“Qualitative and Quantitative Analyses
of Psychological Distress :
Methodological Complementarity
and Ontological Incommensurability”.

Un article publié dans la revue Qualitative Health Research, vol. 10, no 3, may 2000, pp. 411-423. Sage Publications, Inc.

Abstract

Introduction

QUALITATIVE ANALYSIS OF WHAT PSYCHOLOGICAL DISTRESS MEANS IN A NONCLINICAL POPULATION

Distress as Language
Idioms of Distress

QUANTITATIVE METHODS FOR THE VALIDATION OF A PSYCHOLOGICAL DISTRESS MEASUREMENT SCALE

DISCUSSION

Quantification, Reification of Qualitative Constructs, and Loss of Meaning
Convergent or Divergent Epistemologies

Representational Dilemma

A Politico-Representational Crisis

Ontological and Teleological Incommensurabilities : Toward a Paradigm of Tolerance

CONCLUSION

REFERENCES

ABSTRACT

Rigorous qualitative and quantitative methodologies have been used for the development of a multidimensional scale dedicated to the measurement of psychological distress. A comparison between the idioms of distress or the cultural forms through which French Quebecois express their distress (qualitative constructs) and the nonorthogonal factors derived from explanatory and higher order factorial analyses (quantitative constructs) illustrates the possibilities of complementarity between qualitative and quantitative approaches. The comparison shows that these two operationalizations of the concept of psychological distress are founded on incommensurable representations of distress. This article concludes that this representational dilemma of distress as a lived language or as an empirical reified entity leads to an ontological and a teleological incommensurability.


INTRODUCTION

The complementarity of qualitative and quantitative methods and epistemologies has been a key issue in the past two decades. Throughout the past 15 years, I have conducted many research projects that use complementary techniques for data collection and analysis. Although I am an anthropologist by profession, I have worked extensively with epidemiologists and psychometricians in multidisciplinary research teams. As such, I have become interested in the quest for complementarity in the measurement of mental health and the identification of the components of psychological distress. Using the research project described in this article, I have tried to push the potential of the complementarity of qualitative and quantitative analyses beyond its presently known entrenchments. The objective of this project was the development of validated culture-sensitive multidimensional scales capable of being used to assess psychological distress and well-being in epidemiological studies in Quebec (French Canada). Qualitative and quantitative analyses of a pool of more than 2,000 reported distress manifestations show that the quest for meaning and the quest for measurement are incommensurable research objectives. Results show that the process of scales construction and validation leads to the denaturalization, the objectification, and the encapsulation of the meaning of distress. It also leads to the decontextualization of the lived experience of distress and to a disembodied understanding of that concept. It will be shown that beneath this first teleological incommensurability (divergent finalities) lies an ontological incommensurability related to the divergent construction of the nature of the research object (distress) as either a state of health or as a language as proposed by the interpretativist perspective.


QUALITATIVE ANALYSIS OF WHAT PSYCHOLOGICAL
DISTRESS MEANS IN A NONCLINICAL POPULATION

The project discussed in this article was carried out in two successive and complementary stages. A quantitative phase, dedicated to the psychometric analysis and construct validation of a psychological distress measurement scale (discussed later in this article), was preceded by a qualitative phase. Its goal was to create an initial list of distress symptoms to be submitted to psychometric analysis and to reconstruct the lay conception of the psychological distress construct that many national epidemiological studies pretend to assess. Data came from in-depth interviews conducted with a random sample of 179 adult French Quebecois. Following a phenomenological procedure, participants were asked to describe, in detail, a recently lived episode of distress by detailing how it affected them in every aspect of their everyday life. Almost 2,000 pages of distress episode narratives were submitted for computer-based content analysis (using NUD*IST software). Using an iterative inductive approach (based on constant comparison to maximize the grounding of the codes into narrative data), a corpus of 2,200 manifestations of distress was constructed. Thematic content analysis condensed these manifestations into 176 subcategories, which in turn were condensed into 47 categories. Then, using ethnosemantic and phenomenological analyses, we tried to reconstruct the deep and multiple meanings of these distress manifestations. From the start, it became evident that we would have to rid ourselves of the empiricist symptomatic approach. We would have to analyze the signs that people recognized as meaningful, the signs to which they refer to experiment and communicate their distress in a specific sociocultural (French Canadian) and biographical context. We would have to treat distress manifestations as significant, culture-specific items and vehicles of dense, symbolic content as opposed to hypothetical, universal, empirical symptomatic entities.

Distress as Language

We were confronted with the fact that each manifestation of distress took on a plurality of meanings and that these meanings were context dependent. These meanings changed from one episode to another at different times throughout the chronology of the episode in terms of severity or abnormality. It became evident that the best reading of these distress manifestations was through the interpretativist and linguistic perspective. We then looked at distress as a language based on manifestations or signs associated with distress episodes (Massé, 1999). These signs can be seen as the basic, culturally acknowledged vocabulary through which people experience, express, and communicate their distress. This language of distress enables the individual to experience his or her mental suffering on the social scene and to structure the ways that his or her suffering can be enacted. As both a language and a lived experience, psychological distress can no longer be treated as an empirical entity or a mere well-bounded state of mental health. Symptoms themselves are not discrete components of a reified distress but are signs that can take on a plurality of meanings in each individual discourse. In addition, distress should not be defined through a summation of independent decontextualized symptoms but instead through the narratives that link them to the distress experience. These narratives constitute specimens of all potential discourse available to communicate distress in a given culture through acceptable idioms. They do not reflect or portray an empirical entity, but rather, they are some of the possible constructs of the virtual concept of mental distress reconstructed by anthropologists.

Idioms of Distress

Ethnosemantic analysis shows that the regrouping of manifestations into lexicons of signs (the interpretativist version of the symptoms categories) using an empiricist thematic content analysis masks more subtle cultural forms that channel and structure the ways and the means by which people express their distress. Building on these premises, we identified some idioms of distress or modes through which people live, express, and interpret their experience of suffering in a culturally appropriate way (Migliore, 1997 ; Nichter, 1981 ; Parsons & Wakeley, 1991). In Quebec, distress is expressed mainly through six idioms of expression (Massé, 1999).

Demoralization and pessimism toward the future. There is a deep conviction that, in the future, things can only get worse. These feelings are accompanied by a sense of despair and disillusionment, and therefore, they are related to demotivation, disengagement, and a loss of interest in social and professional activities.

Anguish and stress. Distress episodes are also experienced as anxiety, preoccupation, nervous tension, and strain. These anxieties and fears are experienced as an internal suffering and entail a feeling of powerlessness.

Self-depreciation. People doubt their social, affective or professional skills and abilities and their own self-worth. People are very self-critical ; they put the blame on themselves, thinking that they are responsible for what they have become psychologically and socially.

Social withdrawal and social isolation. People isolate themselves and cut off contact with others. They do not want to socialize with others, preferring to escape from social life to cut themselves off from the world.

Somatization. Among the most noteworthy signs are physical exhaustion, loss of energy, and fatigue. These signs are sometimes expressed as an incapability to engage in physical effort. This incapability justifies the individual's social withdrawal and disengagement. This physical fatigue is always associated with a psychological fatigue that, in turn, is linked to demoralization, loss of interest in life, and apathy.

Withdrawal into oneself. The core idiom of distress around which the language of distress is structured (among French Quebecois) seems to be withdrawal into oneself. In the postmodern sociocultural context, distress is, first and foremost, a crisis of the self. This crisis stems from the individual's perceived capacity or incapacity to control his or her life and to adjust to his or her social environment. The values of autonomy, independence, and control over oneself are at the heart of the process of the sociocultural construction of affective or anxiety disorders. Demoralization, self-depreciation, and social disengagement are understood not as consequences of a sociopolitical analysis but as results of the person's inabilities to deal with undesirable life events. Phenomenological appraisal of distress narratives shows that distress is primarily an ill being in the world, an internal suffering, and a suffering of the self. All of these feelings are inherent to a process of self-depreciation and self-condemnation.


QUANTITATIVE METHODS
FOR THE VALIDATION OF A PSYCHOLOGICAL
DISTRESS MEASUREMENT SCALE

These cultural forms or idioms are fundamental keys to any in-depth understanding of what distress really means for any cultural group. However, they are not operational concepts that can be confined to variables that refer to a predefined objectified content. There are two reasons for this : First, an idiom does not refer to a finite set of signs, and second, the meaning of these distress signs are not universal and stable, they fluctuate from one stage of an episode to another, from person to person, and from context to context. Therefore, to reach the primary objective of the development and validation of a distress scale capable of being used in epidemiological studies, we have to rely on classic quantitative methodologies based on multivariate analyses of the covariation between the symptoms included as objectified items in a checklist. A secondary objective was the identification of the underlying components or factorial dimensions of the empiricist construct of psychological distress and their comparison with the idioms derived from the interpretativist approach.

In the qualitative study, 2,182 manifestations of distress were sorted to form 47 lexicons of signs. Based on these lexicons, a list of 73 of the most significant and prevalent manifestations of distress was built, representative of each of the 47 lexicons of signs from which people draw to construct their personally experienced discourse. Using this list of distress manifestations, we conducted a second study with another random sample of 400 people in Quebec. The construction of the final scale was based on data related to the frequency of occurrence of these symptoms (five levels, ranging from never to almost always occur) in the month preceding the home interviews. The result is an instrument called the Psychological Distress Manifestations Measurement Scale (Massé et al., 1998a), which is based on 23 symptoms grouped in four nonorthogonal factorial dimensions derived from exploratory and higher order confirmatory factor analyses (Massé et al., 1998a, 1998b). The internal consistency of the scale is high (Cronbach's alpha = 0.93), and one should note that the same four factors were found on a larger scale containing 45 items from which the 23-item version was derived. This confirms the stability of this factorial solution. The following factors were named :

• "Devalorization" or self-depreciation, which refers to a loss of self-confidence, a feeling of uselessness, low self-esteem, and a tendency to isolate oneself (e.g., I lack selfconfidence ; I had the impression that no one loved me).
  
  
• "Irritability/aggressiveness" tinged with arrogance and conflicts with others (e.g., I was arrogant and even rude toward others ; I was aggressive about every little thing).
  
  
• "Anxiety /depression" mixed with stress (e.g., I felt depressed or down in the dumps ; I felt preoccupied and uneasy).
  
  
• "Demotivation," which refers to a social disengagement and a generalized disinterest toward life (e.g., I was less receptive to the ideas and opinions of others ; I did not feel like doing anything).

Higher order confirmatory factor analyses show that these four factors are the basic components of a latent construct referring to psychological distress, which is a dimension of a third-order construct of mental health (Massé et al., 1998b).

A comparison of the idioms of distress and the factorial constructs points to somewhat divergent results. Technical requirements of factorial and confirmatory analyses and statistical manipulations, unavoidably, do violence to the data in three ways.

First, high interitems correlation and efficiency imperatives result in the elimination of many of the 176 types of signs identified in the content analysis. The remaining 23 symptoms were decontextualized elements of the initial vocabulary of distress. They are but a small subsample in a huge universe of distress manifestations.

Second, even whole semantic lexicons and idioms disappeared from the final instrument. For example, factor analysis made no room for the somatic idioms. In fact, all somatic signs were absent from the final scale. This outcome was to be expected because, although somatic signs were all associated to the same factor, anxiety/ depression, they showed low commonalties and had to be taken out to raise the internal consistency of this subscale.

Third, each of these signs has as many significations as there are semantic axes. Severity, normality, and intensity were some of the semantic axes left aside for the benefit of the frequency of occurrence axis in the scale construction. Exploratory factorial analysis suggested that factorial construals differed significantly in their content and structure when data from different Likert-type measurement scales (intensity, severity, normality, impacts on daily life or on role performance) were included in the database.

The construction of the four psychological distress factorial construals does not end but rather fuels the debates surrounding the nature of distress. These statistical configurations of symptoms are poor representations of the idioms of distress. They do not reflect the 47 lexicons of distress signs identified by the content analysis. The multivariate analysis and other statistical objective criteria used to select and regroup these signs deculturalized the final scales. The fundamental issues we must discuss here are those of the complementarity of etic and emic approaches and the complementarity of the interpretativist and empiricist understanding of the very nature of distress.


DISCUSSION

Exploratory and confirmatory factor analyses are often used to identify underlying dimensions and to test the modelizations of the interrelationship of complex research objects such as social isolation, self-esteem, well-being, and, of course, psychological distress (Gotlib & Cane, 1989 ; Link & Dohrenwend, 1980 ; Weissman, Myers, & Ross, 1988). Results obtained from the confrontation of two independent qualitative and quantitative procedures were divergent at this point. The four factorial dimensions were different from the cultural forms of the distress as language understood as idioms. In particular, components of the lived-experience-of-distress episodes, such as somatization, social withdrawal, rejection of the self, and withdrawal into oneself, were only very partially expressed in the factorial dimensions of devalorization, anxiety/ depression, irritability/ aggressiveness, and demotivation. A very important comment to be made is that the discrepancy between these two readings of the components of psychological distress cannot be explained by an asymmetry in the scientificity of each methodological apparatus. The qualitative approach relied on the in-depth analysis of distress episode narratives of a large random sample. Results of the content analysis and the categorization of hundreds of reported distress manifestations were validated by a phenomenological appraisal of the grounding of these lexicons of distress signs and by an interpretativist reconstruction of the cultural forms (idioms) through which distress was expressed and communicated. Moreover, the initial pool of 73 items submitted for exploratory factorial and higher order confirmatory factor analyses was selected to represent each of the 47 lexicons of signs. The results may claim to rely on highly sophisticated qualitative analyses of textual data and on a rigorous quantitative analysis of a covariance matrix derived from frequency-of-occurrence scales ; however, in this study as in others, the opposition between qualitative and quantitative approaches cannot be resolved on the basis of-different levels of scientificity, logic, or systematicity of operations and reasoning. We agree with Sokal and Bricmont (1997), who establish a difference between a "scientific attitude" and the "scientism" defined as the "illusion that simplistic but renowned 'objective' or 'scientific' methods [are able] to solve complex problems" (p. 193). Both methodological approaches rely on a scientific attitude, even if the sophisticated quantitative exploratory and confirmatory factorial analyses are more often objects of scientists' discourse.

Quantification, Reification of Qualitative Constructs,
and Loss of Meaning

The sources of discrepancy between distress dimensions as identified through qualitative and quantitative methodological devices are deep and complex. The first source of discrepancy lies in the meaning that reductionism (inherent to the reification of distress manifestation) requested for measurement. The process of scale construction and validation results in factorial and latent constructs that, as objectified representations, inevitably denature the very essence of distress as a language based on idioms. We believe that these scales are interesting alternatives to other scales (on a purely psychometric point of view but also as culturally sensitive scales) as responses to the need for mental health assessment and for the identification of at-risk social groups. However, the wealth and sensitivity of our qualitative analyses lead us to admit the flimsiness of the results, the decontextualization of the factorial contruals, and the reification of the flowing reality that constitutes psychological distress. The lived experience of distress embedded in the world of everyday life is subject to the encapsulation of distress manifestations into decontextualized. symptoms. The symptom checklists used in epidemiological studies therefore should not claim to grasp the complexity of psychological distress. They are predictive tools, not analytical ones.

This loss of meaning, correlative to the translation of idioms of distress in measurement devices, is due to three major factors. First, human cases are reduced to sets of variables. In quantitative methodology, these variables are the focus of measurement and statistical analysis : "Individuals are stripped of agency ; they 'endure' but have no initiative, they are sites with measurable attributes but do not 'participate' in or actively produce the events of interest" (Mishler, 1996, p. 79). It is clear that such an encapsulation of distress manifestations into symptoms-as-variables is meant to change a polymorphic, context-dependent construct into a reified object to be reduced to a score on a frequency scale and a covariation matrix. What is at stake then is the "ability to recover persons-to retrieve their individuality-from the matrix of relationships that continue to be established among variables" (Jessor, 1996, p. 4). Constructivists and phenomenologists seem to agree with Mishler (1996) on the task of recovering the missing people lost among the mass of variables and even on the incommensurability of group and individual analyses. However, here again the solution lies only partly in systematic case-based or narrative analyses that tend to contextualize the interactions between reified variables, syntagmatically and paradigmatically, thereby reproduicing into case studies the empiricist, positivist epistemology of quantitative analyses. The challenge of qualitative research is not only to recover the wealth of individual variability from under the dominant normative topics of postpositivist research (such as means, standard deviation from norms, assumptions of linearity, or explained variance) but also to redefine the nature of the dependent variable on an interpretative basis. According to Shweder (1996), qualitative research looks at "qualia" or unobservable subjective experiences (e.g., sensations or feelings) that are nondeducible meanings and properties associated with consciousness, originating in the capacity of qualitative beings to symbolize (i.e., a value, a meaning, a purpose, a rationale). Our study confirms that part of the diverging results lies in the differential treatment of distress manifestations as signs or symptoms, as symbolic elements of a language (qualia) or as objectified realities ("quanta"). Our results are a good example of "the way qualia are underspecified and/or underdetermined by quanta" (Shweder, 1996, p. 180).

Second, factorial analyses are objective accounts of distress, incommensurable with subjective accounts. In Nagel's (1979) understanding, factorial objective representation is

an account of the nature of a thing as it exists independently of our particular awareness-experience-evaluation of it : as it exists without regard to our situated and hence contingent, reaction to it. At the opposite, "subjective" accounts means the way a thing appears or feels for the individual, the way the thing is understood, experienced, or reacted to in this parochial state of mind. (as cited in Shweder, 1996, p. 181)

Third, by their very nature, factorial analyses are deeply incompatible with a consensual social representation of distress. Doise, Clemence, and Lorenzi-Cioldi (1992, pp. 110-111) underline the following paradox : Techniques used in factorial analyses to rebuild common representations are, in fact, based on interindividual variations. They look for covariations between response profiles. Items showing very high intercorrelation must be rejected to make these analyses possible. Factorial analyses summarize variations, offering a structured picture of the ways in which symptoms covariate. They rely, however, on a condensation of the initial data, which implies a loss in the wealth of information given by each individual.

Convergent or Divergent Epistemologies

It follows from what has been said that we cannot fully agree with Jessor (1996), Becker (1996), and others when they say that, although asking different kinds of questions, qualitative and quantitative methods share a common epistemological foundation, with both approaches using more or less explicit quantitative data analyses. Jessor, for example, considers that "it is misleading to cast qualitative and quantitative methods as an irreconcilable antinomy, with each the polar opposite of the other. Qualitative data can be counted and quantified, and quantitative data can be interpreted" (p. 6). A second political argument rejects "the persistent feeling that the methods differ epistemologically" (Becker, 1996, p. 57). Becker suggests that any apparent fundamental epistemological difference between qualitative and quantitative research comes from two things, the prescriptive mode of epistemology as a controversial discipline and the status differential between the dominant quantitative and the dominated qualitative research community. For him, "the similarities between these methods are at least as, and probably more, important and relevant than the differences" (p. 53). To assume that "qualitative and quantitative methods of social inquiry, though often asking different kinds of questions, share a common epistemological foundation and a common philosophy of science" (Jessor, 1996, p. 7) is true insofar as we reduce qualitative approaches to an empiricist/positivist paradigm. If we agree that cultural beliefs or values as well as social relationships are empirical, definite, well-bounded, reified objects treated as variables in interaction, which are directly accessible to the researcher, then it can be argued that qualitative data-collecting techniques and analytical methodologies are epistemologically the same.

Because qualitative methods and procedures attempt to count, calculate, measure, or sample, following the same basic logical rules, it is right to say that "the real and tantalizing clash between quantitative and qualitative approaches ... has something more to do with the ontology of science than with intellectual procedures" (Shweder, 1996, p. 176). The potential systematicity and high validity of qualitative analysis have been convincingly demonstrated, for example, in case studies, by transcendental realism or enlightened empiricism (see Miles & Huberman, 1984 ; Yin, 1984). However, qualitative research is not reducible to a postpositivist paradigm. In the context of the interpretativist-constructivist paradigm, qualitative methodologies share neither common ontological beliefs (related to the very nature of the reality of distress) nor epistemological beliefs (related to the relationship between the knower and the object of analysis) with quantitative approaches enclosed in the postpositivist paradigm. What is measured through symptom checklists is not distress as a language and a contextual sociocultural construct but rather as a reified picture of a standardized form of expression, an objectified constellation of distress manifestations, derived from the original matrix of signs. In spite of the convergence on methodological beliefs, there is incommensurability at the level of the ontological and, therefore, the epistemological beliefs related to the nature of distress.

Representational Dilemma

Our results illustrate what can be called a representational dilemma. We are facing two constructed representations of distress, each one showing a strong methodological and epistemological legitimacy. The first postpositivist representation, influenced by a scientific concern for order, sees distress as an empirically engineered reality. This distress is expressed-through correlation matrices, factors, and latent variables-as artificially and mechanically constructed realms of distress manifestation inferred from factorial structuration of symptom intercorrelations. This factorialbased, representational enterprise may be seen at a deeper level as a response to the need of scientists to respond to disorder by imposing coherence through the use of certain limited conceptual frameworks. As Latour and Woolgar (1979) note, "It is part of our world view that things are ordered, that order is the rule and that disorder should be eliminated wherever possible" (p. 251).

The second interpretativist-constructivist representation is based on cultural forms, which are the idioms that structure the language through which mental suffering is expressed and communicated. This representation is nearer to what Abbott (1992) calls "case narratives" or "fuzzy realities with autonomously defined complex properties ... engaged in a perpetual dialogue with their environment" (p. 65). Signs make sense only in the context of the narrative of the lived contextualized episode of distress. Distress is not a concrete reality whose boundaries and severity can be estimated by the use of sophisticated symptom checklists ; it is the original intangible matrix of signs and idioms from which each individual draws to elaborate an idiosyncratically constructed discourse that is then enacted on the social scene. However, we should not forget that idioms of distress are also, in some ways, a researcher's construct. They represent an alternate strategy to create order through the inference of cultural forms, which are implicit in any given culture. Moreover, the narratives that people reconstructed for the needs of our research are themselves structuring enterprises that bring order upon an experience that has not yet been the object of textualization. Therefore, both representations are products of a process of the objectification of an abstract reality : The first one is based on signs, meanings, and implicit forms ; the second one is based on symptomatological entities and correlation patterns. Both are scientific construals that try to integrate these empiricist and semiotic fragments of distress into coherent representations. Both are scientifically ordered representations.

These representations of distress differ also in their conception of the role played by configurations of symptoms and distress manifestations. Although factorial construals are passive variables, or products derived from a covariation matrix, idioms are active, dynamic structures that channel the way people experience and express distress. They are active, structuring, sociocultural forces as opposed to passive statistically inferred structures. As such, they avoid understating the extent to which individuals can be creative in the process of interpreting suffering. Factorial construals may be seen as common denominators in interitem correlations that are inert pictures of the covariance. They appear as preexisting patterns, passively waiting to be discovered. Idioms are active forces or organizing principles that give form to the lived experience of mental and social suffering. The representational dilemma has a dynamic-passive component.

A Politico-Representational Crisis

Our results illustrate what Denzin (1991) calls a "representational crisis" or the assumption that much, if not all, social science writings are narrative productions, structured by a logic that "separated writer, text, and subject matter" (p. 278). The author becomes the mirror of the world out there, and the subject is always a textual construction that cannot reveal its true nature. This first crisis is linked to a second legitimization crisis or an authority crisis based on the epistemological validity of research results and the power it gives to the researchers. Results presented here illustrate two constructed representations of distress, each with a high epistemological and methodological legitimacy. However, in the context of the utilitarian dominant discourse, the reified representation of distress appears more valuable because it can be used to assess, measure, and identify at-risk groups in epidemiological studies. To see distress as a lived experience of mental suffering enacted on the social scene or as a language and a constellation of signs that makes sense only when situated in specific biographies and the naturalistic sociocultural environments of the cases is to challenge these goals. It rejects the ontological validity of any reified representation of the dependent variable. We are clearly facing not only two incommensurable representations of distress but also a criticism of the dominant, instrumentalist rationality in the management of illness. As Abbott (1992) said about epidemiological statistic indicators,

Beyond these lies a broader, policy-minded audience that is presently under the complete sway of population/analytic methods. Regression coefficients now establish the parameters of public policy by providing both liberals and conservatives with "scientific" evidence of who is doing how well and how badly and what causal forces, lie behind those outcomes. (p. 79)

We agree with Abbott when he stresses the issue of political considerations and the mechanism of prioritization and decision making based on scientific indicators. He underlines another aspect of what might be called a teleological dilemma, that of the ultimate goals served by each representation of distress.

Ontological and Teleological Incommensurabilities :
Toward a Paradigm of Tolerance

Both approaches are confined to the humbleness of their respective means and scope. Each must accept their limits ensuing from the limitation of the universe of the research questions that they are able to address. For these reasons, paradigm dialogue must bear not only on the complementarity of qualitative and quantitative techniques and methods to reach the same given research objective but also on the mutual respect of the value of each respective universe of research questions. What we need is not simply a paradigm dialogue but a paradigm tolerance.

Measurement bias does not invalidate the elaboration and use of psychometric or epidemiological devices to produce the relevant data required by public health agencies to identify at-risk social groups or to evaluate the efficiency of mental health services. Moreover, in the above-mentioned research project, qualitative techniques and methods have allowed the elicitation of an original pool of contextualized distress manifestations with a high level of content validity and the production of a semantic lexicon for each item of the final scale. These results confirm the importance of a methodological complementarity that results in a stronger content validity of the scale. However, what we are now facing in our results are ontological and teleological incommensurabilities. The first ontological issue is related to the true nature of distress represented as either a language or as an objective, measurable state of mental health. The second teleological issue is the incommensurability of the ultimate goals of research, which are to understand instead of to demonstrate, to interpret instead of to objectify or measure, to reveal the plurality of the dimensions of reality instead of to look for consensus and norms. In fact, "we operate within a constructivist research paradigm, while our public health colleagues operate within the framework of logical empiricism" (Coreil, 1997, pp. 252-255). If there is to be a plea for tolerance, it should be at the level of the incommensurability of their ontology, epistemology, and, particularly, their teleology. This tolerance should not be toward their respective ways of practicing science but rather toward ways of seeing the world. Therefore, tolerance does not mean, as suggested by Glanz, Marcus Lewis, and Rimer (1997), that constructivist inquiry should be viewed as contributing primarily to the early stages of research or as helping to define more focused research questions for empirical study. We must transcend this reductionist perspective that confines qualitative research to the role of a preliminary and exploratory inquiry that is useful only as long as it strengthens the validity of quantitative designs, methods, and results. Phenomenological, ethnosemantic, and interpretative methods aim at other incommensurable objectives.

These two representations of distress are incommensurable in the Kuhnian (1970) perspective. To try to translate the language from which scientific work is conducted into the language of a substantially different paradigm leads to a loss of the specificity of the message. In our study, incommensurability can be expressed as a problem of translation between a message on meaning and a message on measure. Therefore, we can fully agree with Kuhn when he says that translation between diverse language communities is "a potent tool for both persuasion and conversion" (p. 202) but without any guarantee of success. Such a translation enterprise may allow the participant "to experience vicariously something of the merits and defects of each other's point of view" (p. 202). However, as he recognizes, "persuasion need not succeed, and if it does, it need not be accompanied or followed by conversion" (p. 202). We should add that the conversion of every postpositivist researcher into interpretativist-constructivist researchers must not be a goal. Instead, we must plead for a paradigm of tolerance toward different ontological and teleological research issues.


CONCLUSION

Qualitative and quantitative methodologies are incommensurable at the level of the finalities of the research but not on the basis of the methods as such. We must not confound teleological incommensurability and methodological incompatibility.

Methods that do not reach the same outcome because they aim at divergent objectives cannot be treated as antinomic. The symptom checklists used in epidemiological studies, including our own, should not claim to grasp the complexity of the virtual concept of psychological distress. They are predictive tools, not analytical ones.

Qualitative research must avoid the pitfalls of postmodern cognitive relativism. Insofar as we agree with the pertinence of research questions directed at the assessment and measurement of sociocultural factors, we must accept, as did Sokal and Bricmont (1997), that "we have to renounce the radical doubts about logic as it pertains to the possibility of learning about the world through experience" (p. 191). We must be careful not to sink into the excess of postmodernism that denies the existence of any reality except that reality expressed through the discourse. We must avoid replacing naive empiricism with a generalized cognitive relativism if we wish to set the tone for a dialogue between these two cultures. In fact, the phenomenological account we have made of these episodes proves that distress is a real suffering and a lived experience ; it is not pure meaning. We cannot deny the reality of distress as a state of mental health expressed through discrete cognitive, somatic, emotive, and behavioral manifestations.

Our use of both qualitative and quantitative methods has revealed more than it has solved the representational and legitimacy crises in the understanding of psychological distress. A few years hence, we will still have to address the task of agreeing on "the question of how to combine emic and etic approaches in an attempt to gain methodologically precise assessments of psychoemotional and somatic complaint" (Parsons & Wakeley, 1991, p. 113).


REFERENCES

Abbott, A. (1992). What do cases do ? Some notes on activity in sociological analysis. In C.C. Ragin & H.S. Becker (Eds.), What is a case ? Exploring the foundations of social inquiry (pp. 60-82). Cambridge, MA : Cambridge University Press.

Becker, H. (1996). The epistemology of qualitative research. In R. Jessor, A. Colby, & R.A. Shweder (Eds.), Ethnography and human development : Context and meaning in social inquiry (pp. 53-72). Chicago : Chicago University Press. Coreil, J. (1997). More thoughts on negotiating relevance. Medical Anthropology Quarterly, 11 (2), 252-255. Denzin, N.K. (1991). Back to Harold and Agnes. Sociological Theory, 9, 278-285.

Doise, W., Clemence, A., & Lorenzi-Cioldi, F. (1992). Représentations sociales et analyses de données [Social representations and data analyses]. Grenoble, France : Presses Universitaires de Grenoble.

Glanz, K., Marcus Lewis, F., & Rimer, B.K. (1997). Linking theory, research, and practice. In K. Glanz, F. Marcus Lewis, & B.K. Rimer (Eds.), Health behavior and health education : Theory, research, and practice (2nd ed., pp. 19-35). San Francisco : Jossey Bass.

Gotlib, I.H., & Cane, D.B. (1989). Self-report assessment of depression and anxiety. In P.C. Kendall & D. Watson (Eds.), Anxiety and depression : Distinctive and overlapping features (pp. 131-163). New York : Academic Press.

Jessor, R. (1996). Ethnographic methods in contemporary perspective. In R. Jessor, A. Colby, & R.A. Shweder (Eds.), Ethnography and human development : Context and meaning in social inquiry (pp. 3-14). Chicago : Chicago University Press.

Kuhn, T. S. (1970). The structure of scientific revolutions (2nd ed.). Chicago : University of Chicago Press. Latour, B., & Woolgar, S. (1979). Laboratory life : The social construction of scientific facts. Beverly Hills, CA : Sage.

Link, B., & Dohrenwend, B.P. (1980). Formulation of hypotheses about the true prevalence of demoralization in the United States. In B.R Dohrenwend, M. S. Gould, B. Link, R. Neugebauer, & R. Wunsch­

Hitzig (Eds.), Mental health in the United States : Epidemiologic estimates (pp. 140-162). New York : Praeger.

Massé, R. (1999). Les conditions d'une anthropologie sémiotique de la détresse psychologique [For a semiotic anthropology of psychological distress]. Recherches Sémiotiques/Semiotic Inquiry, 19, 39-61. [Texte disponible dans Les Classiques des sciences sociales. JMT.]

Massé, R., Poulin, C., Dassa, C., Lambert, J., Bélair, S., & Battaglini, A. (1998a). Élaboration et validation d'un outil de mesure de la détresse psychologique dans une population non clinique de Québécois francophone [Results of the construct and criterion validation of the Psychological Distress Manifestations Measure Scale (PDMMS) is a general population]. Canadian Review of Public Health, 89 (3), 183-189.

Massé, R., Poulin, C., Dassa, C., Lambert, J., Bélair, S., & Battaglini, A. (1998b). The structure of mental health : Higher-order confirmatory factor analyses of psychological distress and well-being measures. Social Indicator Research, 45, 475-504.

Migliore, S. (1997). Mal'uocchiu : Ambiguity, evil eye, and the language of distress. Toronto, Canada : University of Toronto Press.

Miles, M.B., & Huberman, A.M. (1984). Qualitative data analysis : A sourcebook of new methods. Beverly Hills, CA : Sage.

Mishler, E.G. (1996). Missing persons : Recovering developmental stories/histories. In R. Jessor, A. Colby, & R.A. Shweder (Eds.), Ethnography and human development : Context and meaning in social inquiry (pp. 73-99). Chicago : Chicago University Press.

Nagel, T. (1979). Subjective and objective. In T. Nagel (Ed.), Mortal questions (pp. 196-213). New York : Cambridge University Press.

Nichter, M. (1981). Idioms of distress : Alternatives in the expression of psychosocial distress : A case study from south India. Culture, Medicine and Psychiatry, 5, 379-408.

Parsons, C.D.F., & Wakeley, P. (1991). Idioms of distress : Somatic responses to distress in everyday life. Culture, Medicine and Psychiatry, 15, 111-133.

Shweder, R.A. (1996). Quanta and qualia. In R. Jessor, A. Colby, & R.A. Shweder (Eds.), Ethnography and human development : Context and meaning in social inquiry (pp. 175-182). Chicago : Chicago University Press.

Sokal, A., & Bricmont, J. (1997). Impostures intellectuelles. Paris : Odile Jacob Editor.

Weissman, M.M., Myers, J.K., & Ross, C.E. (1988). Community surveys of psychiatric disorders. Piscataway, NJ : Rutgers University Press.

Yin, R. K. (1984). Case study research : Design and methods. Beverly Hills, CA : Sage.

Raymond Massé is a medical anthropologist and a professor in the Department of Anthropology at Laval University, Quebec City, Quebec, Canada. He teaches qualitative methodology and anthropology of health. His main interest is in the sociocultural construction of mental health.