Field Research and Grounded Theory
Gordon G. Darkenwald
This chapter deals with an approach to social research that is little known outside sociology and that departs sharply from the logicodeductive paradigm of experimental science long dominant in educational inquiry. Not only is grounded theory unconventional and unfamiliar, but its methods have yet to be dearly codified and continue to be subject to varying interpretations. Despite these difficulties, grounded theory offers the adult education researcher a new and promising alternative to traditional ways of conceptualizing and conducting empirical studies.
The term grounded theory gained currency with the publication of Glaser and Strauss's seminal book, The Discovery of Grounded Theory (1967), which sets forth the scientific rationale and the data-collection and analysis techniques for this distinctive
style of inquiry. Essentially, grounded theory is an inductive approach to research that focuses on social interaction and relies heavily on data from interviews and observations to build theory grounded in the data rather: than to test theory or simply describe empirical phenomena. It is closely related to the sociological and anthropological fieldwork traditions exemplified by such familiar works as Street Corner Society (Whyte, 1941) and Tally's Corner (Liebow, 1967). Perhaps the major distinction between grounded theory and traditional field work is that grounded theory is less concerned with detailed description and holistic interpretation than with generalized explanations of the social phenomena under study. To facilitate the development of theoretical generalizations, grounded-theory researchers rely heavily on comparative analysis, whereas many social scientists who use field methods confine their research to the intensive study of a single group, tribe, organization, or other social collectivity.
This chapter is addressed primarily to researchers who may be interested in, but unfamiliar with, the grounded-theory approach. Consequently, I have tried to keep the discussion basic and practical. The following pages provide a broad view of the nature of grounded theory and its uses, describe the elements of data collection and analysis, and give examples of the application of grounded-theory methods to research in adult education. Before discussing grounded theory in more detail, however, I must note a few caveats.
First of all, grounded-theory research is probably more difficult than the typical descriptive or experimental study. Researchers need to master its logic and techniques, just as they need to master ,the intricacies of experimental design, measurement, and statistical analysis to undertake more traditional investigations. But there are special problems with grounded theory. Most vexing is the lack of easily understood, codified rules for the collection and analysis of qualitative data and the construction of theory. As in anthropological or historical research, there is more than a little art involved in doing a grounded-theory study. The interested researcher, furthermore, is not likely to find training opportunities or a course in grounded-theory methods, although most universities do offer courses
in fieldwork techniques that would be very helpful. For the most part, however, researchers are left to their own devices. Consequently, a good deal of independent reading, as well as small-scale trial efforts in the field, are essential before a major grounded-theory investigation is launched. Another practical consideration is the amount of effort normally required. Much time must be spent in the field collecting data, which must be transcribed and painstakingly analyzed; and writing reports of grounded-theory research tends to be more exacting than reporting the results of conventional research. The caveats aside, grounded-theory research can be rewarding both personally and professionally. Getting out of the library or computer center and into the world of educational practice can be an exhilarating and enlightening experience.
Nature of Grounded Theory
Much important material, particularly concerning the scientific logic and methodology of grounded theory, is necessarily omitted from this chapter. As I said, the basic reference is Glaser and Strauss (1967); the first five chapters bear rereading several times. Mezirow (1971) has written a provocative article criticizing the assumptions of traditional social science research and outlining the potential of grounded-theory studies for upgrading professional practice in adult education. Grounded theory is usually identified with the sociological tradition of symbolic interactionism that has its epistemological roots in the work of G. H. Mead and John Dewey. The standard work on symbolic interactionism is by Blumer (1969). A critique of symbolic interactionism and grounded theory has been published in the American Sociological Review (Huber, 1973). For a comprehensive, step-by-step treatment of fieldwork techniques and the analysis of qualitative data, the researcher should consult Schatzman and Strauss (1973). This work is invaluable for any researcher preparing to undertake a grounded-theory study. Several methodology texts include material on fieldwork methods, notably Adams and Preiss (1960), Filstead (1970), Hammond (1964), and McCall and Simmons (1969). An orthodox example of a grounded-theory study is Glaser and Strauss's Awareness of Dying
(1965). For examples of doctoral dissertations in adult education that have used grounded-theory techniques, see Beder (1972), Israeli (1973), and MacNeil (1970).
A typical definition of theory is given by Kerlinger: "A theory is a set of interrelated constructs (concepts), definitions, and propositions that present a systematic view of phenomena by specifying relations among variables, with the purpose of explaining and predicting the phenomena" (1973, p. 9). Glaser and Strauss accept this definition, but go beyond it in their contention that good theory should not only explain and predict but also be useful. In their view, the functions of theory are these: "( 1) to enable prediction and explanation of behavior; (2) to be useful in theoretical advance; (3) to be useful in practical applications--prediction and explanation should be able to give the practitioner understanding and some control of situations; (4) to provide a perspective on behavior--a stance to be taken toward data; and (5) to guide and provide a style for research on particular areas of behavior" (1967, p.3).
Assuming that theory should provide understanding and some control of real-life situations, what is the best way to develop it? Glaser and Strauss argue persuasively that theory should be inductively generated through the systematic analysis of empirical data. Furthermore, in their view, the key to successful theory generation is the use of the general comparative method. That is, in order to discover the basic conceptual elements of theory, one must systematically compare similar, and sometimes dissimilar, events or situations. Only in this way can theory be progressively built up so that it is generally applicable to the social behavior under study. Much theory in the social sciences, in contrast, is based not on careful analysis of empirical data but rather on speculation and logical deduction from sometimes dubious a priori assumptions. Such theory is often so abstract that its relationship to the real world is barely discernible, and putting concepts into operation to test the theory is virtually impossible. Theory that fits the world that human beings inhabit and that has endured (for example, Weber on bureaucracy, Durkheim on suicide, and Piaget on child development) has invariably been grounded in empirical inquiry.
Data for the generation of grounded theory can come from any source, including documents and records and previous research. In most cases, however, the bulk of the data is obtained through participant observation, a fieldwork strategy in which the researcher gets to know the situation and the people involved at first hand and collects data largely through careful observation of behavior and informal interviewing.
The conceptual elements (constructs) of grounded theory are referred to as categories and properties. A category is a basic theoretical concept that enables the researcher to explain and predict behavior. A property is a conceptual element of a category that serves to define or elaborate the meaning of the category. Thus, categories and their properties are closely related. Grounded theory also consists of propositions or hypotheses that specify the relationships among categories. A fully developed set of categories, definitions, and propositions, integrated in an analytical scheme, comprises the core of the grounded theory. These basic elements of theory are discussed and illustrated in greater detail in subsequent pages. But it might be noted here that these fundamental elements of grounded theory are the same as those of any other kind of theory. What is different is the way in which grounded theory is generated.
Grounded theory can be either formal or substantive. Substantive theory deals with a particular limited domain of inquiry, such as preschool programs, emergency-room care, or university extension services. A substantive theory is close to the real-world situation. A formal theory, in contrast, deals with a general domain of social science, such as socialization or formal organization, and is necessarily more general and conceptually abstract. The construction of formal theory is the proper concern of academic social scientists, whereas the construction of substantive theory is, or should be, a concern of researchers in applied professional fields such as adult education.
Grounded theory is seldom presented as a tightly knit set of interrelated categories, definitions, and propositions. Rather, the theory is discursively developed in narrative form as categories, and their relationships are defined, elaborated, and illustrated by the data (for example, incidents from field reports) used to generate
them. Thus, the most useful and natural form for the presentation of a grounded-theory analysis is a running discussion that permits full elaboration of the situation or problem under study.
It may be useful at this point to highlight briefly some similarities and differences between grounded theory and other empirical research methods and styles. A fundamental characteristic of grounded-theory research is that it seeks explanations. Experimental or quasi-experimental research, in contrast, is concerned mainly with verifying hypotheses--with testing rather than discovering theory. Of course, grounded-theory researchers are also concerned with verification in that they require evidence to establish the existence of their categories and the validity of their propositions. But in grounded-theory studies, verification is subordinated to discovery, while the reverse is true in experimental investigations. Surveys, though sometimes used in quasi-experimental designs, are often employed for descriptive purposes. Statistical sampling procedures are used in order to ascertain within a specified margin of error the distribution of variables in a population. Grounded-theory researchers are also interested in description, particularly of social processes not amenable to measurement by survey instruments. But description, like verification, is not the primary objective of grounded-theory investigation. Quantitative exploratory research, which often employs powerful statistical techniques such as multiple regression, resembles grounded theory to the extent that it focuses on explanation rather than description or verification. Research of this kind, however, is often problem-oriented and atheoretical. None of the research approaches mentioned above has as its major concern the systematic exploration and illumination of social interaction in real-life settings.
Case studies, including ethnographies, do, however, resemble grounded theory when the researcher's main intent is to develop an analytical description of human behavior in a naturalistic setting. As noted previously, the grounded theorist uses the general comparative method to build substantive theory that has general applicability to the particular type of social process or collectivity under investigation. In contrast, traditional fieldwork tends to focus on the detailed, holistic analysis of a single case. For generating substantive theory, comparative analysis has obvious advantages over
single case studies because emergent categories
and propositions can be elaborated and checked out by collecting data from
a variety of comparison groups. This is not to say that when the grounded-theory
style of research (that is, one characterized by flexibility, an emphasis
on discovery, and an analytical stance toward data) is deftly employed,
case studies cannot yield valuable results. An outstanding example is the
The purpose of developing a substantive theory is to shed light on some aspect of human interaction. Consequently, the grounded-theory researcher is interested in studying a particular type of social process or social collectivity (group, organization unit). Glaser and Strauss (1965), for example, studied the hospital care of dying patients. In the field of adult education, there are numerous topics well suited to grounded-theory analysis, such as literacy education in developing countries, program development in university extension, rural community development, and continuing professional education, to mention only a few. It may sound strange to speak of a "theory" of literacy education. Perhaps theory is too grandiose a term to use for such an undertaking. The intent, nonetheless, is not merely to describe but also to explain and permit prediction of human behavior in a carefully delimited context.
Ultimately, the use of grounded theory in applied fields such as adult education is to improve professional practice through gaining a better understanding of it. It seems self-evident that little improvement can be expected without further systematic knowledge concerning what is actually going on in adult education programs. If the subject matter of the field is the process of adult education, then the actual behavior of students, teachers, and administrators and their interpretations of their experience are of central importance for developing theory and upgrading practice.
Iannaccone has put the problem succinctly in an article deploring the present state of research in educational administration: "Despite our achievements to date, the black box of research in
educational administration is the school and its work, including that of the administrators. We have tested more hypotheses than we have derived from theory. We have related inputs to outputs. We know more about pupil, teacher, and administrator characteristics and about achievement, responses and community reactions than we know about what people do-how they move and live in schools. Given the present state of the art, we know more about the former classes of variables than we can fruitfully use. Until we understand the school as a world of work and why it is what it is, the development of relevant theory in educational administration will continue to falter" (1973, pp. 62-63).
Data Collection and Analysis
Doing any kind of research well requires experience, and this truism applies with particular force to grounded theory. Although grounded theory has its methods, they are not highly codified, in large part because discovery can never be a mechanical process. The researcher who approaches a grounded-theory study without an appetite for exploration and a tolerance for ambiguity is doomed to frustration from the start.
The techniques of grounded-theory data collection and analysis are described in detail in the first chapters of Glaser and Strauss (1967) and also in somewhat more basic form in Schatzman and Strauss (1973). Much of the .material in the former book is likely to be incomprehensible unless one has actually made some attempt to collect and code data. The following pages touch on some of the basic elements of doing grounded-theory research, drawing mostly from Glaser and Strauss, my own experience, and that of my students.
Although it is possible to collect data and later undertake
a grounded-theory analysis, it is far more desirable, for reasons that will become evident later, to collect and analyze data simultaneously from the beginning. This involves "theoretical sampling" and continuous coding and analysis. Theoretical sampling is the "process of data collecting for generating theory whereby the analyst jointly collects, codes, and analyzes his data and decides what data to collect next and where to find them, in order to develop his theory
as it emerges" (Glaser and Strauss, 1967, p. 45) . A statistical sample is drawn in advance of research, but in theoretical sampling the emerging theory dictates where the investigator will go next to collect needed data. For this reason, grounded-theory researchers are never sure at the beginning of a study how many groups or situations they will need to compare. Of course, researchers must start somewhere. Usually, they start by having open-ended interviews with key actors in the situation or by observing some activity. Then comes the difficult task of analyzing the initial data in the research for tentative categories. A category, to reiterate, is a conceptual element in a theory; that is, an analytical concept such as "marginality" that is important in understanding the phenomenon under study.
Normally, investigators begin the search for categories by looking in the data for strategic commonalities and differences. For example, in a study of urban adult basic education (ABE) programs Mezirow, Darkenwald, and Knox (1975) sifted through numerous field reports from participant observers in ABE classrooms. After much tedious work, a number of commonalities and differences in classroom interactions were identified. Teachers' behavior in response to students' failure struck the analysts as surprisingly uniform and at variance with common practice in traditional classrooms. Teachers almost never confronted students with their failure by saying, for example, "No, that's not correct. Does anyone else know the answer?" Instead, the teachers made every effort, in a variety of ingenious ways, to mitigate the students' sense of failure. This category was labeled "failure reduction," and its properties were elaborated. The properties of failure reduction are largely the ways teachers attempt to reduce failure, such as by breaking a problem down into simple, step-by-step components ("working through"), by prompting, and by quickly substituting a new question or assignment.
It takes only a little reflection to see that a category such as failure reduction probably would never be discovered by observing the interaction of students and teachers in only one or two classrooms. Even if an astute analyst identified the category, it would be impossible to be sure of its validity and to fully develop its properties and relationships to other key variables without the aid of theoretical
sampling--that is, without looking for additional incidents of failure-reduction behavior in other classrooms to confirm its existence and elaborate its dimensions.
As the analysis progressed, an explanation of failure-reduction behavior began to take form. The numbers game has a pervasive influence on ABE, indeed on most of adult education. Since budgets are largely predicated on the number of students enrolled, it is vital to the organization to recruit and retain as many clients as possible. Teachers are aware of this situation, and they know that they are evaluated in part on how well they hold their students. If attendance falls below a certain level, the teacher may well be out of a job. Consequently, aware that students fear failure, teachers do everything in their power to relieve this threat in the hope of preventing dropouts.
It is an easy step from the elaboration of the category of failure reduction to the formulation of theoretical propositions. For example, the following hypothesis follows logically: Teachers exhibit greater failure-reduction behavior when the organizational need to increase or maintain student enrollment is high. Certain practical implications of the foregoing analysis (necessarily oversimplified here) are also easy to deduce. If one assumes that the sensitivity to students' needs which is characteristic of failure-reduction behavior is desirable, it follows that eliminating the enrollment economy that typifies ABE programs will have certain undesirable consequences. That is, if ABE programs were subsidized in the same manner as elementary schools, it is reasonable to predict that ABE teachers, secure in their jobs, would begin to behave more like traditional schoolteachers and that students would pay the price.
Identifying categories is one of the most difficult tasks in grounded-theory research. Theoretical sensitivity, as Glaser and Strauss (1967, p. 46) point out, is essential, and it is largely acquired by experience. But qualitative analysis is not an esoteric art. The basic idea is to search for commonalities and differences in the data that seem important in understanding what is going on in the situation at hand. Often, the first categories uncovered are too vague or descriptive to provide much explanatory power. Concepts such as trust, authority, or cooperation may suggest themselves; but obviously they are much too general for illuminating social be-
havior in a particular, substantive context. Analytically potent categories are usually abstracted from the content of the research situation. Failure reduction is an example. In some cases, existing concepts are utilized, provided they fit the data. In the ABE study (Mezirow, Darkenwald, and Knox, 1975), "control" appeared to be an appropriate category to indicate an important dimension of teachers' behavior in ABE classrooms. The meaning of control in the context of ABE was made clear by illustrations from the data and by elaborating its properties, namely the techniques that teachers use to maintain control. As the research unfolds by comparing groups or events to elaborate and test the validity of emergent categories and hypotheses, some of the original categories may be discarded, while others may be reconceptualized to achieve greater specificity and analytical power.
The mechanics of qualitative analysis are relatively straightforward, and adherence to them can preclude a great deal of trouble. A number should be assigned to each emerging category. Incidents in field reports illustrating the category can then be coded in the margin of the report. Glaser and Strauss (1967, pp. 106-107) suggest that when coding an incident for a particular category, the researcher compare it with other incidents previously coded in the same category. This procedure facilitates the generation 6f theoretical properties and the elaboration of the category and its relationship to other categories. I have found that McBee key sort cards greatly facilitate analysis and writing. The researcher simply pastes an incident from a field report on the card and punches the number assigned to the category on the edge of the card. It 'is also important to code and punch for additional information. In the ABE study, each McBee card was punched according to a coding scheme that included categories for staff role (teacher, aide), type of class (basic education, English as a second language), type of facility, and location (city). Once the cards are punched, a simple mechanical procedure allows the researcher to sort out all incidents coded for a particular category or to go even further and sort all incidents for a particular category by staff role, location, and so on.
At various points in the process of analysis, researchers are likely to feel confused, unsure of what to do next, or eager to consolidate their thoughts. It is helpful simply to stop when such a
point comes, take time for reflection, and summarize in writing the main elements of the emerging analysis and the logical next steps in data collection. Such periodic summaries of the research situation are extremely valuable for giving direction to the research and building the analysis systematically and cumulatively. Researchers who fail to begin analysis when they start to collect data and who do not take stock of the research situation periodically run a grave risk of being overwhelmed by a mountain of field reports. In one case, a graduate student spent more than a year collecting data without the direction afforded by continuous analysis and theoretical sampling. Two thousand typewritten pages of field notes were assembled, but the researcher was unable to make sense of the data and finally gave up in despair.
A question that invariably arises in doing grounded-theory research is, Where does it end? How does One know when enough data have been collected? The time to stop collecting data for a particular category is when theoretical saturation has been achieved. A category is saturated when continued data collection yields no new information on the properties of the category-the analyst begins to spin wheels. Since categories are identified at various points in the progress of the research, some are saturated before others. It is important, however, that all the categories believed to be of major importance be fully saturated before one terminates data collection and turns to the task of writing.
The problems of writing a grounded-theory research report can be minimized by following closely the data collection and analysis procedures suggested by Glaser and Strauss (1967). When the fieldwork is finished, the researcher should have a set of memos that cumulatively develop the core of the emerging theory. Moreover, theoretical sampling to collect data in conjunction with ongoing analysis naturally leads to a fairly well-integrated set of categories and propositions by the time major categories are saturated. In writing their theories, researchers rely on their research memos and coded field reports. The memos contain the logic of the analysis, that is, the content of the categories and their properties and an exposition of their interrelationships. Normally, major categories provide. the basic organizing scheme (or themes) of the research report. Coded data from the field reports (actual incidents, quota-
tions) are used to illustrate categories and their properties. For example, the chapter on classroom dynamics in the study of ABE discussed above was organized around the key process categories of attending (patterns of instructional interaction), failure reduction, and control. The properties of each category (for example, the various techniques of failure reduction) were illustrated by excerpts from field reports. The elements of the substantive theory are developed discursively in the research report as categories and their properties are identified, defined, and illustrated by the data.
Integration with Survey Methods
In many cases, the researcher may not wish to generate a fully developed grounded theory, but rather to use grounded-theory techniques for analytical description or to combine grounded theory with survey research. Case studies, as discussed previously, can sometimes contribute to theory and practice if the researcher takes an analytical stance toward the data and seeks not only to describe but also to explain.
Description is useful when little is known about the topic being investigated and when data of adequate generality can be secured. Certainly, little is known about many important aspects of adult education. To ascertain the state of the art, one needs descriptive information. But it is possible to go beyond simple description to achieve more incisive and generalizable findings. For example, when the aim of an investigation is to develop an analytical description of a type of organization or organizational subunit, field research may be combined with survey techniques. This marriage of methods capitalizes on the strengths of each and can yield far richer results than when only one method or the other is used. Sieber, in fact, contends that the integration of field and survey methods constitutes a new style of research that opens "enormous opportunities...for improving our social research strategies" (1973, p. 1340) .
The study of urban ABE programs, discussed previously, attempted to integrate grounded-theory and survey methods to develop an analytical description of urban ABE. Grounded-theory techniques were used when the focus of research was social inter-
action, as in the case of classroom dynamics and the use of paraprofessionals. In these parts of the study the research proceeded along the lines described by Glaser and Strauss (1967). However, for other parts of the study it was necessary to obtain reliable descriptive data by surveying teachers and program directors. The surveys yielded a dependable national picture of program characteristics (for example, size, facilities, client populations) and provided reliable descriptive information about teachers and directors that could not be obtained by field methods. The fieldwork, however, greatly aided in the choice and development of survey items which seemed to offer the possibility of clarifying the role of selected variables. The surveys, in turn, were used to test the universality of certain fieldwork findings.
The joint use of field and survey methods, as noted above, enables the researcher to exploit the advantages of both types of data while minimizing their weaknesses. When accurate description is important, or when the focus is on static variables such as organizational size, survey data tend to be most useful. When the emphasis is on human interaction and the development of an analytical framework to explain it, then qualitative data and grounded-theory analysis are required. Where appropriate, the researcher should consider the advantages of combining fieldwork with at least a modest survey.
In conclusion, grounded theory is no panacea to remedy the lack of systematic knowledge that inhibits meaningful efforts to improve professional practice in adult education. Adult educators need to make more effective use of the full range of social-science research strategies available to them and to continue to borrow relevant findings from the social sciences. Moreover, it is not only empirical research that is woefully inadequate. The dearth of good historical and philosophical inquiry also seriously handicaps the development of the field.
Many research problems can only be addressed by using conventional social-science methods. Others require a more flexible heuristic style, particularly when the emphasis is on understanding social interaction in substantive areas of professional practice. It is time for more researchers in adult education to get out in the field and to try to interpret what is happening there. Ultimately, re-
search in education must be grounded in a firm understanding of the realities of practice. Otherwise, as Iannaccone has warned, studies will continue to be produced that are "methodologically bad, theoretically useless, and...focused on trivial problems" (1973, p. 65). For the educational researcher, grounded theory has pitfalls but also exciting possibilities.
Return to the top of this page.
Return to the first page.