Published on

  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

  1. 1. International Journal of Qualitative Methods 5 (2) June 2006Exploiting Exceptions to Enhance InterpretiveQualitative Health Research: Insights from aStudy of Cancer CommunicationGladys McPherson and Sally ThorneGladys McPherson, RN, PhD(c), University of British Columbia,Vancouver, BC, CanadaSally Thorne, RN, PhD, Professor and Director, University of BritishColumbia School of Nursing, CanadaAbstract: Although it has long been understood that a well-constructed dataset ought to be filled with complexities and contradictions, observations thatchallenge or contradict analytic interpretations are not often given suffi-ciently serious attention in the methodological qualitative health literature.When researchers attempt to produce comprehensive or “holistic” findings,they all too often set aside or gloss over the negative cases that fail to con-form to their emerging interpretive generalizations. In this article, the au-thors challenge fellow qualitative health researchers to engage actively inidentifying and exploiting both actual and theoretical exceptions as a valu-able analytic strategy. They argue that heightened sensitivity for negativecases uncovers the assumptive claims deriving from our various method-ological orientations and illuminates alternative explanations. They proposethat thoughtful attention to contradictory or challenging observations candeepen our expectations about the kinds of knowledge products that qualita-tive research ought to yield, thereby helping us advance the credibility of ourfindings and the ultimate utility of our empirical conclusions.Keywords: qualitative methodology, data analysis, outliers, evidence basedpractice, maximal variationCitationMcPherson, G., & Thorne, S. (2006). Exploiting exceptions to enhanceinterpretive qualitative health research: Insights from a study of cancer Authors’ notecommunication. International Journal of Qualitative Methods, 5(2), The research from which this analy-Article 1. Retrieved [date] from sis was drawn was funded by the National Cancer Institute of Canadahttp://www.ualberta.ca/~iiqm/backissues/5_2/pdf/mcpherson.pdf
  2. 2. 2 McPherson, Thorne EXPLOITING EXCEPTIONS Any given finding usually has exceptions. The (Agresti & Finlay, 1997; Bickman, Rog, & Hedrick, temptation is to smooth them over, ignore them, 1998). Because these observations are unlikely to be or explain them away. But the outlier is your correct according to what is known about the phenom- friend. A good look at the exceptions, or the enon from other independent observations, it becomes ends of a distribution, can test and strengthen the critically important to detect them within data sets, basic finding. eliminate them from the statistic analysis, and nullify —M. B. Miles & A. M. Huberman, 1994, p. 270 their impact on the findings (Osborne & Overbay, 2004). Unless one can simply correct for them by col- nherent in qualitative research processes, particu- lecting large quantities of highly redundant data, theI larly as data analysis proceeds and as preliminary thematic groupings are established, certain observa-tions stand out as being at distinct odds with the major- basic tenets of rigor require that researchers ask prob- ing questions of each outlying instance: Is the observa- tion a mistake in measurement or recording?ity of the data. Analytic challenges within some of our Alternatively, is there something else going on thatrecent inquiries have prompted us to reflect on our ap- does not fit the original study hypotheses? To maintainproach to those data that emerge as seeming exceptions integrity within the research processes, each outlieror negative cases. In various ways, such data bits have cannot simply be discarded without investigation ofallowed us to challenge our tentative thematic interpre- the reasons for its appearance, and the results of that in-tations and call into question the assumptions that we vestigation used to determine how that one observationhave inadvertently embedded in the conceptualizations will be processed.we bring into our projects. What has become apparent The problem of data that lie distant from the major-through this reflective process is that the manner in ity of the data has been dealt with within the qualitativewhich we attend to these observations has significant tradition differently from the way in which it has beenimplications for the credibility of our claims and the ul- handled by quantitative researchers. Although it hastimate utility of our findings. well-established roots within the foundational method- In this article, we expand on these reflections to en- ological tradition of qualitative inquiry (Kuzel, 1999;gage the reader in what constitutes an ongoing discus- Lincoln & Guba, 1985; Patton, 1990), the problem ofsion about the inherent value of that which does not findings that stray from the majority of the observa-conform to our neat and tidy thematic descriptions or tions has generally received less active scrutiny in theour interpretive conceptualizations. We decry what qualitative research literature than in the quantitativeseems to be a recent trend within the qualitative health domain. First, the phenomena of concern to qualitativegenre to overlook such inconvenient variations within researchers are likely to be grounded in the socialdata sets and argue instead that deliberate attention to world rather than in the material world and thereforeobservations contrary to current conceptualizations are much more loosely governed by the kinds estab-deepens our analytic capacity and intensifies the even- lished laws, principles, and theories by which one de-tual credibility and utility of all of our research prod- termines the correctness of a data point within theucts. In the case of qualitative health research, in which “basic sciences.” Second, although the quantitativefindings are proposed not simply for the purpose of outlier becomes problematic because of the profoundtheorizing but more typically to inspire enhanced in- influence it can have on the research results when un-sight and sensitivity for the complexity of a clinical detected, the typical qualitative outlier is immediatelyphenomenon, we argue that understanding the role of apparent to the investigator and is subject instead to thenegative or exceptional cases might be of particularly interpretive maneuvers by which it can be written en-pressing consequence. tirely out of the data set. A third reason for which consideration of the impli- THE CONCEPT OF EXCEPTIONS IN cations of exceptional observations might have been QUALITATIVE RESEARCH relatively absent from the recent qualitative method- ological literature, and the issue to which we orient ourResearchers have long grappled with how best to deal attention in this discussion, is a misconception thatwith observations that are distant from the majority of seems well entrenched within the qualitative health re-the data. Within the theoretical tradition of quantitative searcher community, that the researcher “as interpre-research methodology, these observations have been tive instrument” enjoys the unilateral privilege oflabeled outliers. Outliers are considered to be data determining what do and do not constitute data. In thepoints that are at odds with the majority of the very act of constructing (even co-constructing) data indata—observations that might have a significant (and the qualitative context, that which seems meaningfultypically unjustified) influence on statistical results and relevant is illuminated, whereas that which is not isInternational Journal of Qualitative Methods 5 (2) June 2006http://www.ualberta.ca/~ijqm/
  3. 3. McPherson, Thorne EXPLOITING EXCEPTIONS 3obscured. Despite expectations of methodological pre- ple, they might represent plausible instances of con-cision within the early social science traditions, from tradictions to our tentative thematic and theoreticalwhich the majority of our qualitative health research interpretations of the data, or manifestations of im-methods have derived (Caelli, Ray, & Mill, 2003; portant human diversities discrepant from the domi-Johnson, Long, & White, 2001; Thorne, 1991), we nant discourses. They might be alternative forms ofhave not sustained a strong tradition of reporting that the thing we think we are studying, or entirely differ-which was not studied, asked, sought, or recorded, or ent phenomena that have masqueraded as instancesof questioning the inductive processes by which find- of our focus of inquiry. Most important, they mightings are rendered from the total possible theoretical represent an important window into the complexitiescontext in which we inquire. Although we have fo- of not only that which we study but also the worldcused a great deal of attention on locating the theoreti- within we would hope to introduce our research find-cal positioning and potential “bias” of the ings.researcher-as-instrument, we have paid much less at-tention to an intricate analysis of what it is that the re- EXAMPLES OF EXCEPTIONSsearcher holding that positioning might or might not IN A CANCER COMMUNICATION“see.” Thus, from our perspective, it seems worth con- RESEARCH PROJECTsidering the possibility that the manner in which a re-searcher seeks and handles exceptional cases might be Communication in cancer care has become an in-an important element in ascertaining the credibility of creasingly attractive area of substantive inquiry, inintegrative conclusions made on the basis of qualita- that it reflects a high degree of consumer concern andtive findings. becomes a powerful mediator of the cancer experi- When we obtain qualitative data that, at least at first ence. The goal of an ongoing cancer communicationglance, tend to lie far from the majority of observa- study in which we have been engaged (Thorne,tions, in that they seem outside the conceptual catego- Hislop et al, 2006; Thorne, Kuo et al, 2005) is to lookries into which we strive to sort them, we might be below the surface platitudes of popular health com-tempted to consider them as observations whose effect munication theory and search for deeper principlesought to be nullified. Although there might be cases in that might provide explicit guidance to cancer carewhich these observations represent mistakes in record- practitioners. As our analysis unfolded, we organizeding or interpretation, more often such data can provide preliminary findings around three major themes. Theus with unique opportunities to consider our findings first reflected what we had come to understand as afrom different vantage points at each stage of the ana- fundamental and essential element of communica-lytic process. Assuming they find their way into our tion in cancer care: the sense of “being known.” Adata sets, and given due consideration, observations second theme reflected discourses associated withthat appear to us as exceptions might prompt new ave- references to numerical information and statisticalnues of thinking, push our analyses toward more com- representations, in that these become particularly po-plex and sophisticated conceptualizations of the tent communication forms with a particular relation-phenomena in question, or even prompt us to uncover ship both to the dynamics of communicationassumptions that might revise our core understandings encounters and to critical subjective experiencesof that which we are investigating. such as hope. A third theme was related to outcomes, As Miles and Huberman (1994) have pointed out, what individuals understand to be the benefits oroutliers (or exceptions) can take a variety of forms. We harms associated with cancer communication experi-detect them in our data sets as “discrepant cases, atypi- ences. Each theme addressed a topical commonality,cal settings, unique treatments or unusual events” an issue that arose repeatedly among and between(p. 269). Although our natural inclination might be to study participants as critically important aspects ofrespond to these observations as if they were unfortu- the cancer communication experience. On the basisnate contaminants, artifacts of our design decisions, or of these patterns, we hoped to develop cogent empiri-remnants of an imperfect data collection process dis- cally derived communication standards. However,tracting us from a coherent portrait of the phenomenon we also encountered within each theme exceptionalon which we have set our angle of vision, a more care- observations that presented a considerable challengeful examination of the manner in which they manifest to our ability to articulate even preliminary findingsthemselves within our studies might orient us toward with the confidence that we had originally antici-searching for alternative meanings they might conceiv- pated.ably represent (Kuzel, 1999; Patton, 1990). For exam- International Journal of Qualitative Methods 5 (2) June 2006 http://www.ualberta.ca/~ijqm/
  4. 4. 4 McPherson, Thorne EXPLOITING EXCEPTIONS THE ROLE OF HUMAN CONNECTION vision new angles of possibility as we expanded our understanding of this complex notion of what it is toIt seemed evident within our findings that “human con- “be known” within an intensive health and illness en-nection” is a central ingredient in effective communi- counter.cation between cancer patients and their professionalcare providers. “They know me,” or, conversely, “they THE VALUE OF INFORMATIONdon’t know me” represent the most frequently cited ex-planations for why any particular communication epi- With regard to the second theme, it seemed self-evi-sode was or was not deemed effective for that dent within our findings that information access isparticular person at that particular time in the course of highly valued by a majority of cancer patients and thata cancer illness. This pattern was sufficiently strong those who seek high levels of information consider itsthat we were tempted to articulate the behavioral pat- provision a fundamental professional competency. Ourterns and themes within it as the key to invoking the cultural form places much of the information relevantfeeling of “being known” as a requirement across all to cancer care decision making within a statistical con-cancer care. In general, people prefer to be identified as struction (3 out of 4 people diagnosed with your kindindividuals, with some elements of their unique per- of disease will still be alive 5 years from now; you havesonality or preferences entering into the relational con- a 90% chance of cure; if you obtain this treatment, youtext. They seem better able to transcend the technical will have a 5% better chance of delaying recurrenceand power barriers where human touch, nonmedial dia- within 6 months; and so on). However, even amonglogue, and emotion are permitted or even encouraged those articulate and expressive patients who seek highwithin the clinical encounter. However, in working levels of information, we find a significant range of in-with the data, we became acutely aware that each com- terpretation for such numeric information. Informationmon pattern within this phenomenon of wanting to be described as “too much” might reflect errors of quan-known revealed significant variations that proved in- tity, intensity, or valence, and generosity with such in-structive to our analysis. There were people for whom formation might slip into a volume level that istechnical engagement is most assuring, for whom overwhelming or a detail level that becomes inter-touch and emotional expression are disconcerting, and preted as something somewhat sinister, such as profes-for whom getting down to business is far more impor- sional self-protection. What we see, then, is thattant than getting to know one another as individuals. information-seeking as a primary discourse cannot beFurthermore, there are some individuals for whom the distinguished from hope-building if we are to protectsocial, nontechnical, or emotional elements of dis- the rights of those individuals for whom subtle varia-course might be interpreted as meeting the needs of the tions in information transmission might represent theclinician rather than the patient. Taking these excep- difference between hope and despair. Therefore, con-tions into account, we were led to develop the theme of clusions generalizing the phenomenon of informationbeing known in a manner that transcends the access become problematic. In other words, people aredescriptors by which we most typically measure and decidedly different when it comes to a preference forarticulate it. Being known became a highly individual- adopting a positive attitude to guard against negativity,ized, iterative, and interactive phenomenon within or adopting a negative attitude in the hopes that onewhich cues are detected, interpreted, and applied into a will be pleasantly surprised. Because few of us wouldtheoretically infinite range of effective communicative be sufficiently insightful about our own psychosocialresponses. proclivities to explain that to a clinician, the onus falls In assuming this more global interpretation of the on the health care professional to detect signs of thattheme of being known within the data, we were also led difference. Again, by capitalizing on the presence of ato ask addition questions such as how culture, gender, few exceptional observations within our data set, weage, prior life experience in health care and in authori- begin to question our thematic understandings in atative relationships might influence the desired nature deeper manner, one that might prevent us as analystsof human connection in any individual instance. We from premature closure, thinking we understand thisused those variables as “theoretical outliers” in the matter simply because the common patterns are so con-sense that they not only alerted us to exceptional cases vincing.but also pushed us to consider plausible variations thatcould exist but were not captured within our particular THE IMPACT OF COMMUNICATIONdata set. Using this form of “thoughtful clinician test”in our ongoing reflection and analysis, we begin to en- A third thematic claim had to do with the abiding con- viction of some people (perhaps many, as it requires aInternational Journal of Qualitative Methods 5 (2) June 2006http://www.ualberta.ca/~ijqm/
  5. 5. McPherson, Thorne EXPLOITING EXCEPTIONS 5risk to articulate) that cancer communication is, in and and to some extent the selected methodological ap-of itself, sufficiently powerful in the care experience to proach (Thorne, Joachim, Paterson, & Canam, 2002),influence cancer care outcomes. Although our com- the point of the inductive analytic process is inevitablymon discourse would allow that such “soft” variables to discern what is knowable about the phenomena ofplay a critical role in quality-of-life outcomes, some study, however temporary and tentative that “truth”patients consider them to be of sufficient potency that might be (Emden & Sandelowski, 1999; Sandelowskithey might affect mortality outcomes as well. For such & Barroso, 2003; Thorne, Reimer Kirkham, &persons, attitude (positive or negative) is of critical im- O’Flynn-Magee, 2004). Hence, the cognitive strate-portance in responding to cancer illness at all stages, gies employed during analysis ought to facilitate theand the preservation of a positive attitude becomes portrayal of that truth, drawing attention to variation,consistent with better immune response, inner calling assumptions into question, and pushing the in-strength, and the possibility of influencing molecular terpretation toward the richest and most complex por-changes or cellular proliferation patterns. Such inter- trayal possible (Sandelowski, 2002).pretations, whether explicitly or implicitly expressed, Careful attention to exceptional observations, in-are found with sufficient frequency that they might cluding interrogation of why any observation might belead us to believe that a positive attitude is a univer- categorized an exception and vigorously pursuing thesally preferred position and therefore a standard of ef- theoretical possibility of such negative cases, evenfective cancer care communication. when they do not emerge from the sampling process, However, with the benefit of a large data set, the enhances the likelihood that the assumptions embed-presence of a few unusual observations allows us to ded in our work will be detected and challenged as weglimpse the potential problems that might arise from enact our inductive interpretive processes. In qualita-overgeneralization of common patterns. For a subset of tive health research projects, we inevitably bring somebreast cancer patients in particular, the notion of posi- assumptions about what the findings will looktive attitude has become politically aligned with ac- like—whether they will emerge as core variables thatcepting responsibility for the cancer illness in the first help us understand the diversity we see, or whetherplace. Although many patients are quite capable of si- they might reflect basic social processes that will putmultaneously holding ideas that one might interpret as the actions and tacit understandings of human behav-mutually incompatible, such as the idea that attitude iour into some grand universal context (Miles &had no role in the development of breast cancer but Huberman, 1994). We enter our studies from an ana-might play a role in determining outcome, others find lytic or philosophic perspective that tells us somethingany attempt to acknowledge a role for positive attitude of the nature or structure of the phenomenon we seek toas a direct threat to their fundamental denial of respon- explicate, and we ask research questions that revealsibility for having cancer. For this small subset of indi- strong assumptions about how it will look when we ac-viduals, support for the benefits of positive attitude tually find it (Thorne, Paterson, et al., 2002). Further-might be interpreted as demeaning and patronizing. more, we often do so in the context of an elaborate setThus, here as well, analysis of the exceptions, and ex- of assumptions about the very purpose of phenomenaltrapolating the manner in which they might inform our description and the utility of conceptual knowledgesearch for other theoretical variations we have not yet claims within the foundational theoretical structure ofhad the opportunity to study, helps us understand that an applied health orientation (Paley, 1996).competing conceptualizations of that which has a pos- The examples from the cancer communication re-sibility of influencing outcomes is the central problem. search reveal the presence of a number of such as- sumptions. These include assumptions about what INSIGHTS FROM THE EXCEPTIONS theoretical sampling ought to include in the recruit- ment process, the kinds of questions and prompts usedThese examples, extracted from a single study, illus- during data collection/construction interviews, and thetrate differing ways in which exceptional observations expectations held about the degree to which common-can become quite dynamic components of inductive ality among diverse participants might overshadowanalytic processes. In each instance, these observations some of their differences and therefore make the entireprompt deeper and more complex interpretations of the project worthwhile. The very fact of having decided todata set and provide safeguards against premature the- study cancer care communication betrays a fundamen-oretical claims that might be prejudiced by the assump- tal assumption that there are what might be referred totions embedded in the conceptualization of our as “probable truths” (Johnson, 1996; Kikuchi &research projects. Although the product of qualitative Simmons, 1996; Morse & Mitcham, 2002; Sandel-inquiry will vary depending on the goals of the study, owski, 1996; Sandelowski & Barroso, 2003) about International Journal of Qualitative Methods 5 (2) June 2006 http://www.ualberta.ca/~ijqm/
  6. 6. 6 McPherson, Thorne EXPLOITING EXCEPTIONScommunication in this context that reside “out there” to hensive understanding of that which we are studying.be discerned, and also that patterns and themes de- However, qualitative researchers often enter their stud-tected among individual cases will shed light on such ies with the assumption that such an understanding cantruths. We asked individuals with cancer about their be attained and, in the course of conducting their inqui-experiences of helpful and unhelpful communication ries, might unconsciously search out and privilege thatin their health care encounters. Embedded in this ques- which seems indicative of an evolving comprehensivetion are the basic ideas that something exists in the in- conceptualization. When we fail to appreciate the sin-teraction between individuals and their health care gularity of our perspectives, we too quickly createproviders that matters, and that there is something overly simplified interpretations, and thus the inherentabout that interaction that is important to the well-be- value of the qualitative lens to a larger understandinging of cancer patients. These assumptions follow logi- might be lost. To address the manner in which these as-cally from and exist within the context of the evidence sumptions may play a role in shaping the quality of ourbasis that has been developed by scholars working research products, we turn our attention in particular towithin this field and have been repeatedly articulated a consideration of specific methodological decisionsas fundamental values by clinical leaders within the related to sampling.cancer care context (for example, see Degner, 1998; As evidence of this holistic fallacy and the recentFallowfield, Jenkins, & Beveridge, 2002; Freedman, trend toward oversimplification, Caelli and colleagues2002). (2003) pointed to the “ubiquitous and non-selective” As our case illustration suggests, where we choose invocation of “data saturation” within a significantto study a thing, the very logic within which we justify proportion of qualitative health research reports as aour inquiry typically reveals a profoundly influential justification for discontinuing data collectionassumptive frame that shapes much of what we will en- (p. 18).The strategies of seeking maximal variationcounter when we engage in analysis and interpretation. within sampling and continuing to sample until redun-Interrogation of the exceptions one actually finds dancy have been achieved have solid roots within thewithin a data set, and also those “theoretical outliers” grounded theory tradition (Boychuk, Duchscher, &one can begin to imagine on the basis of extrapolation Morgan, 2004; Glaser, 2002) and have also beento the diversity of the population under consideration, widely cited well beyond that context. The point ofseems therefore an essential ingredient in truly excel- sampling widely to tap diversity is that one reduces thelent inductively generated findings. Our experience risk of misattribution or miscategorization by explic-convinces us that a spirited search for unusual observa- itly searching out alternative explanations (Morse &tions with a data set, combined with an imaginal search Mitcham, 2002). When one claims redundancy, one isfor theoretical possibilities within which other excep- typically situating one’s research as having tapped alltions might reside, also serves as a potent device for possible variations within the theoretical positioninggrappling with the strengths and limitations in our by virtue of evidence that no new conditions or varia-qualitatively derived inquiries and for making induc- tions are likely to arise with continued data collectiontive analytic decisions that are informed by those char- (Glaser & Strauss, 1967).acteristics. A more extreme version of this claim emerges un- der the guise of “theoretical saturation,” a concept that THE IMPACT OF DISREGARDING has a distinct and particular utility in the context of DIVERSITIES conventional grounded theory methodology (e.g., Chenitz & Swanson, 1986; Corbin & Strauss, 1990;Reflection on our study in the context of the larger Glaser & Strauss, 1967) but has crept into the lexiconbody of qualitative inquiry into matters of human of methodological justification across a wide range ofhealth and illness experience has convinced us that qualitative methods. In much of the qualitative healthfailing to attend to actual exceptions and theoretical research today, in which the original doctrine of exten-outliers within our research will have untoward conse- sive data gathering, coding, and theoretical testing isquences for the products of our research, particularly if almost always mentioned but not always taken overlywe intend to portray our findings as holding some ex- seriously (Eaves, 2001), asserting the achievement ofplanatory value. According to Morse and Chung theoretical saturation might reflect a lack of interest in(2003), each of our distinct qualitative research tradi- seeking new diversities within larger populations oftions provides us with one unique angle of vision on representative cases. If we uncritically accept suchwhich to understand a phenomenon. In isolation, one methodological claims as credible research logic, weangle of vision is not normally understood as sufficient do so because of the belief that the overall point of ourto produce holistic knowledge or generate a compre- research is the search for evidence of human universalsInternational Journal of Qualitative Methods 5 (2) June 2006http://www.ualberta.ca/~ijqm/
  7. 7. McPherson, Thorne EXPLOITING EXCEPTIONS 7rather than diversities. Thus, we claim theoretical satu- about theoretical saturation when they rely on smallration most easily when we have restricted our gaze to homogenous samples from which to draw their conclu-that which confirms rather than disconfirms our tenta- sions. In our cancer communication study, we purpos-tive interpretive claims. When we fail to attend to the ively set out to create a large sample (by qualitativeexceptional observations or negative cases, and justify standards) to overcome some of the inherent limita-premature conclusions with the convenient claim of tions of smaller studies and to strive toward findingssaturation, we too readily produce research reports that with potential for general application. To sample fromneither do justice neither to the substantive phenomena a sufficiently diverse and representative sample to bewe wish to understand nor to the credibility of the en- able to theorize intelligently about a complex phenom-tire genre of qualitative health research. In other words, enon such as communication in cancer care, we createdwe offer up forced analysis in the absence of real theo- a data set comprising a sample of 200. Had we relied onretical integration, or what Glaser has referred to as a smaller sample size, we might well have assumed we“conceptual foppery” (2002, p. 24). had achieved saturation around such findings as the A particularly worrisome product of our collective meaning of emotional support cancer patients receivedisregard for the diversity and complexity inherent in from their professional health care providers, the inher-the human phenomena that we study is the preponder- ent value of hopeful information, or the consumer per-ance of published qualitative reports that rely on very spective of cancer care communication influence onsmall samples. Although we fully recognize that it is disease outcome. In all of these instances, the numbertheoretically quite possible to create a powerfully of individuals whose experiences varied from the norm“thick” description on the basis of even a single case was quite small. However, identification of unusualstudy (Sandelowski, 1995), all too often one finds pub- observations within the sample, and further actual andlished reports of very “thin” thematic descriptions de- virtual theoretical sampling led us to a better under-rived from single interview records of a handful of standing of the meaning of the variations within humanconveniently selected individuals (Caelli et al., 2003; connection, hope, and optimism that those observa-Sandelowski, 1995, 2004a). This problem is often fur- tions hinted at. In so doing, it confirmed for us that as-ther exacerbated by an apparent disregard for the pro- suming saturation at a less sophisticated level offound implications of how the actual sampling analysis might well have resulted in findings thatapproach will have influenced the substantive nature of would ring less than true for thoughtful clinicians—thethe data (Eakin & Mykhalovskiy, 2003), leading to very audience we intended to reach.overly grand claims that discredit any sensitizing value Recent inquiries using qualitative metasyn- thesisthat the study findings might otherwise have warranted to interpret the state of knowledge deriving from a gen-(Sandelowski, 1995; Sandelowski & Barroso, 2003). eration of qualitative studies confirm that a preponder- We see the issue of sampling and sample size in ance of smaller studies might have systematicallyqualitative research as problematic. Despite the avail- influenced our current understandings and reduced ourability of a vast body of excellent theory on the logic expectations about the value and utility of qualitativelymodel with which one ought to determine whether one derived knowledge (Paterson, Thorne, Canam, &has achieved sufficient variation within a sample to Jillings, 2001). Where there are many small qualitativeevaluate confirming against disconfirming cases, inter- studies in a particular field, we ought to ask ourselvespret exceptions, and test variations (Guba & Lincoln, whether the multiplicity of such findings means that1989; Higginbottom, 2004; Kuzel, 1999; Lincoln & they are reliable and valid, or whether researchers hareGuba, 1985; Miles & Huberman, 1994; Patton, 1990; simply recreating the same species of analytic error asSandelowski, 1995; 2004a; Schwandt 1997), research- a result of methodological inadequacy (Thorne, Pater-ers all too often justify small sample sizes by citing ex- son, et al., 2002). As a growing body of small qualita-plicit numeric suggestions (such as Kuzel’s [1999] tive studies about a phenomenon begins to confirm asuggestion that 5 to 8 participants are sufficient for ho- homogenous standardized impression of it, we mightmogenous groups or Morse’s [1994] estimate that 6 is a become more easily misled into the conviction thatreasonable number for a phenomenological study). All there is nothing more to know. Thus, under such cir-too often, however, these numbers are cited quite apart cumstances, we might need to ensure that the study offrom the methodological context in which they were important phenomena benefits from the application ofintended as if they represented a more general justifica- a diversity of increasingly sophisticated research meth-tion for the inherent value of studies with small sam- ods to ensure that all relevant angles of vision are ex-ples (Endacott & Botti, 2005). ploited and all reasonable interpretations carefully Paradoxically, researchers are much more likely to considered. The complexity of the empirical underpin-convince themselves of the credibility of grand claims ning of our conceptual knowledge must continue to International Journal of Qualitative Methods 5 (2) June 2006 http://www.ualberta.ca/~ijqm/
  8. 8. 8 McPherson, Thorne EXPLOITING EXCEPTIONSplay an important part in our understanding of its matu- with the capacity to discern variations on each outlierrity. theme, we were able to ask a different set of theoretical questions and begin to propose different answers. Con- TOWARD BEFRIENDING scientious attention to exceptional observations within OUR OUTLIERS a larger study therefore tells us something of the prob- lems inherent in the knowledge gleaned from smallOur experience in exploiting sampling diversity con- qualitative studies in which variation is unlikely orfirms Miles and Huberman’s (1994) classic adage that from quantitative studies in which the quest for proba-“the outlier is your friend” (p. 270) and underscores the bilities may average out our capacity to see complexi-importance of not glossing them over in an attempt to ties. Within a larger qualitative data set, and with aseek commonalities and patterns within the human ex- commitment to capitalizing on what it is that excep-periences we qualitatively investigate. In concert with tions can tell us, we believe we can aspire to a more co-many of our predecessors writing in qualitative meth- herent and defensible evidentiary foundation for ourodology development, we feel strongly that excep- conclusions. In the cancer care communication study,tional observations provide a mechanism for moving by generating conclusions at a conceptual level that ac-below the surface of our findings, and call for a re- commodates both commonality and diversity, we arenewed enthusiasm for exploiting complexity rather able to shift our angle of vision from human connectionthan settling for simplicity within our collective under- to being known, from access to information to preserv-standing of quality criteria. We see this as particularly ing hope, and from supporting the not-yet-rational tosalient for work within the health research arena, attending to meaning. In so doing, we believe that thewhere the tensions between universals and particulari- findings of our qualitative research better approximateties play themselves out in the discourse around the the form and structure that will support clinical wis-role of qualitative research in evidence based practice. dom render our findings amenable to such lofty aspira- tions as bringing qualitatively derived evidence into UNIVERSALS AND DIFFERENCES the evidence based practice agenda.Within all qualitative analytic processes, we recognize EVIDENCE-BASED PRACTICEthat there is an inherent tension between the theoretical“holy grail” of universals and commonalities and the For those of us who conduct qualitative health researcheveryday reality of human diversity and variation. It in an applied context (in contrast to those who studyseems counterintuitive to the human mind not to group, health issues for the explicit purpose of advancing so-organize, and order data in such a manner that seeks cial theorizing), a dominant discourse into which ourcommonalities rather than differences. We seem to re- study findings are seeking entry is that of “evi-quire an understanding of commonality before we can dence-based” practice (Miller & Fredericks, 2003).tolerate considerations of diversity. In the example of Where qualitative research makes a significant contri-the cancer communication study, which had lofty goals bution to the evidence literature is precisely where it at-associated with something fairly generalizable, we tends to a multiplicity of perspectives in a manner thathave been forced to pay serious attention to what the is inaccessible to the practitioner of quantitative re-implications might be for all of those individuals and search, who is forced into constructed categories ofinstances that we might consider exceptional observa- meaning, numeric representations of complex phe-tions if we analytically elevated the strong commonali- nomena, and somewhat artificial population groupingsties we have found into conceptual “truths.” Because in an effort to understand something about a wholewe sought consumer-based evidence in our study to (Barbour, 2000; Cohen, Kahn, & Steeves, 2002;support a practice standard, we began to recognize that Sandelowski, 2004b). Studies using instruments at-seeking the unusual observations and understanding tempt to capture something that, by convention, haswhat they had to tell us about the problem of human di- come to represent an aspect of human health realityversity in this context was of critical importance it we that we wish to study and understand. However, the at-wished to do justice to our eventual claims and conclu- tempt to understand it in this manner invariably re-sions. duces, fragments, and sterilizes it from the Thematically, the exceptional observations we de- contaminating context of the natural world in which itscribed represent angles of the phenomenon that would occurs.not likely have been detectable within small study sam- Qualitative research, appropriately conducted, of-ples, or might have been discounted as irrelevant had it fers a means by which we can relocate the numeric rep-occurred within a single instance. In our larger sample, resentations of reality into their embedded complexityInternational Journal of Qualitative Methods 5 (2) June 2006http://www.ualberta.ca/~ijqm/
  9. 9. McPherson, Thorne EXPLOITING EXCEPTIONS 9and better understand the conditions under which that nuanced and inclusive to make a meaningful contribu-representation does and does not hold (Sandelowski, tion to evidence-based practice.2004b). Not a competing reality but, rather, a comple-mentary one, it makes possible a comprehensive inter- CONCLUSIONSpretation of the full range of knowledge productsrelative to a phenomenon by helping us explain the Because the essential point of applied health researchunique aspects of that phenomenon that can reasonably is to generate knowledge that might be applied in thebe derived from each angle of vision. It not only con- context of individual cases, we are convinced that afirms what statistics will tell us is the majority perspec- thoughtful continuing dialogue about the role of quali-tive but also explains minority views and the social, tative inquiry in the evidence-based context is in order.environmental, physical, or metaphysical interactions Instead of simply declaring our right for equal consid-between them and that which is dominant. In so doing, eration within the evidentiary hierarchy, or assumingit creates a knowledge form in which we can better in- the value of our findings to be self-evident, we whoterpret the contexts and conditions under which evi- wrestle with the genre are best placed to recognize itsdence based practice—the majority opinion— strengths and limitations, untangle its theoreticalbecomes most acceptable (Sandelowski, 1997). claims, challenge its methodological and design as- According to Sandelowski (2004b) “Qualitative sumptions, and set the bar for its quality criteria.health research is the best thing to be happening to evi- In this context, we see exceptional observations asdence-based practice.” She claimed that it unfreezes something of a “heuristic” to keep us collectively hon-and complicates the notion of evidence, foregrounding est with regard to our aspirations, to ensure that we re-its political and ideological underpinnings, and reshap- sist the temptation of neatly packaged theorizing anding it so that it rediscovers its capacity to transfer simplistically coherent conclusions. We therefore ad-knowledge to individual cases, which is, after all, the vocate a vigorous and critically considered applicationoriginal intention. She therefore considers qualitative of the notion of theoretical sampling, with an eye to thehealth research as our best chance of producing “truly really fundamental variations inherent in the phenom-transformative knowledge and fully activating the ena themselves rather than the superficial demographicknowledge transformation cycle foundational to the features by which we might guess at them. We thinkevidence-based practice paradigm” (p. 1382). that in the health field, the search for major and minor If qualitative studies merely count, categorize, and variations ought to be an essential ingredient in allgeneralize, then they will be (and, arguably, ought to qualitative inquiry, perhaps tapping the wisdom ofbe) legitimately discounted as methodologically weak those who have closest access to knowledge of theand unworthy of evidentiary status. In contrast, con- variations by virtue of having seen thousands ofducted it in its full complexity, qualitative research cases—the “thoughtful clinician test,” if you will. As acan, indeed, add substantive value to the evi- scholarly community, we ought to scrutinize carefullydence-based practice discourse. Where our research research reports that seem overly neat and tidy, thatproducts are strong, they engage that which can be dis- gloss over variations, or that look only as far as the sim-cerned through quantitative methods in a manner that ilarities. Presentations of findings that overuse meta-enlightens our comprehension of complex phenomena, phoric representation to simplify complex humanso that we understand more fully the strengths and lim- phenomena might be particularly suspect in this re-its of each of the various knowledge forms available to gard. For us to advance the body of our work, it seemsus in relation to a phenomenon. In this context, large imperative that we attend carefully to subtlety, varia-qualitative sample sizes are not an attempt to compete tion, and depth as essential qualitative quality criteria.with the sampling logic associated with quantitative It is also imperative that we learn how to distinguishstudies; rather, they create the opportunity for us to between analysis for themes (which goes no farthersearch out and capitalize on data that challenge our in- than that) and that which explicitly seeks to conceptu-terpretations—those rarely occurring phenomena and alize complexity within some coherent kind of intellec-the exceptions to dominant patterns—so that we can tual or organizing structure.heighten the complexity of our conceptualizations and Within this evidence-based health culture, it is es-extend the comprehensiveness of our understandings. pecially important that we collectively resist the pres-The insights we obtain from the exceptional observa- sure to present our qualitative findings as trends,tions within a larger study, therefore, become an im- patterns, or probabilities—all forms that tend to be ex-portant mechanism by which to test the degree to pressed with a certitude comparable to the confidencewhich our findings are sufficiently comprehensive, that numerical findings confer within policy and best International Journal of Qualitative Methods 5 (2) June 2006 http://www.ualberta.ca/~ijqm/
  10. 10. 10 McPherson, Thorne EXPLOITING EXCEPTIONSpractice discourse. We must remember that what good Corbin, J., & Strauss, A. (1990). Grounded theory research: Proce- dures, canons, and evaluative criteria. Qualitative Sociology,qualitative studies provide is not a competing form of 13(1), 3-21.uncovering trends, not a better way to document pat- Degner, L. F. (1998). Preferences to participate in treatment deci-terns, but actually a counter to the evidence-based sion making: The adult model. Journal of Pediatric Oncologypractice problematic. What quantitative evidence tells Nursing, 15( Suppl. 1), 3-9.us is the majority, the dominant, the visible, and the Eakin, J. M., & Mykhalovskiy, E. (2003). Reframing the evaluation of qualitative health research: Reflections on a review of ap-overt. What qualitative evidence can provide is the praisal guidelines in the health sciences. Journal of Evaluation innuanced, the subtle, the complex, and the various. In Clinical Practice, 9, 187-194.the currently configured evidence-based culture, that Eaves, Y. D. (2001). A synthesis technique for grounded theorywhich is best served will be that which can be mea- data analysis. Journal of Advanced Nursing, 35(5), 654-663. Emden, C., & Sandelowski, M. (1999). The good, the bad and thesured, reported, and established empirically. What gets relative, part two: Goodness and the criterion problem in qualita-lost will be that which is different, dynamic, abstract, tive research. International Journal of Nursing Practice, 5, 2-7.and tacit. With a renewed enthusiasm for methodologi- Endacott, R., & Botti, M. (2005). Clinical research 3: Sample selec-cal rigor and analytic accountability, as well as an ex- tion. Intensive & Critical Care Nursing, 21, 51-55. Fallowfield, L. J., Jenkins, V. A., & Beveridge, H. A. (2002). Truthplicit humility for the profoundly complex nature of may hurt but deceit hurts more: Communication in palliativethe knowledge we produce, the qualitative health com- care. Palliative Medicine, 16, 297-303.munity can, indeed, serve as a humanizing force within Freedman, T. G. (2002). “The doctor knows best” revisited: Physi-the dominant ideology. cian perspectives. Psycho-Oncology, 11, 327-335. Glaser, B. (2002). Conceptualization: On theory and theorizing us- Our experience with a rather large qualitative study ing grounded theory. International Journal of Qualitativehas allowed us to capitalize on the inherent value of ex- Methods, 1(2), Article 3. Retrieved June 20, 2005, fromceptions as a device that enables us to dig deeper into http://www.ualberta.ca/~ijqm/our data sets, to ponder alternative angles of interpreta- Glaser, B., & Strauss, A. (1967). The discovery of grounded theory: Strategies for qualitative research. Chicago: Aldine.tion, and to account more elegantly for the very real Guba, E. G., & Lincoln, Y. S. (1989). Fourth generation evalua-challenge that human diversity entails. Although com- tion. Newbury Park, CA: Sage.monalities and thematic patterns documented within Higginbottom, G. M. A. (2004). Sampling issues in qualitative re-our data sets can provide us with a solidly grounded set search. Nurse Researcher, 12(1), 7-19. Johnson, J. (1996). Nursing art and prescriptive truths. In J. F.of general principles, it is the exceptions that will often Kikuchi, H. Simmons, & D. Romyn (Eds.), Truth in nursing in-yield the best insights as to how and when we ought to quiry (pp. 36-50). Thousand Oaks, CA: Sage.apply them. On this basis, therefore, we celebrate the Johnson, M., Long, T., & White, A. (2001). Arguments for “Britishexciting role that exceptional observations can play in Pluralism” in qualitative health research. Journal of Advanced Nursing, 33, 243-249.forcing a standard of conceptual clarity and evi- Kikuchi, J. F., & Simmons, H. (1996). The whole truth and prog-dence-grounding within our qualitative health research ress in nursing knowledge development. In J. F. Kikuchi, H.products. Simmons, & D. Romyn (Eds.), Truth in nursing inquiry (pp. 5-17). Thousand Oaks, CA: Sage. Kuzel, A. J. (1999). Sampling in qualitative inquiry. In B. F. Crab- REFERENCES tree & W. L. Miller (Eds.), Doing qualitative research (2nd ed., pp. 33-45). Thousand Oaks, CA: Sage.Agresti, A., & Finlay, B. (1997). Statistical methods for the social Lincoln, Y. S. & Guba, E. G. (1985). Naturalistic inquiry. sciences (3rd ed.). Upper Saddle River, NJ: Prentice-Hall. Newbury Park, CA: Sage.Barbour, R. S. (2000). The role of qualitative research in broaden- Miles, M. B., & Huberman, A. M. (1994). Qualitative data analy- ing the “evidence base” for clinical practice. Journal of Evalua- sis: An expanded sourcebook (2nd ed.). Thousand Oaks, CA: tion in Clinical Practice, 6, 155-163. Sage.Bickman, L., Rog, D. J., & Hedrick, T. E. (Eds.). (1998). Handbook Miller, S., & Fredericks, M. (2003). The nature of “evidence” in of applied social research methods. Thousand Oaks, CA: Sage. qualitative research methods. International Journal of Qualita-Boychuk Duchscher, J. E., & Morgan, D. (2004). Grounded theory: tive Methods, 2(1). Article 4. Retrieved July 25, 2005 from Reflections on the emergence vs. forcing debate. Journal of Ad- http://www.ualberta.ca/~iiqm/backissues/2_1/ html/miller.html vanced Nursing, 48(6), 605–612. Morse, J. M. (1994). Designing qualitative research. In N. K.Caelli, K., Ray, L., & Mill, J. (2003). “Clear as mud”: Toward Denzin & Y. S. Lincoln (Eds.), Handbook of qualitative inquiry greater clarity in generic qualitative research. International Jour- (pp. 220-235). Thousand Oaks, CA: Sage. nal of Qualitative Methods, 2(2), Article 1. Retrieved February 2, Morse, J. M., & Chung, S. E. (2003). Toward holism: The signifi- 2004, from http://www.ualberta.ca/~iiqm/backissues/2_2/html/ cance of methodological pluralism. International Journal of caellietal.htm Qualitative Methods, 2(3), Article 2. Retrieved February 2, 2005,Chenitz, W. C. & Swanson, J. M. (1986). From practice to from http://www.ualberta.ca/~iiqm/backissues/ 2_3final/html/ grounded theory: Qualitative research in nursing. Englewood morsechung.html Cliffs, NJ: Addison-Wesley. Morse, J. M., & Mitcham, C. (2002). Exploring qualitatively de-Cohen, M. Z., Kahn, D. L., & Steeves, R. H. (2002). Making use of rived concepts: Inductive-deductive pitfalls. International Jour- qualitative research. Western Journal of Nursing Research, 24, nal of Qualitative Methods, 1(4), Article 3. Retrieved February 2, 454-471. 2005, from http://www.ualberta.ca/~ijqm/english/engframeset. htmlInternational Journal of Qualitative Methods 5 (2) June 2006http://www.ualberta.ca/~ijqm/
  11. 11. McPherson, Thorne EXPLOITING EXCEPTIONS 11Osborne, J., & Overbay, A. (2004). The power of outliers (and why Schwandt, T. A. (1997). Qualitative inquiry: A dictionary of terms. researchers should always check for them). Practical Assess- Thousand Oaks, CA: Sage. ment, Research and Evaluation, 9(6). Retrieved December 20, Thorne, S. E. (1991). Methodological orthodoxy in qualitative 2004, from http://pareonline.net/getvn.asp?v=9&n=6 nursing research: Analysis of the issues. Qualitative Health Re-Paley, J. (1996). How not to clarify concepts in nursing. Journal of search, 1, 178-199. Advanced Nursing, 24, 572-578. Thorne, S., Hislop, G. T., Kuo, M., & Armstrong, E.-A. (2006).Paterson, B. L., Thorne, S., Canam, C., & Jillings, C. (2001). Hope and probability: Patient perspectives of the meaning of nu- Meta-study of qualitative health research: A practical guide to merical information in cancer communication. Qualitative meta-analysis and meta-synthesis. Thousand Oaks, CA: Sage. Health Research, 16, 318-336.Patton, M. Q. (1990). Qualitative evaluation and research methods Thorne, S., Joachim, G., Paterson, B., & Canam, C. (2002). Influ- (2nd ed.), Newbury Park, CA: Sage. ence of the research frame on qualitatively derived health scienceSandelowski, M. (1995). Sample size in qualitative research. Re- knowledge. International Journal of Qualitative Methods, 1(1), search in Nursing and Health, 18, 179-183. Article 1. Retrieved February 2, 2004, fromSandelowski, M. (1996). Truth/storytelling in nursing inquiry. In http://www.ualberta.ca/~ijqm/ J. F. Kikuchi, H. Simmons, & D. Romyn (Eds.), Truth in nursing Thorne, S. E., Kuo, M., Armstrong, E-A., McPherson, G., Harris, inquiry (pp. 111-124). Thousand Oaks, CA: Sage. S., & Hislop, G. (2005). “Being known”: Patient perspectives onSandelowski, M. (1997). “To be of use”: Enhancing the utility of human connection in cancer care. Psycho-Oncology, 14, qualitative research. Nursing Outlook, 45, 125-132. 887-898.Sandelowski, M. (2002). Reembodying qualitative inquiry. Quali- Thorne, S., Paterson, B., Acorn, S., Canam, C., Joachim, G., & tative Health Research, 12, 104-115. Jillings, C. (2002). Chronic illness experience: Insights from aSandelowski, M. (2004a). Counting cats in Zanzibar. Research in meta-study. Qualitative Health Research, 12, 437-452. Nursing & Health, 27, 215-216. Thorne, S. E., Reimer Kirkham, S., & O’Flynn-Magee, K. (2004).Sandelowski, M. (2004b). Using qualitative research. Qualitative The analytic challenge in interpretive description. International Health Research, 14, 1366-1386. Journal of Qualitative Methodology, 3(1), Article 1. RetrievedSandelowski, M., & Barroso, J. (2003). Classifying the findings in February 2, 2004, from http://www.ualberta.ca/~iiqm/backissues qualitative studies. Qualitative Health Research, 13, 905-923. /3_1/ html/thorneetal.html International Journal of Qualitative Methods 5 (2) June 2006 http://www.ualberta.ca/~ijqm/