Response to "Positivism and Qualitative Nursing Research"

Article excerpt

John Paley argues that "most qualitative research (in nursing, as in many other disciplines) is classically positivist. . . . The main aim of this article . . . is to convince them [qualitative researchers] that they really are positivists, and persuade them that they should strive to be better positivists." There is something correct about this claim and yet it is terribly wrong.

What is correct is that there is no inherent conflict between positivism and qualitative research traditions and methods. Logical positivism developed a fundamentally qualitative approach to scientific methodology and devoted considerable effort to show how quantitative research methods could be grounded in qualitative observation.

What is terribly wrong is the author's portrayal of positivistic philosophy as "pro-observation, pro-induction, pro-plausibility, and pro-subjectivity . . . also anti-cause, anti-realist, anti-explanation, anti-correspondence, anti-truth." It is unclear whether any logical positivist ever held all of those positions. While what Paley champions possibly is a coherent "positivistic" philosophy, it is idiosyncratic and certainly marginal to main threads and developments in logical positivism.

The author's portrait of positivism is synthesized from diverse claims made by various persons in the positivist tradition and other post-positivists such as van Fraassen. Methodologically, its flaw is to assume that positivism is a single monolithic philosophy where all positivists agree methodologically, epistemologically, and metaphysically. That decidedly is not the case for logical positivism-the movement that, judging from citations, Paley largely concerns himself with.

In this response I will try to give an accurate portrait of the diversity of positivistic views impacting on qualitative vs. quantitative issues, carefully set out the core positivistic views that ground quantitative research methods in qualitative observation, and show how the compatibility of qualitative research methods and positivism does not require such objectionable and suspect positions as being "anti-cause, anti-realist, anti-explanation, anti-correspondence, anti-truth."

OVERVIEW OF LOGICAL POSITIVISM AND ITS DIVERSE VERSIONS

Logical positivism is a 20th-century philosophical movement concerned with meaningful scientific discourse and appropriate methods for keeping scientific claims meaningful. The connections between earlier 19th century positivisms such as Auguste Comte's are tenuous and somewhat obscure. Comte's positivism was an evolutionary perspective on human thought that was tied to a rejection of the French Revolution, the founding of sociology, and the attempt to establish a new secular religion with scientists as a new order of priests (Comte, 1830/1987). Most of this has little to do with the issues with which logical positivism concerned itself. The only significant and relevant overlaps are the ideas that scientific knowledge is grounded in the observable and that the methods of the natural sciences can be applied to the social sciences.

The positivist who most influenced logical positivism was Ernst Mach. According to Mach (1886/1959) science should only describe regularities in the sensations scientists experience when observing. Those descriptions should be resolutely antimetaphysical in the sense that no appeal should be made to what cannot be observed. For Mach this excluded even appeal to physical objects. Mach's views strongly influenced Einstein in his development of relativity theory and many viewed relativity as largely vindicating Mach's brand of positivism. There was a serious problem, however: Part of the substantive empirical content of relativity was the Lorenz group of transformations (roughly describing how to translate observations from one frame of reference or perspective into another) and that empirical part of relativity went beyond what could be experienced in sensations. Thus Einstein's success simultaneously showed the promise and the inadequacy of Machean positivism.1

Logical positivism emerged as a specific approach for modifying Machean positivism so as to accommodate the success of relativity theory. In doing so a central focus was restricting the language of science to meaningful discourse by requiring that scientific assertions have observable consequences.

This was done as follows. An ideal scientific language would contain three kinds of terms: observation terms, logical terms, and theoretical terms. The observation terms would refer to things that could be directly experienced (variously construed as Machean sensations or physical things). Singular assertions using observation terms would be true or false in a correspondence sense (they do or do not correctly describe the way things are) and could be verified by simple observation. Logical terms are used to express mathematical and other logical relations, are analytic (in the sense that their truth is determined on the basis of meanings alone), and thus have no empirical content. Theoretical terms are used to make factual assertions about non-observables. They are problematic in that they cannot be directly verified, hence there must be severe restrictions on what theoretical terms are allowed. The basic insight is that to be meaningful theoretical assertions must be suitably tied to verifiable observable conditions via some sort of interpretative sentences. Various proposals were advanced, of which the most important for present purposes is the requirement that theoretical terms must be operationally defined. Such interpretative sentences use logical terms to relate theoretical terms to observable, verifiable conditions.

On this scheme singular statements about observables are given a realist interpretation wherein they describe objective conditions, are true-or-false in a correspondence sense, and are verifiable. Thus contra Paley it is grossly misleading to claim logical positivism is anti-realist, anti-correspondence, and anti-truth.

The status of generalizations about observables is more problematic, and there is a great diversity of views among logical positivists. Most logical positivists held some sort of inductive view about the confirmation of generalizations. Induction typically was viewed probabilistically, that is, a proposition would be confirmed if it enjoyed a suitably high probability. Carnap (1945), however, denied that one could use evidence to give generalizations non-zero probabilities, hence generalizations could not be confirmed in this sense. Rather, the point of induction was to justify applying the generalization to the next instance. Popper (1959) denied the possibility of induction and proposed that generalizations could be falsified but not inductively confirmed.

With regard to theoretical-term assertions some positivists denied that they were true or false, viewing them as calculating instruments (instrumentalism). Others maintained that the full realistic interpretation of the observation-term assertions combined with the interpretative sentences gave theoretical terms an implicit definition which made theoretical assertions refer to nonobservables (realism) and have a truth value in the correspondence sense. They could be tested only indirectly, however, by examining the correctness of observable predictions and could not be verified. The most influential versions of trie positivist analysis of theoretical discourse are due to Carnap (1936-1937,1956a) and Hempel( 1952,1963), and they both give a realistic interpretation of theoretical assertions based in correspondence notions of truth. Such positions are not anti-theoretical (Hempel, 1958).

The logical positivistic account of meaningful scientific discourse is the core of logical positivism, albeit a core with many variations and differences in detail. To that core a variety of other views can be annexed. Carnap and many other logical positivists held a unity of science view wherein all of scientific knowledge could be reduced to basic physics and wherein the same methods were appropriate to all of the sciences including the human sciences (Neurath, Carnap, & Morris, 1936- 1969). But a splinter group, known as General Systems Theory, was resolutely holistic and anti-reductionistic, maintaining that new properties and kinds emerge at various levels (Suppe, 1982; Von Bertalanffy, 1976). Both these reductionistic and holistic positions shared the basic logical positivistic views on meaningful scientific discourse.

Some logical positivists denied that explanation was possible given the restrictions on meaningful scientific discourse. Those that did tended to view explanation as making fundamental recourse to underlying non-observable entities and processes while holding non-realist (e.g., instrumentalist) views about theoretical assertions. But the classic literature on explanation conies from logical positivism. Carl Hempel (1965) developed what is known as the covering law model of explanation wherein explanation is simply post-hoc prediction: To explain an event is to show that the circumstances allow the application of a law which could have been used to predict the event. This is the most influential account of scientific explanation to date, and it embodies and exploits the core logical positivistic view of meaningful scientific discourse in a resolutely realist manner.

Logical positivists spent a lot of time analyzing causal language and attempting to make sense of causal claims in science. Carnap (1956) and others explored causal modal logics. Reichenbach (1956,1957) spent much of his career working on notions of causality in the context of relativity theory. Later post-positivists such as van Fraassen (1969) have been concerned to accommodate causal modalities in the language of science.

Thus much of the logical positivistic corpus is resolutely realist, committed to correspondence notions of truth, views theory and explanation as central to science, and sees causality as having an appropriate place in scientific discourse. And these are the views of the most central, influential positivists such as Carnap, Hempel, Reichenbach, Nagel (1961), and others.

OBSERVATION, OPERATIONAL DEFINITION, AND QUANTITATIVE MEASURES

Logical positivists were concerned with how quantitative assertions could be grounded in observation. To that end they developed theories of extensive measurement. One defined measurement scales in a manner that used arbitrary units tied to some sorts of physical standards such as standard weights or meter rods. The idea was that one looked for congruence between some combination of these standards and a physical thing. To measure length one laid a meter stick and observed what mark was congruent with the far edge of the object. To measure weight one added weights to a balance until the dial was congruent with the balance point. The epistemic heart of measurement was thus observing qualitative congruence (Suppe, 1989).

Logical positivism wasn't the only movement concerned with the nature of quantitative measurement. P. W. Bridgman (1938, 1952) coined the term operationalism for the view that the basic content of science was the operations performed by the scientist. S. S. Stevens (1946) and other behaviorists were concerned with psychological measurement and extended the notion of extensive measurement to include questionnaire and other test items as the basis for defining psychometric scales. Test items have "yes-no" or multiple choice responses, and the assessment of each item is qualitative. From the qualitative responses to the instrument, use of a measurement scale produces quantitative measures of performance. Behaviorist psychology introduced the notion of operational definition to specify the relationships between the qualitative test items and the quantitative measurement scales (Suppe, 1998).

It is important to note that operational definition is not restricted to quantitative or measurement terms. Any qualitative coding scheme for classifying observed behaviors or conditions is an operational definition of the classification categories (Lenz, Gift, Pugh, Mulligan, & Suppe, 1995).

There was considerable interaction between Bridgman, Stevens, Carnap, and Hempel regarding measurement and operational definition. The influence was largely from operationalism and behaviorism to logical positivism, not vice versa (Suppe, 1998). Carnap's (1936-1937) reduction-sentence account of interpretative sentences essentially captures the behaviorist notion of operational definition in cases where intervening variables are allowed.

Whatever the version, measurements constitute a systematic means for applying numerical names to qualitative gradable conditions such as length and mass in a manner that allows relationships between the qualitative situations to be investigated through mathematical analysis using operations defined for the measurement scales in question (Stevens, 1946).

Logical positivism's account of quantitative science is thus tied to a theory of measurement that grounds measurement in qualitative observations of congruence. For the positivists, scientific knowledge is fundamentally qualitative and the quantitative reduces to the qualitative.

Indeed, this proved to be something of a problem for logical positivism, since in physics one encounters real-valued and complex-valued measurement scales, where different measures cannot be correlated uniquely with observable congruences. (The problem is that the number of congruences with standards is countably infinite but the number of real-valued or complex-valued measures is uncountably infinite; hence there is an infinity of different real-valued or complex-valued measures associated with a given congruence with standards.) In practice measurement error and limitations in instrument reliability make this only of theoretical/philosophical concern. But it is troublesome, since it calls into question the basic logical positivistic strategy of making observability the basis for meaningful scientific discourse. For example, it caused Hempel (1952) to abandon the requirement that theoretical terms be operationally defined, because real-valued measure theoretical terms could not be so defined. Instead he imposed the weaker requirement that theories as a whole have some non-trivial observable consequences.

QUALITATIVE RESEARCH NEED NOT BE ANTI-THEORETICAL, ANTI-CAUSE, ANTI-REALIST, ANTI-CORRESPONDENCE, OR ANTI-TRUTH

Paley is correct that there is no incompatibility between positivism and qualitative research. For logical positivism, quantitative research is at bottom qualitative research using systematic numerical names for gradable properties, relations, and conditions. This compatibility rests in the core features of the logical positivist account of meaningful scientific discourse and positivist/behaviorist accounts of measurement.

The compatibility is most easily obtained under versions where theoretical assertions are given a realistic interpretation. For then it is clear to what the systematic names (measures) refer: accelerations, velocities, masses, and so forth. If not interpreted realistically, then they presumably refer to the measurement procedures themselves (Bridgman) or are just a shorthand for some complex qualitative description of observable conditions. And in that case, notions such as construct validity do not seem applicable. (Construct validity concerns determining whether a measurement procedure and scale measures what it purports to when we don't have a pre-existing procedure for measuring it and it typically is unobservable [Cronbach & Meehl, 1955].)

We see, then, that the compatibility of positivism and qualitative research methods need not be tied to questionable post-modern positions such as rejection of truth, correspondence, realism, and the like. Positivism allows both qualitative and quantitative research to be theoretical so long as theoretical assertions have suitable connections to the observable. The covering-law model of explanation is fundamentally qualitative but can be extended to the quantitative through a positivistic/behaviorist account of measurement, so both quantitative and qualitative research have equal potential for being explanatory. So, too for attributions of causality.

The core positivistic position on meaningful scientific discourse is compatible with both reductionistic and holistic accounts of social, psychological, and other higher-order properties and the terms that refer to them. Indeed, as already noted, the account was developed in both versions by the Unity of Science movement and by the General Systems Theory movement, respectively. So, too, the compatibility of qualitative research with logical positivism is consistent with views that insist all science is methodologically the same and Verstehen positions that maintain that the human sciences require methods and explanations distinct from those of the natural sciences (Dallmayr & McCarthy, 1977).

CONCLUSION: THE SPURIOUS QUANTITATIVE VERSUS QUALITATIVE DIVIDE IN NURSING RESEARCH

We have seen why there is no intrinsic conflict between mainstream logical positivism and qualitative research methods. To the extent logical positivism provides a viable account of scientific knowledge and methodology, it shows quantitative research to be fundamentally qualitative. Hence, there is no fundamental conflict between Qualitative and Quantitative Research methods. Whatever conflicts occur are ideological, not epistemological.

Paley suggests, however, that logical positivism fails as a philosophy of the natural sciences. And indeed it does (Suppe, 1977,1989,1996,1998a, 2000). What is the implication of that failure for reconciling the qualitative and the quantitative research methods in nursing?

Positivism has been rejected for three main reasons:

1. The observable-theoretical term distinction is untenable;

2. When that distinction is abandoned the positivistic account of meaningful scientific discourse collapses;

3. Operational definitions do not provide a satisfactory account of the relations between theory and experiment/observation (Suppe, 1997, 2000).

These defects apply equally to qualitative and quantitative research. So the failure of logical positivism as an account of science does not favor either qualitative or quantitative research methods.

Actually, the compatibility of qualitative and quantitative research does not require an observable-nonobservable distinction or the logical positivistic account of meaningful scientific discourse. Operational definitions can account for the relationships between measurements and measurement procedures without giving a satisfactory general account of all the relations between theory and experiment. What is crucial to the positivist account of the compatibility of qualitative and quantitative research methods is the claim that in actual measurement practice the determination of measurement values is based in qualitative assessment of congruence between instruments and attributes of things or persons-rulers and lengths, MMPI or other psychological characteristics and item responses, and so forth.

Any account of scientific practice and knowledge that grounds actual measurement practices in such qualitative assessment has what it takes to accommodate both qualitative and quantitative research methodologies.

To demand that one exclusively use just one of these approaches is spurious and unwarranted. It is to lose sight of the fact that there can be good and bad qualitative research just as there can be good and bad quantitative research. The focus of discussion should be evaluating the appropriateness of specific research designs for specific nursing research problems, not ideological haranguing over the absolute superiority or inferiority of qualitative versus quantitative research.

NOTE

1 For details and documentation of the account being sketched here and below, see Suppe, 1977, 1996, 1998a.

[Reference]

REFERENCES

Bridgman, P. W. (1938). Operational analysis. Philosophy of Science, 5, 114-131.

Bridgman, P. W. (1952). The nature of some of our physical concepts. New York: Philosophical Library.

Carnap, R. (1936-1937). Testability and meaning. Philosophy of Science, 3(4), 420- 468, 1-40.

Carnap, R. (1945). On inductive logic. Philosophy of Science, 12, 92-97.

Carnap, R. (1956). Meaning and necessity (Enlarged ed.). Chicago: University of Chicago Press.

Carnap, R. (1956a). The methodological character of theoretical concepts. In H. Feigl & M. Scriven (Eds.), Minnesota studies in the philosophy of science, Vol. I (pp. 33-76). Minneapolis: University of Minnesota Press.

Comte, A. (1987). The positive philosophy. New York: AMS Press. (Original work published 1830)

Cronbach, L., & Meehl, P. (1955). Construct validity in psychological tests. Psychological Bulletin, 52, 281-302.

Dallmayr, F., & McCarthy, T. (1977). Understanding and social inquiry. Notre Dame: University of Notre Dame Press.

Hempel, C. G. (1952). Fundamentals of concept formation in empirical science. Chicago: University of Chicago Press.

Hempel, C. G. (1958). The theoretician's dilemma. In H. Feigl, M. Scriven, & G. Maxwell (Eds.), Minnesota studies in the philosophy of science, Vol. II (pp. 37-98). Minneapolis: University of Minnesota Press.

Hempel, C. G. (1963). Implications of Carnap's work for the philosophy of science. In P. Schilpp (Ed.), The philosophy of Rudolf Carnap (pp. 685-709). LaSalle, IL: Open Court.

Hempel, C. G. (1965). Aspects of scientific explanation and other essays in the philosophy of science. New York: Free Press.

Lenz, E., Gift, E., Pugh, L., Mulligan, R., & Suppe, F. (1995). Collaborative development of middle-range nursing theories: Toward a theory of unpleasant symptoms. Advances in Nursing Science, 17(3), 1-13.

Mach, E. (1959). The analysis of sensations. New York: Dover. (Original work published 1886)

Nagel, E. (1961). The structure of science. New York: Harcourt, Brace.

Neurath, O., Carnap, R., & Morris, C. (Eds.). (1936-1969). International encyclopedia of unified science (2 Vols.). Chicago: University of Chicago Press.

Popper, K. R. (1959). Logic of scientific discovery. London: Hutchinson.

Reichenbach, H. (1956). The direction of time. Berkeley: University of California Press.

Reichenbach, H. (1957). Philosophy of space and time. New York: Dover. (Original work published 1928)

Stevens, S. S. (1946). On the theory of scales of measurement. Science, 103, 677-680.

Suppe, F. (1977). The structure of scientific theories (2nd ed.). Urbana: University of Illinois Press.

Suppe, F. (1982). Implications of recent developments in philosophy of science for nursing theory. In E. R. Lenz (Ed.), Proceedings of the Fifth Biennial Eastern Conference on Nursing Research (pp. 9-16). Baltimore: University of Maryland School of Nursing.

Suppe, F. (1989). The semantic conception of theories and scientific realism. Urbana: University of Illinois Press.

Suppe, F. (1996). Scientific theories. In D. Borchert (Ed.), The Encyclopedia of Philosophy Supplement (pp. 521-524). New York: Macmillan Reference USA.

Suppe, F. (1998). Operationalism. In E. Craig (Ed.), Routledge Encyclopedia of Philosophy, Vol. 7 (pp. 131-136). New York and London: Routledge.

Suppe, F. (1998a). Scientific theories. In E. Craig (Ed.), Routledge Encyclopedia of Philosophy, Vol. 9 (pp. 344-355). New York and London: Routledge.

Suppe, F. (2000). Understanding scientific theories: An assessment of developments 1969-1999. Philosophy of Science, 67(Suppl.), S102-S115.

van Fraassen, B. (1969). Meaning relations and modalities. Nous, 3. 155-167.

Von Bertalanffy, L. (1976) General system theory: Foundations, development, applications (Rev. ed.). New York: George Brazillier.

[Author Affiliation]

Frederick Suppe, PhD

Texas Tech University, Lubbock

[Author Affiliation]

Offprints. Request for offprints should be directed to Frederick Suppe, PhD, Philosophy Department, Box 43092, Texas Tech University, Lubbock, TX 79409-3092.