Grounded Theory and Organizational Research

17 downloads 0 Views 1MB Size Report
PATRICIA YANCEY MARTIN. BARRY A. TURNER. Grounded ... ofthe department of sociology, and dean offaculties .... services (Martin, 1984), and the occupa-.
Grounded Theory and Organizational Research*

PATRICIA YANCEY MARTIN BARRY A. TURNER

Grounded theory is an inductive, theory discovery me tho do log^^ that allows the researcher to develop a theoretical account of the general features of a topic while simultaneously grounding the account in empirical observations or data (Glaser & Strauss, 1967). This article explicates the utility of a grounded theory approach to research on work organizations. Following a general introduction to the grounded theory method, the authors'review of the organizational literature using grounded theory illustrates the variety of issues and topics studied through this approach. The authors describe andexplain specific strategiesfor conducting grounded theory research in and on organizations, including note taking and note writing, concept discovery, and concept definition and preliminary elaboration of theory. Throughout the article emphasis is placed on grounded theoryk ability tofacilitate understanding and to identify desirable improvements in work contexts.

Work organizations are complex entities that operate in divergent and often conflicting ways (Clegg & Dunkerly, 1979). Therefore, organizational researchers and consultants should value research that produces a multifaceted picture and an analysis of the dynamics occurring both within and across hierarchical levels. As indicated by Alderfer and Smith (1982), The Journal of Applied Behavioral Science Volume 22, Number 2, pages 141-157. Copyright @ 1986 by NTL Institute. All rights of reproduction in any form reserved. ISSN: 0021-8863.

we suggest that research methods must incorporate the complexities of the organizational context-rather than ignore or simplify them-to produce accurate or *Theauthors thank Gordon Sutton, Nik Pidgeon, Diane Blum Norton, and two anonymous JABS reviewers for helpful comments on earlier drafts of this article.

Porricio Yoncey Morrin is a professor of social work and an affilioredprofessor of sociology at the Florido Store Universiy, Tollohossee. Florido 32306. Barry A. Turner is0 reader of sociology, head ofthe department of sociology, and dean offaculties or the Universiryof Exeter. Exeter. Devon. England.

142

THE JOURNAL OF APPLIED BEHAVIORAL SCIENCE Vol. 2UNo. U1986

useful results. We believe that a method of research known as grounded theory is appropriate (Glaser & Strauss, 1967). Turner (1983, p. 348) has used grounded theory to study organizations for more than a decade because it has enabled him to". ..produce theoretical accounts which are understandable to those in the area studied and which are useful in giving them a superior understanding of the nature of their own situation." Such understanding can also help employees and managers identify and institute changes to bring about organizational improvements (see Pasmore & Friedlander, 1982). In this article, we advocate grounded theory as a method useful for analyzing qualitative data. In particular, we explicate the relevance of grounded theory for organizational research and identify and explain aspects of the grounded theory craft. Available materials o n logicodeductive research aimed at theory verification-e.g., handbooks on measurement, instrument construction, statistical sampling and analysis, and the likeabound, whereas explicit guides to the practice of inductive research (i.e., that intended to discover and develop theory) are rare. This article adds to the organizational literature relative to grounded theory by (I) providinga general introduction to the method, (2) reviewing the organizational literature that uses grounded theory and illustrating the variety of issues and topics previously studied with this approach. and (3) describing and explaining specific strategies for conducting grounded theory research in or on organizations. The third point is particularly important because of the dearth of available "how-to" and technical reference sources. Besides Glaser (1978). recent work by Turner (198 1, 1983) provides the most definitive guidance on the conduct of grounded theory, and even these efforts leave

numerous topics only partially developed for comprehension by the novice. The section on strategies should add to the arsenal of techniques and knowledge of grounded theory as a craft or set of research skills.

GROUNDED THEORY RESEARCH: DEF-N~T~oNs A N D USES Sometimes a social researcher can approach an area of inquiry with a prior, well-formulated theory that so accurately describes it that the research can concentrate on the accumulation of information applicable to the existing theory. Many inquiries, however, d o not fit this pattern. Frequently, no relevant theory exists at all, and even when theories concerned with a topic d o exist, they may be too remote or abstract to offer much detailed guidance and assistance. Under such circumstances. the researcher will want to develop a theoretical account that facilitates discussion of the general features of the topic under study and is firmly based or grounded in the data collected-a grounded theory. Researchers thus generate grounded theory when they are concerned, to use Glaser and Strauss's (1967, p. I) somewhat provocative phrase, with "the discovery of theory from data" rather than with the testing or verification of existing theories. The grounded theorist initially approaches an inquiry with a fairly open mind as to the kind of general theoretical account likely to emerge from the particular investigation. Preconceptions cannot, of course, be wholly abandoned, and we d o not suggest that they should be. Wedo, however, encourage the investigator to commence by concentrating on a detailed description of the features of the data collected before attempting to produce

Grounded Theory and Organizational Research

more general theoretical statements. Once an adequate stock of accurate descriptions of relevant social phenomena has been compiled, the researcher can begin to perceive or hypothesize about relationships among them, relationships that subsequently may be tested using other portions of the data. From the growing accumulation of data indicating such relationships the researcher develops or "discovers" the grounded theory. An emerging grounded theory primarily justifies itself by providing a detailed and carefully crafted account of the area under investigation. This theoretical account not only aids the investigator's understanding, but provides a means of communicating findings to those in the area studied, either as a basis for discussion or as a vehicle for implementing change. This account also enables the researcher to ask questions about the similarities and differences between this theory and other more general theories in the field, especially with respect to goodness of fit and scope of coverage. Such an exercise generally provides not only a perception that some elements discernible in the locally based grounded theory may be relevant or applicable t o a wider context, but also a n enrichment of available general theories as they are evaluated in the light of questions drawn from a new, firmly based, and locally grounded theory. For many years an assumption has prevailed that all worthwhile research is quantitative, pursued by means ofsurveys or other methods of mass investigation. Interest in grounded theory derives in part from a disillusionment with the universal usefulness of quantitative research methods (Beteille, 1976; Phillips, 1971; Selye, 1964). particularly for certain kinds of applied research. T h e processes of grounded theory generation have relevance for qualitative inquiries, whether

'

these are seen as ends in themselves o r as pilot stages of investigations to be pursued later by quantitative means. This stress upon qualitative aspects, however, does not imply that grounded theory is concerned with vague statements, poetic intuitions about society, o r the kind of imprecise handling of data suggested by the term "soft science." O n the contrary, the discipline urged upon grounded theorists through the processes of description, definition, and specification of relationships pushes such investigators toward a high degree of rigor in the handling and interpretation of data.

APPLICATIONS T O ORGANIZATIONAL RESEARCH In the field of organizational studies, grounded theory is likely to interest those concerned with the pilot stages of largescale survey inquiries, those conducting case studies of organizational behavior who wish to produce more than a n impressionistic account from their inquiries, those interested in features of the organizational world-such as corporate cultures-that lend themselves particularly well to qualitative investigation (see Uttal, 1983), and those concerned about carrying out the detailed, locally based fact gathering and interpretation essential to conducting excellent organizational research. Grounded theory does not offer a panacea, a solution to all research problems. Turner, while noting its limited usefulness for dealing with large-scale structural features of society, such as demographic trends or systems of social stratification, does suggest, however, that grounded theory is particularly well suited to dealing with "qualitative data of the kind gathered from participant observation, from the observation of face-to-face interaction, from semi-structured or un-

144

THE JOURNAL OF APPLIED BEHAVIORAL SCIENCE Vol. 22,No. 2/1986

structured interviews, from case-study material or from certain kinds of documentary sources" (1981, p. 227). Typically, these particular kinds of inquiry generate large amounts of data, which accumulate in nonstandard and unpredictable formats. The grounded theory approach offers the researcher a strategy for sifting and analyzing material of this kind. The term grounded theory was developed by Glaser and Strauss as a result of their research into American health institutions. This approach thus has its origins in studies of professional and organizational settings. Work carried out by teams of researchers investigating psychiatric institutions (Strauss, Schatzman, Bucher, Erlich, & Sabshin, 1964) and the organization of the American medical profession (Becker, 1961) provided the background for the development of this methodology, which was first used explicitly in the study of patterns of organizational behavior exhibited in hospitals with respect t o dying patients (Glaser & Strauss, 1964, 1965a). Since this initial study, grounded theoretical approaches have been used in examinations of industrial organizations and cultures (Turner, 197I), batch production scheduling organizations (Reeves & Turner, 1972), hospital planning procedures (Trimble, Cherns, Jupp, & Turner, 1972), the organization of family planning services (Riley & Sermsri, 1974), academic activities (Conrad, 1978), corporate growth (Johnson, 1981), organizational change (Dunn & Swierczek, 1977), antinuclear organizations (Blum, 1982), collectivist organizations and work place democracy (Rothschild-Whitt, 1979; Rothschild-Whitt & Whitt, forthcoming), union conflicts in the field of social services (Martin, 1984), and the occupational behavior of nurses (Hawker, 1982). We do not pretend this list is complete,

particularly since many users of grounded theory d o not discuss the method used to analyze their data and develop theories. If they acknowledge using the approach a t all, they generally d o so only in footnotes, preferring to let others assess their findings based on the quality of the arguments presented rather than on the methodological approach adopted. For these reasons, a researcher will learn little about methodology by studying the substantive theoretical writings of others, for such writings typically concentrate upon findings and conclusions and devote little-if any-space to discussions of the manner in which the analyses were developed. One may learn much about collectivist organizations from a study of RothschildWhitt's work (Rothschild-Whitt, 1979; Rothschild-Whitt & Whitt, forthcoming), about nursing behavior from Hawker (1982), or about corporate growth from J o h n s o n (1981), but t o learn h o w grounded theory may be pursued, one must turn to more explicitly didactic accounts, such as those by Glaser and Strauss (1967), Glaser (1978), and Turner (1981, 1983). Organizationalcultures

As we noted above, the approach and methods of grounded theory are appropriate for use in investigatingand working with organizational cultures. Indeed, the approach was used in a n early investigation of this kind of phenomenon (Turner, 1971). The mixture of participant observation and semistructured interviews typically used to illuminate the elements of a n organizational culture produces data that researchers can readily handle via grounded theoretical strategies. Grounded theory is a systematic way of dealing with such nonstandard data, and-skillfully handled-can produce accounts of a corporate culture that are

Grounded Theory and Organizational Research

recognizable to the members of that corporation and may therefore serve as a structure for discussions within the company on cultural themes. Such discussions may also lead to changes in a corporate culture, although we share recently expressed reservations (Uttal, 1983) as to the extent to which outside consultants can consciously manipulate and modify a corporate culture. lfchange is regarded as desirable and possible, however, the use of grounded theory to analyze the data directing the change will likely result in such modifications being carried out upon a firmer basis than that provided either by the monitoring of the opinions of a few senior managers or by the impressionistic assessment of more widespread datagathering activities (for a similar view, see Pasmore & Friedlander, 1982).

STRATEGIES FOR GROUNDED THEORY RESEARCH IN A N D O N ORGANIZATIONS We will now discuss three aspects of the grounded theory craft: note writing, discovery or identification of concepts, and development of concept definition by means of theoretical memoranda. Each section is brief but nevertheless offers guidance beyond that presently available in the literature. Note writing

The taking and writing of notes are two distinct steps in the grounded theory process. During the course of a morning's observations and conversations, many comments are heard, many events observed. In vivo (Glaser, 1978) notes taken on the spot are, as suggested by Lofland (197 I), primarily brief records of key ideas or a means of triggering memory, useful when later writing a full set of notes. The demands of note writing are as great as

those of any part of the grounded theory process. Note writing is time consuming, often taking as long or longer than the observation itself. One cannot wait more than a day or two after the observation to write the notes or else the majority of details and nuances will be forgotten. Finally, if done correctly, the writing of notes requires considerable investments in thought and energy by the researcher. T o be fully useful, notes must be richly detailed and descriptive of the context. Table I provides examples of poor notes and good notes, based on the following criteria.' First, good notes provide full or complete explanations. They describe and explain the context of a comment or event and identify all speakers or actors by position, title, location in the organization under study, and so forth. T o be most useful, notes should be rich in detail and "story like" in explaining the topic described. Poor notes, in contrast, often contain cryptic phrases, summary statements, and generalizations that are incomprehensible to anyone other than the researcher and, after a short period of time, require considerable guesswork on the researcher's part to fill in what was actually said, done, or observed. Statements shown in the "poor notes" section of Table 1 followed by footnote "a" fail to explicate fully the substance of the situation described. Second, good notes provide more than a chronology of events or comments and more than a verbatim report of who said what to whom. Although verbatim comments are desirable and useful-they help provide rich, descriptive detail-alone they are insufficient. As indicated above, quotations and items related to chronology must be placed in context and the context of the data must be fully described. Statements in Table 1 followed by footnote "b" fail to provide a full, contextual explanation of verbatim

THE JOURNAL OF APPLIED BEHAVIORAL SCIENCE Vol. 22/No. 211986

Table I Examples of Poor and Good Notes for Grounded Theory Research Poor Notes Boss came in to discuss a position. Informed Marilyn Bem that external response was not good; seemed to want to go internal to fill position." He acted like a bossc.. .Another worker dropped by t o discuss fire near a foster care home near one of the clients. Marilyn had been there They both seemed to use event as a way t o reinforce professional identity.< Have cooperative management of the client. There are no case managers although Marilyn sees this as ideal. Resents push o n part of mental health to take this role because they treat social workers as cab drivers, "they should get into field Marilyn finds it difficult to imagine "integrated services" be-

.

cause it is too complicated even with a lower caseload." Barbara Mason and Carl Marx performed as co-leaders but no corner on market of ideas."" Every problem received some active dispositiona.. . . Meeting adjourned at 11:35. People lingered for one to one and small group discussions." Sounded healthy and productive."" This collaborative effort engages most participants and appears to produce services.< Telephone incoming. Dick McVee, supervisor at Adult Services Unit, did not respond as promised, caller says. Barbara calls Dick; excellent handling of supervisory nudge to activate and renew commitment to send requested material.""

Good Notes

After drawing the authority structure (see the drawing on Page 17, both his and my redraft) in his department for me, Dick McVee made several points about how it works. First, there are six separate service units under Dick's immediate supervisor (Barbara Mason who is a Section Supervisor); five of these six are child protective services units and only one is adult services (i.e., which Dick's is). This means, in Dick's view, that Barbara is not very much "up on" adult service program rules and procedures. making her less useful to him than he would prefer. (Nore: Some of this is my conclusion.' Dick was very circumspect in what he had to say and I don't want to attribute things to him.) Second, there are no direct communication lines between social workers in Dick's adult services unit and those in the other five units under Barbara's supervision. Thus, if there is a coordination problem between two social workers in two different units (one in adult service and one in child protection), each social worker must go to hislher immediate supervisor (e.g., Dick) and then each immediate supervisor must go to Barbara Mason t o resolve the problem. Third, suppose the two social workers are under different Section Supervisors-e.g., one

under Barbara and one under Andy Watson who is over seven other units in the same building. If there is a conflict now, the problem must go t o Dick and his equivalent in the other unit, then to Barbara and Andy, and then to their supervisor (Mary Sims, the Division Supervisor) for resolution. Fourth, Dick emphasized that he was not thinking particularly about "problem resolution" in making his comments but rather about mere communication-about the ability of different front-line workers to simply share information about plans of action o r clients. His point is that thestructure mostly gets in the way of direct service workers' having much time o r opportunity to communicate with each other. Dick noted by way of illustration that if a client has four problems-e.g.. is mentally retarded, mentally disturbed, financially dependent, and cannot manage on his/her own, then that client is likely to have four different social workers-and none of these will know what the other three are doing. When I asked "Why?" he drew me the chart and explained how the formal arrangements affect relationships and interaction among service workers. He also stressed that there is a lot of pressure on people at State HSO not to go around o r outside formal authority lines.

'The description ofthe observation. event, or incident is too brief or too general (broad). Not enough details are provided regarding who said what, titles or positions, what comments were about. or the like. b~ chronology or verbatim report is given without any explanation ofthe context ora full explanation of what was involved. 'The note writer is editorializing rather than providing a full. rich description of what actually occurred.

Grounded Theory and Organizational Research

quotes or sufficient substantive details regarding the chronology of events observed. Third, good notes require only minimal editorial comments. Any editorial comments made should be explicitly identified as such, as should questions regarding the interpretation of events or the researcher's possible lack of comprehension (e.g., incomplete understanding or information). Editorial comments, indicated by footnote "c" in Table 1, evaluate a situation or event rather than merely describe or report it. The poor notes in Table 1 are rife with editorial comments that have little utility to the grounded theorist without accompanying detailed descriptions. The lower portion of Table I provides examples of better, though far from perfect, notes. These notes also contain editorial comments and no doubt reflect the imperfections and biases of the note taker. They are, however, richly detailed and contextually explicit and clear. Not only can the person who wrote them use them to remember a year (or more) later exactly what was observed, but other members of the research team can use these notes to discover concepts. In short, good notes are not only detailed and contextually complete, but are fairly selfexplanatory and remain useful even after a considerable period of time. Concept discovery

Concept discovery refers t o the strategic process of moving from data to abstract categories (Glaser, 1978; Glaser & Strauss, 1967), labels (Turner, 198I), o r concepts. (Note: These three terms are used interchangeably and can be interpreted generally t o mean concepts.) Turner (1981) pays extensive attention to this process, providing a strategy for the researcher to shift attention from written, qualitative notes-such a s those in

Table I-to the identification or discovery of concepts capturing the ideas o r phenomena described by the notes. The points made below address mistakes typical of graduate students in research seminars and colleagues in collaborative research projects. We offer them to fill gaps in understanding of what concept discovery entails and how to accomplish it. The basic strategy Movement from data to concept is appropriately viewed as a movement across levels of abstraction, not a numerical tabulation of incidents associated with a discovered concept. The researcher seeks to discover (identify) a slightly higher level of abstraction-higher than the data themselves-that allows the application of a name t o the action or object observed or referred to. Incidents (see Turner, 1981) accumulated under a label are used later to develop a statement of the concept's nominal definition, theoretical meaning, or substantive content. Table 2 illustrates what a concept "card" looks like, providing a sample label and multiple incidents taken from numerous sets of data. When an incident is recorded on a concept card (Turner, 1981), the theorist assumes this incident has something in common with other incidents cited on the card-making them all referents of the same concept. Later, when working out definitional statements (see below), one either explains explicitly what these common qualities are or drops from the card those incidents no longer considered illustrative of the concept. No addition or subtraction of incidents is to be considered irreversible. For this and other reasons, Glaser ( 1978) and Turner (1981) recommend recording each incident in more than one concept category, particularly during the early stages of concept generation.

THE JOURNAL OF APPLIED BEHAVIORAL SCIENCE Vol. W N o . 2/1986

Table 2 A Sample Concept Card With Multiple Incidents From Field Notes* Card 17: Coordination/Linkage Among Different Program Areas (I) NS I I:*' 11-12

Dick McVee (who is a unit supervisor of an adult services unit) drew a figure illustrating the formal structure of a service division in which his unit is lodged and explained how problems at the service point are "officially" resolved. He noted that they are solved vertically rather than horizontally. for the most pan. He noted also that the structure actually prevents workers in separate units from knowing what workers with the same client in other unitsare doing. Marilyn (a social worker in adult services unit under Dick McVee)saysthat because workers and clients are "constantly changing," and because it would be so complicated to have "integrated services,"it wouldn't work-even if workers had reduced caseloads. (Nore:Some of these notes are summarized rather than developed-so be careful.) Marilyn thinks the idea of having a "case manager" is a good one-some person could cooperatively manage the different social workers having a single client. But she says thereare none now. She also resents the push on the part of mental health workers to act like "case managers"and to treat social workers as if their majorjob is that ofcabdriver. She notes that mental health workers can "close out" patients who are "uncooperative," but adult service workers cannot. This complicates herjob. Dick McVee says, "Adult services get the 'sewage'of the larger agency; we get the people no one else will take and this causes a morale problem with our workers. Also. our workers are not treated or viewed as professionals by workers in other program areas. Then, you add problems like Lora mentioned this morning-no way to be promoted or have your pay raised after I I years (whereas in food stamps there is a way), and you get morale problems." Sue Jones, head of planning and quality control in the adult services section of the division. says: "We need Title XX money for training re. how to respond to various crisis behaviors of clients. Sam Douglas, my equivalent in mental health. is working with us on this. Adult services is getting to have a lot of input into the mental health plan for the coming year. Before giving money to mental health. adult servicesisgoing to be sure mental health emphasizesthe areaswe need."She reports that Sam D o u-~ l a s. ( i nmental health) thinks that: "What we need isday careW(i.e..for theelderly). Sue Jonessays this is not her priority for care for theelderly. "We need to be able to respond to crises ourselves.. .mental healthS crisis teams will not go out unless theclient requests the team!"All present guffaw at this-and say this is a problem. If the client is saneenough to request help, they wouldn't havesucha problem to begin with. When the mental health crisis team won't go out unless the client requests it, this forces the adult services worker to take clients to the psychiatric facility by force (sheriff, or police. often in handcuffs). Then the client hates you as the worker.. ."you can just hang it up." A directive from the state office (of State HSO) has come to the division mandating formation ofan interdisciplinary team composed of representatives of health, mental health, licensure and certification, and adult services to deal with health and mental health issues of residents in adult congregate living facilities. The directive specifies not only the team's formation but that it meet every twoweeks. Ada. social worker in adult services, spent all of one morning this week standing in line in the food stamps office around the corner from her office-on behalf of one of her clients. She saysshe tried for a week or more to get through by phone to a food stampseligibility worker but the phone wasalways busy. She had, therefore, no moreaccesstothe FS workerthan the client did (but the client was not well enough to stand in line for herself).

*For suggestions on organizingconcept cards. see Turner (1981). **NS = Note Set: numbers following the colon stand for page numbers in that note set.

Grounded Theory and Organizational Research

Placement of an incident in more than one category clearly distinguishes the activity ofconcept generation from that of tabulation in theory verification research. A hallmark of measurement in verification research is the independence of one concept's referents from those of any other concept (Campbell & Fiske, 1959). In theory discovery research, too, a single referent-embodying a single idea-is ultimately assigned to one and only one abstract concept. Until the researcher nioves beyond the intuitive stage of incident placement, however, multiple placement helps one avoid eliminating options too early and stimulates ideas with respect to connections (causal, contextual, or other types) among different conceptual categories.

set of notes, the grounded theorist does not begin to generate an entirely new set of concepts for subsequent note sets. Instead, one mostly adds incidents from later notes to the concept cards developed earlier. Glaser and Strauss (1967) note that concepts emerge fairly quickly, particularly core concepts (Glaser, 1978). Our own experience verifies this. By the time three or four sets of data have been analyzed, the majority of useful concepts will have been discovered. Level of abstraction

Concept utility

All concepts, including those discovered by grounded theorists, are only more or less useful, not more or less true or valid (Cohen & Nagel, 1934). As one's theory emerges, more useful concepts will remain and less helpful ones will fall into disuse. If a concept is useful, one (typically) discovers it often and members of the organization under study can recognize it and relate it to their experiences (Glaser & Strauss, 1967; Turner, 1983). The same holds true for one's entire set of concepts and their relationships-that is, one's theory. Theories are more or less useful-for explaining, predicting, and the like. If they are of no use, one will eventually recognize this. The grounded theory methodology encourages the researcher to take steps, through the rotating cycle of collection and analysis (Glaser & Strauss, 1967), to determine if one's theory is genuinely useful for addressing the phenomena under study. Coding multiple sets of notes

Following the analysis o r coding (Glaser, 1978) of incidents included in one

'

In preparing labels for one's concept cards, the initial aim is to find a level of abstraction high enough for one to avoid creating a separate concept card for every "fact" observed but low enough to ensure that the discovered concept relates explicitly to the substantive phenomenon under study. If a label is insufficiently abstract or general, too few observations will fall into that category. One will add few incidents to each concept card as one analyzes the data. In such cases, the label merely restates or rephrases the data. To "work" (Glaser & Strauss, 1967), a conceptual label must occupy a higher level of abstraction than the incidents (facts, observations) it is intended to classify. If the concept label is too abstract, however, too much information will fall into that category. For example, a concept card labeled "interactionWor"exchangeWmight include every observed instance of persons talking to one another, all communications among organizational offices, units, or departments, and interorganizational linkages and ties. Distinctions regarding forms and types of social intercourse-and the substance or content of a transaction-would be missed. T o develop a substantive theory relevant to a particular organization or organizational issue, one should devise labels clearly

150

THE JOURNAL OF APPLIED BEHAVIORAL SCIENCE Vol. 2 2 N o . 21986

relevant to that organization, a t least initially. In Glaser and Strauss's view (1967), this process leads to the discovery of substantive-rather than formaltheory. We again remind readers that little in the grounded theory process is irreversible. If one's concepts are too specific o r too general this can be remedied, in the first case by combining specific categories into more general categories and, in the latter case, by breaking down a category that is too general into its more specific dimensions or aspects. One project of the senior author's (Martin, 1984) produced fewer than 100 concepts from a massive set of field notes (including interviews, official documents, participant observation notes, and the like). Furthermore. fewer than 40 of these concepts proved very useful, and of these about ten have formed the bases for analysis to date. Sufficiently complete description When entering incidents o n cards (Turner, 198 I), these incidents should be described fully enough to avoid the need for backtracking to the notes t o recall substantive details. In this regard, Turner's (198 1, p. 234) use of brief phrases may be misleading to the novice coder. T o assure sufficient detail, some find it useful either to quote extensively from one's notes or to cut and paste material from extra copies of the notes. Glaser (1978) recommends that researchers simply write concept labels in the margins of their notes, whereas Turner (1981) prefers the transference of descriptions of incidents to cards because this facilitates the comparing of incidents as one develops concept definitions. When all descriptions of potentially relevant incidents have been put in one location (i.e., on a card or set of cards), the difficult task of determining what incidents have in common is made somewhat easier.

Nonlinear process The grounded theory researcher should not expect to proceed in a linear fashion from raw data to concept cards to preliminary writing on theory to the final theory (Glaser, 1978; Turner, 198 1). In a single day, one may discover a new concept, draft a complex theoretical memorandum on ideas prompted by that discovery, begin the work of clarifying a concept's nominal definition, make notes for further data collection, and draft a n elementary memorandum regarding what the incidents on a concept card have in common. In short, contrary to depictions (generally idealized) of logicodeductive research that suggest a linear progression from theory to data to analysis to interpretation, the grounded theory process is self-consciously and intentionally nonlinear and iterative (Glaser & Strauss, 1967). Concept definition and the preliminary elaboration of theory

Glaser (1978) plays down the importance of concept definition because the nominal definitions (Hemple, 1952) of concepts discovered via grounded theory remain in flux-that is, the continuing discovery of new incidents generally modifies a concept's meaning. Turner (1981), however, views concept definition as vital because it forces the theorist to formulate previously intuitive thoughts and wrestle with the difficult task of expressing-explicitly and at length-the themes inherent in incidents previously placed in a conceptual category on the basis of global associations. A practical approach to developing a definition is to select a concept to which the researcher has assigned several incidents and hopes has central relevance to the topic under study. The order in which concepts are defined o r developed is

Grounded Theory and Organizational Research

arbitrary because, as noted earlier, the more useful concept categories will persist and the less useful ones will be discarded as the theory matures. One can use many methods to develop definitional statements; we present below one that has worked for us. Read the descriptions of incidents on a particular concept card and ask what common threads, if any, tie these incidents together. (One can do this either by examining each incident and its relationship to the others or by viewing the incidents as a group.) Once a theme is perceived, entitle a theoretical memorandum (Glaser, 1978; also see below) by that theme and write in a free-flowing manner-free of any self-editing-about any and all ideas that seem embodied in the theme. Glaser (1978) stresses the importance of allowing oneself to write whatever comes to mind. Turner (1981) also supports unedited writing, even when one has the feeling of "making it up." The value of the theoretical memorandum can be assessed at a later time, for any and all materials can be tossed out, revised, added to, or rewritten. A worthy goal is to have one's ideas, insofar as possible, correspond well with the data, but beyond that one should identify and develop themes and concepts suggested by the data on whatever level of abstraction they tend to emerge. In developing the theoretical substance of a concept or proposition, one or more themes may emerge that are not precisely components of the concept but rather are conditions under which it occurs, possible consequences, effects of it, or the like. This occurrence is a primary product and benefit of theoretical "memoing." The relationship between a concept and particular empirical incidents must eventually be made explicit (Turner, 1981). As this is attempted, the memo's title o r label is often revised, incidents reassigned, and/

or new concepts identified. Such events are typical, not unusual. Early discoveries are neither sacred nor constraining t o later ones. The process of concept definition often results in insights, new discoveries, and clarification of complexities and contradictions in the data-and in reality itself-not previously perceived. Theoretical memoranda

Despite his equivocation with respect to the value of concept definition, Glaser equivocates not at all when discussing the value of theoretical memoranda (1978, p. 83): The core stage is the process of generating theory, the bedrock of theory generation. its true product is the writing of theoretical memos. If the analyst skips this stage by going directly from coding to sorting or to writing. he [sic] is not doing grounded theory . Memos are the theorizing u,rite-up of ideas about codes and their relationships as they strike rhe analyst whik coding.. . .

Memoing is a constant process that begins when first coding data, and continues through reading memos or literature, sortingand writing papers or monographs to the very end.

T o illustrate theoretical memoranda, Table 3 shows a memo aimed a t identifying a common theme among several incidents described on Card 17 (see Table 2). T o begin, the researcher asks the following questions: D o any common themes connect the incidents on the card in question? Do any two or more incidents seem to address the same ideas o r phenomena? If so, what is the theme they reflect? We find that the third, fourth, fifth, and seventh incidents on Card.17 (Table 2) have at least one theme in common, which we label "conceptions of adult service workers by workers in other program areas." We entitle a theoretical memo with this phrase (see Table 3) and begin, using free association, to speculate

THE JOURNAL OF APPLIED BEHAVIORAL SCIENCE Vol. 2UNo. 2/1986

Table 3 Sample Memorandum On a Lower Level of Theoretical Generality Memo 7 Conceptions ojadult service workers (and/or their mandate or appropriate tasks) by workers in otherprograms Links: Cardl7;Memos3, 12,13,14,15,16,19,28,34 Based on incidents 3. 4, 5. and 7 of Card 17, workers in adult services feel they are misperceived. not viewed with respect or as equals by service workers in other program areas. Mental health workers view them as "cab drivers" and people who are to be managed by them (rather than, one presumes. as equals). This suggests that adult service workers perceive they are treated with less respect than they deserve by workers in other program areas. The example of the adult services worker standing in line toenroll a client in the food stamps program suggests that adult service workersdo not personally know workers in other program areas-even when they are housed in the same building and are close geographically (only a few hundred feet apart). This latter incident may reflect more the ac~ualrelationships among workers i t ~dijj'jerent program areas and not just perceptions or conceptions.* Ifso, this needs to be developed in another memo. Incident 5 suggests that a high official in mental health may have already decided what program option workers in adult services should pursue for the elderly with mental health problems. This is perceived by adult services staff a s a lack of understanding of their jobs, their tasks, and their relationships with clients. They also perceive a lack of respect as professionals of equal status. A question could be raised as to why this is the case: wha!mo).contribute to thesemisperceptions and lack of respect? One theme suggested by incident 3 is that workers in (for example) mental health have more options in handling cases than d o workers in adult services, and greater options in "closing cases" gives the mental health worker an advantage over the adult service worker. If options

are resources (and the literature on HSOs suggest that job discretion is a major determinant of staff performance in such agencies; see Whiddon. 1982). then mental health workers are factually advantaged vis-A-vis adult services workers. This not only results in their being perceived as advantaged: they are advantaged. To the extent factual advantage shapesperceptions.advantaged workersnodoubt perceive themselvesand their jobs as superior. Similarly, comments in item four suggest that adult services constitutes a catchall category of service provision-which is mandated to serve adults who fail to fit criteria for other programs. Again, this suggests less "control" by (adult services) staff over client selection or termination. and to the extent that job discretion is important for staff performance, morale. etc. it would be expected that it is lower in adult services than some other places. if gain, see Whiddon. 1982, for findings on this.) Suggested consequences of feeling that one has fewer options than others and that one is less valued than othersare lowerwork moraleand, the literature would suggest, a lower actual level ofjob performance. If adult service workers in fact have less discretion than other workers and must accept other programs'rejected clients, and if the workers are viewed as less professional in the eyes of other workers and are respected less, then the chances for adult service workers to influence these other areas is less than the reverse. Their comparative status and discretion may actually disadvantage them relative to other program areas. On a more general level. relationships between program areas in a large HSO may be less positive when status . and discretion are asymmetrical.

*Underlined phrases indicate topics that differ from the memo's label or title and need to be developed in other theoretical memoranda.

or specify exactly what these incidents have in common conceptually. T h e grounded theorist's goal is to represent conceptually what the data reflect empirically. This conceptual representation, when complete, is the primary product of

the discovery process and becomes one's theory. Several points can be made from Table 3. First, until fairly late in the memo, Memo 7 consists primarily of a restatement of the data on Card 7. This is charac-

Grounded Theory and Organizational Research

teristic of early theoretical memoranda and, in our experience, represents a phase grounded theorists typically go through before progressing to later, more theoretical stages. In writing about data in early memos, one's goal is minimally to move up a notch in level of abstraction. Rather than name particular people who were observed, one writes about their organizational roles, positions, or activities. In doing this, one attempts to see the behavior or comments of persons as indicative of those occupying their o r similar positions. In this way, even rehash memos advance the grounded theorist's thinking by moving it from particular incidents to potential general insights regarding relationships among positions, departments, and/ or organizations. Note also that Memo 7 contains other conceptual ideas and themes besides those reflected in its title. This, too, is typical of early memos. As one writes about a concept or theme, other topics may emerge. While it is preferable to have a separate memo for each concept o r proposition, early memos are unlikely t o reflect such discipline. Glaser (1978) advises, therefore, that concepts or themes differing from those reflected in the memo's title be underlined to highlight their presence. Later, separate memos can be written on each new topic, assuming these concepts prove useful o r relevant to the emerging theory. Some topics that emerge in early memos may, as suggested earlier, either cause or result from the concept(s) embodied in a memo's title. This occurred in Memo 7. Dimensions of "conceptions of adult service workersn appear, as d o potential causes and consequences of these conceptions. The possible causes include, for example, the job discretion of workers in different program areas. Those with greater discretion may view themselves and be viewed by others as ad-

vantaged, perhaps even superior. Ideas such as these are important and require further research-often in the form of new data collection-and further development in later memoranda, as Table 4 illustrates. Memo 28 (in Table 4) illustrates an effort to develop further the topic of interprogram relationships among service workers. In comparison to Memo 7, Memo 28 is more general-that is, less explicitly tied to particular incidents in the data. As does Memo 7, however, Memo 28 contains selfdirections and caveats by the author regarding the need for confirmation, further data collection, additional theoretical development, and the like. Both Memo 7 and Memo 28 relate some emerging concepts and propositions to the extant literature. The integration (Glaser & Strauss, 1967; Turner, 1981) of theory with the literature is required prior to completion of the theory's formulation. At early stages, the theorist primarily notes that similar or possibly related concepts o r propositions have appeared in the literature and that final stages of development should take this into account. Typical memoranda at later stages are detailed accounts of the emergent theory's confirmation or contradiction of existing theory. As noted earlier, preliminary ideas such as many of those in Memo 7 lead to subsequent memoranda (and data analysis). (Memo 28 [see Table 41 represents a deeper analysis of certain aspects of Memo 7.) Additional memoranda stemming from incidents on Card 17 (Table 2) might discuss the following: the nature and role of formal structure in interprogram coordination and/or linkage; the nature of coordination and/or linkage a t different hierarchical levels; staff morale, including its definition and causes (e.g., comparative status, the role of discretion, the effects of society's evaluations of

THE JOURNAL OF APPLIED BEHAVIORAL SCIENCE Vol. 22/No. 2/1986

Table 4 Sample Memorandum On a Higher Level of Theoretical Generality* Memo 28 Issues #ding

the form a n d quality o j interprogram relationships among service workers a n d the consequences ojthese relationships Links: Card 17; Memos 7,12,13,20,22,23,24

Two concepts seem particularly important for tying work conditions and structural arrangements to service outcomes: (a) the extent to which (or whether) workers in one program area interact with those in other program areas, and (b) perceptions of the relative status of one's own job vis-8-vis other (for definitions, see earlier memos). These two things may. in turn, affect feelings toward and/or evaluations of workers in other program areas, worker morale, and possibly quality ofjob performance (see Whiddon, 1982). Factors that may influence the extent of contact and perceptions of relative status include the following. (Note: The following are rank-ordered in terms of their suspected significance for predicting the two foregoing variables.) XI = Timeand task constraints(adivision-of-laborissue)of the workers'jobs. Is timeavailable for interacting with other workers?(Even if one is not required to interact, it would be hard to d o so on one's own initiative if a person's job demands every second of hislher time). Does fulfillment of one's job require contact with workers in other program areas? If so, workers with such jobs will have considerably more contacts than those with jobs not requiring it.

XZ= The actual amount of discretion of workers in one program area relative to workers in another program area regarding (a) the ability to accept or reject clients for their program and (b) the ability to close cases as they see fit or at their own choosing. The suggestion here is that workers in two different areas with similar discretion levels will have more positive relationships with one another than will those with very different discretion levels.

X I = The extent to which workers in one program area have accurate (vs. inaccurate) perceptions of the job tasks and constraints of workers in other (or related) areas. Accurate knowledge of the tasks and constraints of the jobs of workers in another program may affect the assessments of the value or worth of the service those workers provide and the worth or status of the workers themselves (i.e.. greater accuracy of knowledge will lead to higher assessments). XI = Authority structure of the organization. This is related to the foregoing factor, but is not identical to it. Given the organization's authoritystructure, are the routes by which people officially resolve conflicts with workers in other program areas long vs. short, simple vs. complex, direct vs. indirect, etc.? D o supervisors in one program area have extensive (or few) contacts or interactions with supervisors in other program areas? When they have more, routine contact among direct service workers should occur more.

XS= "Valuations" by workers in one program area of the worthiness or nature of clientele served by workers in another area. Although 1 suspect this has minimal effect, society's "valuing"ofchildren over the old, of mentally ill over the retarded,etc. may affect (tosome degree) workers' comparative assessments of each other because of the tendency to evaluate workers on the basis of the status of their clients (see Hasenfeld, 1983; Martin. 1984; Rosenthal. 1974). Reasons for the rank ordering are as follows. If workers are basically constrained in the performance of their own jobs because of time or "permission" to interact with others in other program areas. I suspect there will be little of this.. .. *The memo from which this is taken is I 2 pages long.

Grounded Theory and Organizational Research

clients, or the significance of opportunities for advancement). Less obvious theoretical memoranda might also be developed, depending on the direction the theory takes or the theorist's interests. Related memo topics might include the following: a definition of coordination per se; a definition of interprogram linkages; the role of negotiation in interprogram relations; or a definition of interprogram planning. Highly abstract o r general memoranda are easier to develop after other, more specific materials have been developed, for memoranda of a low level, of abstraction facilitate the development of more general ones. The grounded theorist usually develops a low-level memo, stops to draft one of a higher level, then returns to work on the original or a different low-level memo. Additionally, memos on the relationship of the theory to the literature are appropriate a t any and all stages of the theory's developmentalthough both Glaser (1978) and Turner (1981) encourage the creation of such memos later rather than earlier so as t o minimize the imposition of apriori concepts onto the data and because the themes in the emergent theory often d o not become obvious until well into the grounded theory process. Theoretical memos, both specific and general, are valuable because the material they represent is permanently available to the researcher and can be consulted, reviewed, revised, developed, a n d / o r deleted later. Theoretical memoranda free the grounded theorist from having to remember everything observed o r thought. With one's ideas on paper, the theorist can move on to other tasks, assured that ideas developed earlier will be available when the time for more development arrives. Memos 7 and 28 (in Tables 3 and 4) may strike the reader as jumbled, ill phrased, ill organized, disconnected.

These qualities are typical, particularly of early memos. We could have reorganized and polished them, but this would have represented the process as neater and more coherent than it is. Hand-written comments in the margins of typed memos are typical. Such notes remind the theorist of additional points in need of development, the need for more data collection, and so forth. Theoretical memos are preferably typed double or triple spaced to leave room for additional thoughts, editing, and extensions of points and ideas.

DISCUSSION In this article, we have attempted to provide a r a t i o n a l e . f o r the use of grounded theory in organizational research -both basic and applied -and to illustrate various technical points and procedures that have received little attention in the literature. Alone, the material we present will not enable someone t o engage in grounded theory research. In combination with recent work by Glaser (1978) and by Turner (I98 1, 1983), however, the article addresses some important gaps in the grounded theory literature. Perhaps the major contribution of this article is the way it explicitly applies both the topic and methodology of grounded theory research to organizational analysis. We believe that grounded theory has potential far beyond its use to date. When used with other methods designed t o "make manageable" seemingly unmanageable qualitative data (see, for example, Turner, 1983) and/or aid in the discovery of useful and important concepts for subsequent research and development, grounded theory research in organizations can provide important components for the researcher's "kit of tools" for making sense of-and improving-organizational reality.

156

THE JOURNAL OF APPLIED BEHAVIORAL SCIENCE Vol. 2UNo. 2,1986

NOTE I . The data and memoranda in the tables come from a study of a large, structurally integrated human service organization in Florida (see Frumkin. Imenhein, Chackerian. & Martin, 1983; Imershein. Chackerian. Martin. & Frumkin. 1983; Martin, Chackerian, Imershein, & Frumkin, 1983). The research was primarily intended to assess the extent and form of cross-program coordination among workers assigned to different program categoriesand the effects of this on the quality of service provision.

REFERENCES Alderfer. C.P., & S m i t h , K.K. (1982). Studying intergroup relations embedded in organizations. Adminisrrarive Science Quarrerly, 27. 35-65. Becker. H.S. (1961). B0.w in white. Chicago: University of Chicago Press. Beteille. A. (1976). The dangen of research methodology. Inrernarional Social Science Journal, 27. 195-197. Blum. D.B. (1982). tije span changes in an alremarive social movemenr organizarion: The case of anri-nuclear alliance. Unpublished doctoral dissertation, the Florida State University, Tallahassee. Campbell, D.T., & Fiske, D.W. (1959). Convergent and discriminant validation by the multitraitmultimethod matrix. Psychological Bullerin. 56.81-105. Clegg. S., & Dunkerley, D. (1979). Organizarion and ronrrol. London: Routledge & Kegan Paul. Cohen. M.R.. & Nagel. E. (1934). An inrroducrion ro logic and scienrijic method. New York: Harcourt, Brace, and World. Conrad. C.F. (1978). A grounded theory of academic change. Sociology of Education, 51, 101-1 12. Dunn. W.N.. & Swierczek. F.W. (1977). Planned organizational change: Toward grounded theory. Journalof Applied BehavioralScience, 13. 135-157. Frumkin. M.. Imershein, A.. Chackerian, R.. & Martin, P.Y. (1983). Evaluating state level integration of human services. Adminisrrarion in Social Work, 7 . 13-24. Glaser. B.G. (1978). Theorerical sensiriviry: Advances in rhe merhodology of grounded theory. Mill Valley. CA: Sociology Press. Glaser. B.G.,& Strauss. A.L. (1964). The social loss

ofdying patients. American Journalof Nursing. 64, 1 19-122. Glaser. B.G., & Strauss, A.L. (1965a). Awarenessof dying. Chicago: Aldine. Glaser. B.G.. & Strauss, A.L. (1965b). Temporal aspects of dying as a non-scheduled status package. American Journal of Sociology, 71. 48-59. Glaser, B.G.. & Strauss. A.L. (1967). Thediscovery ofgrounded theory. Chicago: Aldine. Hasenfeld, Y. (1983). Human service organizarions. Englewood Cliffs. NJ: Prentice-Hall. Hawker, R. (1982). The inreracrion between nurses and parienrs' relarives. Unpublished doctoral dissertation. University of Exeter, England. *Hemple. C.G. (1952). Fundamentals of concept formation in empirical science. Inrernarional Encyclopedia of Un9ed Science, 2(7). Imershein. A.W.. Chackerian. R., Martin, P.. & Frumkin. M. (1983). Measuring organizational change in human services. New England Journal of Human Services. 4.2 1-28. Johnson, G. (1981). The applicarion of grounded theory to a srudy of rorporaregrowrh (working paper 212). Birmingham, England: University of Aston Management Centre. Lofland. J. (1971). Analyzing social serrings: A guide to qualirarive observarion.and analysis. Belmont. CA: Wadsworth. Martin. P.Y.. Chackerian, R., Imershein, A.. & Frumkin, M. (1983). The concept of "integrated" services reconsidered. Social Science Quarrerly, 64.747-763. Martin. P.Y. (1984). Trade unions, conflict, and the nature of work in residential service organizations. Organizarion Studies. 5. 169-185. Pasmore, W., & Friedlander, F. (1982). An actionresearch program for increasing employee involvement in problem solving. Adminisrrarive Science Quarrerly. 27.343-362. Philips, D.L. (1971). KnowledgeJrom whar? Theories and methods in social research. Chicago: Rand McNally. Reeves, T.K.. & Turner. B.A. (1972). A theory of organization in batch production factories. Adminisrrarive Science Quarrerly. 17.81-98. Riley. J.N., & Sermsri. S. (1974). The variegated Thai medical system as a context for birth control services (working paper no. 6). Bangkok. Thailand: Institute for Population and Social Research. Mahidol University. Rosenthal, D.E. (1974). Lawyer and clienr: WhoS in charge?New York: Russell Sage Foundation. Rothschild-Whitt, J. (1979). The collectivist organization: An alternative t o rational-

Grounded Theory and Organizational Research

bureaucratic models. American Sociological Reviews, 44,509-527. Rothschild-Whitt. J.,& Whitt, J.A. (Forthcoming). Work wirhour bosses: Conditionsand dilemmos of orgonizarional democracy in grassroots cooperatives. Cambridge, MA: Cambridge University Press. Selye. H. (1964). From dreom to discover)*:On being a scienrist. New York: McGraw-Hill. Strauss. A.. Schatzman. L., Bucher, R.. Erlich, D.. & Sabshin. M. (1964). Psychiatric ideologies and insrirurions. New York: Free Press of Glencoe. Trimble. E.G.. Cherns. A.B.. Jupp. B.C..&Turner. B.A. (1972). The eflecriveness of cost planning and orher cost control rechniques in hospital consrruction (final report to Department of Health and Social Security). Loughborough. England: L o u g h b o r o u g h University of Technology.

Turner. B.A. (1971). Exploring the indusrriol subculture. London: Macmillan. Turner, B.A. (1978). Man-modedisosrers. London: Wykeham Press. Turner. B.A. (1981). Some practical aspects of qualitativedata analysis: One way of organizing the cognitive processes associated with the generation of grounded theory. Qualiry and Quanriry, I S . 225-247. Turner, B.A. (1983). The use of grounded theory for the qualitative analysis of organizational behaviour. Journal of Manogemenr Studies, 20. 333-348. Uttal. B. (1983. October 17). The corporate culture vultures. Fortune. pp. 66-72. Whiddon, B. (1982). The eflecr ojcongruence on the relorionships berw~eenporriciporion/job discretion and srofl perJormance. Unpublished doctoral dissertation. the Florida State University. Tallahassee.