Abstract
Research that examines coaching approaches for special education teachers is very limited. This study, a secondary analysis of a wait-list controlled, randomized trial (106 teachers, 2,195 students, 18 schools), investigated the effects of a data-driven coaching that integrated observational assessment and performance feedback on general education (GE) versus special education (SE) teacher practices and student outcomes in high-poverty urban elementary schools. Coaches used observational data via the Classroom Strategies Assessment System to identify practice needs, set goals, create plans, and monitor progress toward goals. Prior to coaching, GE and SE teachers were observed using evidence-based instructional and behavior management practices; however, some practices were at rates lower than recommended by the research literature. Results suggest that goal selection and frequency and quality of practices were generally comparable between GE and SE teachers. However, SE teachers used 30% fewer behavior corrective feedback statements, on average, than GE teachers (p = .04). Overall, the effect of the coaching intervention did not differ across GE and SE teachers; both had significantly improved instructional and behavior management practices and student outcomes when compared with teachers in the control condition. Limitations and future directions for research and practice are discussed.
Keywords
The implementation of Every Student Succeeds Act (2015) has placed greater accountability on state education departments and school districts to identify and implement effective teaching evaluation and professional development approaches that enhance teachers’ use of research-based practices. School leaders must make strategic decisions related to staffing and resources to provide professional development that improves teacher instruction and student academic performance (Reddy et al., 2019; Sawchuk, 2016). At the center of these initiatives includes examining how school personnel use data to support improvements in teacher practices and promote positive learning environments for all learners, including students with disabilities.
One professional development approach, instructional coaching, has been adopted by schools to overcome some of the limitations to workshop-based professional development (Denton & Hasbrouck, 2009; Gulamhussein, 2013; Shernoff et al., 2015). For decades, scholars and school practitioners have viewed coaching as a mechanism for translating research into practice via iteratively supporting teacher effectiveness and professional growth (Gersten et al., 1995; Joyce & Showers, 2002). Coaching includes job-embedded, individualized, sustained support to promote enduring changes in classroom practices and student learning and behavior (Denton & Hasbrouck, 2009; Glover & Reddy, 2017). Although coaching approaches vary, many share common elements and processes. These include prioritizing instructional needs, establishing goals, creating plans for implementation, modeling, and providing ongoing performance feedback (Erchul, 2015; Kurz et al., 2017). The quality of coach and teacher interactions and relationship is critical for forging engagement, motivation, and receptivity in using data to modify classroom practices and interventions to better meet student needs (Knight, 2009; Reddy et al., 2019).
While instructional coaching is important for improving instruction, empirical support for this model of professional development is limited. In a meta-analysis of 60 coaching randomized controlled trials (RCTs; Kraft et al., 2018), the overall effect of coaching on teacher instructional practices was d = 0.49 and the overall effect of coaching on student achievement was d = 0.18. Studies included in the meta-analysis focused primarily on literacy coaching among general educators working in elementary schools. Very few studies examined sets of strategies for instructional improvement or examined the impact of coaching on universal instructional and behavior management practices (Fabiano et al., 2018; Reddy et al., in press). It is noteworthy that only two of the 60 studies reviewed in the meta-analysis included special educators in the teacher sample, resulting in fewer than 5% of the total teachers sampled in the meta-analysis (Hemmeter et al., 2016; Matsumara et al., 2010).
The need to bolster the inclusion of special education teachers in research samples has been noted as a critical step in developing and validating professional development innovations (Sindelar et al., 2010). Likewise, no studies in the Kraft et al. (2018) meta-analysis examined the effect of coaching on general versus special education practices and student educational outcomes. Taken together, there are significant gaps in the literature regarding assessment and professional development interventions that enhance general and special education teacher practices. Rigorous investigations are needed to assess coaching processes and components that benefit general and special education teachers—school personnel who naturally collaborate to support student learning and behavioral needs.
Advancing Research on Coaching for Special Education Teachers
Research findings related to professional development for general educators should not be expected to generalize to special educators. This is, in part, because of the different range of student needs that must be addressed by special educators along with substantive differences in the content and philosophy of general and special education teacher preparation programs (e.g., Flower et al., 2017; Sindelar et al., 2010).
Research on general and special education teacher preparation programs suggests substantive differences in the philosophies and content covered between general and special education. Aspiring teachers in general education preparation programs are more likely to engage in coursework and practical experiences based on a constructivist philosophy of education. The constructivist philosophy typically emphasizes some instructional strategies (IS; for example, inquiry-based or student-centered instruction) over others (e.g., direct instructional modalities; Brownell et al., 2005). These differences in philosophical orientation may contribute to substantive differences in the content or focus of training. For example, Flower et al. (2017) identified significantly greater coverage of behavior assessment and intervention techniques for special versus general educators. Oliver and Reschly (2010) also reported that the majority of coursework for preservice special educators omitted classroom behavior management.
Special education teachers may also have different needs for, and responses to, coaching. These differences may have implications for students with disabilities. Education experts have long voiced concern regarding limited growth in achievement by students in response to special education services (e.g., Chudowsky et al., 2009; Feng & Sass, 2013; Kavale, 2007; Thurlow et al., 2016). Thus, the nature and effectiveness of professional development offered to educators working with special education students must undergo close examination.
Scholars have asserted different opinions regarding the core features of effective professional development for general and special educators. For example, Desimone (2009) suggested a set of features for the context of general education, including content focus, active learning opportunities, coherence, collective participation, and sufficient duration. Special education scholars have further added to these features by spotlighting differences in the content of professional development for special educators, including a greater emphasis on the implementation of innovations (e.g., collaborative strategic reading, classwide peer tutoring, daily report cards) and evaluation of fidelity (Boudah et al., 2003; Sindelar et al., 2010; Volpe & Fabiano, 2013). Despite this noteworthy work, limited empirical research has been conducted on effective professional development for special educators (e.g., Harry & Lipsky, 2014; Kraft et al., 2018; Sindelar et al., 2010). The larger literature on effective teaching highlights a range of practices that could be included in professional development models (e.g., Hattie, 2012; Hattie & Clarke, 2018; Marzano et al., 2001; What Works Clearinghouse, 2012). Examples include strategies teachers use to actively engage students in learning practices, deliver academic content, convey information to students, activate student thinking about lesson material, and deliver performance feedback to students.
Furthermore, research assessing the efficacy of professional development interventions (i.e., coaching) for general and special education teachers in high-poverty schools is particularly warranted. Teachers employed in high-poverty schools have higher rates of attrition and stress in comparison with teachers working in other school contexts (e.g., Borman & Dowling, 2008; Boyd et al., 2012). Furthermore, research has suggested that the effectiveness of teachers in high-poverty schools is less than teachers in lower poverty schools (e.g., Sass et al., 2012), reducing the learning of students, in particular students with disabilities in high-poverty schools. Thus, coaching research is warranted for special education and general education teachers in this context. Indeed, teachers can benefit from evidence-based, practical, job-embedded professional development that supports their implementation and evaluation of research-based instructional and behavior management practices that maximize student learning and behavior (Reddy et al., in press).
Rationale and Purpose of the Present Investigation
The present study addresses gaps between research and practice related to coaching general and special education teachers serving elementary school students in high-poverty schools. The present study includes secondary analyses of data from a wait list (WL) controlled RCT study that investigated a data-driven coaching model—the Classroom Strategies Coaching Model (CSC). The CSC Model integrates teachers’ formative assessment and individual performance feedback to enhance universal practices in elementary classrooms (Reddy et al., in press). The main RCT used a randomized block design (Addelman, 1969) in which teachers within each school (i.e., block) were randomly assigned to either the experimental (CSC Model) or WL control condition among 14 participating elementary schools. We examined the extent to which there were statistically and clinically significant differences between conditions related to the frequency and quality with which all teachers implemented evidence-based instructional and behavior management practices and classwide student academic engagement. We also examined between-group differences related to teacher ratings of student academic and behavioral functioning and perceived supports and stress. Thus, the main RCT was designed to test the efficacy of the CSC Model on classwide practices and classwide student academic engagement. We hypothesized that there would be statistically significant differences favoring teachers in the coaching condition related to observed use of instructional and behavior management practices and classwide student academic engagement. We also hypothesized that CSC teachers would have improved ratings of instructional and emotional support and stress relative to WL teachers. Finally, we hypothesized that teachers would rate the CSC acceptability positively as a professional development support. Our theory of change guiding the main RCT study includes the key components of the CSC Model described hereafter and proximal outcomes associated with coaching (i.e., improved evidence-based instructional and behavior management practices, frequency and quality, and teachers’ perceived supports and stress). We theorized that the CSC Model would improve classroom practices (proximal teacher outcomes) and perceived supports that would lead to increased classwide student academic engagement (distal student outcomes).
Overall, findings from the main RCT were favorable (Reddy et al., in press). Direct observations of teacher practices indicated that relative to the WL control condition, teachers in the coaching condition demonstrated significant improvements in instructional quality (d = 0.60), behavior management (d = 0.52), and student engagement (d = 0.41). Teachers reported significant improvements in student academic (d = 0.96) and behavioral functioning (d = 1.24) along with instructional (d = 1.04) and emotional support (d = 0.90, respectively). No change was found for teacher stress.
The present study builds on the findings from the main RCT by examining the processes and outcomes of CSC implementation between general education and special education teachers. Based on the larger coaching literature (e.g., Kraft et al., 2018; Kurz et al., 2017), the current study serves as the first investigation of the processes and outcomes of coaching for general versus special education teachers in urban high-poverty elementary schools or other settings. Investigations examining coaching processes that benefit general and special education teachers are warranted as these school personnel commonly collaborate to support student learning and behavioral needs in inclusion classrooms. Likewise, research on coaching for general and special education teachers in high-poverty schools is needed as teachers in this context have higher rates of burnout, attrition, and stress in comparison with teachers working in other school contexts (e.g., Borman & Dowling, 2008; Boyd et al., 2012). Taken together, the present study offers a first step in addressing significant gaps in the literature regarding assessment and professional development interventions that enhance general and special education teacher practices.
In addition, the CSC Model has several innovative features that advance the science and practice of coaching in elementary schools. First, the CSC Model uses a unique teacher formative assessment tool (Classroom Strategies Assessment System; CSAS) to measure teachers’ use of evidence-based instructional and behavior management practices known to improve student learning and behavior—including students with or at risk of disabilities. Second, coaches in this study assess teacher practice needs, set goals, and monitor improvements through two metrics: (a) the frequency of strategy use (CSAS Strategy Counts) and (b) the quality of strategy use (CSAS Strategy Rating Discrepancy Scores described in detail in the methods). CSAS scores have evidence of reliability and validity, as well as sensitive to change following coaching (Fabiano et al., 2018; Lekwa et al., 2020; Reddy & Dudek, 2014) and predict student achievement and behavior outcomes (e.g., Dudek et al., 2019; Lekwa, Reddy, Dudek, & Hua, 2019a; Lekwa, Reddy, & Shernoff, 2019; Reddy et al., 2013b, 2020). Third, the CSC Model is designed to improve universal instructional and behavior management practices. Fourth, the CSC Model targets sets of evidence-based strategies rather than single targets for change in lessons. Finally, the CSC Model uses CSAS scores to guide coach and teacher decisions and actions, including identifying practice needs, setting goals, and monitoring and evaluating teacher progress via graphic performance feedback (Reddy & Dudek, 2014; Reddy et al., 2013a).
The present study builds on findings summarized in the main RCT by examining the processes and outcomes of CSC implementation with a distinct focus on general education and special education teachers in addition to more closely examining the types of practice goals that general and special education teachers select in the context of participating in a coaching intervention. Specifically, the present study examined the following questions:
Method
Participants
Participants included 106 kindergarten through fifth-grade teachers (85 certified general education and 21 certified special education teachers) randomly assigned to a coaching or WL control condition. Teachers served 2,195 students within 14 high-poverty elementary schools located in a medium-sized district in a Northeastern state. Table 1 illustrates teacher characteristics by condition.
Teacher Characteristics by Intervention Condition and Teacher Group.
Note. There were no significant differences between groups for any of these demographic variables. CSC = Classroom Strategies Coaching; Gen. ed. = general education; Sp. ed. = special education; IEP = Individualized Education Program.
In the present study, participating teachers, assigned to coaching or WL condition, were leading instructional opportunities for students and were not serving as classroom assistants. Special educators taught in general educational settings and divided their instructional responsibilities with the general education teachers. Coaches conducted observations during times in which participating teachers (general and special educators) were directly delivering instruction within a district that has fully adopted an inclusive model of special education service delivery.
Students included in the study were placed in general education classrooms where approximately 5% of students were receiving special education services. Participating classroom teachers had approximately three students with IEPs in their classroom. Approximately 80% of students in the district qualified to receive free or reduced-price lunches. Approximately 32% of students were Latinx, 41% were Black, 12% were Asian, 12% European American, and the remaining 3% were Native American, multiracial, or other. Teacher recruitment included project flyers mailed to schools and presentations to principals and teachers during staff meetings. Eligible teachers who expressed an interest in participating were contacted by project staff to review and complete informed consent approved by the Institutional Review Board.
Measures
CSAS
The CSAS, a multimethod and multidimensional classroom observation measure, guided the coaching process and assessed changes in teachers’ use of evidence-based instructional and behavior management practices (Reddy & Dudek, 2014; Reddy et al., 2013a, 2013b). The CSAS includes strategies based on findings from more than 50 years of effective instruction and classroom management literatures (e.g., Brophy & Good, 1986; Hattie, 2012; Hattie & Clarke, 2018; Marzano et al., 2001; What Works Clearinghouse, 2012). The CSAS has an Observer and Teacher Form. The CSAS Observer Form assessed the frequency and quality of strategy use. The CSAS consists of Strategy Counts, Strategy Rating Scales, and a Classroom Checklist. During two pre- and two post-30-min observations, observers completed the Strategy Counts (i.e., total number of times teachers used eight instructional and behavior management strategies; see Table 2). During observations, observers took specific notes (sources of evidence) for completing the Strategy Rating Scales.
Descriptions of the CSAS Strategy Counts.
Note. CSAS = Classroom Strategies Assessment System.
A total of 29 trained independent observers performed the baseline and postassessment observations for the study. Observers were undergraduate and graduate students enrolled in psychology and education programs who were blind to condition and conducting observations independently from the coach. Observers were primarily female and European American (80%). Coaches consisted of 12 supervised MA-level school practitioners or advanced graduate students with prior experience in high-poverty elementary schools hired to work on the study. Coaches were predominantly female (75%) and European American (83%) with a mean age of 28 (SD = 5) years.
The Instructional Strategies (IS) and Behavioral Management Strategies (BMS) Rating Scales were completed after the observation (see Table 3). The Rating Scales include a total score (IS + BMS) and composite scores of IS total (five subscales) and BMS total (four subscales) each, which include subscales. Using a 7-point Likert-type rating scale, observers rate (a) how often (observed frequency rating; 1 = never used, 3 = sometimes used, 7 = always used) the teacher used specific IS and BMS, and (b) how often the teacher should have used each strategy (recommended frequency) using the same 7-point scale. The dual rating scales of observed frequency and recommended frequency yield IS and BMS discrepancy scores, in which absolute values of the difference between observed and recommended frequency (recommended frequency ratings − observed frequency ratings) indicate need for change in classroom practices (larger scores suggesting greater need for change). Observers rated observed and recommended frequency based on the following information: (a) strategy counts frequency data; (b) notes of evidence on the IS and BMS Rating Scales nine dimensions (subscales); (c) notes on the lesson objective, content, and activities, lesson flow, student responses, reactions, and behaviors; and (d) CSAS training on the effective instruction literature (see training section).
CSAS IS and BMS Rating Scale Definitions.
Note. CSAS = Classroom Strategies Assessment System; IS = Instructional Strategies; BMS = Behavior Management Strategies.
The psychometric evidence of the CSAS has been examined in numerous investigations (e.g., Reddy et al., 2013a, 2013b, 2015; Reddy, Glover, et al., 2019; Reddy et al., 2019). The CSAS evidences good psychometric properties that include reliability, content, construct, convergent, and predictive validity. The IS and BMS Rating Scales are theoretically and factor analytically derived via confirmatory factor analysis. The CSAS evidenced good internal consistency (Cronbach’s α values above .90), fair-to-good test–retest reliability across a 2- to 3-week span (r values above .70), plus good interrater reliability for Strategy Counts (r = .94), the IS and BMS Rating Scales (r = .80 and .72), and the Classroom Checklist (r = .86). Also, the CSAS is free from item bias using differential item functioning analyses and convergent and discriminant validity with observational tools such as the Classroom Assessment Scoring System (CLASS; Pianta et al., 2008; Reddy et al., 2013a), Danielson Framework for Teaching (Reddy et al., 2019), and student ratings of instructional environment (Nelson et al., 2017). The CSAS IS and BMS discrepancy scores predict student achievement (e.g., Dudek et al., 2019; Lekwa et al., 2019a; Reddy et al., 2013c; Reddy et al., 2020; Reddy, Glover, et al., 2019; Reddy, Shernoff, et al., 2019) and student behavior (Lekwa et al., 2019b).
Cooperative Learning Observational Code for Kids (CLOCK)
The CLOCK (Volpe & DiPerna, 2010) is a direct observational measure used for assessment of classroom academic engagement. The CLOCK assesses five observable behaviors (i.e., active engagement, passive engagement, positive social interactions, nonphysical aggression, and interference) and was collected by trained independent observers at baseline and postassessment blind to condition. The effect of CSC on classwide academic behavior engagement, active engagement time (AET), and passive engagement time (PET) scores was included in the analyses. AET was defined as attending to instructionally relevant stimuli and engaging in instructionally relevant behavior (e.g., students raising hands to ask or answer questions, writing in a notebook, working on a learning task with a peer). PET was defined as attending to instructionally relevant stimuli (e.g., looking at the teacher during instruction). CLOCK observations begin by randomly selecting 12 students to observe at baseline and postassessment. Observers record the occurrence of AET and PET using a momentary time sampling procedure (based on 15-s intervals) that alternates between the 12 students after every four intervals. After observing each of the 12 randomly selected students, observers repeat the same sequence for a total of at least 36 min per classroom. CLOCK observers listened to prerecorded audio files during their observations to ensure they were coding the correct intervals for the CLOCK, which requires 15-s intervals of coding during live observations.
Independent observers rotated observations of all students in the classroom selected at random using a time sampling procedure that has been found to be a reliable method for estimating classwide academic and behavior trends (e.g., Briesch et al., 2014; Dart et al., 2016). The CLOCK yields high levels of interobserver agreement, ranging from 89% to 92% during intervention studies (Volpe et al., 2012).
Teacher self-ratings of strategy use and student functioning
Participating teachers rated the degree to which their use of instructional and behavior management strategies had changed over the previous 4 weeks (at baseline) or over the previous 8 weeks (postcoaching) using a 7-point Likert-type rating scale that ranged from 1 = very much worse to 4 = unchanged, and 7 = very much improved (five items; α = .93) using a measure created by the lead author. Teachers also rated the degree to which student behavior (seven items; α = .92) and academic functioning (eight items; α = .88) changed at baseline and postassessment using the same scale.
Teacher Stress and Support Assessment (TSSA)
The TSSA is a multidimensional assessment of teachers’ perceived instructional and emotional support and stress related to classroom teaching. The TSSA was created by the authors and evidences good reliability, content, and construct validity (Reddy et al., 2016). The TSSA consists of three factors (factor loadings ranging from .471 to .893 for hypothesized scale) with 27 nested items. The Instrumental Support scale (eight items; α =.92) evaluates forms of practical guidance, such as input from peers and instructional leaders, which lead to improvement in classroom practices. The Emotional Support scale (six items; α = .94) evaluates feelings of acknowledgment, receipt of respect, and encouragement. The Stress scale (13 items; α = .88) evaluates feelings of stress from the school context. For the Instrumental and Emotional Support Scales, teachers rated their level of agreement (1 = strongly disagree to 5 = strongly agree), and for the Stress Scale, teachers rated their level of stress (1 = no stress to 5 = extremely high stress). The item ratings within each scale were averaged to compute a total scale score (Lekwa et al., 2019b).
Teacher Coaching Evaluation Scale
Teachers who received coaching completed the Teacher Coaching Evaluation Scale, a 14-item Likert-type scale assessing teachers’ satisfaction with CSC using a 7-point Likert-type scale (1 = strongly disagree to 7 = strongly agree). The scale demonstrated good internal consistency (α = .87) and was used in other coaching studies (e.g., Fabiano et al., 2018; Reddy et al., 2017).
Procedures
Informed consent was obtained from all participating teachers in the early fall or prior to the school year. All teachers were randomly assigned to one of two conditions: (a) CSC Model or (b) WL control, with each condition following three phases: baseline assessment, coaching or a period of noncontact for WL teachers, and postassessment. Coaching was implemented in the late fall and commenced in the spring of the same academic year.
Observer training
A total of 29 trained independent observers performed baseline and postassessment observations. Observers were trained to criterion on the CSAS and the CLOCK and then conducted observations independently from the coach. CSAS observer training included 3.5 days of training focused on theory, research-based instruction and behavior management practices, assessment design, evidence, and scoring. Training focused on taking targeted notes that assessed evidence of practice use and student–teacher interactions, knowledge tests, and coding videos with feedback from a Master CSAS coder. At the conclusion of the training, all observers independently coded five classroom videos to criterion (i.e., 80%), similar to criteria used for other classroom observation tools, including the Framework for Teaching (Danielson, 2013; Joe et al., 2013) and the CLASS (Pianta et al., 2008).
CLOCK training included instruction on the basic concepts of systemic, direct observation of student behavior (e.g., methods of time sampling, explanations of operational definitions) and definitions of targeted behaviors. Training entailed the use of videos of classroom instruction and coding of student behavior therein. This enabled opportunities for group and individual practice and feedback based on observer coding of behavior in the training videos. All observers independently coded three classroom videos to criterion (i.e., 90%).
Observation data collection
Participating general and special education teachers were observed for 30-min live observations by pairs of independent observers (i.e., one observer completed the CSAS and the other completed the CLOCK concurrently). Observations were scheduled to occur no more than 1 week apart. A total of two CSAS observations per teacher, as well as a minimum of 36 min of observation using the CLOCK (three observations), were completed at baseline and postassessment. Interrater reliability for the CSAS was assessed separately in a set of observations (26% of the total sample), in which observer pairs completed CSAS observations at the same time for 30 min. Average rates of agreement on the IS and BMS Rating Scales ranged between .82 and .84. Likewise, interrater reliability for the CLOCK was assessed by observer pairs that completed at least 24 min of CLOCK observation for a group of 12 students selected at random. During interrater reliability observations of the CLOCK, observers used ear-bud splitters and listened to the same 15-s interval recording stored on one audio device shared by both individuals to ensure observers coded behavior for the same student at precisely the same intervals.
Interobserver agreement of CLOCK and CSAS was examined prior to analyses using a subset of 26% of the baseline and postassessment observations. For the CLOCK, average rates of agreement of .89 (AET) and .86 (PET) were obtained. Average ranges of agreement on the CSAS Strategy Counts were 73% and IS and BMS Rating Scales were .82 and .84.
Scores were averaged across the two observations to determine the baseline CSAS scores. Following the baseline assessment period, the coaching began for the CSC condition. The WL group did not receive CSC and did not interact with study staff until the postassessment.
For all teachers, independent observers completed the CSAS during the same classroom lessons (the same time of the school day) as the baseline assessment. When possible, the same observers were assigned to assess the same teachers at baseline and postassessment. Teachers who received coaching completed the TSSA, Strategy Usage Questionnaire, and Teacher Coaching Evaluation Scale. Both CSAS and WL teachers were provided US$100 as a token of appreciation for their time during the study.
CSC model and coach training
The CSC Model is a data-driven coaching model based on coaching, consultation, and teacher formative assessment literatures (Bergan & Kratochwill, 1990; Kurz et al., 2017; Reddy et al., 2013a). The CSC Model uses CSAS observational data to guide coach–teacher actions for improving instructional and classroom management practices and the classroom environment for all students, including those with disabilities. The CSC Model consists of eight, 30-min coaching sessions (4 hr) that iteratively used CSAS scores and online visual performance feedback to quantify, target, encourage, and support teacher implementation of research-based practices. Specifically, the CSC Model uses data to inform the implementation of research-based instructional and classroom management practices by (a) identifying teachers’ needs and resources, (b) establishing data-based goals, (c) developing implementation plans to achieve those goals, (d) modeling and practicing implementation steps, (e) providing ongoing verbal and visual performance feedback, (f) evaluating implementation and goal attainment, and (g) revising action plans as needed for implementation and generalization (e.g., Reddy et al., 2017, in press). Coaches engage in an interactive cycle of classroom observation, modeling, practice, encouragement, and performance feedback to support teachers’ implementation of research-based practices toward professional development goals. Teacher engagement and implementation fidelity are enhanced through collaborative decision-making, relationship alliance building, self-assessment, and coach actions of modeling, practice, feedback, and encouragement. The CSC Model includes five phases of coaching: identify practice needs, formulate practice goals, design plans and prepare for implementation, support implementation, and evaluate implementation and goal progress. Table 4 describes the five phases of coaching and objectives by session.
CSC Model Phases, Sessions, and Objectives.
Note. CSC used cycles of CSAS scores, visual performance feedback graphics, and lesson-specific anecdotal observations to guide coach–teacher professional development conversations during coaching phases (see Reddy, Shernoff, & Lekwa, in press). CSC = Classroom Strategies Coaching; CSAS = Classroom Strategies Assessment System.
The CSC process included teacher self-assessment on their weekly implementation of targeted classroom strategies during Coaching Sessions 4 through 8. At the beginning of Sessions 4 through 8, coaches asked teachers to self-assess (rate on a 5-point frequency scale, that is, 1 = never to 5 = almost always) how often and how well they implemented targeted strategies during the past week. Teachers’ responses were then briefly discussed in the coaching session. Data were used for descriptive purposes (process data) not to assess coaching effects. Specifically, the average self-assessed frequency of strategy use for CSC teachers (n = 53) was 3.87 (.77) and general education (n = 44) versus special education teachers (n = 9) was 3.87 and 3.86. Also, the average ratings of how well CSC teachers (n = 52) implemented targeted strategies were 3.71 (.75) and general education and special education teachers were 3.68 and 3.78. Overall, teachers’ self-assessed strategy implementation after Session 3 (i.e., established goals and implementation plans) was consistent across Coaching Sessions 4 to 8; similarly, no differences were noted between general and special education teachers.
The CSC Model has promising evidence of effectiveness based on two previous RCTs (Fabiano et al., 2018; Reddy et al., in press). In the present RCT, coach and teacher meetings occurred approximately weekly over 10 weeks, with two coach observations (30 min each) conducted between coaching sessions. For further detail about the model and examples of interactions between coaches and teachers, readers are referred to a detailed case coaching study (Reddy et al., 2019).
Coaches participated in a 3-day training on the purpose and structure of CSC Model and were trained to reliability on the CSAS (Reddy & Dudek, 2014). All coaches were required to complete a CSAS reliability test of 70% or higher agreement on five master-coded videos. Coaches were trained on a manual-based approach that included a session-specific procedural checklists. Coaches were also trained on student, school, and community contextual factors learned from the a qualitative study that included focus groups representing K–5 grade teachers from the district in advance of the RCT (see Shernoff et al., 2017).
Coaches received weekly supervision by the investigators to maximize fidelity and problem-solve barriers to coaching implementation. Supervision included role-playing and modeling coaching sessions, reviewing CSS data and progress in meeting goals, problem-solving barriers to implementation of teacher professional development goals, and the integration of memory strategies (i.e., ways to help teachers remember to implement their plans) to enhance teachers’ implementation fidelity.
All coaching sessions were audio recorded when teachers consented and at least 65% of coaching cases were audio recorded (eight planned sessions) and reviewed by PhD-level supervisors using a fidelity CSC checklist aligned to each session. Coaching fidelity was computed based on the percent of procedural steps completed by the coach within and across coaching sessions per case. Approximately 99% of CSC session-specific components were implemented. The high degree of coaching fidelity in this study is attributed to the manualized coaching model and coach training aligned to the coaching checklists prior to implementation. In addition, the supervision model included coach supervision before and after each coaching session and audio recording of coaching sessions, which allowed for review, discussion, and feedback to coaches during supervision to maximize implementation fidelity.
Data Management and Analytic Approach
The RCT from which these data were collected employed a three-cohort, randomized block design carried out by blocking consented teachers within a school and then randomly assigning this block of teachers to either the CSC or WL control condition. Teachers were observed with the CSAS approximately 1 week prior to the start of coaching (baseline) and approximately 10 weeks later (postassessment). There were no statistically significant differences between general or special educators in terms of demographic variables between experimental conditions (see Table 1).
Two CSAS observations were conducted by independent observers at baseline and postassessment each. For baseline, the Strategy Count totals for each behavior were calculated separately for Observations 1 and 2 and then averaged across the two observations. For the Strategy Rating Scales of IS and BMS, scale scores (frequency scores) were calculated separately for Observations 1 and 2 and then averaged across the two observations. The same procedures were repeated for the postassessment scores.
IS and BMS discrepancy scores were computed as follows: recommended frequency − observed frequency ratings. Absolute value discrepancy scores were first calculated at the item level for the IS and BMS scales before being summed to create discrepancy scale scores. Item and scale score calculations were performed separately for Classroom Observations 1 and 2 prior to averaging across the two observations to obtain the average absolute value discrepancy scale scores. Aggregate IS and BMS discrepancy score procedures were conducted for the baseline and postassessments. Thus, average absolute value discrepancy scores in this study indicate whether the observer determined whether any change was needed in the teacher’s classroom practices (Reddy et al., 2013a, 2015).
Differences in baseline rates of strategy use (e.g., CSAS Strategy Counts and IS and BMS Rating Scale scores prior to random assignment) were compared between general and special educators. Generalized linear models for data following a negative-binomial distribution were used to examine differences in strategy counts between general and special educators (Beaujean & Morgan, 2016; Gelman & Hill, 2007). Differences in baseline quality of strategy use, as measured by IS and BMS Rating Scales total scores and subscale scores, were compared via linear models of each strategy using educator type (general or special) as a sole predictor. The frequency with which teacher strategies were selected as goals for coaching by general and special educators was compared using Fisher’s exact test (Upton, 1992).
The present investigation analyzed coaching outcomes as direct observation of changes in teacher practices and student behavior, as well as a set of teacher ratings of student functioning, strategy improvement, and school-related supports from baseline to postassessment. First, Strategy Counts selected by the teacher and coach as areas in need of improvement (practice goals) were used as outcomes (see Tables 2, 5, and 6). We analyzed Strategy Count data as a family of strategies (instead of a group of distinct, independent strategies) for two reasons. First, how often teachers use individual strategies can depend in part on how often they use complementary strategies (e.g., higher use of Academic Response Opportunities associated with higher use of Academic Praise). Second, variation in selection of goals is based on individual teachers’ needs. For these reasons, we compared overall IS and BMS strategy counts between pre- and postcoaching, followed by analysis of individual strategy counts to fully explore nuance of changes between teacher type and condition.
CSAS Descriptive Statistics for Baseline Strategy Use.
Note. CSAS = Classroom Strategies Assessment System; Gen. ed. = general education; Sp. ed. = special education.
Negative binomial models were fit to CSAS Strategy Count data; general linear model was applied to CSAS Strategy Rating Scale discrepancy scores.
Frequency of Selected Teacher Practice Goals (N = 53).
Note. CSAS = Classroom Strategies Assessment System; Gen. ed. = general education; Sp. ed. = special education.
Participating teachers who were in the coaching condition.
The distribution of Strategy Rating scales practice targets (goals) were evenly distributed within the IS and BMS domains (see Table 6). Specifically, IS ratings revealed that Direct Instruction, Promotes Student Thinking, and Academic Performance Feedback were most often selected as goals, and BMS ratings indicated that Proactive Methods, Directives, and Praise were most often selected by teachers and coaches. Given this pattern, we analyzed Strategy Rating discrepancy scores at the IS and BMS Totals. Third, CLOCK student observation data and teacher rating scales were analyzed at the total scale score for each outcome.
Finally, analysis of covariance (ANCOVA), covarying baseline scores on measures for all outcomes, was computed (Dimitrov & Rumrill, 2003; Van Brueklen, 2006). Coaching outcomes were examined using the Strategy Rating Scales—IS and BMS total discrepancy scores from independent observers at baseline and postassessment. In the present study, assumptions of the ANCOVA procedure were met (i.e., linear relationship between covariate and outcome, independence of baseline data and treatment, homogeneity of variances and slopes), with the exception of the normality of the residuals for the Strategy Counts, although we expect that this did not impact Type 1 error, given the robustness of F tests to nonnormality (e.g., Blanca et al., 2017).
Results
This investigation focused on several key baseline assessments and coaching outcomes that included (a) observed frequency of specific instructional and behavior management strategies via CSAS Strategy Counts, (b) observed quality of strategy use via CSAS IS and BMS discrepancy scores, and (c) student academic behavior engagement via the CLOCK. In addition, teacher ratings of student academic and behavior functioning, improved strategy use, perceived instructional and emotional supports, and stress were outcomes of interest.
Research Question 1: Frequency and Quality of Classroom Practices Prior to Coaching
Descriptive results related to the frequency and quality of teachers’ use of specific instructional and behavior management practices are displayed in Table 5. General and special educators in general did not differ in their average use of discretely observable instructional and behavior management strategies (differences of 0.5 to 5 uses per 30-min observation). Negative binomial models revealed no significant differences between general and special education teachers’ rates of use of discrete IS during the 30-min observations (Strategy Counts) prior to coaching. Findings revealed no significant differences between general and special education teachers’ average use of directives (i.e., Clear One to Two Steps or Vague Directives), and Behavior Praise. Special education teachers, however, used 30% fewer Behavior Corrective Feedback statements on average (p = .04) when compared with general education teachers prior to coaching (see Table 5). Results from general linear modeling indicated that the qualities of teacher practices, as measured by Strategy Rating scale, IS and BMS Total and subscale discrepancy scores, did not differ significantly between general and special education teachers.
Overall, participating teachers were observed implementing evidence-based instructional and behavior management practices at a range of frequencies prior to coaching. Strategy Counts (frequency) for instruction reflected relatively low rates of Concept Summarizes, Academic Praise and Corrective Feedback, whereas Academic Response Opportunities reflected relatively moderate rates for both groups (Reddy et al., 2020). Likewise, Strategy Counts for behavior management reflected relatively low rates of Behavior Praise and Vague Directive, whereas Clear, One- to Two-Step Directives and Behavior Corrective Feedback reflected relatively moderate rates for both groups (Reddy et al., 2012, 2013c).
Common practice ratios were also examined. Specifically, general education teachers, as a group, demonstrated a ratio of approximately 3.19 Academic Response Opportunities to each Academic Praise Statement and a ratio of 5.02 Academic Response Opportunities to every Academic Correct Feedback; the same ratios for special educators were 2.47 to 1 and 5.11 to 1, respectively. General education teachers, as a group, demonstrated a ratio of approximately 4.98 Behavior Corrective Feedback Statements to each Behavior Praise, and special educators, as a group, demonstrated a ratio of 3.03 Behavior Corrective Feedback to each Behavior Praise. Academic Praise to Academic Corrective Feedback ratio reflected a 1.57 to 1 ratio for general education and a 2.07 to 1 ratio for special educators, respectively.
Research Question 2: Goal Selection by Teacher Group During Coaching
Fisher’s exact tests (see Table 6) examined the frequency with which general versus special educators selected specific instructional and behavior management strategy goals and revealed no significant differences between groups. Results revealed special and general educators in this study generally focused on goals related to frequency of use of Concept Summaries and Academic Praise (Strategy Counts of IS), as well as strategies involved in Academic Performance Feedback, Direct Instruction, and Promotion of Students’ Thinking (IS Rating scales). General and special education teachers selected Behavior Praise most often within behavior management strategies (CSAS Strategy Counts and BMS Rating scales).
Research Question 3: Effects of Coaching by Intervention Condition and Teacher Group
Observed classroom practices and student behavior
The effect of CSC coaching did not differ between general and special education teachers (all p values for interaction terms between coaching effect and teacher type exceeded .08). Similarly, significant effects of coaching on children’s behavior engagement observed in the main RCT (Reddy et al., in press) did not vary significantly between general and special education teachers. As reported in the main RCT, the effect of coaching on teacher practices was moderate to large (d values ranging from 0.52 to 0.74), and the effect of coaching on student behavior was moderate (d = 0.41).
Teacher-rated student functioning, classroom practices, supports, and stress
No significant differences were observed between teacher type (i.e., general vs. special educator) for student academics (p = .37) or behavioral functioning (p = .35). General education teachers’ self-rated change in strategy use was slightly greater at postcoaching than that of special education teachers, yet this did not achieve statistical significance (p = .07). The effect of coaching on general and special education teachers’ perceived instrumental support, emotional support, and stress did not differ (p values for interaction terms each in excess of .20). Finally, mean ratings from both teacher groups who received coaching indicated strong satisfaction with the CSC Model (M = 6.15, SD = 1.24); mean ratings between general and special education teachers did not differ significantly (p = .14).
Discussion
This study represents the first RCT of a data-driven coaching approach on universal classroom practices of general versus special education teachers and student behavior in high-poverty elementary schools. As a secondary analysis of a WL, controlled, coaching efficacy trial, the CSC Model appeared to produce comparable improvements in general education and special education teachers’ classroom practices and student outcomes. Findings from this study are promising, as the main RCT (Reddy et al., in press) found that teachers in the intervention condition (n = 53) evidenced statistically and clinically meaningful improvements in the frequency and quality of their use of evidence-based classroom practices (ds ranged from 0.52 to 0.74) and classwide academic engagement (d of 0.41) when compared with WL control teachers (n = 53). Furthermore, CSC teachers in the main RCT self-reported significant improvements in their instructional and emotional support (ds of 1.04 and 0.90) and student functioning (i.e., academic and behavior; ds of 0.96 and 1.24) when compared with control teachers. Teachers in the main RCT reported a high level of acceptability and satisfaction with coaching. Thus, the present study builds on prior evaluations of the CSC Model by further illustrating the utility of integrating teacher formative assessment in instructional coaching with promising outcomes observed in both general and special education teachers’ universal classroom practices and classwide academic engagement. Specifically, CSAS assessment data guided coaching actions such as identification and prioritization of needs, setting practice goals, creating plans for implementation of strategies, and evaluation of progress toward goals (see Table 4). CSAS scores reflecting general and special education teachers’ practices changed, following the brief coaching intervention (i.e., eight sessions, 4 hr) delivered in high-poverty elementary schools.
Prior to coaching, participating general and special education teachers in the current sample used a range of research-based instructional and behavior management practices (see Table 5) with similar frequencies and levels of quality. Although these are promising findings, some practices used were not aligned with recommendations from the research literature. In contrast to the long-standing recommended 3:1 ratio of Behavior Praise to Behavior Corrective Feedback in general education contexts and 5:1 ratio in special education learning contexts (e.g., Partin et al., 2010; Shernoff et al., 2020), general education teachers in this sample demonstrated a ratio of approximately 1 Behavior Praise to 5 Behavior Corrective Feedback statements prior to coaching, and special educators demonstrated a ratio of 1 Behavior Praise to 3 Behavior Corrective Feedback statements prior to coaching. However, in this study, Academic Praise to Academic Corrective Feedback ratio reflected slightly higher rates, with approximately 1.6:1 ratio for general education and a 2:1 ratio for special educators. Low rates of academic and behavior praise have been well documented (Brophy & Good, 1986; Fabiano et al., 2018; Shernoff et al., 2020; White, 1975), with rates of praise decreasing as grade level increases (Reddy et al., 2013). In this study, findings suggest that CSC implementation may have contributed to improving (reversing) the ratio of Behavior Praise to Behavior Corrective Behavior Feedback for both general and special education teachers at postassessment. Similarly, both groups of teachers evidenced rates of Academic Response Opportunities that appeared to be below recommended frequencies for the instruction of new material (four to six response opportunities per minute of instruction) or reviewing material previously learned (nine to 12 response opportunities per minute of instruction; for example, Council for Exceptional Children, 1987; Sutherland et al., 2003).
Similarly, baseline CSAS data suggest higher rates of Academic Response Opportunities in relation to Academic Praise or Behavior Praise. General education teachers demonstrated a ratio of Academic Response Opportunities to Behavior Praise of 11:1 and special education teachers demonstrated a ratio of 8:1. These ratios perhaps represent missed opportunities for both teacher groups to provide meaningful feedback (e.g., praise) to students after they participated verbally or nonverbally (i.e., answers on white boards) to learning opportunities in class. Research on opportunity to respond (OTR) has found that high frequencies of OTR and behavior praise can increase student engagement, learning, and time on task and reduce disruptive behaviors for students with or without disabilities (e.g., Sutherland et al., 2002; Sutherland & Wehby, 2001). In sum, baseline CSAS data reflect teachers’ use of a range of research-based practices, but, in general, at rates below recommended standards. Modest rates of strategy use offer opportunities for researchers and school personnel to design and validate new assessments and professional development models that lead to teacher professional growth and effectiveness over time.
In the current study, coaches and teachers collaboratively used CSAS data to reflect on practices and formulate goals for improvement of specific practices during coaching. These findings offer insights regarding general and special education teachers’ beliefs regarding important practice areas for which they require professional development. Descriptive statistics suggested that general education teachers in the CSC condition prioritized Concept Summaries, Academic, and Behavior Praise as their top three goal choices (frequency as measured by CSAS Strategy Counts). Also, among general education teachers in the CSC condition, Direct Instruction, Promotes Student Thinking, and Academic Performance Feedback (IS Rating subscale discrepancy scores) were selected as their top three goals for instructional quality improvements, whereas Proactive Methods, Directives, and Praise were selected as their top three goals for behavior management quality improvements (BMS Rating subscale discrepancy scores, that is, larger scores reflect greater need for change in practices). Special education teachers in the CSC condition identified Concept Summarizes and Behavior Praise (Strategy Counts) as priorities for support and Direct Instruction, Academic Performance Feedback and Behavior Praise as targets for instructional and behavior management quality improvements (IS and BMS Rating Scale discrepancy scores).
In addition to having similar baseline frequencies and qualities in teaching practices, general and special education teachers assigned to the CSC condition evidenced similar benefits from coaching—in terms of change in classwide instructional and behavior management strategies, as well as independent observers’ coding of student behavior. Similarities at baseline related to the frequency and quality of instructional and behavior management strategy use, goal selection, and response to coaching between general and special education teachers contrast with prior work in this area (e.g., Brownell et al., 2005; Flower et al., 2017; Harry & Lipsky, 2014; Sindelar et al., 2010). Scholars in teacher education have noted philosophical and practical differences between general and special educators. General education teachers, for example, might traditionally adopt a constructivist teaching approach while special education teacher training often centers on behavioral principles and direct instruction methods (e.g., Brownell et al., 2005; Burns & Ysseldyke, 2009). Results of this study did not bear out this trend, which suggests at least partial overlap between professional development needs of general and special education teachers that can be addressed through brief coaching focused on formative assessment of specific classwide instructional and behavior management strategies. This finding is encouraging in the context of inclusive classrooms where general and special educators are working alongside each other to plan for and deliver instruction to students with and without disabilities.
Limitations
Findings from this investigation should be interpreted in light of limitations regarding setting, sample, and data collection. First, participating teachers were predominately female and European American and teaching in a large school district in a northeastern state. We were unable to collect detailed information on teacher credentials (i.e., quality of prior education, preservice training, or exposure to other professional development supports). Thus, we are unable to assess the influence of teacher credentials and characteristics or other services received on classroom practices or coaching outcomes. Second, coaching was delivered primarily in general education classrooms in urban high-poverty elementary schools in a district committed to inclusive instruction. Thus, results from this study may not be generalizable to other communities (e.g., affluent, suburban, rural), schools (e.g., middle or high schools), teachers, and classroom contexts (e.g., self-contained). Future research on the effect of the CSC Model with special education teachers in self-contained settings would be beneficial. Third, participating general and special education teachers were not selected based on professional development needs or concerns and it remains unclear whether coaching supports would produce similar findings with teachers struggling in instruction. Fourth, it is unclear whether teachers may have responded differently to coaching if the CSC Model was required by their principals. Fifth, CSC was delivered by coaches external to the school district, and thus, it remains unknown whether CSC-trained coaches employed by the school district would yield comparable findings. Sixth, the present study did not assess the long-term effects of coaching for general and special education teachers and their students, and thus, teacher and student outcomes may have varied based on teacher type. Sixth, this RCT did not randomly stratify participants based on teacher type (i.e., general vs. special education) and context of instructional delivery (general education, resource, self-contained classrooms). As a result, our sample sizes differed with fewer special education teachers (n = 21) overall despite meeting data analytic assumptions; detection of perhaps nuanced differences between teacher types will require prospective analysis of data from larger groups of both kinds of teachers. Seventh, this study examined the effects of coaching on universal classroom practices and classroom-level student academic engagement, precluding student-level analyses based on specific special education classifications. Likewise, this study was not designed to assess the effect of coaching on individual student academic and/or linguistic skills. Specifically, the study was not designed to evaluate the effects of coaching on individual or small groups of students who struggle with specific academic areas such as English language, reading, writing, mathematics, science, and/or foreign language. Eighth, the presence of independent observers in the classroom may have influenced teacher instruction along with student behavior and/or engagement in that instruction. Ninth, student academic functioning was not assessed with standardized achievement testing in this study, but rather teacher ratings of perceived improvement in their students’ behavior and academic functioning. Tenth, studies examining the effects of coaching with larger populations of students with learning disabilities are warranted. Finally, despite the randomized design, teachers within the same school were randomized by block to WL and CSC conditions, and this design does not preclude the possibility of contamination between conditions (e.g., teachers discussing the study or practices embedded within the coaching model). The main RCT employed procedures to help reduce the possibility of contamination (i.e., distributed written guidelines for teacher participants at recruitment, discussed importance of not sharing information with colleagues while receiving coaching). Finally, due to project sample size, we were unable to examine whether changes in the quality of instructional and/or behavior management practices mediated the relationship between coaching for general versus special education on student outcomes. This is an area for future research to help elucidate the active ingredients that are carrying coaching intervention programs for diverse teachers.
Directions for Research
Replication of these findings with a larger number of special education teachers is warranted. Such replication will allow not only for greater statistical power, and potential identification of small differences in effects of coaching between teaching certification types, but might also enable analyses of the extent to which classroom settings also influence teachers’ baseline strategy use and response to coaching. For example, special education teachers in the current sample were primarily following a “push-in” or “co-teaching” (e.g., Friend et al., 2010) format, in which instructional responsibilities may be somewhat blended between special and general education teachers for students with and without Individual Education Programs (IEPs). Although the magnitude of special and general education teachers’ responses to the CSC Model did not appear to differ in these analyses, it remains possible that the rate at which their practices changed in response to coaching differed in meaningful ways. Future studies incorporating formative assessment of teacher practice in data-based models of coaching might enable use of comparative growth models to evaluate differences in the magnitude or functional form of teachers’ rates of acquisition or improvement of specific practices. Also, the quality of prior preservice teacher education training warrants investigation. It is possible that exposure to high-quality didactic training on effective teaching and behavior management practices, as well as the provision of targeted ongoing performance feedback by mentor teachers during preservice teaching placements, may influence the frequency and quality of universal practices observed during future employment in schools.
Findings from this study, which included urban high-poverty schools, may be of particular importance for teachers who work in similar contexts. Educators who teach students in high-poverty schools have higher rates of attrition and stress and higher student achievement gaps in comparison with teachers working in other school contexts (e.g., Borman & Dowling, 2008; Boyd et al., 2012). Recent data indicate that teachers in such contexts can benefit from evidence-based and practical job-embedded professional development that support their implementation and evaluation of research-based instructional and behavior management practices that maximize student learning and behavior (Reddy et al., in press). However, additional research that assesses the efficacy of professional development interventions (i.e., coaching) for general and special education teachers in high-poverty schools is needed. Also, future investigations of the CSC Model that use school-based coaches would be beneficial for validation efforts. Research that examines the influence of race/ethnicity match or mismatch between classroom observers and teachers warrants investigation.
Research is needed that measures the effects of coaching on special education teachers’ practices and individual or small groups of students who struggle with specific academic areas such as English language, reading, writing, mathematics, science, and/or foreign language. Also, investigations that examine the effect of coaching on student response to interventions via systematic direct observation and standardized academic assessments are warranted.
Finally, a comprehensive understanding of the effects of any teacher professional development or coaching model requires assessment and analysis of students’ response to change in teacher practices, an area of particular interest as research on professional development for special educator progresses. For example, studies have found differential benefits of instructional approaches based on student baseline skill levels (e.g., skill by treatment interactions; Connor et al., 2004a, 2004b; Doabler et al., 2018).
Conclusion
Findings from the current investigation advance knowledge on the practice needs and goals of general and special education teachers receiving data-driven coaching in high-poverty elementary school settings. The current RCT examined the processes and outcomes of the CSC between general and special education teachers’ instructional and behavior management practices, as well as classwide student academic engagement. Overall, findings from this study were promising and spotlight the need for further validation with larger teacher samples and possible new avenues for school personnel preparation and professional development.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The research reported here was supported by the Brady Education Foundation to Rutgers University. The opinions expressed are those of the authors and do not represent views of the Foundation.
