It is indeed encouraging to see debate about concept analysis and concept development and, in that spirit, the response by Duncan et al. (2009) is welcome. Criticism, however, depends on shared understanding. Before we can move to productive discussion of the issues, it is important to correct some of Duncan et al.’s misinterpretations of my JAN article (Risjord 2009).

First, ‘Rethinking Concept Analysis’ has an unrelenting commitment to contextualism: ‘In this study, I will argue that these questions can best be answered by reaffirming the idea that concepts must be related to contexts, such as theories, discourses or speech communities’ (Risjord 2009, p. 685). Given such an unambiguous commitment, it is startling to see my essay characterized as arguing ‘against concept “contextualization”’ (Duncan et al. 2009, p. 1985) and in favour of ‘concept analysis and theory construction divorced from contextual reality’ (p. 1986). In addition, Duncan et al. attribute to the essay the assertion that ‘concepts are solely contextualized by theory’ (p. 1985). As these attributions are the opposite of what were my assertions, the criticisms based on them are non-starters.

Second, Duncan et al. (2009) impose a hierarchy of scientific and non-scientific knowledge on the essay. Contrary to the various assertions in their response, ‘Rethinking Concept Analysis’ puts the analysis of theoretical discourse and colloquial speech on the same epistemic grounds: ‘The justification for a concept analysis in a scientific context is, therefore, largely the same as the justification for the analysis of a colloquial term’ (Risjord 2009, p. 687). The difference between theoretical and colloquial concept analyses is not driven by some imagined difference between ‘scientific’ and ‘non-scientific’ knowledge. The difference is simply that scientific theories often explicitly define terms and limit their meaning, whereas in common speech we do not. This does not imply that ordinary concepts are merely a ‘potential knowledge source’ (Duncan et al. 2009, p. 1985). It would be absurd to suggest that I only ‘potentially know’ that my child is in pain because I have failed to operationally define pain. Conceptual content is determined by context, and there are many different kinds of context (not just theoretical and colloquial). It is simplistic to insist that every difference creates a hierarchy.

Unpacking the meaning of concepts in common use requires methods associated with qualitative research, and my essay agrees with Duncan et al. (2009, p. 1986) that qualitative research is ‘scientific and methodologically sound’. It is a little odd, however, that Duncan et al. support this statement on the authority of Morse (2004). As important as Morse’s work has been, the methodological soundness of qualitative research was established long before her contribution to the field. Have the contributions of Leininger and Benner been forgotten so quickly? And should we not recognize Quint (1963) as a nursing pioneer in this area? Furthermore, these nurse scholars recognized that they were applying methods whose soundness was demonstrated 100 years ago (see Risjord 2000, 2007, 2010 for a discussion of qualitative methods both inside and outside of nursing). Complaints about the ‘ghettoization’ of qualitative research are thus based on ideas introduced by Duncan et al., not on any statements made in my JAN paper.

Why did Duncan et al. base their critique on an interpretation that flies in the face of explicit and unambiguous statements in my original essay? The answer, I suggest, is that they assume that realism precludes contextualism. Because ‘Rethinking Concept Analysis’ argues for a form of realism, they misread the essay’s commitment to contexualism and imposed a cluster of ideas they associate with realism. The essay argues directly against this assumption, and therefore, while Duncan et al. do not respond to (or even mention) this argument, it is at the heart of our disagreement.

In Duncan et al.’s (2007)JAN paper and in their response now to my own, they oppose ‘relativism’ to ‘realism’. Duncan et al. are never clear about what is relative to context, and as a result, at least three different doctrines are conflated under the label of ‘relativism’:

  • • 
    Concepts are relative to context.
  • • 
    Truth is relative to context.
  • • 
    Knowledge is relative to context.

Denying these doctrines produces three different kinds of realism: realism about concepts, realism about truth and realism about knowledge. The three doctrines (and their relativist correlates) are not equivalent, so the relationships among them need to be carefully examined. The first sense of relativism is synonymous with ‘contextualism’ as used in my essay (Risjord 2009). Realism about concepts would hold that conceptual content is not relative to context. Presumably, this is the position that Duncan et al. describe when they talk about concepts ‘transcending’ context, or there being ‘objective universal concepts’. As my essay is committed to contextualism, it stands with Duncan et al.’s position in rejecting realism about concepts.

‘Rethinking Concept Analysis’ argues that ‘moderate realism’ as defined by Kikuchi (2003) is consistent with contextualism (relativism about concepts). Moderate realism is a very modest view. Some readers might notice its similarity to the ‘natural ontological attitude’ in Fine (1984). It holds only that ‘reality exists independent[ly] of the human mind’ (Kikuchi 2003, p. 12). The truth or falsity of a statement is a matter of how well it represents reality. Judgments of truth are always provisional and subject to revision, never certain or fixed. Moderate realism thus denies relativism about truth. ‘Rethinking Concept Analysis’ (Risjord 2009, p. 688) argues that contextualism is consistent with moderate realism. There is no reason to repeat the argument here, as no criticism of that argument is offered by Duncan et al.

It is perhaps worth mentioning that the combination of contextualism and moderate realism does not have many of the features that Duncan et al. associate with ‘realism’. It introduces no hierarchy between scientific and non-scientific knowledge, or between qualitative and quantitative research. Because concepts are contextual while truth is not, there is no ‘simplistic and linear relationship between concepts, theory and reality’ (Duncan et al. 2009, p. 1986). Nor need it take ‘objective “detached” research’ as an objective (p. 1986). Other forms of realism may commit these sins, but moderate realism is innocent.

To conclude, let us take Duncan et al.’s advice and examine the issues surrounding the concepts of compliance/adherence/concordance. As they present it, the conflict is between an operationalization of ‘compliance’ in terms of measurable patient behaviour, and a richer concept that is sensitive to the alliances between practitioners and clients. Duncan et al. treat the conflict as a difference between ontological perspectives. The patient management context is supposed to be realist about both concepts and truth, while the alliance context is relativist about both. The problem is that their relativism about truth robs them of the resources for criticism. As they admit, ‘It is not our intention to argue the relative value of any particular ontological perspective’ (Duncan et al. 2007, p. 296). They are unable to argue that the behaviourist operationalization of compliance is a misrepresentation, because doing so would presuppose that there is something to be represented well or badly. Duncan et al. thus leave us with an irreconcilable difference between ontological perspectives, and no way forward.

Duncan et al. (2007, p. 298) start in the right place when they recognize that ‘compliance is both an “ordinary” and “scientific” concept’. Because compliance is a concept with more than one contextual home, there are many dimensions to the meaning of ‘compliance’. It is just the kind of concept that needs the treatment recommended in my essay. To understand what it means in the context of specific theories, a theoretical concept analysis is required. Its meaning to nurses and their clients, on the other hand, is best uncovered by a colloquial concept analysis that uses interviews, focus groups, participant observation and the other methods of qualitative research.

To move forward, we need to compare and evaluate these concepts of ‘compliance’ (as well as related concepts). The combination of contextualism and moderate realism provides tools that resolve the dilemma. The dispute is not about truth or reality, it is about concepts. A commitment to moderate realism means that we can compare, evaluate and modify these concepts. The behavioural conceptualization is, no doubt, limited and morally suspect. Does it teach us anything of value? If so, we should keep the useful elements of meaning, enrich them with what we have learned from the colloquial analysis and create a new concept that is more adequate. If not, we should reject such theories of compliance because they misrepresent the phenomenon—that is, because they are false—and begin to develop a better understanding on the basis of what we have learned from nurses’ and their clients’ experiences. In either case, we develop more adequate concepts by revising them as parts of their theoretical, discursive and practical context.


  1. Top of page
  2. References