articles grounded theory: a promising · pdf filearticles grounded theory: a promising...

ARTICLES

GROUNDED THEORY: A PROMISING APPROACH TOCONCEPTUALIZATION IN PSYCHOLOGY?

DAVID L. RENNIEYork University

JEFFREY R. PHILLIPSNorth Bay Psychiatric Hospital

GEORGIA K. QUARTAROGeorge Brown College

ABSTRACTThere is a growing body of opinion that psychology suffers from an elaborate researchtechnology that overemphasizes theory verification and impairs thinking and discovery.Grounded theory is advanced as an approach to research that can address this crisis ofmethod in psychology. The grounded approach is described and illustrated in terms ofits application to psychotherapy process research. The emphasis on theory creation charac-terizing the approach is examined within the history of induction. The challenges to andlimitations of grounded theory are discussed.

There are growing indications that psychologyas a discipline is undergoing a crisis of confi-dence about its research methods. These rum-blings of discontent were felt two decades agowhen Bakan (1967) characterized psychologistsas playing at science in much the way childrenplay at being cowboys, imitating every aspect ofcowboy life except the one which is most cen-tral — taking care of cows. In much the sameway, Bakan contended, psychologists have beenconcerned with developing an elaborate researchtechnology while ignoring the main work ofscience — thinking and discovery.

The fault does not lie entirely with contem-porary psychologists. It has been pointed out thatpsychology imitated physics with its emphasis onhypothetico-deductive research while inretrospect the fledgling science might have foundastronomy or zoology, with their emphasis on

Research leading to this paper was supported by SocialSciences and Humanities Research Council of Canada Grants451-83-3642 and 410-83-1264 to David Rennie.

We gratefully acknowledge the assistance of D. Bakan,G. Cupchik, K. Danziger, R. Goranson, L. Heshusius,R. Kroger, A. Mahrer, J. Martin, S. Mor, A. Shaul, I. Sil-verman, J. Weiser, M. Westcott, and L. Wood who com-mented on earlier drafts of the paper.

Reprint requests may be sent to David L. Rennie, Depart-ment of Psychology, York University, North York, Ontario,M3J 1P3.

description, more appropriate models (Endler,1984; Kendler, 1986). As a result, the ensuingtechnology has increasingly dictated the kinds ofstudies that are done and has made theorizing lesscommon and less respectable than energetic buttrivial research (Bakan, 1967; Brandt, 1982; En-dler, 1984; Gergen, 1982; Koch, 1981; Secord,1982; Silverman, 1977). Many theories that havebeen developed have been tested by disjointedinvestigations of irrelevant hypotheses (Arthur,1983). Even when the hypotheses of studies havebeen more closely tied to theories, testing hasbeen contaminated by research participantsresponding to their own inner hypotheses aboutthe experiments in which they are engaged(Orne, 1962; Silverman, 1977). Furthermore,speculation that investigators themselves are in-capable of shedding biases despite the mostrigorous experimental approaches (Kuhn, 1970;Polanyi, 1969) has received empirical support(Pyke, 1982).

Adair (1981) proposes that six research modelshave been developed in response to criticism oflaboratory research. At one end of this continu-um are supporters of laboratory experimentationeither in its present form (e.g., Kruglanski, 1973)or as a buttressed approach (e.g., Orne, 1973;Rosenthal, 1967). At the other extreme are skep-tics who advocate either temporarily abandoning

Canadian Psychology/Psychologie Canadienne, 1988, 29:2 139

140 Canadian Psychology/Psychologie Canadienne, 1988, 29:2

the laboratory experiment (e.g., Barker, 1965;Harre & Secord, 1972; Silverman, 1977;) orreplacing it with observational, humanisticresearch strategies (e.g., Giorgi, 1970).

There is another approach which falls towardthe observational, humanistic end of this continu-um and which has relevance for research inpsychology. The grounded theory method(Glaser, 1978; Glaser & Strauss, 1967) providesan opportunity to create theory in subject areasthat are difficult to access with traditionalresearch methods. Glaser and Strauss developedthis approach in response to a perceived crisisin theory development in sociology. In theirview, sociology suffers from deductive, arm-chair theorizing in which data are forced to fittheories in lieu of a more appropriate generationof theories from data. As an antidote, theyworked out a systematic, inductive approach totheory building that has shown promise insociology and education. However, regionalboundaries separating disciplines have occludedpsychologists' awareness of grounded theory.

We became introduced to grounded theory inour search for a way to understand psycho-therapy process from the client's point of view.As we surveyed the literature, we found thatattempts to portray therapy from the client's per-spective mainly entailed theory-driven, quantita-tive approaches entailing instrumentation (e.g.,Elliot, 1984; Hill, Helms, Spiegel, & Tichenor,in press; Orlinsky & Howard, 1975; Strupp,Walloch & Wogan, 1964), content analyses ofclients' accounts of the psychotherapy experience(e.g., Mayer &Timms, 1970; Oldfield, 1983),and case studies (e.g., Rogers, 1951; Yalom &Elkin, 1974). With few exceptions (e.g., Fessler,1978), there had been no attempt to systemati-cally contend with a hermeneutic analysis ofclients' verbatim accounts. Yet we felt that thisapproach would be required to understand theclient's perspective in a way which is relativelyuncontaminated by theory derived from thetherapist's perspective, given that most psycho-therapy researchers are psychotherapists.

Having come from the hypothetico-deductiveresearch tradition, we were apprehensive aboutshifting to a qualitative approach. After consider-ing the approach outlined by Bogdan (Bogdan &Biklen, 1982; Bogdan & Taylor, 1975), theAmerican phenomenological approach (e.g.,Giorgi, 1970; Keen, 1975), and the British newparadigm research (e.g., Harre & Secord, 1972;Reason & Rowan, 1981), we chose the grounded

theory method. We were attracted to it becauseit seemed relatively more systematic than any ofthe other approaches and because, compared tothe phenomenological and new paradigmapproaches, it places less emphasis on the roleof the researcher in co-constructing the respon-dent's accounts. We viewed the reducedemphasis on co-construction as being two-edged:it meant that a grounded analysis would be rela-tively less intense but, as a compensation, wouldbe a way of studying a relatively larger numberof individuals. We were swayed by the lastfeature because of its implication for generaliz-ability.

In adopting the new method, we have had toabandon most of the canons of the hypothetico-deductive approach, particularly random assign-ment, the use of large numbers of participants,hypothesis testing, experimental control, andstatistics. This abandonment led us into what wenow refer to as "the period of darkness" whereinwe were not sure that what we contemplateddoing was credible, and were tormented with adread that, even if we felt that the returns fromthe new approach were credible, peers within ourdiscipline would rule otherwise.

To a certain extent we have now run this gaunt-let and have emerged with the conviction thatthere is a place for this approach within psycho-therapy process research. Encouraged by thereturns in our own field of endeavour, and but-tressed by successful application of the approachin related disciplines, we raise the possibility thatit holds promise for psychology as a whole.

The purpose of this paper is two-fold. The firstgoal is to depict an implementation of thegrounded approach by drawing upon our ownwork. This objective is addressed in the mainbody of the paper, wherein we provide an over-view of the grounded approach and then depictwhat Glaser and Strauss refer to as the "cons-tant comparison method." The second goal is toconsider the main issues pertaining to groundedtheory, at least as we see them. Four issues areraised in the Discussion section: philosophicaldifferences about the virtues of theory genera-tion versus theory verification in induction; andcredibility of the approach in the light ofresearcher subjectivity, the use of verbal reportsas data, and generalizabiliry of findings. It is ourhope that, in describing the approach and inaddressing contentions surrounding it, we willprovide a bridge for psychological investigators

Grounded Theory 141

who are contemplating a transition from a quan-titative to a qualitative approach to inquiry.

Description of the Approach

Overview

The emphasis in the grounded theory approachis on the generation of theory through the induc-tive examination of information. This emphasisis contrasted with the more traditional approachin sociology (and psychology) of using informa-tion to verify existing theory. This is not to saythat verificational activities hold no place in thegrounded approach. M. H. Marx (1963) advo-cates a "functional" interplay between inductionand deduction in social science. In a similar vein,grounded theorists may choose to test a generatedtheory through traditional research techniques.However, this use of deduction holds a minorplace in the approach. Furthermore, although themethod can be applied to quantitative informa-tion, it has usually been used to analyse qualita-tive data such as interview protocols, records,or bibliographic material.

As inductivists, grounded researchers arefaced with a paradox. They attempt to rid them-selves of preconceptions about the phenomenonunder investigation so that its "true" nature willbe allowed to emerge in the analysis. At thesame time, they believe that this Husserlian"phenomenological reduction" (see Giorgi,1970; Spiegelberg, 1972) can never be achieved.This paradox has both practical and philosophicalimplications.

In terms of practice, grounded researchersgenerally avoid reading pertinent literature untilthe investigation is finished and their groundedtheories are in place. Once a theory is set,existing theories are evaluated and those that fitthe grounded theory are integrated with it so thatit may be further enriched and elaborated. Asecond practical implication is that researchersattempt to identify and record their biases atevery stage of the development of a groundedtheory. This explicit acknowledgement ofassumptions helps to contain their influence.Furthermore, if the biases are indicated in thefinal write-up of the grounded theory, readersof the report are able to evaluate it more effec-tively.

Philosophically, it is recognized that theresearcher is a mediator of the phenomenonunder investigation and that different

investigators might develop somewhat differentviews of the same phenomenon, each of whichmay be credible within its own limits. Thegrounded approach forces investigators to stayclose to their data, so that somewhat differenttheories arising from the same data are the resultof the different analysts emphasing differentaspects of them. Hence, the reactive impact thatinvestigators have upon their data bears more onthe scope than on the credibility of an emergingtheory. The technique that forces investigatorsto stay close to their data, and which constitutesthe systematization of the approach, is the cons-tant comparative method.

The Constant Comparative Method

Glaser and Strauss (1967) maintain that thereare three general approaches to qualitativeresearch. In the first approach, investigators codedata and crudely quantify the codes in an attemptto prove a theoretical proposition1. In thesecond approach, researchers are interested onlyin creating theories. They inspect data to detectnew properties of theoretical categories. In thethird approach, which entails Glaser's andStrauss's constant comparative method, the firsttwo approaches are combined. Investigators sys-tematically categorize data and limit theorizinguntil patterns in the data emerge from thecategorizing operation. This method requiresdata collection, open categorizing, memoing,moving toward parsimony through the determi-nation of a core category, recycling of earliersteps in terms of the core category, sorting ofmemos, and the write-up of the theory in termsof the picture arrived at through the last step(Glaser, 1978).

In the following description of the method, weshall be deriving illustrations from our researchon the client's experience of psychotherapy.

Data Collection. One of the main features ofthe grounded approach is that data collection is

'Glaser and Strauss use the terms "codes", "coding" and"categories". We feel that the term "coding" connotesa transformation of information and, in research, properlydescribes the operation of assigning qualitative material tocategories as part of a content analysis. In this procedure,the qualitative features of an item are lost once it is assignedto a category, since it then becomes a frequency count. Asshall become apparent, the grounded method involves theassignment of material to categories wherein the qualitativefeatures of the material are preserved. For this operation,we prefer the term "categorizing."


influenced by the outcomes of the emerging anal-ysis. The collection proceeds through successivestages which are determined by changes incriteria for selecting data sources (e.g., inter-viewees) according to what has been learnedfrom previous data sources. Initially, theresearcher focuses on learning what is central andcrucial to the phenomenon. Hence, participantsare selected who seem likely to represent thephenomenon and who are relatively similar. Thisis done in order to maximize the chances thataspects of the phenomenon will emerge clearly.This approach facilitates the generation ofcategories and consensus about their properties(see below).

Next, the researcher seeks to clarify variabilitywithin this focal area. Variability is defined asthe appearance of respondents' attributes that arepotential qualifiers of the emerging theory. Inorder to examine the impact of such attributeson the theory, data sources representing theattributes are selected. This selection requiresone or more comparison groups. For example,in a study of the experience of psychotherapy asrecalled by former clients (Phillips, 1984; 1985),the analysis of the accounts of an initial set ofrespondents gave rise to a tentative theory thatthe primary experience of clients in therapy isone of active self focus. However, these respon-dents had participated in eclectic, humanistictherapy. The generalizability of the model waschallenged by interviewing clients of other typesof therapists, such as those with a behaviouralor psychoanalytic orientation.

The selection of new data sources on the basisof the emerging theory is what Glaser and Straussrefer to as theoretical sampling or, as we preferto call it, theory-based data selection. As anapproach to data acquisition, it is very flexible.When two or more groups are compared in tradi-tional research, attempts are made to hold cons-tant all variables other than those defining thecomparison. However, in the groundedapproach, groups can be compared on the basisof even a single dimension if it is judged to begermane to the emerging theory. As the numberof comparison groups increases, the conditionsand limitations of the theory unfold.

Categorization. The choice of an analytic unitis somewhat arbitrary but, once defined by agiven set of investigators, should be clearly expli-cated and consistently used. For example, Glaser(1978) recommends that interview materialshould be analyzed line by line. However, in our

work we have found that the method is moreworkable if a transcript is broken up intomeaning units of individual concepts conveyedby the interviewees (see below).

The mechanics of the analysis of data may varyamong investigators. Glaser and Strauss indicatethat they like to make their analytic notes on themargins of their research protocols. We preferto condense the contents of meaning units andto record each condensation on an index card.For example, here is a meaning unit derived froma transcript of an inquiry interview with a youngman who had just emerged from a psychotherapysession, therapy tape in hand, and who com-mented on the therapy as the tape was replayed.

Int. Can you recall what was going onbetween you and your therapist duringthis part of the interview?

Cl. Well, he was just listening intently. Hewasn't commenting very much. I neverreally let him. Sometimes I interrupt himall the time because I have more to add,you know.

The meaning unit was condensed in two steps.The first reduction was as follows:

C (client) reported that T (therapist) waslistening intently and wasn't commenting toomuch because C wouldn't let him. C wouldinterrupt because he had more to add.

This reduction was further rendered to a sum-mary that could be typed on one line on anotherindex card:

T listened intently; C interrupted; more to add.Units of analysis are sorted into clusters on the

basis of the meaning embedded in the items. Themeaning tying items to a given cluster is lexi-cally symbolized, and this symbol constitutes acategory. For example, here are a number ofsecond-order reductions of meaning units, der-ived from the responses of several interviewees,that have the same general meaning as thatembedded in our first example above.

An articulating word by T helps C continuewith thought.T's metaphor disrupted C's train of thought(Interviewee B).T's self-disclosure did not throw C off(Interviewee C).T's security enables C to correct him toretrack.T's leading is important when C needs it; ithits the main focus (Interviewee D).When T is off track, C feels C is not com-municating very well.

Grounded Theory 143

Although it doesn't seem like it, T must knowthat C is moving toward something(Interviewee E).

For the early stage in the analysis, it is recom-mended that category generation be descriptive,so that the name of the category closely reflectsthe language used by the respondents. Thisprocedure serves as a check against straying fromthe substance of the data. For example,the meaning units above are assignable to adescriptive category labelled The Client's Track.This category is descriptive since many inter-viewees used words and phrases like "retrack"and "offtrack."2

As categories continue to emerge, units ofanalysis are compared to each category. Eachunit is assigned to as many categories as possible.If no categories fit a given unit, a new categoryis developed to represent it. The assignment ofa given unit to as many categories as possibleis referred to as open categorizing. This tech-nique separates the grounded method from con-tent analysis since, in the latter approach, a givendatum is assigned to only one category. As anexample of open categorizing, some of the aboveincidents that were first assigned to The Client'sTrack subsequently were assigned to othercategories as follows:

"C would interrupt T because C had more toadd" was also assigned to Sidetracking theTherapist;

"Talking is like daydreaming — make someimportant discoveries" was additionally assignedto Client Evaluates Own Processes and toClient's Sharpening of Preconscious Awareness;and

"Made connections alone but it was not sopleasant" was categorized under Client's Shar-pening of Preconscious Awareness and Relation-ship with Therapist.

Continuing with the example, units that wereinitially assigned to other categories were sub-sequently assigned to The Client's Track:

' 'C looks for signs from T that C is on the righttrack" had been assigned first to Faith in theTherapist;

"C sidetracked T by asking her about herbandaged finger" had been categorized firstunder Client Sidetracking the Therapist; and

"Rude to comment when not getting any-where; there's more to it than that" initially hadbeen placed under Fear of Criticizing theTherapist.

Open categorizing permits the researcher to

preserve subtle nuances of the data and suppliesthe groundwork for the development of richtheory.

A number of commentators have indicated thatqualitative researchers should actively interprettheir findings. They should not merely generatedescriptive categories but should also constructcategories that help to explain the descriptivecategories and the relationships among them (seeCollaizi, 1979; Sullivan, 1984). Glaser andStrauss subscribe to this viewpoint. A groundedtheory is typically a blend of descriptive and con-structed categories with the former often sub-sumable under the latter. As an example of aconstructed category, the first meaning unitabove ("C interrupted the T") is also pertinentto the category Client Defensiveness. This is aconstructed category since the intervieweesseldom used the term "defensive" whendescribing their interactions with their therapists,yet the concept seemed to capture certain aspectsof the interactions.

As the analysis proceeds, it becomes clear thatsome categories are defining characteristics orproperties of other categories. It also becomesclear that the categories saturate, which meansthat the analysis of additional protocols revealsno new categories, properties, or relationshipsamong them. Saturation often occurs after theanalysis of 5 to 10 protocols (Conrad, 1978;Glaser & Strauss, 1967; Jones, 1980; Pen-nington, 1983; Phillips, 1984; Quartaro, 1985;Rennie, 1984).

Researchers using the grounded method mustcategorize their own data which means that theapproach is labour intensive, particularly duringthe early stages of an analysis. As categories andtheir properties begin to saturate, the categori-zation of new data quickens. Independentcategorizing by research collaborators can beuseful as a check on the perceptual field of theprimary investigator.2Our illustration of categorizing makes it appear thatthe actual assignment of categories is based on a scrutinyof the second-level reductions. They are in tact based on carefulstudy of the first-level reductions. This is an important point,since the first-level reductions have fuller meaning than do thesecond-level reductions. However, we find that, once thecategorizing decisions have been made on thebasis of analysis of the first-level reductions, the second-level reductions reintegrate our memories of themeanings contained in the first-level reductions (andhence in the transcripts). The advantage in working withthe second-level reductions is that they can be easilyarrayed. The arrays give compact portrayals of the meaningsembedded within categories, and of the relationships betweencategories.


Memoing. As analysts collect and analyze data,ideas inevitably occur about the data, thecategories and the theory. In the groundedapproach, these ideas are systematically recordedin the form of memos. For example,

Jan 5/84. I like the category Politics of theRelationship but it is as yet not part of the tax-onomy. I'm holding it in reserve. Most of itsproperties are embedded in Criticism of T byC. I'm reluctant to quickly adopt the Politicstitle because the term implies transaction andI don't have the therapist's side. But I do findthe title appealing. "Game" is anotherpossibility.

These memos have several functions. They helpthe analyst to obtain insight into tacit, guidingassumptions. They raise the conceptual level ofthe research by encouraging the analyst to thinkbeyond single incidents to themes and patternsin the data. They capture speculations about theproperties of categories, or relationships amongcategories, or possible criteria for the selectionof further data sources. They enable theresearcher to preserve ideas that have potentialvalue but which may be premature. They areuseful if gaps in the relation of theory to dataarise, for they provide a record of theresearcher's ideas about the analysis and can beused to trace the development of a category.They are used to note thoughts about thesimilarity of the emerging theory to establishedtheories or concepts. Finally, as shall be seenbelow, they play a key role in the write-up ofthe theory.

The Movement Toward Parsimony. After satu-ration has been achieved, the researcher's focusshifts to the relationships among categories.Some categories are deemed central becausethey have links with many other categories asa result of the multiple categorization of items(as illustrated above). It becomes apparent thatdie network of linked categories forms a hier-archical structure in which central categoriessubsume lower-order categories. Depending onthe conceptual nature of the categories, thestructure can have several levels, with thecategories in each level serving as propertiesof the category or categories at the next highestlevel. At this stage in the analysis, judge-ments about the pertinence of categories aremade on the basis of the extent to which theycontribute to the emerging structure. Categoriesthat have few connections with it are eitherdropped or collapsed into other categories. For

example, one respondent interviewed by thefirst author had a therapist with a gestaltorientation, and another participant's therapisthad a transactional analytic orientation. In theirinquiries, both of these clients commented onthe therapist's orientation to such an extentthat the categories of Gestalt Techniques andTransactional Analytical Techniques weredemanded by the data. However, as the studycame to a conclusion, the endorsements of thesecategories were localized in the two clientsgiving rise to them. Consequently, bothcategories were pooled into the categoryTherapist Directiveness.

As part of the process of developing centralcategories, effort is directed toward determiningthe most central, or core category. This isthe category that is most densely related tothe other categories and their properties. It istypically an abstract category but it is not vague.It is clearly defined by its properties, which arethe categories it subsumes. It is sensitive to newinformation in the analysis because it isassociated with many other categories. It is thusdie last category to saturate and usually emergeslate in the analysis.

Turning to our work to illustrate corecategories, Phillips's (1984; 1985) Self Focuscategory captured much of the meaningexpressed in the accounts of former clients asthey reflected upon the experience of psycho-therapy. This is a rich category, understandablein terms of six (constructed) underlying processesserving as its properties. These properties arein turn supported by a number of descriptivecategories pertaining to the therapy experience.Secondly, in analyzing clients' comments onthe playback of their own psychotherapy audio-tapes, Rennie (1984; 1985a,b; 1987) foundthat the clients: often had private reactions totherapist operations; typically metacognitivelyappraised their own affective and cognitiveprocesses, again in privacy; and often covertlyengaged in executive manoeuvres in order toinfluence the course of the therapy and therelationship with the therapist. These privateprocesses frequently constituted the realtherapeutic work, as phenomenologicallyexperienced. This finding led to the developmentof a core category called Client Agency, sup-ported by many other categories on severallevels. For example, one property of Agency isClient Feeds the Process. This property in turnis supported by The Client's Track, Client

Grounded Theory 145

Narration/Regurgitation, Confessions by Client,Discriminating Use of the Therapist, UnfinishedBusiness, and Client Evaluates Own Processes.Another property of Agency is Client Deference.This property has its own properties consistingof, for example, Concern About the Therapist'sApproach, Client Understanding the Therapist'sFrame of Reference, and Fear of Criticizing theTherapist. In this analysis, Client Agency has 8properties which in turn have a total of 37 sub-properties. The main feature of this model of theclient's experience of psychotherapy is that itbrings the role of metacognition into sharperrelief when compared to theory-driven modelsderived from the hypothetico-deductive approachto psychotherapy process research.

Glaser (1978) suggests that should more thanone highly central category arise, effort shouldbe made to determine if only one of them is acore category to which the others are sub-sumable. Failing this, he recommends that thenonsubsumable categories be split off from theanalysis and reserved for additional groundedtheories, lest the analysis becomes overlycomplex and burdensome.

Writing the Theory. In the grounded approach,the main repository of the analyst's interpretiveactivity is the research memo. It is the memo thatenables the investigator to record ideas aboutpotential central categories and about relation-ships among categories. This conceptual materialis the basis of the grounded theory. During thisadvanced stage of the analysis, the researchmemos are sorted, and new memos are createdin response to the insights and speculationsproduced by initial memo sorts. Additionalmemoing contributes to the generation of the corecategory and the specification of the structure ofits properties and the relationships among them.This meaning system provides the organizationalstructure for the write-up of the theory.

There are four criteria for a grounded theory.It should be believable in that it should seem tothe reader to be a plausible explanation. It shouldbe adequate in that it should present a compre-hensive account that does not omit large orimportant portions of the data. It should begrounded in terms of the appropriate proceduresand thereby inductively tied to the data. Finally,it should be applicable and should lead tohypotheses and additional investigation (Glaser,1978).

DiscussionThe grounded theory approach is a systematic

and potentially powerful strategy for theorydevelopment. It provides investigators with aheady freedom in exploring complex pheno-mena. Having the method in hand is like carryinga flashlight that can be beamed on any aspect ofa cluttered attic. In principle, anything can beaddressed so long as it is represented in colliga-tive symbols. The method can illuminate aspectsof human existence as varied as organizationalcommunication (Browning, 1978); marital dis-solution (Laner, 1978); the ways of being ofcancer patients who outlived their life expectancy(Pennington, 1983); the experience of gettingsettled in a new city (Jones, 1980); the processesof academic change (Conrad, 1978; Newcombe& Conrad, 1981); graduate students' subjectiveexperience of being blocked when attempting todo a thesis (Rennie & Brewer, 1987); and, aswe have seen, the client's subjective experienceof psychotherapy.

In making the transition from a quantitative,hypothetico-deductive approach to a qualitative,grounded one we have undergone a transforma-tion of our relationship to the research process.We do not seem to mind how much time this typeof research requires because of the saliency ofthe data. We have the sense that we are gettingclose to the bone and in conference presentationswe are getting feedback that reinforces thatimpression. Furthermore, when we attend to thework of other qualitative researchers, we feel asimilar sense of cogency as we relate to theirfindings, and encounter an excitement about thisapproach to inquiry that matches our own.

At the same time, the approach is contentious(Brown, 1973; Layder, 1982). In our view, thereare four main issues: the place of theory-generation as opposed to theory-verification ininduction; and the credibility of the groundedapproach in terms of the researcher subjectivityit entails, in terms of its utilization of verbalreports as data, and in terms of the generaliza-bility of findings. Let us look at these issues inturn.

Contentions About theGrounded Theory Approach

Philosophical Differences about Theory-Generation versus Theory-Verification. Inductionis generally characterized as the formulation ofgeneral laws from particular instances. In ahistory that extends to Aristotle, there has beencontroversy about how such propositions areformed. There has also been debate on whether


it is the development of propositions or the proofof them that should be given primary importance.For example, Whewell (1847/1967; 1860/1971)maintains that propositions are formed bybringing imagination, or "conceptions" to bearon facts. The exact processes of this creativeinferencing are poorly understood. Nevertheless,the inferencing is vital because without it theinvestigator is awash in a sea of facts. It is thispower of conceptions to colligate facts that leadsWhewell to conclude that the formation of theoryis a more critical aspect of induction than is theproof of it.

J.S. Mill (1872/1973-1974) argues withWhewell on both counts. Mill grants that colli-gation of facts is an essential part of induction,and he agrees that conceptualization in inductionis "skillful guessing." However, he asserts thata conception is something seen in the facts, notsomething added to the facts. It is perhaps thisdownplay of the role of creative inferencing ininduction that leads Mill to conclude that theproof of inductive propositions is more essen-tial than their generation.

The grounded approach is more in keepingwith the philosophy of Whewell than of Mill. Itthus runs counter to prevailing sentiment inpsychology about the proper praxis of induction.One of the virtues of the hypothetico-deductiveapproach is that products of its use have the trap-pings of scientific credibility. The groundedapproach is less scientistic. Here theories aredeveloped through the use of conceptualizationto colligate facts, rather than through a recur-ring cycle of inferencing and hypothesis testing.Exemption from the constraint of the verifica-tional approach frees grounded researchers toaddress highly complex meaning structures thatmight otherwise be beyond the pale. However,the release is won at the expense of a numberof threats to the credibility of the approach.These threats bear both on the intrinsic worth ofa grounded theory and on its generalizability.

Threats to Credibility. Subjectivity: Agrounded theory wins its credibility through itspersuasiveness. This in turn is a function of theextent to which the inferencing processes of thegrounded analyst have been demystified. It isalso a function of the extent to which thecategories underpinning the theory have beendocumented. There are forces at play thatthreaten sufficient realization of both of theserequirements.

It is difficult for grounded researchers to fully

realize the goal of identifying and disclosing then-implicit assumptions. The norm of total objec-tivity, one of the legacies from positivism, is stilleffective. Grounded theorists receive little com-fort from the attention that has been drawn tothe role of subjective factors in traditionalresearch approaches (e.g., Kuhn, 1970; Polanyi,1969; Pyke, 1982) in light of the dearth ofevidence that such criticisms have influenced theways in which scientific results are presented inmainstream publications. Consequently, it isdifficult to be candid about the subjective factorsthat were at work as a given grounded theory wasformulated. There are fears that the opendisclosure of such factors will impede acceptanceof the research effort by the scientificcommunity.

Editorial constraints on the length of publish-able articles impose limits on the extent to whichgrounded researchers can document theircategories. Various solutions to this problemhave been suggested. Glaser and Strauss claimthat grounded researchers need not dwell ondocumentation when writing up their theories,because categories emerging from trulygrounded analyses will resonate with the readerand will not require much illustration. J. Shorter(personal communication, October 14, 1983)maintains that the detailed presentation of arepresentative example from each category is onesolution. In line with Snorter's suggestion,Rennie and Brewer (1987) gained editorialapproval by giving one example per category foreach of two comparison groups.

Hence, candor about the reactive impact of theinvestigator on the investigation is constrainedby the climate of positivism in psychologicalresearch. However, there is a change in thewind. New humanistic and interdisciplinarysocieties (e.g., the International Human ScienceResearch Conference) and new journals (e.g.,The Humanistic Psychologist, Phenomenologyand Pedagogy, and Methods) are being formedwhich encourage such openness.

Verbal Reports as Data: Another area of threatto the intrinsic worth of grounded theoriesderives from the fact that the grounded analysisof accounts typically entails the use of verbalreports as data. This means that groundedresearchers may not gain access to internalprocesses of which participants are unaware. Theinvestigators may also be misled by participantswho misrepresent processes of which they areaware. Nisbett and Wilson (1977) adduce

Grounded Theory 147

evidence to support their conclusion that, whenattempting to report on cognitive processes,individuals do not truly introspect but insteaddraw upon a priori, implicit causal theories aboutthe extent to which a given stimulus is a plau-sible cause of a given response. This claim hasgiven rise to a number of challenges (see Adair& Spinner, 1981; Cotton, 1980; Ericsson &Simon, 1980; Morris, 1981; Rich, 1979; Smith& Miller, 1978; White, 1980). It is beyond thescope of this article to review all of the opposingarguments. However, we would like to brieflysummarize the points made by Ericsson andSimon (1980) and Morris (1981), which areespecially compelling.

Ericsson and Simon examine verbal reports asdata in terms of a model derived from informa-tion theory and they cite evidence supportingtheir prediction that verbal reports can beaccurate if they address processes that areactually attended to and registered in short-termmemory. Along similar lines, Morris (1981)takes the logical position that verbal reports candraw only on conscious material, and that whatis conscious is intentionality.

The existence of levels of cognitive processingrequires grounded researchers, who rely onverbal accounts as data sources, to carefullyconsider which levels are appropriate for verbalreport. It also behooves them to use techniquesthat facilitate accurate reporting whenappropriate levels are being tapped. Meichen-baum and Butler (1979) maintain that improve-ments over the typical post-experiment inquirycan be made in the form of intensive interviewsand videotape recall. In using these techniques,we have found that interviewees can subjectivelyjudge the extent to which they are either remem-bering a past event or constructing it (see Spence[1982] for a discussion of this distinction inthe context of psychoanalysis). We have addi-tionally found that participants can derive sub-jective estimates of the extent to which an internalattribution is felt to be true. The absence ofexternal criteria makes it impossible to validatethe truth value of individual verbal reports thatare evaluated through these procedures.However, the use of the constant comparativemethod to demonstrate that different individualssay the same thing increases the credibility ofindividual accounts.

Generalizability: The fact that grounded ana-lyses are typically conducted on a small numberof selected participants is problematic. Ideally,

grounded researchers extend the generalizabilityof their emerging theories by systematicallycomparing a series of contrasting groups,selecting participants from each group until satu-ration ensues. However, the arduous nature ofthis work limits the conduct of extensive com-parisons. Thus the generalizability of groundedtheories is often in question.

The replication of findings across a smallnumber of participants places the groundedapproach between the individual case study andtraditional group approach to psychologicalresearch. Like the case study, it provides theinvestigator with the opportunity to exploresubjective, idiographic events. Unlike the casestudy, it emphasizes the necessity to replicatethe evidence of such events by addressing morethan one individual. In replicating individualfindings across as many people as are necessaryfor emergent categories to saturate, the investi-gator gets a foothold on a commonly experiencedphenomenon. The resultant theory typicallycommunicates to the reader this contact withcommonality. It is intimacy with thephenomenon that grounded theorists seek muchmore than external criteria of adequacy such ashard evidence of generalizability derived froma random sampling of a large number ofindividuals. Once again, the object of theapproach is to create new theory that is directlytied to the reality of individuals. The objectis not to verify the theory so generated beyondthe verification yielded by saturation ofcategories. Additional verification is deliberatelyleft to subsequent studies and/or otherinvestigators.

Summing up this discussion of challenges tothe grounded approach, candor about investi-gator reactivity and full documentation offindings are being facilitated by the developmentof professional societies and editorial policiessympathetic to qualitative research. Scrutiny ofthe usefulness of verbal reports as data indicatesthat they can yield valid information about inten-tionality. Finally, the problem of limited gener-alizability of grounded findings is not resolvedbut is accepted by grounded researchers as alegitimate price to pay for research that is inti-mately tied to the phenomena it addresses.

The approach yields access to aspects ofhuman experience which are difficult, if notimpossible, to address with traditionalapproaches to psychological research yet areinherent in the subject matter of psychology. We


have found that the hermeneutic analysis ofpsychotherapy clients' accounts has exposed thecovert worlds of clients as they participate intherapy and has demonstrated the extent to whichthey are metacognitively aware and active withinthat privacy. We are accordingly developingmodels of this dimension of the client'sexperience of therapy which complementexisting models derived from alternative researchapproaches.

In our view, there is nothing special aboutpsychotherapy clients' accounts which makesthem suitable for a grounded analysis. As indi-cated above, in principle any type of verbalreport is appropriate for this research approach,which opens it up to a vast domain of subjectmatter. It is our conclusion that the groundedtheory approach does hold promise as a usefulresearch strategy that could be broadly appliedwithin the discipline of psychology.

RESUMEOn avance de plus en plus l'opinion que la psychologie est la proie d'une technologie derecherche eiaboree qui surestime la verification de la theorie et nuit a la pensee et a ladecouverte. La theorie de base est proposee comme approche a la recherche qui peutresoudre cette crise de methode en psychologie. L'approche de base est decrite et illustreeen termes de son application au processus de la recherche en psychothe'rapie. On met l'accentsur la creation de la theorie caracterisant cette approche qui est examinee a l'interieurde 1'histoire de l'induction. Les defis et les possibility de la theorie de base sont discutees.

ReferencesAdair, J.G. (1981, June). The laboratory experiment: After

the paradigm crisis and 20 years of research on the socialpsychology of the psychological experiment. Paperpresented at the Interamerican Congress of Psychology,Santa Domingo Republic.

Adair, ]., & Spinner, B. (1981). Subjects' access to cogni-tive process: Demand characteristics and verbal report.Journal of the Theory of Social Behaviour, 11, 31-52.

Arthur, A.Z. (1983). A science in need of change. Cana-dian Psychology, 24, 223-224.

Bakan, D. (1967). On method: Toward a reconstruction ofpsychological investigation. San Francisco: Jossey-Bass.

Barker, R. (1965). Explorations in ecological psychology.American Psychologist, 20, 1-14.

Bogdan, R., & Biklen, S.K. (1982). Qualitative research foreducation: An introduction to theory and methods. Boston:Allyn and Bacon.

Bogdan, R., & Taylor, S.J. (19T'5).Introduction to qualita-tive research methods: A phenomenological approach tothe social sciences. Toronto: Wiley.

Brandt, W.L. (1982). Psychologists caught: A psycho-logicof psychology. Toronto: University of Toronto Press.

Brown, G.W. (1973). Some thoughts on grounded theory.Sociology, 7, 1-16.

Browning, L.D. (1978). A grounded organizational commu-nication theory derived from qualitative data. Commu-nication Monographs, 43, 93-109.

Collaizi, P.F. (1979). Psychological research as Ihe phenome-nologist views it. In R.S. Valle & M. King (Eds.).Existential phenomenological alternatives for psychology.New York: Oxford University Press.

Conrad, C.F. (1978). A grounded theory of academic change.Sociology of Education, 51, 101-112.

Cotton, J.L. (1980). Verbal reports on mental processes:Ignoring data for the sake of theory? Personality andSocial Psychology Bulletin, 6, 278-281.

Elliot, R. (1984). A discovery oriented approach to signifi-cant change events in psychotherapy: Interpersonal

process recall and comprehensive process analysis. InL.N. Rice & L.S. Greenberg (Eds.). Patterns of change:Intensive analysis of psychotherapy processipp. 246-286).New York: Guilford.

Endler, N.E. (1984). To know or not to know: That is themethod. Department of Psychology Report No. 142, YorkUniversity, North York, Ontario, M3J 1P3.

Ericsson, K.A., & Simon, H.A. (1980). Verbal reports asdata. Psychological Reviews, 87, 215-251.

Fessler, R. (1978). A phenomenological investigation of psy-chotherapeutic interpretation. Dissertation Abstracts Inter-national, 39, 2981B.

Gergen, K. (1982). Toward a transformation in socialknowledge. New York: Springer-Verlage.

Giorgi, A. (1970). Psychology as a human science: AphenomenologicaUy based approach. New York: Harper& Row.

Glaser, B.G. (1978). Theoretical sensitivity: Advances in themethodology of grounded theory. Mill Valley, CA: TheSociology Press.

Glaser, B.G., & Strauss, A. (1967). The discovery ofgrounded theory: Strategies for qualitative research.Chicago: Aldine.

Harrg, R., & Secord, P.F. (1972). The explanation of socialbehaviour. Oxford: Blackwell.

Hill, C , Helms, J.E., Spiegel, S.B. & Tichenor, V. (inpress). Development of a system for categorizing clientreactions to therapist interventions. Journal of CounselingPsychology.

Jones, W.L. (1980). Newcomers' biographical explanations:The self as an adjustment process. Symbolic Interaction,3, 83-94.

Keen, E. (1975). A primer in phenomenological psychology.Washington, DC: University Press of America.

Kendler, H.H. (1986). Interview of Howard H. Kendler. InB.J. Baars, The cognitive revolution in psychology (pp.111-122). New York: Guilford.

Koch, S. (1981). The nature and limits of psychologicalknowledge. American Psychologist, 36, 257-267.

Grounded Theory 149

Kruglanski, A.W. (1973). The human subject in the psy-chology experiment: Fact and Artifact. In L. Berkowitz(Ed.), Advances in Experimental and Social Psychology,8, 101-147.

Kuhn, T. (1970). The structure of scientific revolutions(2nd ed.). Chicago: University of ChicagoPress.

Laner, M.R. (1978). Love's labors lost: A theory of mar-ital dissolution. Journal of Divorce, 1, 213-232.

Layder, D. (1982). Grounded theory: A constructive cri-tique. Journal for the Theory of Social Behavior, 12,103-123.

Marx, M.H. (1963). The general nature of theory construc-tion. In M.H. Marx (Ed.), Theories in contemporary psy-chology (pp. 4-46). New York: Collier-Macmillan.

Mayer, J.E., & Timms, N. (1973). The client speaks:Working impressions of casework. London: Routledge,& Kegan Paul.

Meichenbaum, D., & Butler, L. (1979). Cognitive ethology:Assessing the streams of cognition and emotion. In K.Blankstein, P. Pliner, and J. Polivy (Eds.), Advances inthe study of communication and behavior, (Vol. 6). NewYork: Plenum Press.

Mill, J.S. (1973-1974). A system of logic: Ratiocinative andinductive; being a connected view of the principles ofevidence and methods of scientific investigation. In J.M.Robson (Ed.), The collected works of John Stuart Mill,(Vols. 7-8). Toronto: University of Toronto Press,(Original work published 1872).

Morris, P. (1981). The cognitive psychology of self-reports.In C. Antaki (Ed.), The psychology of ordinary expla-nations of social behaviour. London: Academic Press.

Newcombe, J.P., & Conrad, C.P. (1981). A theory of man-dated academic change. Journal of Higher Education, 52,555-577.

Nisbett, R.E., & Wilson, T.D. (1977). Telling more thanwe know: Verbal reports on mental processes. Psycho-logical Review, 84, 231-259.

Oldfield, S. (1983). The counselling relationship: A studyof the client's experience. London: Routledge, & KeganPaul.

Orlinsky, D.E., & Howard, K.I. (1975). Varieties ofpsy-chotherapeutic experience: Multivariate analyses ofpatients' and therapists' reports. New York: Teacher'sCollege Press.

One, M.T. (1962). On the social psychology of the psy-chological experiment: With particular reference todemand characteristics. American Psychologist, 17,776-783.

Orne, M.T. (1973). Communication by the total experimentalsituation: Why it is important, how it is evaluated, andits significance for the ecological validity of findings. InP. Pliner, L. Krames, & T. Alloway (Eds.), Communi-cation and affect. New York: Academic Press.

Pennington, S. (1983). Living or dying: An investigation ofthe balance point. Dissertation Abstracts International,44, 441A.

Phillips, J.R. (1984). "Influences on personal growth asviewed by former psychotherapy patients". DissertationAbstracts International, 46, 2820B.

Phillips, J.R. (1985, May). The influence on personal growthas viewed by former psychotherapy patients. In D. Rennie(Chair). The phenomenological experience of psycholog-ical treatment. Symposium conducted at the annualmeeting of the International Human Science ResearchConference, Edmonton, Alberta.

Polanyi, M. (1969). Knowing and being. Chicago: Univer-sity of Chicago Press.

Pyke, S.W. (1982). Confessions of a reluctant ideologist.Canadian Psychology, 23, 125-134.

Quartaro, G. (1985, May). Using psychology self-help books:A grounded theory perspective. In D. Rennie (Chair).The phenomenological experience of psychological treat-ment. Symposium conducted at the annual meeting of theInternational Human Science Research Conference,Edmonton, Alberta.

Reason, P., & Rowan, J. (Eds.), (1981). Human inquiry:A sourcebook of new paradigm research. Toronto: JohnWiley & Sons.

Rennie, D.L. (1984, May). Clients' tape-assisted recall ofpsychotherapy: A qualitative analysis. In D. Rennie(Chair). Recent advances in psychotherapy research: Theexperience of the client. Symposium conducted at theannual meeting of the Canadian Psychological Associa-tion, Ottawa, Ontario.

Rennie, D.L. (1985a, February). An early return from inter-views with clients about their therapy interviews: Thefunctions of the narrative. In P. O'Grady (Chair). Recentdevelopments in psychotherapy research: Focus on theclient. Symposium conducted at the annual meeting ofthe Ontario Psychological Association, Ottawa.

Rennie, D.L. (1985b, June). Client deference in the psy-chotherapy relationship. In D. Rennie (Chair). Theclient's experience of psychotherapy. Symposium con-ducted at the annual meeting of the Society for Psy-chotherapy Research, Evanston, Illinois.

Rennie, D.L. (1987, May). A model of the client's experienceof psychotherapy. Presented at the annual meeting of theInternational Human Science Research Conference,Ottawa.

Rennie, D.L., & Brewer, L. (1987). A grounded theory ofthesis blocking. Teaching of Psychology, 14, 10-16.

Rich, M.C. (1979). Verbal reports on mental processes:Issues of accuracy and awareness. Journal for the Theoryof Social Behaviour, 9, 29-37.

Rogers, C.R. (1951). Client-centered therapy: Its currentpractice, implications and theory. Boston: HoughtonMifflin.

Rosenthal, R. (1967). Covert communication and tacit under-standing in the psychological experiment. PsychologicalBulletin, 67, 356-367.

Secord, P. (Ed.). (1982). Explaining human behavior.Beverly Hills: Sage.

Silverman, I. (1977). The human subject in the psycholog-ical laboratory. New York: Pergamon Press.

Smith, E.R., & Miller, F.D. (1978). Limits on perceptionof cognitive processes: A reply to Nisbett arid Wilson.Psychological Review, 85, 355-362.

Spence, D.P. (1982). Narrative truth and historical truth:Meaning and interpretation in psychoanalysis. New York:W.W. Norton.

Spiegelberg, H. (1972). Phenomenology in psychology andpsychiatry. Evanston: University of Chicago Press.

Strupp, H.H., Walloch, M.S., & Wogan, M. (1964). Psy-chotherapy experience in retrospect: Questionnaire surveyof former therapy patients and their therapists. Psycho-logical Monographs, 78, (11, Whole No. 588).

Sullivan, E. (1984). A critical psychology. New York:Plenum Press.

Whewell, W. (1967). Philosophy of the inductive sciences,(Parts 1-2). London: Frank Cass. (Original work pub-lished 1847).


Whewell, W. (1971). On the philosophy of discovery. events: A refutation of Nisbett and Wilson and of Bern.New York: Burt Franklin. (Original work published Psychological Review, 87, 105-112.1860). Yalom, I.D., & Elkin, G. (1974). Every day gels a little

White, P. (1980). Limitations on verbal reports of internal closer. A twice-told therapy. New York: Basic Books.

articles grounded theory: a promising · pdf filearticles grounded theory: a promising...

Documents