Paper

I utilized a three-stage case study methodology, with nine peer-nominated teacher leaders in three elementary schools as primary data sources. The multi-stage design for data collection and analysis permitted modification, enhancement, and expansion of a categorization of supports and barriers to teacher leadership at three distinct points during the data collection process (Yin, 1991). Recursive data analysis and modification of the categorization allowed tailoring of interview protocols at each stage in order to seek confirmatory or disconfirmatory data and strengthen the findings. In this section, I describe the methodology in slightly more detail. For more information, I refer the reader to Zinn (1997).

Prior to data collection, I developed an a priori categorization of supports and barriers to teacher leadership that was based on existing research on teacher leadership. This categorization consisted of a sixcelled matrix listing supports and barriers within three arenas: (a) conditions within the educational context; (b) conditions outside the educational context; and (c) internal, intellectual and psycho-social factors. I used this matrix as a starting point from which to expand my understanding of factors influencing teacher leadership.

Each teacher and the principal at potential sites nominated three teachers in their schools whom they felt best exhibited a given set of teacher leader charactenstics (Table 1). After weighting the ranked nominations, I selected for participation three teacher leaders in each school with the highest weighted scores. One teacher leader at each site was interviewed in the first round of data collection, and the two remaining teacher leaders at each site were interviewed in the second round of data collection. All participants engaged in a focus group interview in the final round of data collection.

Characteristics of Teacher Leaders Nominated for the Study

Teacher leaders in this study have the following characteristics:	Teacher leaders in this study do not necessarily have the following characteristics:
Maintain some level of teaching responsibility Participate actively in school-wide or district decision-making on a regular basis Are held in high regard by their peers Engender a high level of trust Are teachers to whom others go for advice, information, or assistance	Do not necessarily have formal titles or roles as teacher leaders Do not necessarily chair or facilitate committees Do not necessarily get paid extra for extra work

Data Collection and Analysis

The design for data collection and analysis was based on Yin's (1991) recommendations for multi-stage, case study design. Data collection and analysis took place in three distinct rounds. First round data collection and analyses were based on individual interviews with a single teacher leaders from each school. Second round data collection and analyses were based on single, individual interviews with the remaining two teacher leaders from each school. Final round data collection and analyses were based on a focus group interview with all nine participants. In the following section, I describe each of the three phases of the study in greater depth.

In the first round of data collection and analysis, one teacher leader from each school took part in a series of three open-ended, conversational, individual interviews (Seidman, 1991). I analyzed interview data in two ways. First, I developed and annotated personal narratives (Mishler, 1995, Polkinghorne, 1995). In these, I condensed each participant's interview transcripts to tell her story in chronological order, emphasizing supports and barriers to her own leadership Alongside the personal narratives, I added annotations highlighting key points and connecting selected text with statements by other Round One participants or concepts in the literature. In the second form of analysis, I labeled transcripts with codes indicative of various supports and barriers descabed by participants, clustered related codes and labeled clusters with descaptors, and placed code clusters with descriptors into separate, blank, six-celled matrices for each participant. Next, I combined the three matrices into one. Using my analyses, I developed a cross-case analysis grounded in the data for Round One (Miles & Huberman, 1994; Patton, 1990, Yin, 1991). Although I had planned to use first round findings to enhance the a priori categorization, in practice, I discovered I needed to work the other way around. I used the a priori matrix to enhance a categorization grounded in the Round One data.

In the second round of data collection and analysis, the six remaining participants took part in single, individual interviews. I utilized a semi-structured interview protocol, developed at the conclusion of Round One and designed to permit enrichment of data within each cell of the draft conceptualization. For the Round Two interviews, I used two forms of data analysis. First, I coded individual transcnpts, clustered codes, and added cluster descriptors. With codes and descriptors placed in individual blank matnces, I developed individual case reports. Second. I developed a cross- case analysis for Round Two, in which I compared and contrasted responses of second round interviewees with each other, with those of the three Round One respondents, and with existing literature on teacher leadership. To develop the cross-case analysis, I collected provisional cluster descriptors for all six cases in yet another blank matnx and grouped descriptors by category, provisionally labeling each set of grouped descriptors. This process yielded a categorization grounded in the issues raised by the secondary respondents. Melded into the matrix developed at the conclusion of Round One, the result was an enhanced conceptualization based on findings from two phases of data collection and analysis, as well as existing literature.

The final round of data collection consisted of a single focus group interview held with all nine teacher leaders. The interview guide, based on Krueger's (1991) design and developed at the conclusion of Round Two, had two main purposes: (a) to test accuracy of m* constructions to this point, and (b) to hear from the group implications and findings their experiences might have for teacher leadership. Data were analyzed by cross-checking participants' comments in each portion of the interview with the draft categorization. This focus group interview permitted additional revisions and refinements to the categorization. In addition, it informed subsequent discussion of implications.

Internal and external consistency were enhanced through use of multiple data sources, categories derived both deductively and inductively; and participant checks of data, personal narratives, and individual case reports. By grounding the study in a categorization based on existing research and by using three, separate rounds of data collection and analysis, I was able to test and retest my evolving understanding of sources of support and barriers to teacher leadership.

Participants taught in elementary schools in separate school districts. Of the nine participants, only three were classroom teachers. The remaining six, while serving in a variety of capacities, shared three sign)ficant characteristics. Their regular job responsibilities: (a) did not tie them to a single classroom; (b) allowed relative flexibility and discretion in the wav in which their time was used. and (c) demanded considerable contact with their peers and their pnncipals. In order to form a basis for understanding the perspectives of the participants, I briefly acquaint the reader with the three participants from each school. For a more detailed description of participants and their educational contexts, I refer readers to Zinn (1997).

Lea, Maya, and Sonny taught at Marshall Elementary, a dark and overcrowded school located in a small, relatively rural, community with a diverse population comprised mostly of farmers and residents who commuted to work in a large city nearby. Over half the students at Marshall are migrant or are students for whom English is a second language, and the high risk nature of the school population has had sign)ficant impact on school goals. Lea, the school's Title I teacher and the Round One participant from Marshall Elementary, had taught at Marshall for all but three of her twenty-five years teaching. Maya, an Hispanic woman and one of two Round Two respondents from Marshall, was a second grade teacher. For her entire, twenty-one year career, she had taught primary grades at Marshall. The other second round participant, Sonnv, an Hispanic and the only male in the study, worked as a school counselor in two schools and as the district's Title I parent liaison. He, also, had spent most of his career in the same district. With thirty-one years in teaching. he was one of the two most experienced participants.

Mary Ellen, Jo, and Anne were teachers at Thayer Elementary, located in a large metropolitan area and the most urban of the three sites. Thayer served some of the most disadvantaged students in the district and struggled with a high student turnover rate. Mary Ellen, a Title I teacher and a first round interviewee, had significant leadership experience outside her school district. with much of work in policy-setting at both state and national levels. Jo and Anne were interviewed in the second round of data collection. Like Mary Ellen, Jo had a long history of leadership involvement both within this district and in her former district in another state. She had multiple roles at Thayer as: classroom support teacher, mentoring and coaching colleagues; instructional resource teacher, providing materials and co-teaching units with classroom teachers, and gifted and talented teacher. She had considerable experience within her district as a consultant and staff developer and, at the time of the study, was deeply involved in the district's mentoring program for beginning teachers. Anne, a woman of Japanese descent, was her school's only kindergarten teacher. Her story was heavily influenced by her cultural background.

The last three participants, Madison, Ashley, and Fran, worked at Ransom Elementary, a bright, clean, recently remodeled school, located in a small city, in which the primary employers were two large computer firms, the city, small industries, and a meat processing plant. Ransom Elementary is located in a primarily low-income neighborhood with a sign)ficant number of single parent families, families in turmoil, and families with limited English proficiency. Staff at Ransom must deal with a constantly changing student population, resulting, in large part, from family instability. All teacher leaders selected for participation had worked at Ransom for five or fewer years, although all were extremely experienced teachers. Madison, the first round participant from Ransom, was one of two study participants working in more than one school as part of her regular job responsibilities. She was one of four district literacy consultants in the mornings and a school-based literacy teacher during her afternoons. Ashley and Fran were involved in the Round Two interviews. Ashley, whose Quaker background has had sign)ficant impact on her leadership style, was a fifth grade classroom teacher. As a classroom teacher, she has felt compelled to restrict her leadership roles to those which did not demand sign)ficant blocks of time away from her classroom. Fran was a resource teacher who was teaching reading to at-risk students in Ransom Elementary's literacy lab. Although it was her nature to get involved, she was cautious about over-committing to leadership roles.

Despite many commonalities, each participant s story was unique. By relating their experiences and discussing their perceptions, each teacher leader s observations enhanced the resulting findings. In the next section, I report the major findings as they contributed to the development of the categorization of key sources of support and barriers to teacher leadership.