© Please note: Shirley Freed owns the copyright for the dissertation below. You may download and print one copy for educational purposes only. These pages may not be duplicated, distributed, redistributed or republished in any manner without express permission from the author.

CHAPTER THREE

METHOD

Introduction

Since change (the learning of a new teaching method) is a process, not an event, and requires on-going observation to understand what actually takes place, the chosen research design is a focused ethnographic approach. The process is described in terms of the person(s) going through the change and the context of the change. The description of the context includes the cognitive states of the teachers, their working conditions, their level of utilization of a direct instruction approach to reading (ECRI), and details acquired through observations and interviews.
The research was accomplished in three phases:
Phase One: Observation and description of one teacher, using ECRI methods and her/his students a pilot study.
Phase Two: Observation and description of 15-20 teachers learning how to use ECRI as part of the course work in EDTE492 and EDTE436 the summer of 1990 at Midwest University.
Phase Three: Follow-up and description of utilization of the methods taught during the summer. The teachers were studied during the 1990-91 school year.
In studying the process of change as teachers experience learning an instructional strategy, the most appropriate method would be a qualitative methodology. As Bogdan and Biklen (1982) clarify, qualitative researchers are not concerned with products, they are concerned with processes and the meanings which people attach to their lives. The interest of this study is in the process of change as it is affected by specific contexts: e.g., high conceptual level or difficult working conditions.
Another reason for using a qualitative design is that it matches the process of implementation. Both require interpretation. Carson (1983) states that
for teachers, implementation is an interpretative act . . . [which] depends upon the teacher's stock of knowledge and beliefs about how children learn, what society wants, what the future will be like and the teacher's understanding of the organizational context of their work. . . . The interpretative act is the effort by the teacher to fuse the horizon of the curriculum plan as text, with the horizon of teaching as a lived experience. (p. 20)
Aoki (1983) suggests that implementation has been viewed primarily as an instrumental action and as such teachers are reduced to beings-as-things. In contrast, he believes implementation should be portrayed as "situational praxis." He uses "praxis" in the contemporary sense meaning reflection and action. In this framework "implementation is constituted by the inter-subjective actions of beings-as-humans, oriented towards interest in mutual understanding, and . . . seen as competance in communicative action and reflection" (p. 16). Going beyond the immediate level of interpretation, the teacher discovers his/her own reality. Thus, a qualitative design humanizes teachers and permits their subjective reality to become the focus. The analysis of collected data is also an interpretative activity, so in this study both the researcher and the teachers are involved in interpretation.
From the earlier chapters, it is clear that reformers, staff developers, and researchers of the change process are centering their attention more and more on teachers. As they do so, they need data which describe the teachers' world as they (teachers) view it. This can best be acquired through the methodologies of qualitative research participant observation and interviews. These techniques have the potential for providing answers to the questions posed by this study.

Features of Qualitative Research
Case Studies
One of the common forms of qualitative research is the case study. This form of inquiry provides special opportunities to understand an individual's reality. "Instead of explanations based upon statistical associations between operationally defined constructs, interpretations are drawn directly from actual circumstances, events, behaviors, and expressed sentiments as played out daily by people in the context of their work" (Sirotnik, 1989, p. 95). Some of the characteristics of case studies are given below.
Smith (in Merriam, 1988) suggests that a bounded system is necessary for a case-study approach. By this it is meant that "a specific phenomenon such as a program, an event, a person, a process, an institution, or a social group" (p. 9) is needed for the case study. Lincoln and Guba (1985) believe that "no inquiry, regardless of which paradigm may guide it, can be conducted in the absence of a focus . . . [and] such focusing establishes the boundaries for a study; it defines the terrain, as it were, that is to be considered the proper territory of the inquiry" (pp. 226, 227). Wilson (1979) uses the term particularistic to describe case studies. This suggests that the study focuses on a particular situation, event, program, or phenomenon. Shaw (1978) has said that case studies
concentrate attention on the way particular groups of people confront specific problems, taking a holistic view of the situation. They are problem centered, small scale, entrepreneurial endeavours; data is collected by being on the spot with expectations and directed vision, but there is a readiness to reconceptualize the problem as the data accumulates and to take account of the broad slice of social reality. (p. 2)

Another descriptor of case studies used by Stake (1981) is that they are heuristic. This means that the case study brings new light to the situation being studied. It may indicate new relationships, confirm known entities, and extend the known. Wilson (1979) also uses the term longitudinal as a characteristic of case studies. Studying a specific phenomenon over a period of time allows the researcher to study the interaction of variables.
These characteristics are evidence that a case-study approach to studying change is appropriate and gives results that will benefit the educational community. The bounded system studied is the process of change, the particular situation is the change of teachers learning a direct instruction reading method during the summer of 1990 and following. The study is longitudinal over a period of approximately a year, including the three-week training session. The results are heuristic, as expected, and certainly can sensitize educators to the process of change from a teacher's viewpoint.

The Context
Overview of Context
Guba (in Butler, 1984, p. 4) stated that "human behavior is rarely if ever context free," and it is because the context or situational elements need to be considered that a qualitative-naturalistic research design is used. In 1978, Hunt realized the importance of context in planning inservice activity. He said planning "must give first priority to the context the specific school, the specific concerns of the teachers, and the specific relations among the people engaged in the activity" (p. 240). He suggested that the theory-to-practice process tends to ignore the context while research-based programs are believed to be universally acceptable. Butt (1984) believes that "prior to any sort of inter-vention it would be imperative for the outsider to learn about the teachers [sic] own professional reality, style, beliefs as well as functional metaphors, images, and rules of teaching; to understand the personal practical and professional knowledge and context within which changes might be made, and successfully embedded" (p. 16). Mishler (1979) also addresses the issue of context and shows that the experimental approach is a "context-stripping" method which leaves gaps in one's understanding of human behavior. He proposes alternative approaches like qualitative research.

Defining the Context in This Study
Since researchers (the third section of chapter 2) have pointed out certain teacher characteristics which are ubiquitous to change, some of those characteristics are used to define the context by which an individual relates to the process of change. This study describes behavior in relationship to the various contexts in which it is known to affect change, e.g., working conditions and conceptual level. It also endeavors to expand the knowledge base of staff development by focusing on cognition and the way different cognitive patterns affect the learning of a direct instruction reading approach.
While it is known that teachers can change (implement new strategies) if they are self-actualizing and working in a democratic school climate (Hopkins, 1990), it is not known what the change process is like in those particular situations. If a teacher under difficult working conditions is able to implement, one needs to understand the process by which she/he was able to change.
Wilson (1979) states that learning from cases "involves getting a feel for how things `come down' in schools, for sensing what happens in improvement activities, and for generally extending one's experience, thus gaining both the power and the sensitivity to act" (p. 455). Learning a direct instruction method may require drastic changes on the part of a teacher. In that process of changing roles, beliefs, values, perceptions, preferences, and behaviors, one needs an understanding of how the context affects change.
Therefore, I collected data on Gregorc's mind styles, Piagetian operational level, Hunt's conceptual level, and Szabo's working conditions (these instruments are described and documented below). Since research has shown that conceptual level affects implementation (Showers, 1984), an instrument to indicate levels of this construct accepted in the field of educational research was used. Although Hopkins' (1990) study used Gregorc's Mind Styles, no conclusions were drawn concerning its relationship to implementation. I have chosen Gregorc's Mind Styles hypothesizing that it may provide explanations for observed behavior. For example, do teachers with a dominance in sequencing (Gregorc) find it easier to learn a direct instruction approach (with each specific step defined) than those who are dominant in random ordering? A similar line of reasoning was used in the choice of Piaget's Operational Levels. It seems possible that an individual at the level of `formal operations' according to Piaget may find it easier to implement an approach which has many interrelated components. Instruments have been chosen because of their research-backing, acceptance in the field of education, and usefulness to the study.
As this study proceeded, I looked for patterns within each specific context. For example, how do high operational level teachers relate to change in comparison with low operational level teachers? Discovery of the aspects of the change process which are unique to high conceptual level, etc., was anticipated.
In summary, rather than describing the general behavior of an individual who is seeking to learn a direct instruction reading method, I studied his/her behavior in reference to his/her particular context as defined by the instruments listed below.

Description of Instruments Used to Define Context
Gregorc Style Delineator
Developed by Anthony Gregorc, this style delineator is a self-analysis tool that assesses two types of mediation abilities: perception and ordering (Gregorc, 1982). Perception abilities emerge as two qualities: abstractness and concreteness, while ordering abilities emerge as sequence and randomness. By ordering 10 different sets of words from 4 most like you to 1 least like you, a profile emerges which shows four possible transaction ability channels: concrete/sequential (CS); abstract/sequential (AS); abstract/random (AR); and concrete/random (CR). Construct validity and reliability for the style delineator may be found in Gregorc (1984).
Every individual "will be strongly oriented toward one, two, or even three of the four style channels." A score above 26 indicates a dominance in that channel and signifies "that those mediation qualities are powerful means of transaction" for that individual (Gregorc, 1982, p. 14). Every individual has some preference in each of the four qualities: concreteness, abstraction, sequence, and randomness, but feels more comfortable when in circumstances where they are able to function in their dominant mode.

Paragraph Completion Method (PCM)
Conceptual Level (CL) is a developmental construct which views individuals on a continuum "of increasing conceptual complexity, self-responsibility, and independence" (Hunt, 1977-78, p. 78). On a scale of 0-3, Conceptual Level is measured by a semi-projective Paragraph Completion Method (PCM) (Hunt, Butler, Noy, & Rosser, 1978). Although Hunt would recommend standard ranges of scores with low defined as 0-1.1, moderate CL as 1.2-1.9, and high as 2.0+, many studies define CL in whatever way is expedient for their particular study (Miller, 1981). The measure consists of five 2-minute timed responses to stems such as "when I am not sure . . ." Using a 3-point scale, each response is rated and the CL is the mean of the highest three scores. For example, one response to the above stem may be "I just don't know what to do." This response is rated "0" because it indicates no tolerance for ambiguity or indecision. In contrast, another individual may respond, "I try to look at all alternatives [or] ask advice of knowledgeable persons. If still undecided, I'll go with my intuition, or past experience, or desired results, then make the necessary decision." This response is rated "3" because the emphasis is "upon utilizing one's own knowledge, finding information, and seeking as much advice as the individual feels is necessary to arrive at a satisfactory conclusion" (Hunt, Butler, Noy, and Rosser, 1978, pp. 26-28). Information concerning construct validity and reliability can be found in Hunt, et al. (1978).
The construct of Conceptual Level has been expanded to consider matching instruction with CL. The premise is that low CL persons (dependent on external standards and incapable of generating their own concepts) should profit more from a highly structured approach while high CL persons (capable of generating new concepts and holding internal standards) should either profit more from low structure, or be less affected by variations in structure (Hunt, 1975). In Miller's (1981) review of 29 studies of matching procedures, 19 showed main effects of CL; however, research designs were varied and few were in typical classroom settings. Differences in instruction are usually conceptualized by the degree of autonomy afforded the learner. At the opposite ends of a continuum, one would expect to find discovery student-centered approaches and didactic teacher-controlled approaches.
Showers (1984) found that high CL individuals transfer training more often than low CL; however, CL is not so significant when the teachers are coached. Showers found also that low CL-coached teachers surpassed high CL-uncoached teachers in implementation.

An Inventory of Piaget's
Developmental Tasks (IPDT)
An Inventory of Piaget's Developmental Tasks (IPDT) is an experimental inventory which was designed by Hans Furth, James Youniss, and Bruce Ross at Catholic University to translate some of Piaget's concrete and formal operational tasks into an objective, quick, standardized paper-pencil format (Furth, 1970). It has been used in studies of computer programmers (Cafolla, 1987-88), chemists (Milakofsky & Patterson, 1979), and teacher educators (Reyes, 1987; Reyes & Alter, 1990). Patterson and Milakofsky (1980) have reported on reliability and validity. The test has 18 subtests each with four items. Teachers show mastery of a subtest by scoring three or four of the items correctly. The subtests are further arranged under five problem areas: classification, conservation, imagery, proportional reasoning, and relations. Mastery of a problem area is indicated by scoring all subtests under the problem area correctly. Formal operational level is indicated by mastery of five problem areas. A transitional level is indicated by mastery of three or four problem areas and concrete operational level is indicated by mastery of zero to two problem areas.

Szabo's Typology of Working Conditions
Developed originally with new teachers in high-school settings, the framework provides a structure from which teachers may assess their working conditions. The typology considers two aspects: level of difficulty and level of support. In Szabo's (1989) study, teachers perceived a high level of support in their working situations when they were respected by the veteran teachers and administrators, the administration provided avenues for safe help in planning/materials, valued feedback was provided, and they participated in planning. High levels of difficulty were associated with uncertainty about employment status, number of preps or room changes, type of students, and adequacy of written information (textbooks, guides, etc.). Szabo found that a person in a situation of low support and low difficulty was "ignored" whereas a teacher in a highly difficult situation with low support was "tested." Those teachers with high support and low difficulty were "nurtured," while those with high support and high difficulty were "coached"
(Figure 1). In assessing the teachers' working conditions, I used a focused interview (Appendix A) and endeavored to define the situation according to the criterion used in Szabo's study.

 

Level of Support
Level
of
Difficulty		    Low  High
 Low  Ignored  Nurtured
High   Tested  Coached

Figure 1. A typology of working conditions. (Figure based on Szabo's 1989 study).

 

Other Instruments
I acquired other information in this study by using Joyce and McKibbin's Growth States (1982) and measuring Stages of Concern about the Innovation by Hall, George, and Rutherford (1986). These instruments helped broaden the picture of change.

Trustworthiness Issues
Comparison with Quantitative Research
First, all researchers are concerned with trustworthiness issues. They want to believe that there is "truth value" to their conclusions. Quantitative and qualitative researchers have different ways of convincing their public that what they have found can be trusted.
Lincoln and Guba (1985, p. 300) state that "the four terms `credibility,' `transferability,' `dependability,' and `confirmability' are, then, the naturalist's equivalents for the conventional terms, `internal validity,' `external validity,' `reliability,' and `objectivity.'" To be credible, an inquiry needs to have prolonged engagement, persistent observation, and triangulation (explained in detail below). The findings need to be approved by the individuals being studied.
Concerning transferability, the naturalist cannot set up relationships that can be generalized across different people, settings, or time as the quantitative researcher attempts to do. The most he/she can do is provide an adequate data base of the widest possible range of information. It is the responsibility of those who read the "thick description" to decide whether the data are transferable applicable to their situation.
Dependability is similar to reliability. How consistent are the findings? Lincoln and Guba (1985, pp. 316-317) suggest that "since there can be no validity without reliability (and thus no credibility without dependability), a demonstration of the former is sufficient to establish the latter." However, they recognize this is a weak argument and would encourage naturalists to increase dependability by a process of replication which involves more than one researcher or to have an auditor check the findings for accuracy in relationship to the original data. This would involve an examination of the process of the inquiry as well as the findings. Confirmability is established when the auditor finds the product and the data to be internally coherent. Guba (1981) suggests that triangulation and the keeping of a reflective journal also add to confirmability. The journal is a means whereby the auditor is able to track decisions made by the researcher as he/she progressed with his/her study.
A naturalistic study has its own unique ways of establishing trust-worthiness. The ones which I used in this study are described in the following section.

Specific Strategies for Increasing Trustworthiness
Any research design must have provisions built into the design which make the research credible. In the final analysis, one first needs to have confidence that the findings are of value and then one needs to be able to convince the readers of this. The following strategies add credence to my study.
1. Triangulation. Denzin (1970) suggests four different types of triangulation: methods, data sources, investigators, and theories. The different methods I used are defined in Table 1
(p. 70). Each of the main topics: working conditions, cognitive process, and utilization are assessed in at least two different ways. If the data did not agree, more investigation was necessary. More than one source of data were used to verify, cross-check, and provide negative cases. While I intended to be the primary investigator, other individuals helped in data collection and I compared the findings of different investigators. As I proceeded through the process of data collection and data analysis, I was aware of the different models of change and tried to ascertain how my data fit the various frameworks discussed in Chapter 2.
2. Member Checks. The member-check concept is explained by Guba and Lincoln (1981, p. 316). It is a way of endeavoring to verify data collected and interpretations by having the original interviewees read the analysis. It does not certify that all errors will be discovered, but it does help verify the data. By the use of pseudonyms, anonymity was guaranteed and sensitive material could be used.
3. Repeated Observations. Over a long period of time, observations are viewed by Merriam (1988, p. 169) as another way to increase the validity-credibility of the findings. I remained in contact with the teachers in this study for approximately one year.
4. Audit Trail. Guba and Lincoln (1981, p. 122) explain the audit-trail process whereby an independent investigator could come in and reconstruct the process whereby categories were established and decisions made. In the original field notes, I continually recorded my own reactions and reflections on what I observed. I also included decisions as the plan emerged. Establishing an audit trail increased the dependability or confirmability of the study.

The Data
Techniques for Data Collection
The primary techniques which I used were:
1. The Ethnographic Interview. This interview sought to understand the process by which teachers are able to learn a direct instruction method.
2. Participant Observation. This step required the researcher to become involved as a participant observing the behavior of the teachers who sought to learn the ECRI method under Dr. Eugene Haskell's instruction the summer of 1990 when they enrolled in EDTE492 Teaching Developmental Reading and EDTE436 Elementary Language Arts Methods at Midwest University.
3. Measurements. Measures of mind styles, Piagetian operational levels, Hunt's conceptual levels, and working conditions were used to assist in defining the context of the study.
4. Questionnaires. During the three-week training it was imperative that data be collected from all teachers, since I could not predict who would implement ECRI. Because twenty-five teachers were initially involved, I used a series of four questionnaires (Appendix A) to gather data during the training.
Data were collected under the following main topics: the working conditions of the teacher, the cognitive processing of the teacher, and the utilization of the ECRI program.
1. In evaluating the working conditions of the teacher a focused interview was used. Szabo's typology (1989) provided the framework for a description of working conditions and observations validated information acquired in an interview.
2. In describing cognitive processing, measures of mind style (Gregorc, 1982), operational level (Milakofsky & Patterson, 1979), and conceptual flexibility (Hunt, 1977-78) were used as well as participant observation to describe behaviors related to cognitive processing.
3. The individual teacher's utilization of ECRI was assessed using all three techniques: the Levels of Use interview (Loucks, Newlove & Hall, 1975), a checklist of ECRI critical components, and participant observation to validate data from the interview and questionnaire.
These data-collecting techniques are summarized in Table 1:

TABLE 1

DATA-COLLECTING TECHNIQUES

  				     Interview	Inventory	Observe
Working Conditions	    *		    -		    *
Cognitive Process	    *	    	*		    *
Utilization		        *		    *		    *

The Data File
The data file was formed by a compilation of all pertinent information concerning each individual. Results from the inventories, answers to questionnaires, transcriptions of taped interviews, summaries of telephone conversations, notes from observations, and reflections of the researcher are organized into a chronological record of events for each teacher. Also included in the data file are various documents acquired from Dr. E. Haskell, the professor in charge of the training. These include cards that each teacher filled in on the first day of class, and answers to test items on two different examinations. Each individual's profile also has examples of the materials needed to teach using ECRI. This provided a picture of the way the teacher has arranged the classroom to facilitate the use of ECRI.
Organized in three volumes, the data collected during the summer are incorporated in the first volume while the second volume contains information related primarily to the four cases. In total, over 400 pages of transcribed interviews, questionnaires, and observations make up this portion of the data file. Also, each teacher kept a journal, the contents of which were coded in the original. Finally, in the third volume, approximately 100 pages of documents are used as supporting data in the case studies.
This format of individual characterizations facilitated the cross-comparison of the different contexts. For example, the comparison of findings of all high CL teachers validated the observations with respect to one high CL teacher. These were then compared with low CL teachers. In the same way, cross-comparisons were made with high operational level, low operational level, and random and sequential mind styles.

Selection of Cases
In the selection of individuals for the case studies I employed a system of purposive sampling (Patton, 1987). It was important to have a wide variety of characteristics represented in the case studies; therefore, experience, gender, race, conceptual level, formal operations, and mind styles were considered. A number of teachers were immediately deleted because they were undergraduate education students and did not intend to implement ECRI in the fall. Several more had never taught reading and had no intention of implementing.
From the outset, I was concerned that my study not be an "appraisal of non-events" (Charters & Jones, 1973). Since I was studying implementation, I wanted to collect data on teachers who were at least attempting the use of ECRI. Table 2 summarizes the information which I considered in selecting the case-study informants.
Of the 25 individuals who registered for the course, three dropped out after the first week and two were so irregular in attendance that I could not gather the necessary data from them. From the remaining 20, I chose four teachers: Becky, Deana, Debbie, and Tammy for my case studies. Although four men took the training, only one, Ken, tried to implement ECRI; however, he indicated he was overwhelmed with the amount of preparation involved and discontinued after several weeks.

Data Collection for Case Studies
In addition to the data that were collected on all teachers in the training session, in-depth interviews and observations were conducted on the four teachers chosen for the case studies. While an effort was made to establish an interview and observation schedule, it soon became clear that taking time to create an atmosphere of trust was more important than following a pre-arranged schedule. This meant, for example, that with Becky I was unable to gain entrance into her classroom for observation until November. She had not implemented ECRI at the beginning of the school year, and it took a while for her to feel secure enough to have me in the room. With Debbie, I visited her classroom in September and realized it was too soon since she had not really established routines, so I adjusted my visits according to what appeared most reasonable.

TABLE 2

TABLE FOR PURPOSIVE SAMPLING

        Teacher	    I	Ex	Ra	G	CL	OL	MS
   * 	Alerie	    I	17	C	F	L	L	CS-AS
	    Barbara	    X	 S	Sp	F	-	M	CS-AS
 o *	Becky	    I	11	C	F	H	L	AS-CS
	    Bonnie	    X	 S	C	F	-	L	CS
	    Carmen	    X	 S	C	F	-	M	AR-CR
	    Clarence	X	 4	C	M	-	H	CS
 o *	Deana	    I	 4	C	F	H	H	AR-CR
 o *	Debbie	    I	 1	B	F	H	M	AR-CR
	    Harriet	    I	16	B	F	L	M	AR
   *	Jackie	    I	 4	C	F	H	M	CS-CR
   	    Janey	    X	 S	C	F	-	H	CR-CS
	    Joan	    I	10	C	F	H	H	AR
   *	Joyce	    I	 3	B	F	M	L	CR-AR
	    Ken	        X	 6	C	M	L	H	CS-AS
   *	Lisa	    I	 3	C	F	M	H	AR-CR
	    Mary Lane	X	 7	B	F	-	M	CS-AS
	    Randy	    X	 S	C	M	-	H	AR-CS
	    Ray	        X	 5	C	M	-	H	AS-CS
   *	Sophia	    I	 ½	C	F	L	H	AR
 o *	Tammy	    I	 1	C	F	L	H	CS-AS

* Reasonable Possibility had implemented and was still using in October
o Final Choice for Case Studies
I Tried to Implement
Ex Experience in Years
Ra Race
G Gender
CL Conceptual Level
OL Operational Level
MS Mind Style

S = Student Sp = Spanish H = High
C = Caucasian M = Medium
B = Black L = Low

An outline of data collection for the case studies is given in Table 3. The observations represent from 3-5 hours in the classroom. These were taped sessions, and the tapes were used in helping to create an understanding of the progress made by the teacher in using ECRI. Each interview was from 1-3 hours and was transcribed to make up part of the data file. Interviews were primarily unstructured, however, a guide was used for the last two interviews. These are found in Appendix A with the questionnaires used during the training. Telephone conversations were summarized and were an important source of data since several of the teachers were 3-7 hours' drive from Midwest University. The case-study data were collected over ten months, beginning in June 1990 and ending in March 1991.

TABLE 3

SCHEDULE OF DATA COLLECTION

Name		June-July		August-December	  January-March


Tammy		Q Q Q Q I O		T T O O I I		  T T I O 
Debbie	    Q Q Q Q I O		T T I O I O		  T T I O
Deana		Q Q Q Q I O 	T I O I O O I I	  T T I O
Becky		Q Q Q Q I O		T T T I O O I I	  T T I O


Q = Questionnaire		I = Interview
O = Observation		T = Telephone Conversation

Techniques for Analysis of Data
The data were interpreted analytically using coding, categorizing, and discovering cultural themes. Cross-case analysis was also done resulting in a multiple-case study. Using a procedure explained by Yin (1984), I chose specific cases according to the particular context which I wanted to focus on and used other cases as I endeavored to search for patterns (described in previous section). Since the primary purpose of this study was the description of a process, there is limited statistical analysis.
In analyzing the data, a combination of techniques were used. First, segments of the data file were coded using a process of repeated readings and comparison with previously coded items. An effort was made to develop definitions for each code. It was then noted that these segments fell into several large categories (Strauss & Corbin, 1990) representing themes which were verified deductively by going back to the data to ascertain whether statements fit the themes.
Using techniques suggested by van Manen (1990), deeper meanings emerged. I used his "wholistic" and "selective" approaches to make summary statements of large amounts of data. For example, using the "wholistic" method I read through one person's data file and tried to capture the essence of that person's lived experience in a sentence or two. Using the "selective" approach I identified what the person was really saying. For example, Alerie said:

With so much repetition, I have wondered if the students are able to react as robots and not really think about what they are doing.
This segment had previously been coded "values" and in brackets underneath, I wrote, "thinking is important to this teacher."
Miles and Huberman (1984) suggest that when data are displayed in some standardized format and are assembled coherently in one place that "this is the first deep dive into cross-set analysis." In Appendix B are examples of the clustering which such displays permit. Having identified the occurrence of certain coded segments for each person, I was able to unite that person's `profile' with others to get a group picture. With this clustering I was able to search for trends among sequential individuals in contrast with random individuals, high conceptual level individuals in contrast with low conceptual level individuals, and high operational level individuals in contrast with low operational level individuals.
The teachers which I used in these clusters are those who had implemented ECRI and were still using it in October 1990 (Appendix B). Although nine teachers were available for these comparisons, not all nine are on each table. Because I compared opposite extremes, the middle teachers are not in the comparisons. The coded segments were taken only from the four questionnaires given during training. They did not include observation during training nor do they include any data collection following the training.

Role of the Researcher
In qualitative research "the researcher is the primary instrument for data collection and analysis" (Merriam, 1988, p. 52). This was a role which I found both fascinating and frustrating.
As a participant observer, I registered for one of the 4-hour courses which made up the three-week training session: EDTE492 Teaching Developmental Reading. Therefore, I came to the training with a "dual purpose": to engage in all the activities as a regular student and to "observe the activities, people, and physical aspects of the situation" (Spradley, 1980, p. 54). This dual role was often complicated by another role that of trainer. From the beginning some teachers thought I was one of the trainers:
I wish to learn all the methods you present. I wish to make the best grade possible because I need this course for certification. (V. 1, p. 136)

From time to time I did actively play the role of a trainer, checking off directives and doing evaluations of the skills section. During this time, I was able to document some physical manifestations which may not have been part of my observations had I not played that role. However, at one point, I chose to no longer act as a trainer because I felt it was putting distance between me and those I had come to observe.
Maintaining the "insider/outsider" balance was a juggling act for me through the entire three weeks. Spradley (1980) suggests that "the participant observer . . . will experience being both an insider and outsider simultaneously" (p. 57). As an insider I was privy to criticisms and jokes about the trainers; however, as an outsider, the trainers shared valuable information with me. I remember one day when one of the trainers said to me, "We just sent Clarissa (the other trainer) into that room to observe, we could not stand it any longer the teachers are going so slow!"
As an insider I experienced many of the same stresses the other teachers had. I, too, had nightmares and found myself relying on Diovol to settle my stomach. The fact that I am a certified teacher was an asset to my role as participant-observer. I was able to empathize with the teachers as they sorted through the various aspects of the ECRI program, endeavoring to make sense of it. I could respond to their concerns about how to collect materials, how to gain approval from principals and superintendents, how to grade a mastery approach, how to group in a multi-grade situation, etc. Guba and Lincoln (1981) view empathy as "the characteristic most applicable in any naturalistic study" (p. 140).
I did not address the problem of researcher-effect until after the three-week training. When teachers began to implement ECRI in September and October, I regularly kept in touch with them by telephone:
I'm so glad you called, you're just what I needed. I just do not know how to organize my classes. I have four levels. (V. 2, p. 129)

I called Shirley today, wanted to talk about grouping. (V. 2, p. 132)

I'm so glad you phoned because I have this problem. I do not know how to do the unit test. (V. 2, p. 141)

I knew I could not remain neutral when teachers were calling for assistance. I recognized vividly that my influence would bias the study, that teachers would, in fact, implement differently had I not been a participant in their process of change. As one verified:
Your visit, Shirley, renewed my enthusiasm for ECRI. I have now had two conferences with students a small start, but at least a start. (V. 2, p. 45)

I decided to handle this seeming conflict of interest by carefully documenting all my conversations with the teachers. As Metz (1981) points out, these possibilities for variability can be corrected by "giving full and fair accounting of the instrument and its use so that readers may perceive and allow for the various biases which will inevitably creep into both the field work and the later interpretation of it" (p. 1).
I realize I have interpreted this experience through my own world view, values, and perspective, and that this is in agreement with "one of the philosophical assumptions underlying this type of research . . . that reality is not an objective reality; rather, there are multiple interpretations of reality" (Merriam, 1988, p. 39).
Finally, the experience of "being the instrument" in this study was a most satisfying one. Teachers assured me that they "thought about their profession in a new way." I discovered implicit meanings and have been sensitized to the lived experience of myself and others as we implement new strategies in our classrooms.

The Strategy
The teaching strategy studied was the Ethna Reid method taught at the Exemplary Center for Reading Instruction (ECRI) in Salt Lake City, Utah. The ECRI method was started by Ethna Reid in the 1960s. Working originally in a clinical setting, Reid later analyzed the critical behaviors of effective teachers. These were found to include abilities to elicit correct responses from non-responding pupils, to establish high mastery levels of response with performance and rate as criteria, to correlate language-arts activities to increase responses and save time, to utilize effective management and monitoring systems, and to diagnose and prescribe instantly when errors or no responses occur (Educational Programs that Work, 1980). Since that time, Reid (1986) confirms, "ECRI has functioned as a consulting firm and a developer-demonstrator project designed to instruct teachers in the effective and efficient use of classroom time. The center is built upon the belief that most, if not all, students can learn to read and write successfully if they are properly taught to do so" (p. 510). The program is planned in such a way that it can be used with any basal program, but the basal program needs reformatting for use. Students learn to "read, write, and spell each word, understand its meaning, and use it in sentences and stories (and) as they encounter it in their reading exercises." What are the unique characteristics of an ECRI classroom?
1. Teachers elicit overt, accurate, and rapid responses from students.
2. Teachers ignore negative behavior but strongly reinforce positive behavior with a point system or some other measure.
3. Students practice until they have mastery of spelling lists.
4. Students read stories three times.
5. Teachers conduct individual conferences with some students daily.
6. Discussion groups result in high-level questioning and responding.
7. Students are encouraged to use creative writing to use the vocabulary words they have learned.
The ECRI program is recommended by Rosenshine (1983), Christenson, Thurlow, and Ysseldyke (1987), and the U.S. Department of Education (1982) as an alternative instructional strategy for handicapped, minority, and normal children. Two doctoral dissertations (Hughes, 1983; Miller, 1983) showed no apparent advantages of using the ECRI method. In Hughes' (1983) study, three ECRI teachers and three basal teachers were used, and in the Miller (1983) study, 68 subjects were studied. Reid in an interview with Brandt (1990) contends that "in recent years we've had teachers in about 800 new schools per year adopt the program. Our staff and certified trainers offer inservice training to 5,000 or 6,000 teachers a year" (p. 78).
In light of the research on direct instruction (see chapter 2), because ECRI is listed in Educational Programs that Work (1987), knowing that ECRI has a high proportion of the characteristics suggested in Becoming a Nation of Readers, and realizing that ECRI has many of the characteristics of teaching effectiveness research, the study is based on ECRI for the following reasons:
1. It has reported phenomenal growth records. Brandt (1990) related the progress of one girl who moved from first grade to fifth grade in silent reading in one month and Liske (1990) in Sandy Lake Elementary School told of two children who progressed from non-reading to third grade in seven months.
2. It has been reported (Reid, 1986) to be successful regardless of sex or race. This is significant since Campbell (1985) and Erion (1983) have shown that girls have an academic advantage over boys in the early years and numerous studies (Bowman, 1988; Christensen, Gerber, & Everhart, 1986; Gregory, 1986; Magliocca & Rinaldi, 1982; Smith, 1983; Tucker, 1980; Wright & Santa Cruz, 1983) have found disproportionately high numbers of minority children identified as disabled.
3. It is a `teacher control' method. This is defined by Hahn (1988) to include initial skills instruction (to improve reading comprehension) where the teacher explains, models, questions, and corrects. Many studies (Darch & Kameenui, 1987; Hall, 1988; Judy et al., 1988; Knight, Waxman, & Padron, 1989; Pearson & Dole 1988) have found direct instruction more beneficial than other strategies for skill-oriented subjects. Hahn (1988) has identified the relationship between teacher control and direct instruction. He believes that direct instruction implies the idea that learning comes through some social interaction (teacher) before it becomes internalized or student controlled. Direct instruction is most useful in sequential learning and where lower level thinking skills are used. Also, teachers seem to be more comfortable in a controlled situation. Garner (in Hahn, 1988) found only three in eight teachers were able to release responsibility for learning to the students. It seems reasonable to use a method which teachers may more `naturally' accept.

Summary
The method used in this study is a qualitative approach which allowed for sustained observations over approximately a year. The context of the change was established by various inventories. Triangulation and member checks added credibility to this study. The end result is the description of the process of learning a direct instruction reading approach by several different individuals.