Running Head: Changing Principals’ Leadership through Feedback and Coaching Changing Principals’ Leadership through Feedback and Coaching Corresponding Author: Peter T. Goff University of Wisconsin-Madison pgoff@wisc.edu Contributing Authors: Se Woong Lee University of Wisconsin-Madison James Guthrie, Ellen Goldring, & Leonard Bickman Vanderbilt University Abstract In this study we used a multiyear, Institute of Education Sciences (IES)-funded randomized experiment to investigate the impact of feedback and coaching intervention on principals’ leadership behavior. In our sample of 52 elementary and middle schools, half of the principals received feedback only, while half received both feedback and coaching. We analyzed changes in teachers’ perceptions of principals’ actions as well as principals’ perceptions of their own actions. The findings show that principals who receive coaching are more likely to integrate teacher feedback into their leadership practices. However, we were unable to find evidence of principals influencing teachers’ instruction development. Our study concludes with a discussion of the policy implications of how coaching and feedback may be coupled with disseminated existing initiatives to achieve sustained school improvement. DRAFT COPY ONLY PLEASE DO NOT CITE WITHOUT PERMISSION COMMENTS AND FEEDBACK ARE WELCOME This paper is supported by grant R305A070298 to Leonard Bickman from the U.S. Department of Education's Institute of Education Sciences. All opinions expressed in this paper represent those of the authors and not necessarily the institutions with which they are affiliated or the U.S. Department of Education. 1 Running Head: Changing Principals’ Leadership through Feedback and Coaching Introduction According to U.S. Education Secretary Arne Duncan, “There’s no such thing as a high-performing school without a great principal . . . .You simply can’t overstate their importance in driving student achievement, in attracting and retaining great talent to the school” (Connelly, 2010, p. 35). As school districts across the United States seek to improve the performance of their schools, high quality principals are becoming increasingly essential. When it comes to principals, the real question is not the quantity, but the quality of principals (U.S. Department of Education 2004). This is not to imply that there is a shortage of individuals who have administrative certification, but rather there is a perceived lack of well-qualified leaders who are equipped with knowledge and skills to improve student performance (DarlingHammond, Meyerson, LaPointe, & Orr, 2009). There are two mechanisms to improving the quality of school leaders: (a) improve the quality of individuals applying for administrative positions and (b) develop the capacity of existing principals. This paper focuses on the later option. In order to prepare principals with the ability to carry on their complex roles, principals need a valid and reliable system of feedback from which principals could understand and enhance their leadership (Goldring et al., 2009). However, not many principals find this evaluation process helpful or inspiring, primarily because few principals receive performance feedback beyond ritualistic and perfunctory evaluation (Thomas, Holdaway, & Ward, 2000, Reeves, 2004). Even when there is a meaningful feedback, however, principals experience difficulty understanding the results and integrating them into their practice (Martin, 2010). Furthermore, many people strengthen their existing self-concept by denying or rationalizing negative feedback (Kombarakaran et al., 2008; Thach, 2002; Smither et al., 2005; Bowles et al., 2007). In the private sector, coaching has been given to leaders in order to facilitate 2 Running Head: Changing Principals’ Leadership through Feedback and Coaching better understanding of feedback. Coaches were able to help leaders to understand negative feedback and give an advice on how to act based on feedback. Successful coaching models in the private sector offer hope for similar results in educational settings (Kombarakaran et al., 2008; Thach, 2002; Smither et al., 2003; Bowles et al., 2007). Showers & Joyce (1995) found that without coaching only 5 to 15 percent of learning has been transferred into staff development programs. Also, DarlingHammond (2010), found that principals’ most preferred form of professional development was coaching, even though actual experience of this was rare. Although surprisingly little empirical support for the efficacy of coaching exists, many practitioners and scholars recommend coaching as a viable tool for improving principals’ leadership (Wise & Jocobo, 2010). This study uses a multiyear, IES-funded randomized experiment to investigate the impact of a feedback and coaching intervention on urban principals’ leadership behavior. This study is based on a theoretical framework suggesting that behavior is altered or regulated when feedback is given with coaching. For this study, we developed and implemented a new approach to principal development that entails both feedback to principals from their teachers and coaching sessions with former administrators specifically trained in developing instructional leadership. Evidence of Leadership and Evaluation There is no doubt that the principal plays an important role in enhancing student achievement (Goldring & Rallis, 2000; Spillane, 2006). Waters, Marzano & McNulty (2003) have suggested that school leadership is the second most important school factor that impacts student learning, after classroom instruction. Moreover, the impact of principal leadership is even greater in low-performing schools (Branch, Hanushek, & Rivkin, 2009). Despite increasing attention to improving principal leadership, 3 Running Head: Changing Principals’ Leadership through Feedback and Coaching school districts often use inaccurate and inconsistent measures for principal evaluation (Goldring et al., 2009). Furthermore, the complexity of the role of principals, different political contexts in each state, and debates regarding “What” and “How” to evaluate leadership have created a climate in which it is very difficult to research principal evaluation (Glasman & Heck, 1992; Marcoulides, Larsen, & Heck, 1995; Oyinlade, 2006). Complexities arising from the political and policy context are compounded by the multifaceted nature of school leadership, which resists a simple, uni-dimensional outcome-based measure of effectiveness. To ensure highly effective leaders in every school, many researchers argue that well structured performance assessment for principals can reinforce and strengthen leadership practices (Goldring et al., 2009; Halverson, Kelley, & Kimball, 2004; Kimball, Milanowski, & McKinney, 2007). However, Kempher and Cooper (2002) note that because most principal evaluation policies are locally developed, there is rarely coherence among states policies. Also, Reeves (2009) characterizes most leadership assessments as “infrequent, late, unhelpful, and largely a source of administrative bother” (p.x). To overcome these problems and to accommodate the complex nature of principal leadership, scholars emphasize the importance of linking principal evaluation to a set of clear standards for performance that can be easily understood by principals (Goldring et al., 2009; Derrington and Sanders, 2011). Many states have tried to integrate set clear standards into pre-service programs, in-service training, and evaluation by adopting one or more sets of national professional practice standards such as those from the Interstate School Leadership Licensure Consortium (ISLLC), National Board for Professional Teaching Standards (NBPTS), and National Association of Elementary School Principals. Among those national standards, the ISLLC has been adopted by more than 40 states for 4 Running Head: Changing Principals’ Leadership through Feedback and Coaching performance evaluation and preparation purposes (Anthes, 2005; Hale & Moorman, 2003). The ISLLC was developed by the Council of Chief State School Officers (CCSSO) in order to create a common language and to help strengthen school leadership (CCSSO, 2008). One of the assessments that successfully integrate these standards to assess principals’ leadership is the Vanderbilt Assessment of Leadership in Education (VALED). VAL-ED is a multi-component assessment system that measures critical leadership behavior, performance feedback, and personal decision-making (Vanderbilt University, 2008). Also, VAL-ED aligns their core components using the ISLLC standards, which allow them to produce a quantitative diagnostic profile that is linked to the ISLLC standards (Vanderbilt University, 2008). To be specific, it assesses the intersection of what principals must accomplish to improve academic and social learning for all students (the core components), and how they create those core components (the key processes) (Vanderbilt University, 2008). Although principal evaluation holds great promise, only 25 percent of principals have reported that their district’s principal evaluation system was aligned with professional standards (Kimbell et al., 2007). In addition, Goldring et al., (2009) found that nearly 50 percent of district evaluation protocols were not directly aligned with professional standards. Considering the potential impact of the principal, providing valuable information that can be used to improve leadership purposes is a critical component of school reform (Reeves, 2005). Feedback A key element of evaluation for behavioral change is meaningful feedback that leaders perceive to be both valid and reliable (Portin, 2009). Over several decades, 5 Running Head: Changing Principals’ Leadership through Feedback and Coaching feedback has been articulated in many different ways. Hattie & Timperley ( 2007, p.102) define feedback as “information provided by an agent (e.g. teacher, peer, book, parent, self, o r experience) regarding aspects of one's performance or understanding ... feedback is thus a consequence of performance.” Robertson (2008) noted that leaders rarely, if ever, have to confront with their own leadership, and because of that they have no urgent need or motivation to change their practice. Thus, providing feedback to leaders not only promotes understanding of their own leadership but also can eventually lead to changed leadership. Among various feedback approaches, multi-source feedback holds prominence in the private sector. Multi-source feedback is composed of a leader’s self-assessment as well as assessments from subordinates, peers, and/or superiors (Moore, 2009). The integration of subordinate feedback in the multi-source process facilitates communication (Tornow & London, 1998) and serves as a valid source of information (Smither et al., 2005). Furthermore, by using subordinate feedback, leaders can acknowledge one’s practice (e.g., Hesketh et al., 2005) and improve overall performance (Kluger & DeNisi, 1996; Walker & Smither, 1999). While the majority of studies focusing on feedback between employees and employers are in business settings, several studies have dealt with the use of feedback to support principals in education. Through a meta-analysis Hattie (1999) found that feedback (from teachers to leaders, from leaders to teachers, or from students to teachers) was one of the top five factors influencing student achievement. Also, Daw and Gage (1967) found that principals who received feedback from teachers try to align their behavior with teachers’ perception. Yet among the limited research that has focused on feedback in the education field, most focuses on the teacher-student feedback loop, not teacher feedback as a means to enhance principals’ leadership. 6 Running Head: Changing Principals’ Leadership through Feedback and Coaching This is mainly because, in practice, principals rarely receive systematic feedback regarding their performance and, therefore, remain blind about their practice (Martin, 2010). Considering that principals not only overcome obstacles (Darling-Hammond, 1997; Schmoker, 2006), but also drive school improvement, it is crucial to provide systematic and valid feedback to principals (Marks & Printy, 2003). After years of research, scholars have found that not all forms of feedback are equal. In many cases, feedback is positive, and it tends to motivate the recipient (Jaworski & Kohli, 1991). However, there is also negative feedback. Negative feedback is difficult to handle and people tend to reject or disagree with it (Larson, 1984; Veiga, 1988). Furthermore, scholars have pointed out that people tend to “accept feedback from others that is consistent with the way we see ourselves and to reject feedback that is inconsistent with the way we see ourselves” (Goldsmith, 2004, p.7). Even though feedback is essential in improving leadership, it is not sufficient to change one’s behavior (Cannon & Witherspoon, 2005). Leaders need adequate support in order to interpret feedback correctly and change their behavior (Alicke & Sedikides, 2009). Coaching Coaching can help leaders reflect adequately on their particular experience, which many leaders rarely find easy (Bloom, 2004). Peterson and Hicks (1995) define coaching as a “process of equipping people with the tools, knowledge, and opportunities they need to develop themselves and become more effective” (as cited by Feldman & Lankau, 2005, p. 841). Also, Gray (2006) noted that “[c]oaching is now emerging as one of the most significant approaches to the professional 7 Running Head: Changing Principals’ Leadership through Feedback and Coaching development of senior management and executives” (p. 475). In 2001, Manchester, Inc, found that coaching had significant impact and showed a 529% return on investment and a 788% overall return when factoring employee retention. The results of the executive coaching program delivered an overall, estimated return of 5.7 times the initial investment (McGovern, Lindemann, & Vergara, 2001). That is in part why at least 59 percent of major companies currently offer coaching (London, 2002) and 70 percent of organizations prefer coaching as a major leadership development (Zenger and Stinnett, 2006). Compared to consulting, coaching has emerged fairly recently in leadership development (Schein, 2009). However, unlike consulting, coaching facilitates the client’s active engagement, learning, and internal commitment to a course of action (Bacon and Spear, 2003). A coach is a highly skilled professional who help leaders to be aware of the problem or situation and help leaders to set goals to resolve the problem. There is no rule of thumb on how coaches engage but there are steps and phases coaches follow. The basic phases in a coaching engagement are: 1) Groundwork, which involves building relationships base on trust; 2) Assessment and Feedback, which involves helping principals to understand feedback and explore its meaning; 3) Goal-setting, which involves assisting principals to select a meaningful target for change; 4) Action planning, in which, designing a specific path or set of concrete steps that, if followed, will lead to achievement of the goal; and 5) Ongoing assessment and support, which involves measuring progress over time, addressing any challenges that emerge, providing encouragement and support to build motivation, and keeping the principal on track (Madeline & Mark, 2009). The key goals of these steps are to monitor how progress is being made and maintained and to sustain change, building on positive changes over time in a process of continual 8 Running Head: Changing Principals’ Leadership through Feedback and Coaching improvement. Coaching requires directed focus and has shown more promising effects when coupled with feedback. Leaders who worked with coaches following feedback improved more than those who did not work with coaches (Smither, London, Flautt, and Vargas, 2006). Above all, coaching helps individuals to deal with negative feedback more constructively (Brett and Atwater, 2001). This is why many companies have begun to combine feedback with coaching and it has become one of the fastest growing leadership development strategies in private sector (American Management Association, 2008; Luthans & Peterson, 2004). Thach (2002) found that this combination of feedback and coaching increased leadership effectiveness by up to 60 percent. This coaching with feedback model is becoming a common strategy available to leaders in the private sector. Many scholars in education believes that coaching can also help school leaders to enhance their leadership in order to improve schools and elevate districts to higher levels of achievement (Thach, 2002). In the past several years, there has been a growing interest in principal coaching as a significant component of principal professional development (Hobson, 2003, Karla, 2006). Unlike other professional development programs, coaching can respond to the needs of the principal directly and enhance their ability to solve the complex problems they face every day (Neufeld and Roper, 2003). However, even with increased interest in principal coaching there is little empirical evidence to support the use of leadership coaching in school settings. Thus, Grant (2001) concluded that more research is needed to demonstrate the effectiveness of leadership coaching. One example of this need is that the dose of coaching can impact the effectiveness (Reeves, 2009). Wise (2008) found that typically principals receive coaching on 9 Running Head: Changing Principals’ Leadership through Feedback and Coaching average two –four hours a month, which is less than 2% of the average principal’s time at work. In order for coaching to be a catalyst of change (Vygotsky, 1978), sufficient time with coaches has to be guaranteed (Wise & Javobo, 2010). Many pilot initiatives regarding coaching have been executed throughout the country. However, there has not been solid and rigorous research to identify the effectiveness and efficacy of coaching. In this paper, we use a randomized experiment to explore changes in teachers’ perceptions of principals’ actions as well as principals’ perceptions of their own actions as a result of feedback and coaching. The central question addressed by this study is: Does coaching, together with feedback from teachers, change principals’ leadership practices relative to those who only receive feedback? Method This study uses a multiyear (2008 to 2010) IES-funded randomized experiment to investigate the impact of feedback and coaching intervention on urban principals’ leadership behavior. We gave principals in the treatment group feedback with coaching but only gave feedback to control group principal. Description (Participants) The experiment was implemented in a large urban school district in the Southeastern United States. We excluded high schools as well as schools for specific student populations (e.g., alternative schools, special education) due to the unique nature of leadership in these contexts. Out of the 108 regular elementary or middle 10 Running Head: Changing Principals’ Leadership through Feedback and Coaching schools that were eligible to participate, the 76 that participated were randomly assigned to the treatment and control groups. Of the 76 schools, 24 schools were dropped in the second year due to principals who retired, changed schools, or declined to participate further in the project for personal reasons1. We draw from data collected during the second year of the project in which a treatment group of twenty-six principals received both coaching and feedback while the twenty-six principals in the control group received only teacher feedback. It is in this second year of the study that a set of action items (described in the following section) was added to the survey instrument. In our sample principals had been leaders at their current schools an average of 2.4 years and had been principals in the same school district an average of 5.6 years. The average percentage of students in the free and reduced price lunch program was 68%. Furthermore, 88% and 85% of students were proficient or advanced on their standardized test in reading and math, respectively. To identify measures of leadership practice reported by teachers and principals we used multilevel exploratory factor analysis for binary data. Also, to test the significance of treatment on measures identified through factor analysis, we used OLS estimation with cluster-adjusted standard errors. Both of these methods are elaborated upon in the following sections. Feedback For this study, the project team developed and implemented a new approach to principal development that entails feedback to principals from their teachers. Data for 1 Notable attrition problem between years one and two will be discussed in the Analysis section. 11 Running Head: Changing Principals’ Leadership through Feedback and Coaching the feedback were collected through surveys taken in the fall, winter, and spring of the academic year. In these surveys teachers reported principals’ leadership across a number of measures, including a modified 36-item learning-centered leadership scale (adapted from the VAL-ED) and the effective leadership action items, the dependent variable of this study. Principals, in turn, completed parallel surveys reporting their own leadership practices. The surveys were compiled into feedback reports and sent to principals and their coaches. These action items were anchored by and aligned with the ISLLC standards and VAL-ED assessments. Also, these action-items were a primary component of the feedback principals received from teachers in the intervention, as well as discussion and data points for the coaching sessions. Feedback was presented in reports that included a narrative as well as graphs of the result of teacher and principal survey responses for each wave of the data. The data on all schools in the sample were averaged so that principals could compare their results to those of other participating principals. Teacher surveys were completed by 1,779 teachers within 52 schools in one district. About 34 teachers on average per school provided feedback data. Coaching The leadership coaches selected for this study were former educational administrators and instructional leaders specifically trained for school leadership coaching. Principals selected for treatment (coaching and feedback) met two to four times between each wave of survey collection. Each session focused on feedback from teacher surveys and ranged from 40 to 90 minutes in duration. Coaches were not limited to a set of procedures or protocols for each session but rather focused on developing goals, interpreting feedback, assessing and motivating. Furthermore, coaches were assigned based on geographic location in order to encourage coaching 12 Running Head: Changing Principals’ Leadership through Feedback and Coaching sessions. Multilevel Factor Analysis A multilevel factor analysis was conducted to determine the factor structure for the action-item data at both the within-person level and the between-person levels. We conducted a factor analysis from the second-year teacher survey, shown below. Table 1: Action Items for Factor Analysis Among 19 items, first we excluded item 7 since it was very similar with item 8. Both of the questions were addressing about the principals’ participation on school leadership team. Item 12 also was not included because it was not associated with principal’s support regarding teachers’ instructional improvement. After deleting 2 items, we conducted Exploratory Factor Analysis (EFA) using Geomin rotation on 17 action items through Mplus ver. 6.11. Decision about the number of factors to retain in each case was made based on a combination of empirical considerations and the need for scales that would be substantively meaningful and informative. We then generated scales by calculating the average item level response across the items 13 Running Head: Changing Principals’ Leadership through Feedback and Coaching assigned to a scale for each factor. In order to determine the best-ο¬tting model (a) the root mean square error of approximation (RMSEA; Steiger, 1990), and (b) the standardized root mean square residual (SRMR) were used. The use of these two descriptive ο¬t indices is consistent with recent measurement research (see Millsap & Kwok, 2004). We used nineteen items in both of teacher and principal surveys to identify the impact of treatment on two domains of principal action: Principal Leadership Development (PLD) & Teacher Instructional Development (TID). Based on the indices, two within and two between model showed a relatively good fit (χ2 (df=206, N = 4385) = 1110.713*, p < .01, RMSEA = .050, and SRMR 0.089) according to both descriptive fit indices and theoretical background. Of the 17 items 5 items were dropped as their loadings indicated they were not appropriate to integrate into a factor (Stevens, 1996). Of the 12 remaining items, the first factor consisted of seven items that we labeled as Principal Leadership Development (PLD), which indicates that principal’s work with teachers to develop principals’ own leadership practice. The PLD factor included items such as sharing feedback, planning action based on feedback, and discussing with teachers regarding their own goals and leadership. The second factor consisted of five items that we labeled as Teacher Instructional Development (TID), which represents that principals engage teachers to develop teachers’ instructional practices. The TID factor included items such as observing teachers’ teaching, giving meaningful feedback regarding teaching, discussing students work, and how teachers can engage with parents more effectively. 14 Running Head: Changing Principals’ Leadership through Feedback and Coaching Table 2 Factor Means (Standard Deviation) Factor Fall Winter Spring Principal Leadership Development 3.98 4.41 4.26 (1.10) (1.20) (1.15) 3.41 3.64 3.56 (0.57) (0.53) (0.62) Teacher Instructional Development Threat to Validity 2 Principals’ freedom in scheduling coaching sessions resulted in notable variation in the number of coaching sessions each principal received (see figure 1, below). In the second year of the project, from which our data is drawn, treatment principals averaged eight coaching sessions, though this number ranges from a minimum of five to a maximum of fourteen. To account for this variation in dose, we define treatment in our analysis as the number of coaching sessions each principal received rather than as a binary variable representing assignment to the treatment group. Handling 2 How threat of validity was handled will be discussed in the Analysis section. 15 Running Head: Changing Principals’ Leadership through Feedback and Coaching Figure 1 Frequency of Coaching Sessions Analysis: In this paper we have set out to identify the impact of treatment on two domains of principal action: principal’s leadership development (PLD) and teachers’ instructional development (TID). Barring any outside threats, a simple regression controlling for multiple observations on each principal would be an adequate analytic approach. However, two validity threats preclude using this approach. First, treatment consisted of coaching sessions that are arranged, in part, by principals. When principals have influence regarding the treatment dose, an endogenous relationship is introduced between treatment and outcomes. It is easy to imagine a scenario where ineffective principals may also be less likely to meet with coaches; such a scenario would induce an upwards bias on the treatment effect. To mitigate this threat we use an instrumental variables approach where assignment to treatment (which was randomly assigned two years prior, at the start of the project) is the excluded instrument. The logic of this 16 Running Head: Changing Principals’ Leadership through Feedback and Coaching approach is that variation in assignment is uncorrelated with the number of coaching sessions undertaken by treatment principals. The two-stage least squares models are shown below: πΆπππβπππππ‘ = πΌ0 + πΌ1 (ππ₯)ππ‘ + π’ππ‘ Μ )ππ‘ + πππ‘ π΄ππ‘πππππ‘ = π½0 + π½1 (πΆπππβπππ In these models Coaching represents the cumulative number of coaching sessions by principal i at time t; Tx is a dummy variable indicating assignment to treatment; Action represents either the Principal’s Leadership Development (PLD) or the Teachers’ Instructional Development (TID); Coaching (hat) is the predicted number of coaching sessions based on the first-stage estimate; and u and e are stochastic error terms. The second threat to a simplistic approach pertains to differential attrition before the second year of the study. As mentioned previously, this analysis pertains to the second year of the intervention and uses measures that were not available during the first year. Differential attrition between year one and year two would invalidate the randomization and negate the assumption of strong ignorability of treatment assignment (Rosenbaum & Rubin, 1983). Again, it is not difficult to imagine a scenario where the ineffective principals were also more likely to view coaching as threatening and leave the program. Under such scenarios a differential exit of principals from the control group would induce an upward bias in the treatment effect. In our study there was notable attrition between years one and two, however there is little evidence of differential attrition between the treatment and control principals. In total, 19 of the 71 year-one principals left the study. Nine of these were from the treatment group. Of these 19 principals, 11 were removed from the study by the authors because they accepted a position in a school not participating in the study. 17 Running Head: Changing Principals’ Leadership through Feedback and Coaching Two principals retired. The remaining six principals declined to participate in the second year of the study (four control, two treatment). An attrition analysis where all 19 exiting principals were included is shown in Table 1 below.β 1 The coefficients represent odds-ratios from a logistic regression where one identified a principal who left the sample. Table 3 Attrition Analysis 0.002 5.083 1.311 0.039* 0.559 0.420 4.651 0.079 0.648 Standard Error (0.019) (13.986) (2.449) (0.064) (0.536) (0.550) (7.059) (0.360) (1.621) 15.260 (28.954) 3.291 (4.026) 7.209 0.199 67.097 (11.882) (0.377) (492.405) Coefficient Treatment Academic Press Teacher-Teacher Trust Teachers' Perceptions of Leadership Effectiveness Principal's Perceptions of Leadership Effectiveness Principal Self-Efficacy Principal's Trust of Teachers Treatment*Academic Press Treatment*Teacher-Teacher Trust Treatment*Teachers' Perceptions of Leadership Effectiveness Treatment*Principal's Perceptions of Leadership Effectiveness Treatment*Principal Self-Efficacy Treatment*Principal's Trust of Teachers Constant The measures used in the attrition analysis include several measures of school climate and leadership, including teachers’ perceptions of academic press, teacherteacher trust, teacher perceptions of leadership effectiveness within the school, principal’s self-perceptions of leadership effectiveness, principal self-efficacy, and principal’s trust of teachers. These are the baseline measures, taken in the first year of the study before any feedback or coaching was provided. Of these, the only measure that appears related to attrition is teachers’ perceptions of leadership effectiveness (0.039*, p=0.048), indicating that principals who left the study have significantly lower measures of leadership effectiveness (mean teacher ratings of 3.53 versus 3.33). 18 Running Head: Changing Principals’ Leadership through Feedback and Coaching However, the interaction term between teachers’ perception of leadership effectiveness and treatment is not significant (15.260), suggesting that there is not a discernable difference on this measure between those who left the study from the treatment group as compared to those from the control group. This analysis provided evidence that the initial randomization held through the second year. Nonetheless, to further insulate against this threat we have included a model in each analysis that also includes the above six baseline measures of school climate and leadership in addition to dummy variables for whether the data was collected during the fall, winter, or spring (models 3, 4, 7, and 8 in tables 2 and 3). These variables were included as excluded instruments in the IV model as well. Results: The analytical results pertaining to the effect of treatment on actions related to principal’s leadership development are presented in Table 4. Model 1 presents naïve OLS results, model 2 presents the 2-stage least squares instrumental variables (IV) approach, model 3 is a naïve OLS model that includes school climate, leadership measures, and dummy variables for the wave of data collection, and model 4 replicates model 3 with the addition of the instrumental variables. In each of the four parameterizations we see that coaching has a significant positive effect on principals’ activities surrounding leadership development. The most constrained model reports an increase of 0.103 on the Principal’s Leadership Development scale for every additional coaching session. This corresponds to an effect size of 0.15. The higher IV estimates (Model 1 < Model 2, Model 3 < Model 4) imply that the naïve estimates were under-estimating rather than overestimating the 19 Running Head: Changing Principals’ Leadership through Feedback and Coaching treatment effect. Table 4 The Effect of Treatment on Principal's Leadership Development Coaching & Feedback Intercept N Wave Dummies Baseline Covariates IV Model 1 0.117** (0.038) 3.931*** (0.188) 156 no Model 2 0.128** (0.052) 3.901*** (0.214) 156 no Model 3 0.086* (0.033) 2.022 (2.224) 156 yes Model 4 0.103* (0.040) 2.132 (2.127) 156 yes no no yes yes no yes no yes The second line of inquiry pertains to the impact of treatment on the ways in which principals’ influence teachers’ instructional development (TID, factor 2). Table 5 provides consistent evidence through all four modeling approaches that treatment made no discernible impact on principal actions in this domain. Table 5 The Effect of Treatment on Teachers' Instructional Development Coaching & Feedback Intercept N Wave Dummies Baseline Covariates IV Model 5 0.001 (0.019) 3.223*** (0.018) 156 no Model 6 -0.013 (0.027) 3.26*** (0.094) 156 no Model 7 -0.014 (0.018) 2.774* (1.148) 156 yes Model 8 -0.010 (0.024) 3.063** (1.110) 156 yes no no yes yes no yes no yes Discussion: This study is motivated by a need to provide school leaders with an iterative, sustainable, and efficient mechanism to integrate formative feedback with changes in leadership practices. Feedback reports serve as a discrepant event that in turn stimulates cognitive dissonance in the principal. Under this framework the actions 20 Running Head: Changing Principals’ Leadership through Feedback and Coaching taken by the principal are initiated in order to resolve the dissonance. Coaching serves two functions in this process: to facilitate dissonance and to direct/motivate subsequent actions. First, the findings from principal’s leadership development presented here suggest that coaching and feedback together facilitates principals’ ability to engage with their faculty regarding their own leadership development. Based on teachers’ feedback, coaches facilitate principals’ self-reflection in order to initiate leadership change. This is promoted by helping principal to clarify and prioritize issues in their schools, interpret feedback that they could address through their leadership practices, and provide skills, which principals need to enhance their overall leadership. This finding is consistent with existing research, which identified strong effectiveness of combination of coaching and feedback (Hobson, 2003; Smither, London, Flautt, and Vargas, 2006). However, the second finding suggests that it does not impact teachers’ instructional development. The question raised by these findings is why we observed growth in the first construct but not the second? Interpretation of these findings can be grouped as logistic effects, focus effects, or dissonance augmenting effects. Logistic effects may be the result of information being conveyed more efficiently to groups rather than to individual teachers. The specific actions that contribute to the Principal’s Leadership Development scale include “sharing feedback with teachers,” “Discussing specific actions planned to take as a result of feedback” and “Discussing goals of leadership with teachers.” These are actions that may lend themselves to large group settings where principals can address the entire faculty in one sitting. 21 Running Head: Changing Principals’ Leadership through Feedback and Coaching In contrast, actions that contribute to the Teachers’ Instructional Development scale include “Discussed with a teacher the quality of student work” and “Gave a teacher feedback on the quality of his or her teaching.” These actions almost certainly require principals to engage with teachers one-on-one and are more logistically challenging. The differing logistical constraints of these actions suggests that, when aggregated to the school level, we are likely observe a stronger signal from changes in the Principal’s Leadership Development scale than we are from Teachers’ Instructional Development scale. These results may also speak to a focus effect; an effort on the part of coaches to push principals to be more strategic in their interactions with teachers. High quality learning-centered leadership is a time-consuming venture. Indeed, learning-centered leadership is an area towards which principals would like to devote more time, and yet it tends to be the first domain to be curtailed under the press of external demands. When coaches encourage principals to focus their interactions on a focused group of teachers (e.g., new teachers or a grade-level team) rather than across the entire school, we may see a null (or perhaps even negative) effect at the school level. This may be the case even when principals are actively taking actions to change their leadership behaviors. As a result of concessions made with the district, teacher responses were anonymous and we cannot track teachers over time to determine if principals are using such an approach to effectively target the marginal teachers. We can also interpret these results as evidence for the dissonance augmenting function of coaching. In keeping with prior work on feedback and dissonance, qualitative work on this project suggests that some principals are apt to dismiss the feedback. Coaches report that principals’ initial responses to the feedback is frequently defensive, where principals claim, “I know it’s just a few teachers who are 22 Running Head: Changing Principals’ Leadership through Feedback and Coaching dissatisfied and I know who they are” or “I’m doing all I can – what more can they expect me to do?” We have also found that coaches are instrumental in helping principals engage the dissonance process by overcoming their resistance to the feedback. Principals are unlikely to share feedback regarding their leadership if they feel that the feedback itself is invalid. Thus, the act of sharing feedback becomes an active acknowledgement of the feedback validity. Soliciting additional input from teachers on their leadership is another mechanism to legitimate the validity of teacher perspectives. One of the goals of coaching is to facilitate the dissonance process, which begins when resistant principals come to see the feedback as valid. This legitimating process allows cognitive dissonance to take place and only then can principals begin altering their behavior to bring their leadership views in line with those of their faculty. In this light, we can see that the above results to be evidence for the dissonance augmenting aspect of the coaching intervention. Lastly, one year of coaching may not have been enough for principals to change their leadership practice, and at the same time it might not have been enough to alter teachers’ perception of principals’ instructional support. In order for coaches to bring desired and ongoing change in principals, an adequate amount of time has to be devoted for a plan of action. Even though principals in this study received average of eight coaching session per wave; however, each session was less than two hours, which might not have been enough time to sustain desirable leadership behavior. In addition, besides for time of coaching received by principals, teachers also need time to acknowledge the change of leadership. However, not all teachers 23 Running Head: Changing Principals’ Leadership through Feedback and Coaching have adequate time during school hours to interact with principals to be aware of the change. Seeing principal in the classroom once or twice a semester may not be enough. Conclusion Providing meaningful feedback through principal assessment, or helping them to adequately interpret feedback through coaching can serve as a viable tool to improve leadership practice. This study addresses the gap in the literature by both measuring the quantitative impact of feedback and coaching and providing preliminary evidence to support their combined efficacy. To inform practitioners and policy makers, research in this area must go further and directly examine the impact of coaching and feedback on leadership. Investigating how reflection from teachers help principal to improve their leadership and whether coaching promotes this knowledge into practices and eventually improve students outcomes are important avenues for future inquiry. 24 Running Head: Changing Principals’ Leadership through Feedback and Coaching Reference Alicke, M. D., & Sedikides, C. (2009). Self-enhancement and self-protection: What they are and what they do. European Review of Social Psychology. Anthes, K. (2005). What's Happening in School and District Leadership? Assessing the Effectiveness of School Leaders. (2009). Assessing the Effectiveness of School Leaders. Association, A. M. (2008). Coaching: A global study of successful practices. Retrieved October. Barnett, B. G., & O'Mahoney, G. (2002). One for the To-Do List: Slow Down and Think. Bloom, G. S. (2004). Emotionally intelligent principals. School Administrator, 61, 14–16. Bowles, S., Cunningham, C. J. L., Rosa, G. M. D. L., & Picano, J. (2007). Coaching leaders in middle and executive management: goals, performance, buy-in. Leadership & Organization Development Journal, 28(5), 388–408. doi:10.1108/01437730710761715 Branch, G., Hanushek, E., & Rivkin, S. (2009). Estimating Principal Effectiveness. Working Paper 32. Brett, J. F. J., & Atwater, L. E. L. (2001). 360 degree feedback: accuracy, reactions, and perceptions of usefulness. Journal of Applied Psychology, 86(5), 930–942. Cannon, M. D., & Witherspoon, R. (2005). Actionable feedback: Unlocking the power of learning and performance improvement. The Academy of Management Executive, 19(2), 120-134. 25 Running Head: Changing Principals’ Leadership through Feedback and Coaching Connelly, G. (2010). A Conversation with Secretary of Education Arne Duncan. Principal, 90(2), 34–38. Council of Chief State School Officers (CCSSO). (2008). Educational Leadership Policy Standards: ISLLC 2008 As Adopted by the National Policy Board for Educational Administration. (Washington, DC: Council of Chief State School Officers) Darling-Hammond, L. (1997). The Right To Learn: A Blueprint for Creating Schools That Work. The Jossey-Bass Education Series. Darling-Hammond, L. (2010). The Flat World and Education: How America’s Commitment to Equity Will Determine Our Future. Teachers College Press, NY Darling-Hammond, L., Meyerson, D., LaPointe, M., & Orr, M. T. (2009). Preparing Principals for a Changing World. Jossey-Bass. Daw, R. W. R., & Gage, N. L. N. (1967). Effect of feedback from teachers to principals. Journal of Educational Psychology, 58(3), 181–188. doi:10.1037/h0024678 Derrington, M. L., & Sanders, K. (2011). Conceptualizing a System for Principal Evaluation. Feldman, D. C., & Lankau, M. J. (2005). Executive Coaching: A Review and Agenda for Future Research. Journal of Management, 31(6), 829–848. doi:10.1177/0149206305279599 Fox, J. J. (2009). Building capacity in urban schools by coaching principal practice toward greater student achievement. University of Southern California). ProQuest Dissertations and Theses, , 183. Retrieved from http://search.proquest.com.ezproxy.library.wisc.edu/docview/304998353?accou ntid=465. (304998353). 26 Running Head: Changing Principals’ Leadership through Feedback and Coaching Glasman, N. S., & Heck, R. H. (1990). The changing leadership role of the principal: Implications for principal assessment. Peabody Journal of Education, 68(1), 5– 24. Goldring, E. B., & Rallis, S. F. (2000). Principals of dynamic schools: Taking charge of change. Goldring, E., Cravens, X. C., Murphy, J., Porter, A. C., Elliott, S. N., & Carson, B. (2009). The Evaluation of Principals: What and How Do States and Urban Districts Assess Leadership? The Elementary School Journal, 110(1), 19–39. doi:10.1086/598841 Grant, A. M. (2002). Towards a psychology of coaching: The impact of coaching on metacognition, mental health and goal attainment. Macquarie University (Australia)). ProQuest Dissertations and Theses, , 327-327 p. Retrieved from http://search.proquest.com.ezproxy.library.wisc.edu/docview/305462830?accou ntid=465. (305462830). Gray, D. E. (2006). Executive coaching: Towards a dynamic alliance of psychotherapy and transformative learning processes. Management Learning. Hale, E. L., & Moorman, H. N. (2003). Preparing School Principals: A National Perspective on Policy and Program Innovations. Hallinger, P., & Heck, R. H. (1998). Exploring the Principal's Contribution to School Effectiveness: 1980β 1995∗ . School Effectiveness and School Improvement, 9(2), 157–191. doi:10.1080/0924345980090203 Halverson, R., Kelley, C., & Kimball, S. (2004). Implementing teacher evaluation systems: How principals make sense of complex artifacts to shape local 27 Running Head: Changing Principals’ Leadership through Feedback and Coaching instructional practice. Educational Administration, Policy, and Reform: Research and Measurement, 153–188. Hattie, J. (1999). Influences on student learning. Inaugural Lecture Given on August. Hattie, J., & Timperley, H. (2007). The Power of Feedback. Review of Educational Research, 77(1), 81–112. Hesketh, E. A. E., Anderson, F. F., Bagnall, G. M. G., Driver, C. P. C., Johnston, D. A. D., Marshall, D. D., et al. (2005). Using a 360 degrees diagnostic screening tool to provide an evidence trail of junior doctor performance throughout their first postgraduate year. Medical Teacher, 27(3), 219–233. doi:10.1080/01421590500098776 Hobson, A. (2003). Mentoring and Coaching for New Leaders. Improvement, U. S. D. O. E. O. O. I. A. (2004). Innovative Pathways to School Leadership. Jaworski, B. J., & Kohli, A. K. (1991). Supervisory Feedback: Alternative Types and Their Impact on Salespeople's Performance and Satisfaction. Journal of Marketing Research, 28(2), 190–201. Kempher, L. L. (2001). A document analysis of state-mandated principal evaluation programs in the north central association states and best practice in effective principal characteristics. Northern Illinois University). ProQuest Dissertations and Theses,, 147-147 p. Retrieved from http://search.proquest.com.ezproxy.library.wisc.edu/docview/304714739?accou ntid=465. (304714739). Kilburg, R. R. (2006). Conclusion. American Psychological Association. 28 Running Head: Changing Principals’ Leadership through Feedback and Coaching Kimball, S. M., Milanowski, A. T., & McKinney, S. (2007). Implementation of standards-based principal evaluation in one school district: First year results from randomized trial. Kluger, A. N., & DeNisi, A. (1996). The effects of feedback interventions on performance: A historical review, a meta-analysis, and a preliminary feedback intervention theory. Psychological Bulletin. Kombarakaran, F. A., Yang, J. A., Baker, M. N., & Fernandes, P. B. (2008). Executive coaching: It works! Consulting Psychology Journal: Practice and Research, 60(1), 78–90. doi:10.1037/1065-9293.60.1.78 Larson, J. R. (1984). The performance feedback process: A preliminary model. Organizational Behavior and Human Performance. Leithwood, K., Louis, K. S., Anderson, S., & Wahlstrom, K. (2004). How leadership influences student learning. Center for Applied Research and Educational Improvement, University of Minnesota, 289-342. London, M. (2001). Leadership Development. Psychology Press. Luthans, F., & Peterson, S. J. (2004). 360β degree feedback with systematic coaching: Empirical analysis suggests a winning combination. Human Resource Management. Madeline C. M., & Mark C. (2009). Coaching Principals: A Model for Leadership Development. Unpublished paper presented at University Counsel For Educational Administration Convention, Anaheim, CA. Marcoulides, G. A., Larsen, T. J., & Heck, R. H. (1995). Examining the generalizability of a leadership model: Issues for assessing administrator performance. International Journal of Educational Management, 9(6), 4–9. 29 Running Head: Changing Principals’ Leadership through Feedback and Coaching Marks, H. M., & Printy, S. M. (2003). Principal Leadership and School Performance: An Integration of Transformational and Instructional Leadership. St. Martin, K. (2010). The knowing-doing gap: The influence of teacher feedback in changing principal behavior---a mixed methods study. Western Michigan University). ProQuest Dissertations and Theses, , 289. Retrieved from http://search.proquest.com.ezproxy.library.wisc.edu/docview/848607893?accou ntid=465. (848607893). Matsumura, L. C., Garnier, H. E., Correnti, R., Junker, B., & DiPrima Bickel, D. (2010). Investigating the Effectiveness of a Comprehensive Literacy Coaching Program in Schools with High Teacher Mobility. The Elementary School Journal, 111(1), 35–62. doi:10.1086/653469 Levenson, A. (2009). Measuring and maximizing the business impact of executive coaching. Consulting Psychology Journal: Practice and Research,61(2), 103. Moore, B. (2009). Improving the evaluation and feedback process for principals. Reeves, D. B. (2006). The learning leader. Ascd-Association for Supervision and Curriculum Develoment. Neufeld, B., & Roper, D. (2003). Coaching: A Strategy for Developing Instructional Capacity: Promises & Practicalities. Aspen Institute. Oyinlade, A. O. (2006). A Method of Assessing Leadership Effectiveness: Introducing the Essential Behavioral Leadership Qualities Approach. Peterson, D. B., & Hicks, M. D. (1995). Development first. Pdi Ninth House. Reeves, D. B. (2009). Leading change in your school. Alexandria, VA: Association for Supervision and Curriculum Development Robertson, J. (2008). Coaching Educational Leadership. Sage Publications Limited. 30 Running Head: Changing Principals’ Leadership through Feedback and Coaching Schmoker, M. (2006). Results now: How we can achieve unprecedented improvements in teaching and learning. ASCD, Alexandria, VA. Showers, B., & Joyce, B. (1996). The evolution of peer coaching. Educational Leadership, 53, 12–16. Smither, J. W., London, M., & Reilly, R. R. (2005). Does Performance Improve Following Multisource Feedback? A Theoretical Model, Meta-Analysis, and Review of Empirical Findings. Personnel Psychology, 58(1), 33–66. Smither, J. W., London, M., Flautt, R., Vargas, Y., & Kucine, I. (2006). Can working with an executive coach improve multisource feedback ratings over time? A quasi-experimental field study. Personnel Psychology, 56(1), 23-44. Spillane, J. P. (2012). Distributed leadership (Vol. 4). Jossey-Bass. Thach, E. C. (2002). The impact of executive coaching and 360 feedback on leadership effectiveness. Leadership & Organization Development Journal, 23(4), 205–214. doi:10.1108/01437730210429070 Thomas, D. W., Holdaway, E. A., & Ward, K. L. (2000). Policies and practices involved in the evaluation of school principals. Journal of personnel evaluation in education, 14(3), 215-240. Tornow, W. W., & London, M. (1998). Maximizing the Value of 360-Degree Feedback: A Process for Successful Individual and Organizational Development. Van Meter, E., & Murphy, J. (1997). Using ISLLC standards to strengthen preparation programs in school administration. Washington, DC: Council of Chief State School Officers. Veiga, J. F. (1988). Face Your Problem Subordinates Now! The Academy of Management Executive (1987-1989), 2(2), 145–152. 31 Running Head: Changing Principals’ Leadership through Feedback and Coaching Vygotsky, L.S. (1978). Tool and symbols in children’s development. In Mind in society: Development of higher psychological processes, ed M. Cole, V. JohnSteiner, S. Scribner, and E. Souberman, 12-92. Cambridge, MA: Harvard University Press. Walker, A. G., & Smither, J. W. (2006). A Five-year Study of Upward Feedback: What managers do with their results matters. Personnel Psychology. Waters, T., & Grubb, S. (2004). Leading schools: Distinguishing the essential from the important. Aurora, CO: Mid-continent Research for Education and Learning. Waters, T., Marzano, R. J., & McNulty, B. (2003). Balanced Leadership: What 30 Years of Research Tells Us about the Effect of Leadership on Student Achievement. A Working Paper. Wise, D., & Jacobo, A. (2010). Towards a framework for leadership coaching. School Leadership & Management, 30(2), 159–169. doi:10.1080/13632431003663206 Wise, D. (2008). The effect of selected coaching components utilized on the performance of California school principals. Paper presented at the annual conference of the National Association of Professors of Education Administration (NCPEA), San Diego, CA. Zenger, J. H., & Stinnett, K. (2006). Leadership coaching: Developing effective executives. Chief Learning Officer. 32