Edison's management

⁴ This includes methods to increase student motivation to achieve (e.g., displaying exemplary work), to involve parents in supporting the school (e.g., advisory councils), and to recognize and reward staff for performance. See Gill et al. (2005) for further discussion

⁵ On a reading/math implementation scale ranging from one to two, Edison schools with strong instructional leaders had a mean score of 1.89, whereas schools without strong instructional leaders had a mean score of 1.61 (N = 18). On a nontested subjects implementation scale ranging from one to two, Edison schools with strong instructional leaders had a mean score of 1.68, while schools without strong instructional leaders had a mean score of 1.29 (N = 18). In both cases, differences were statistically significant at p < .05

⁶ Schools without substantial local constraints had a mean score of 1.77 on the professional environment index (for which scores ranged from one to two), whereas schools with substantial local constraints had a mean score of 1.57. The difference is statistically significant at p < .05

⁷ One of these schools was a very troubled 1st-year Edison school in Philadelphia, whereas the other was a long-time Edison school, which not long after our visit ended its contract. Both schools had serious problems with leadership and morale

⁸ Edison schools typically reported some sort of teacher turnover rate, but reported rates were based on local definitions of turnover and were therefore not necessarily comparable across Edison schools. We were unable to calculate an Edison-wide teacher turnover rate with confidence

⁹ Five of eight charter schools in the sample were coded as having strong instructional principals, whereas only two of ten district schools were rated with strong instructional principals. (In five schools, we lacked sufficient information to make a judgment about instructional leadership.

¹⁰ In the small sample of case study schools for which we were able to rate instructional leadership, only one of seven (14%) schools that later ended relationships with Edison had strong instructional leaders, whereas six of 11 (55%) schools that remained with Edison had strong instructional leaders

¹¹ The sample size for these analyses is less than the total number of case study schools because complete achievement data were not available for all case study schools

¹² Similarly, Zhang, Shkolnik, and Fashola (2005) found that schools that had been implementing a comprehensive reformmodel for three to five years and that were rated as strong implementers achieved larger test-score gains than schools of similar vintage that were judged to be low implementing

¹³ Note that this effect size cannot be directly compared to the achievement Z-score scale, which is standardized relative to a different distribution

¹⁴ We examined the relationship between instructional leadership and achievement both for the year of the visit and across all operation years (controlling for Edison-wide operation year trends), on the rationale that principal's instructional leadership might affect both the current level of the school's achievement and its deviation from general Edison trends in all operation years. Apparent effects on overall trends controlling for operation year are comparable to apparent effects in the operation year of the visit. Sample sizes in Table 3 are somewhat smaller than in other case study analyses because we lack instructional leadership ratings for a few principals (as well as lacking achievement results for some schools)