History of Q Methodology

The History and Principles of Q Methodology in Psychology and the Social Sciences
Steven R. Brown Department of Political Science Kent State University Kent, Ohio (USA)

Abstract. Q methodology was innovated by British physicist/psychologist William Stephenson (1902-1989), but has been applied and has continued to evolve primarily in the United States and outside academic psychology, most notably in the fields of communication and political science, and more recently in the health sciences. The principles of Q methodology are restated, and contrasts are drawn with the earlier understandings prevalent in 1930s British psychology, with contemporary illustrative applications drawn from a variety of disciplines. The conclusion is reached that adherence to an outdated Newtonianism plus concern for psychometric assessment led British psychology to embrace R methodology and to miss Q's parallels with quantum theory and its implications for a science of subjectivity, and that postmodern developments have enabled social scientists, including a new generation of British psychologists, to reestablish contact with Q methodology and to take advantage of the leverage which it provides in understanding human behavior.

What has come to be referred to as Q methodology was introduced in a letter to Nature, written by William Stephenson (1935a), a physicist (Ph.D. 1926, University of Durham) and psychologist (Ph.D. 1929, University of London), who served as the last assistant to Charles Spearman, the inventor of factor analysis. Spearman once referred to his protégé as the most creative statistician in psychology, but from virtually the moment of its inception, the broader considerations of Q as a methodology were destined to be controversial and to be shunned by most of academic psychology. Today, Q methodology is being widely adopted in the social sciences, but for the most part is little remembered in psychology itself, save (in the United States) for the technical procedure of Q sorting. Only recently has there been evidence that a younger generation of psychologists is rediscovering Q methodology and becoming acquainted with the vision which William Stephenson promoted for more than a half century.

Stephenson's most celebrated work was The Study of Behavior: Q-technique and Its Methodology (1953), and something of the controversy surrounding Q and its originator can be glimpsed by revisiting some of the reviews which greeted this publication. Charlotte Banks (1954) of Britain, for instance, while noting "Stephenson's cheerful belligerence," his "lively and entertaining style," and his "new and original ideas," also implied that some of his innovations had been preceded by Stern and Burt in particular. Banks may have been encouraged in this regard by Burt (1955) himself, who also referred to Stern's prior work as a way to assert his own position contra Stephenson--namely, that "if we confine ourselves to measurements obtained on a single occasion, we may either average the persons and correlate the traits, or average the traits and correlate the persons." Thus, for Burt, there was always only a single matrix of data that was at issue, and multiple ways to average across that matrix. The most stinging criticism perhaps came from Eysenck (1954), who accused Stephenson of "a somewhat disingenuous tendency to change the meaning of the term Q-technique over the years, whilst pretending that what he now means by it is what he has meant by it all along."

In the United States, McNemar (1954) criticized the author of The Study of Behavior for being obscure and for attacking "such intellects as Godfrey Thomson and Cyril Burt," and was especially skeptical of the value of single-case studies. In the most thorough critique, to which Stephenson (1954a) was invited to respond, Cronbach and Gleser (1954) summarized technical innovations such as card sorting and the incorporation of Fisher's principles of experimental design, but then issued a warning: "It is imperative to discourage students of personality and social psychology from copying Stephenson's designs as he presents them" (p. 330, emphasis in original). Stephenson's alleged showiness and carelessness were also noted. Finally, Turner (1955) accused the author of Q methodology of "misplaced contentiousness," "repetitiousness," of "dwelling on irrelevancies" and making "excessive claims," and of "apparent unfamiliarity with much work others have done," but also concluded that Q would "undoubtedly stand with Guttman scaling as one of the two most important recent contributions to technique."

Once a step is taken outside academic psychology and psychometry, however, the mood changes somewhat. Psychiatrist Bernard Glueck (1954) welcomed Q as furnishing "the long-awaited stable and dependable frame of reference" for addressing the "universality of uniqueness." Russell Ackoff (1955), one of the founders of operations research, predicted that "this book will have to be taken into account in psychological methodology for a long time to come." And psychotherapist Lyman Wynne, while noting Stephenson's "florid and rather megalomanic style of writing," also stated that the book would have "widespread, immediate appeal to the clinical investigator in psychiatry and related fields." Social worker Gershenson (1955), on the other hand, while complaining that The Study of Behavior constituted "a running argument between the author and his critics," acknowledged that much of what seemed to him incomprehensible stemmed from the lack of statistical sophistication of his field.

The Roots of the Q-R Controversy

Much of the confusion concerning the relationship between Q and R as factor-analytic procedures arose due to the fact that what Cyril Burt meant by "correlating persons" (Burt, 1937, 1940) and what Stephenson meant by it (Stephenson, 1935b, 1953) were, in at least two respects, quite separate things, a matter about which at one point they agreed to disagree (Burt & Stephenson, 1939):

For Burt, both factor systems had foundations in one and only one basic data matrix which was typically filled with scores from objective tests, e.g., of intelligence, reading speed, or mathematical ability; whereas,

For Stephenson, two separate data matrices were at issue, the one containing objective measures (R) and the other containing data of a subjective kind (Q).

When Burt insisted on the reciprocity of Q and R, therefore, a principle to which he adhered until the very end (Burt, 1972), he was referring to two complementary matrix manipulations (one by columns, the other by rows) of the same transposed data matrix, whereas for Stephenson (1953), "There never was a single matrix of scores to which both R and Q apply" (p. 15).

From his very first pronouncement, Stephenson (1935a) made clear that R methodology referred to "a selected population of n individuals each of whom has been measured in m tests" (note the passive verb has been measured), and that Q methodology referred to "a population of n different tests (or essays, pictures, traits or other measurable material), each of which is measured or scaled by m individuals" (note the active verb is measured). In the former case, something is done to the person, as when we take blood pressure or measure height: This is the objective mode and the person's stance relative to measurement is passive. In the latter case, the person actively does something, i.e., measures or scales a population of measurable material: This is the subjective mode insofar as measurement is from the person's standpoint. As Stephenson (1935b) later stated, "Previously individuals obtained scores; now the tests get them instead, due to the operation of the individuals upon them" (p. 19), and in the first worked example which he presented, a group of persons rank-ordered a set of colors in terms of their pleasingness, which is clearly subjective. These two modes are incommensurate (Brown, 1972) and in this sense nonreciprocal, and it is also in this sense that Q methodology provides the basis for a science of subjectivity (Brown, 1994-1995).

What is fundamentally at issue was originally suggested by Stephenson (1936a) and has been most clearly demonstrated in a study of 25 bodily measurements--arm length, thigh width, foot length, etc.--each in inches for 20 persons (for details, consult Brown, 1972, 1980, p. 13). When submitted to an R factor analysis, the outcome was a segmentation of body parts across eight factors, with shoulder width, arm length, width of palm, and foot length (among others) significant in one factor, chest and waist width in another factor, and so forth. Breaking wholes into parts (analysis) is what R factor analysis does. When the original data matrix was transposed, however, and then refactored, the result was a single dimension accounting for 99% of total variance, a reflection of the fact that people are physically about the same. Keeping parts together in their interrelation (synthesis) is what Q factor analysis does. After correlating and factoring persons, the factor scores (now associated with body parts) were given to an artist who was asked to draw a body proportionate to the scores, and the recognizably-human result is shown in Figure 1a.


Figure 1a	Figure 1b

The prior results rely on the same data matrix, and the measurements are in the objective mode: A person's arm is so-many inches long and there is very little that can be done about that fact, and this holds true whether correlating traits or persons. This is Burt's version of correlating persons. Stephenson's version, however, involves a separate matrix of subjective data, and this was effected in this instance by instructing the same 20 individuals to assess the same body parts in terms of their "significance for me." The four resulting Q factors revealed different saliencies relative to the body parts included in the sample: The first factor, for example, placed great emphasis on eyes, head, and mouth, which received the highest factor scores; whereas the second factor placed most emphasis on trunk, hips, and chest. As before, the factor scores for the first Q factor were also submitted to an artist, whose depiction is in Figure 1b.

It is noteworthy that the outcome of the R factor analysis above cannot be given artistic form in the same manner as Figure 1 due to the fact that R dissembles reality so that all the King's artists cannot put it back together again. However, neither can the grotesqueness of Figure 1b relative to 1a be considered indicative of the accuracy of Burt's approach to correlating persons compared to Stephenson's; rather, it arises from the expression in material form of two phenomena, only one of which is actually material. Physical extension, such as arm length, is easily pictured in spatial terms, but subjectivity is not spatial in this physical sense, and much of Psychology's difficulty with Q methodology can be attributed to this conceptual distinction. Stephenson was concerned to bring the elusiveness of subjectivity under the suzereignty of the same science that he had come to know so well as a physicist, and to employ the technology of factor analysis to this end. Academic Psychology, meanwhile, was preoccupied with a search for objective abilities and universals, such as Spearman's g and Eysenck's introversion-extraversion, which are also expressed as factors but which leave no room for subjectivity (except as error variance).

Burt's version of correlating persons had the additional advantage of retaining a familiar psychometric world: The objective measurements that provided the basis for correlating traits were also the basis for correlating persons, and so no new concepts were necessary, only a different way of looking at them. As Stephenson was aware, however, the study of subjectivity required a different way of thinking that moved away from conventional factor analysis, with its roots in Newtonian dimensions and causality, and in the direction of new developments in science and alternative ways of conceiving human behavior that are only now beginning to come into vogue. The major difference between Burt and Stephenson, therefore, turned not on the mechanics of factor analysis, but on what it was that was to be measured, and how.

The Current Status of the Q-R Controversy

More than a half century has elapsed since Cyril Burt and William Stephenson squared off against one another, and there is little doubt that Burt's version of the connection between Q and R has had the more pronounced effect: It makes sense to a scientific worldview populated by causes, effects, and variables. His elaboration of P, Q, R, and other techniques (all related to one and the same basic data matrix) became foundational for other eminent factorists, such as Raymond Cattell, Hans Eysenck, J.P. Guilford, Godfrey Thomson, and L.L. Thurstone, no less than the two generations of psychometricians who followed their lead, so that it has become virtually impossible to read a standard text on factor analysis without learning that Q and R are based on transpositions of the same data matrix, hence are reciprocal.

This dominant point of view has come to be more widely questioned during the past two decades, however, as a new generation of health, social, and psychological scientists has gained a deeper appreciation of what Stephenson was trying to say. The current status of the Q-R debate is revealed in a recent study which utilizes Q methodology to examine the diverse understandings of Q methodology itself. The inquiry is reported in detail elsewhere (Brown, 1993), and it had its beginnings in differences of opinion about the nature of Q methodology which were posted on one of the many electronic discussion groups found on the Internet. Most of the comments expressed will come as little surprise to those familiar with the issues, and include the following:

It is intended to get at patterning within individuals (case-wise) rather than simply across individuals (factor-wise sorting).

It uses an ipsative technique of sorting a representative set of subjective statements drawn from a concourse of possible feelings or reactions about a subjective condition.

Row standardization is implicit in computer correlations between row vectors. Data are examined relative to the individual's mean.

SPSS (with its Flip routine) and SAS are generally inappropriate since they assume reciprocity.

Statements of this kind are subjective in the sense in which Stephenson meant the term inasmuch as they emanate from a particular vantagepoint: They are assertions and as such are rarely subject to proof or disproof, which is not to say that evidence and argument cannot be amassed in favor of each. The volume of statements on a topic was originally referred to as a population or trait universe (Stephenson, 1950, 1953), but has been rechristened concourse (Stephenson, 1978) to connote the running together of ideas in thought. Concourse is the common coinage of societies large and small, and is designed to cover everything from community gossip and public opinion to the esoteric discussions of scientists and philosophers.

    From the larger body of statements is drawn a Q sample designed to represent the range of subjective communicability in the concourse and to be used for experimental purposes. To assist in achieving representativeness in the Q sample, Stephenson typically relied on principles of experimental design (Fisher, 1935)--for instance, in the above case, by drawing statements representing the views of (a) Stephenson and (b) Burt, respectively, and also by including statements addressing © technical as opposed to (d) philosophical matters, all statements being incorporated into a 2×2 factorial with four cells: ac, ad, bc, and bd. m=5 replications of each combination resulted in a Q sample of size N=20 which, although somewhat small by most standards, is quite adequate for demonstrating the principles involved in the application of Q methodology as currently employed.

    The procedure of Q sorting was considered something new in Banks' (1954), Eysenck's (1954), and Cronbach and Gleser's (1954) reviews of The Study of Behavior, although the only thing truly new was the term itself: More than a decade before it even had a name, Stephenson employed what he came to call the Q sort in his very first published illustration (Stephenson, 1935b), in which a group of persons sorted a set of 60 colors from those they liked best to those they liked least. Later, 15 adults sorted a set of 50 postcards (containing pictures of vases) from most to least aesthetically pleasing (Stephenson, 1936b), and 40 boys and 40 girls sorted the titles of 60 school examinations in terms of which they would prefer most/least to sit for (Stephenson, 1936c). The operation was therefore not new in 1953.

    Fundamentally, Q sorting calls for a person to rank-order a set of stimuli according to an explicit rule (condition of instruction), usually from agree (+5) to disagree (-5), with scale scores provided to assist the participant in thinking about the task. The operation is inescapably subjective in the sense that the participant is sorting the cards from his or her own point of view, and it is this subjectivity which seems to have escaped Burt (1972): Referring to a 1915 study of his own in which observers scored children for the frequency with which they displayed certain emotions, he averred that "this method, I fancy, broadly corresponds with what Stephenson calls a 'Q sort'" (p. 42), but the frequency with which an emotion occurs is objective and quite different from a person's asserting (via Q technique or some other instrumental means) a preference for statement x over statement y, which is subjective.

    In the study under consideration, the Q sample consisted of the 20 statements drawn from the concourse noted above. In a typical Q study, a sample of participants (P set) would be invited to represent their own views about the Q-R debate by Q sorting the statements from agree (+3) to disagree (-3) in a forced, quasinormal distribution single-centered around each person's own mean of zero. For purposes of demonstration, however, actual participants were foregone and instead various viewpoints were simulated as Q sorts: A Q sort hypothetically representing Stephenson's position was created, for example, as well as one representing Burt's point of view, and also one representing the view of Kerlinger (1964), whose chapter-length treatment of Q was the first in a major social science text. Theoretical Q sorts were likewise constructed to represent the views of four main contributors (designated A, B, C, and D below) to the aforementioned electronic mail discussion about Q methodology from which the concourse was drawn. A Q sort was also theoretically composed to represent the point of view about Q and R typically found in statistics texts, and another to represent a "quantum" standpoint, about which more will be said. Finally, my own point of view (the only nontheoretical Q sort) was added so as to make explicit the location of the observer's perspective relative to all others once the data had been factor analyzed.

    The above nine Q sorts were intercorrelated, resulting in a 9×9 correlation matrix among persons that reveals the degree of association of each Q sort with all the others. It is worth emphasizing that the basic data matrix upon which the correlation matrix depends is wholly subjective in character, i.e., each Q sort is simply the perspective (simulated in this illustration) of the person whose Q sort it is, and this is different from a data matrix filled with objective characteristics such as the frequency with which an emotion occurs, or intelligence as measured by the WISC-R, or reading ability as determined by the number of words read per minute, and so forth. Unlike objective tests and traits, subjectivity is self-referential, i.e., it is "I" who believes that such-and-such is the case and who registers that belief by placing the statement toward the +3 pole of the Q-sort scoring continuum.

    Likewise, the factor analysis of a correlation matrix such as described above leads to factors of operant subjectivity (Stephenson, 1977), so named by virtue of the fact that their emergence is in no way dependent upon effects built into the measuring device. Stephenson, in part, borrowed the concept of operant from the American behaviorist B.F. Skinner (1938, p. 20), who used it to refer to behavior that is not under the immediate control of an eliciting stimulus. Stephenson's real inspiration, however, he traced to an earlier comment by Charles Spearman (1927): "... if only a test could be so fashioned as to eliminate all possible difference in the subjects' manner of procedure, then this single test might by itself conceivably afford a perfect measure of g" (p. 241). Spearman was looking for some way to measure that general intellective ability (g) present in all cognitive activities from tying shoelaces to writing philosophical treatises; the problem was that specific tests (e.g., of mathematical ability, reading ability, etc.) required specific abilities of their own, hence there was no way to measure g without involving the effects of specific tests used in measuring it. This remains the bane of conventional R factor analysis, and also of any by-person factor analysis based on the same data matrix, for these factor outcomes must invariably fall back on the tests included in the analysis in order to explain the factors. In this respect, R factor analysis skates dangerously near tautology.

    This is not the case in Q methodology, and the reason it is not inheres in the character of subjectivity. Whereas it is true that statements for a Q sample are typically selected in terms of a factorial or other representation of a theory, the supposed a priori meaning of the statements does not necessarily enter into the Q sorter's considerations when evaluating them: Participants inject statements with their own understandings. Objective measures (e.g., IQ tests) have right answers, but this is not the case within the realm of subjectivity, and it is for this reason that Stephenson always utilized factor analysis rather than variance analysis in analyzing data obtained from Q technique: Variance analysis relies on the a priori meanings built into a sample of Q statements, hence is tied to prestructured effects in precisely the same way as R methodology; Q factor analysis, on the other hand, more nearly retains the meanings which participants give to the statements by preserving the operations involved in ordering the statements, i.e., it is the participants' actual operations in Q sorting that give rise to the factors. The factors which emerge are therefore categories of operant subjectivity, and consequently have a pristine quality missing in the factors of R methodology.

    The factors which emanated from the study under consideration are shown in Table 1, which shows two factors, one of them bipolar. Q sorts A to D represent the views of four individuals who were actively involved in the electronic mail discussion in which the nature of Q methodology was being considered. Three of them (A, C, and D) join with Burt and with that version of Q most often encountered in textbook discussions of factor analysis: The positive pole of Factor I, then, is the conventional position which has become canonized in texts and fixed in the minds of the current generation of psychometricians, whereas the opposite end of the factor represents Stephenson's position. The leading ideas of these two vantagepoints can be seen in those statements which received the highest factor scores in the two poles:

Conventional view: (11) The idea is to come up with a set of traits that characterize individuals, then compare individuals for the distribution of these sets.... (13) It is intended to get at patterning within individuals (case-wise) rather than simply across individuals (factor-wise sorting).... (16) Factor scores can be tough to come by because the correlations are of reduced rank.... (19) The frequencies in the piles must be restricted to the frequencies that would be expected if you had a normal curve, with each pile corresponding to an area of a normal curve.

Stephenson's position: (5) Centroid factor analysis is recommended since its indeterminacy is compatible with quantum theory and, at the rotational stage, with interbehavioral principles.... (18) Q has never involved the correlation and factor analysis by rows of the same matrix of data that is analyzed by columns in R methodology.... (2) Q methodology is a set of procedures, theory, and philosophy supporting the study of the same kind of subjectivity that is the focal point of much qualitative research... (9) Cluster analysis is really something quite different and has no commitment to that subjectivity which is central to Q methodology.

Table 1
Perspectives on Q Methodology

Factors

Q Sorts	I	II

1 A	82	32
2 B	14	86
3 C	93	05
5 Stephenson	90	04
6 Burt	83	29
7 Kerlinger	31	89
8 textbook	84	30
9 quantum	- 83	-14
10 Brown	- 87	13

Significant loadings in boldface; others
insignificant. (Source: Brown, 1993)

Hence, the conventional end of factor I agrees strongly with statements 11 and 13 and disagrees with 5 and 18, whereas the reverse is the case at the negative end of the factor. At the Burt end of the factor we see the familiar emphasis on traits and on measurement based on individual differences (statement 11), as well as an assumption of a single data matrix underlying case-wise vs. trait-wise patterns (13, 16). At the Stephenson end of the factor, by way of contrast, there is a denial of reciprocity (18, 9), an embracing of subjectivity, and an expressed preference for centroid factor analysis due to its compatibility with quantum theory and interbehaviorism (5), the latter associated with behaviorist J.R. Kantor (1959).

Before moving on to a fuller consideration of Stephenson's less-understood position, it is worth bringing factor II into the picture. This factor, as shown in Table 1, is orthogonal to the Burt-Stephenson debate and is defined by Kerlinger and also by one of the participants in the electronic mail discussion who was familiar with Kerlinger's (1964) chapter, which stresses the importance of the variance design of Q samples and of establishing the unambiguous meaning of statements. The character of this position can be seen in those statements with high positive and negative factor scores:

Factor II most agree: (1) It permits the a priori structuring of hypotheses in the design of the Q set to be sorted.... (3) The method can be coupled with analysis of variance to test hypotheses.... (4) The interpretation of factors is more difficult if the Q sorts are internally inconsistent than when they are based on structured Q sets representing testable scientific hypotheses.... (20) It uses an ipsative technique of sorting a representative set of subjective statements drawn from a concourse of possible feelings or reactions about a subjective condition.

Factor II most disagree: (5) Centroid factor analysis is recommended since its indeterminacy is compatible with quantum theory and, at the rotational stage, with interbehavioral principles.... (10) Variance designs are only used to represent theory. Testing is in terms of dependency factor analysis.... (6) "Ipsative" generally applies to patterns of objective scores for persons, and has little to do with the subjectivity intrinsic to Q methodology.

Factor II therefore represents something of a hybrid, which is why it is orthogonal to factor I. There is a certain recognition that Q methodology is useful for the study of subjectivity (statement 20), a point which escaped the British factorists and their progeny, but at the same time this is mixed in with hypothesis-testing based on statement meaning, variance analysis, and adoption of the normative/ipsative nomenclature, all of which deviate from Q methodology as a science of subjectivity. This position, as noted, is associated in particular with the work of Kerlinger, with whom Stephenson was acquainted while the former was working on his doctorate at New York University and Stephenson was involved in advertising in the New York City area following his years at Chicago, and just prior to his becoming Distinguished Professor of Advertising Research in the University of Missouri School of Journalism. As is apparent, Kerlinger caught some of the spirit of Q methodology, but failed to free himself of many of the presuppositions of R methodology.

Main Principles of Q Methodology

There was little room for Stephenson's views when he originally enunciated them in the latter half of the 1930s. This was in part due to Cyril Burt's overshadowing influence in British psychology (which helps account for why Stephenson took his family to the United States following the war), and partly to the character and complexity of what Stephenson had to say. With a long list of eminent factor analysts arrayed against him--including the likes of Burt, Banks, Cattell, Eysenck, Thomson (to a lesser extent), McNemar, and others--it became an easy matter to dismiss him as unclear, careless, and lacking in other intellectual virtues. There is mounting evidence, however, that the tide is turning and that a new generation of British psychologists in particular, and also a growing number of health and social scientists, are beginning to take an interest in Stephenson's ideas and to glimpse the vision that escaped his contemporaries (see, for example, Curt, 1994; R. Stainton Rogers, 1995; W. Stainton Rogers, 1991; Stenner & Marshall, 1995).

Space precludes going into great detail concerning the principles of Q methodology, but it is useful to touch base with at least some of the more fundamental. As will be seen throughout, the concepts and principles which Stephenson advanced were influenced by his training in both physics and psychology, as Logan (1991) has made clear.

Subjective Communicability

First and foremost is the axiom of subjectivity and its centrality in human affairs. Subjectivity is everywhere, from the loftiest philosophizing and diplomatic negotiating to the street talk of the juvenile gang and the self-talk of the daydreamer, and it is the purpose of Q technique to enable the person to represent his or her vantagepoint for purposes of holding it constant for inspection and comparison. Communicability of this kind is typically shared, i.e., a matter of consciring (Stephenson, 1980), and is consequently about fairly ordinary things--about soccer, yesterday's debate in Parliament, the scandal surrounding President Clinton's fund-raising activities, the opening of a new play, and anything else under the sun. What is considered "ordinary" will, of course, depend on context, so that even the above study about Q methodology was about a fairly ordinary topic among those entering into that discussion: Each participant generally understood what the others were talking about.

R methodology, on the other hand, has almost wholly to do with assessments of one kind or another (of intelligence, mathematical ability, social anxiety, and such), and these assessments are typically of traits about which the participant is at best only dimly aware--i.e., the measurements are not about the kinds of ordinary things that enter into shared communicability; rather, they represent specialized knowledge and stand outside life as it is lived moment to moment. Two individuals may be assessed for their intelligence, for instance, yet be none the wiser about what the assessment was for; afterwards, however, they may enter into all manner of discussion about their experiences--e.g., "I didn't like the math part, but then I never was good at numbers," or "I kind of liked putting those blocks together," or "What did you think about those pictures? I know we were supposed to say something about them, but my mind just went blank," and on and on. Q models this communicative situation and provides measures for it, just as R provides measures for intelligence and other traits, and neither is convertible into the other through transposition or any other matrix manipulation.

Q technique can, of course, be used for assessment purposes, as indicated in Block's (1961) well-known monograph and as Stephenson (1954b) himself has shown, but this can be accomplished in at least two different ways. When Block's California Q-Set is employed, it is invariably the psychologist who uses it as a way to provide an overall portrayal of the person under scrutiny, as Miller, Prior, and Springer (1987), for example, have demonstrated: This provides an "external" perspective on the person, but says nothing directly about the structure of subjectivity (except, perhaps, that of the psychologist who performed the Q sort!). The alternative is to include the "internal" perspective by permitting the person to provide Q sorts that become part of the assessment, thereby including the subjective record alongside the person's other "vital signs" (Stephenson, 1985).

Quantum Theoretical Aspects

Related to subjective considerations was Stephenson's recognition, as well as Burt's (1938), that the mathematics of factor analysis and Heisenberg's matrix mechanics were virtually identical. Subjective communicability is indeterminant, as are the Q factors which document that subjectivity. During a typical discussion over dinner or a game of cards, for instance, it is impossible to predict who will speak next or what that person will say, and the same is true at the individual level, e.g., in day dreaming, in the free associations of the psychoanalytic hour, or in the soliloquy of a Molly Bloom in Joyce's Ulysses (Stephenson, 1991). For Burt, factor analysis and quantum theory were linked by analogy only, and this was due in large part to the phenomena being measured: R methodology measures objective variables, such as intelligence, whereas quantum mechanics measures not variables, but states (of energy). R methodology therefore remains necessarily within the classical paradigm of Newtonian mechanics and of determinant causation (hence of reductionism), and this remains true even when its basic data matrix is transposed and recorrelated by rows: As shown in Figure 1a (supra), Burt's version of correlating persons keeps the previously understood world intact.

In Q methodology, by way of contrast, the quantum character of subjectivity is not an analogy, but a reality, although its theory is rooted in Spearman's factor theory rather than in physics; moreover, it gains meaning in relation to actual measurements (Brown, 1992; Stephenson, 1988, p. 180). In classical mechanics, each effect is determined by a cause, and this remains intrinsic to the logic of dependent and independent variables; in quantum theory, however, there are no quantities that determine an individual subatomic collision, and similarly in Q methodology "there is no quantity hitherto put forward to explain a psychological event that determines operant factors" (Stephenson, 1995-1996, p. 3n). In short, Q technique does not measure variables as such, but states of mind; and when Q studies are made of single cases, several factors are typically shown to exist simultaneously in a state of complementarity (Stephenson, 1986), i.e., communicability exists in various states of probability. Moreover, the complementarity at issue in Q methodology, as in particle physics, is a function of measurement rather than a vague metaphor: It is the Q factors which are in a relationship of complementarity. Finally, measurement and meaning are as inextricably entwined in Q as in Heisenberg's uncertainty principle: The observer of the person's subjectivity is the person him- or herself (rather than the external scientist), and it is the person who also provides the Q-sort measurements.

Operantcy and Interbehaviorism

Operantcy, as noted previously, can be traced to Skinner, and before that to Spearman and Watson, who were on the trail of this idea even before it became a central principle in physics. Science deals with operations associated with confrontable events, and in Q methodology self and subjectivity are rendered operational through Q technique. In the process of Q sorting, the person operates with statements or other measurable stimuli by rank-ordering them under some experimental condition. The operation is subjective inasmuch as it is me rather than someone else who is providing a measure of my point of view, and the factors which emerge are therefore categories of operant subjectivity (Stephenson, 1977).

The factors are also naturalistic in the sense that they are naturally-occurring events (Brown, forthcoming): The statements, unlike scale items, are naturally rendered (usually in the course of interviews) and in no way implicate variables or pretend to be a test of anything, and the Q sorts are a function of the person's understanding; hence, the factors which emerge from this process must, of necessity, be a natural consequence of all that preceded, and relatively uncontaminated by the scientist's intrusions. It is this naturalism, in part, that drew Stephenson to the interbehavioral psychology of Kantor (1959) and to the confluence of Kantor, Spearman, and quantum theory (Stephenson, 1982, 1984). Kantor wished to avoid the nonsecular metaphysics of surface effects as indirect indicators of hidden causes and constructs, and to base human science on confrontable events and field conditions, and this comported well with the presuppositions of a science of subjectivity (Smith & Smith, 1996). Kantor's was a thorough-going behaviorism, yet not of the "decontextualizing" variety about which Stenner and Marshall (1995, p. 634) complain; rather, field and contextual principles were central to it.

Interbehaviorism also entered into Q methodology itself at the stage of factor rotation. Unlike any other major factor analyst, Stephenson retained the centroid method of factor analysis for essentially the same reason that others abandoned it--namely, because of its statistically imprecise nature. Given their imprecision, centroid factors were typically rotated to simple structure so as to render them more determinant, but Stephenson saw in their imprecision the opportunity to save this particular method of factor analysis from all of the determinant presumptions of the other methods and to use it as a way to probe subjective space. Just as a Q sorter interbehaves with statements, therefore, so does the factor analyst interbehave with factor space, so that theory and data can interact within their own field also. It was because of this that Thompson (1962) was able to see that Stephenson's view of factor analysis, unlike those of Burt, Cattell, Eysenck, and others, could never be routinized and reduced to generally applicable algorithms.

The Future of Q Methodology

It is perhaps too early for a definitive statement, but there are indications that adherents of Q methodology have reached a critical mass and that interest in Stephenson's innovation will continue to rise in at least the short run. This renewed interest in Q technique and its methodology has been fortified by a number of recent initiatives:

The beginning in 1977 of Operant Subjectivity, now in its nineteenth volume, a journal/newsletter devoted to pursuing the implications of Stephenson's ideas and to providing a link among users of Q methodology in diverse fields.

The establishment in 1985 of the William Stephenson Communication Research Center in the School of Journalism at the University of Missouri, USA.

The founding in 1989 of the International Society for the Scientific Study of Subjectivity for the expressed purpose of pursuing Stephenson's ideas. (This event represented the formalization of annual meetings that began in 1985; the fourteenth such meeting is scheduled for the autumn of 1998 in Seoul, Korea.) This event was preceded in the same year by a conference at the University of Reading--attended by Stephenson just two months prior to his death--in which studies utilizing Q in a British context were prominently displayed.

The initiation in 1991 of the Q-Method electronic mail discussion list, to which almost 300 persons worldwide now subscribe. This facility has been supplemented by the QArchive (for the archiving of electronic versions of texts, software programs, etc.) and by the Q-Method page on the World Wide Web.

The development in 1992 of the QMethod freeware package for mainframe computers, which now takes its place alongside the PCQ and QUANAL programs. QMethod has since been expanded to provide PC versions for both IBM and Macintosh platforms.

    In addition, Q methodology has begun to be featured at special panels and symposia in various fields as interest in it has accelerated, and clusters of self-designated Q methodologists are beginning to appear in various locales--e.g., in the psychology departments at the universities of Reading and East London (UK) and Central Oklahoma, Arizona, and Eastern Michigan (USA); in the political science departments at Kent State (USA) and Melbourne (Australia); in medical schools at Maryland and Illinois-Chicago; and in departments and schools of communication at Missouri, Nebraska, Syracuse, and Iowa (USA) and in Windsor (Canada).

    But most important, perhaps, there has been a noticeable acceleration in the use of Stephenson's methodological innovation in studies of increasing sophistication in an expanding number of intellectual fields--e.g., in marketing (Kleine, Kleine, & Allen, 1995), pharmacy (Mrtek, Tafesse, & Wigger, 1996), political science (Brown, 1980; Gillespie, 1993; Peritore, 1990), child psychology (Taylor, Delprato, & Knapp, 1994), nursing and medicine (Dennis & Goldberg, 1996), psychoanalysis (Edelson, 1989), public policy (Durning & Osuna, 1994), religion (Braswell, 1994; Parker, 1994-1995), public administration (Brown, Durning, & Selden, forthcoming; Sun, 1992), literary interpretation (Thomas & Baas, 1994) and communication, the latter having been stimulated primarily by Stephenson's (1967) main contribution to the field. In addition, Q has turned out to be a significant method of choice among those pursuing newer intellectual developments, such as postmodernism (Grosswiler, 1997), deconstruction (Rebekah Stainton Rogers, 1993; Thomas, McCoy, & McBride, 1993), social construction (Curt, 1994; W. Stainton Rogers, 1991), discourse analysis (Dryzek, 1994), identity theory (Davis, 1997), narrative analysis (Felkins & Goldman, 1993; Knight & Doan, 1994), feminism and women's issues (Febbraro, 1995; Gallivan, 1994; Senn, 1993), and qualitative methods (Brown, 1996). A Q bibliography that stood at approximately 600 entries three decades ago (Brown, 1968) has now blossomed to nearly 2500.

    William Stephenson was ahead of his time when he introduced his bright idea in 1935, and he remains in advance of much contemporary theorizing about the wellspring of human action and the methodological principles required for its study. The record is beginning to show that he was not the muddled thinker that his critics pictured him to be, rather that he was surprisingly consistent from the very start, although the focus on subjectivity became sharper as he became more familiar with his invention. It is astonishing in retrospect that Stephenson was able to persevere with his views given the strong opposition to them and the lack of understanding all around him, and that he was able to do so is testament to an unusually strong spirit and uncommon reserves. Now that the main intellectual combatants have passed on, the significance of his contribution is coming into sharper focus, and it is a pleasure for those who knew him to see him increasingly restored to his proper place in the history of the human sciences.

References

Ackoff, R.L. (1955). [Review of the book The study of behavior]. Philosophy of Science, 22, 67

Banks, C. (1954). [Review of the book The study of behavior]. British Journal of Statistical Psychology, 7, 62-63.

Block, J. (1961). The Q-sort method in personality assessment and psychiatric research. Springfield, IL: Thomas.

Braswell, R.D. (1994). Passing down Pentecost. Paraclete: A Journal of Pentecostal Studies, 28(3), 1-11.

Brown, S.R. (1968). Bibliography on Q technique and its methodology. Perceptual and Motor Skills, 26, 587-613.

Brown, S.R. (1972). A fundamental incommensurability between objectivity and subjectivity. In S.R. Brown & D.J. Brenner (Eds.), Science, psychology, and communication: Essays honoring William Stephenson (pp. 57-94). New York: Teachers College Press.

Brown, S.R. (1980). Political subjectivity: Applications of Q methodology in political science. New Haven, CT: Yale University Press.

Brown, S.R. (1992, October 23). Q methodology and quantum theory: Analogies and realities. Read at a meeting of the International Society for the Scientific Study of Subjectivity, Columbia, MO.

Brown, S.R. (1993). A primer on Q methodology. Operant Subjectivity, 16, 91-138.

Brown, S.R. (1994-1995). Q methodology as the foundation for a science of subjectivity. Operant Subjectivity, 18, 1-16.

Brown, S.R. (1996). Q methodology and qualitative research. Qualitative Health Research, 6, 561-567.

Brown, S.R. (forthcoming). Q methodology and naturalistic subjectivity. In B.D. Midgley & E.K. Morris (Eds.), Modern perspectives on J.R. Kantor and interbehaviorism. Westport, CT: Greenwood.

Brown, S.R., Durning, D., & Selden, S.C. (forthcoming). Q methodology. In G.R. Miller & M.L. Whicker (Eds.), Handbook of data analysis and quantitative methods in public administration. New York: Marcel Dekker, 1998.

Burt, C. (1937). Correlations between persons. British Journal of Psychology, 28, 59-96.

Burt, C. (1938). The unit hierarchy and its properties. Psychometrika, 3, 151-168.

Burt, C. (1940). The factors of the mind: An introduction to factor-analysis in psychology. London: University of London Press.

Burt, C. (1955). [Review of the book The study of behavior]. Occupational Psychology, 29, 58.

Burt, C. (1972). The reciprocity principle. In S.R. Brown & D.J. Brenner (Eds.), Science, psychology, and communication: Essays honoring William Stephenson (pp. 39-56). New York: Teachers College Press.

Burt, C. & Stephenson, W. (1939). Alternative views on correlations between persons. Psychometrika, 4, 269-281.

Cronbach, L.J. & Gleser, G.C. (1954). [Review of the book The study of behavior]. Psychometrika 19, 327-330.

Curt, B.C. (1994). Textuality and tectonics: Troubling social and psychological science. Buckingham: Open University Press.

Davis, T.C. (1997). Patterns of identity: Basques and the Basque nation. Nationalism & Ethnic Politics, 3, 61-88.

Dennis, K.E. & Goldberg, A.P. (1996). Weight control self-efficacy types and transitions affect weight-loss outcomes in obese women. Addictive Behaviors, 21, 103-116.

Dryzek, J.S. (1994). Australian discourses of democracy. Australian Journal of Political Science, 29, 221-239.

Durning, D. & Osuna, W. (1994). Policy analysts' roles and value orientations: An empirical investigation using Q methodology. Journal of Policy Analysis and Management, 13, 629-657.

Edelson, M. (1989). The nature of psychoanalytic theory. Psychoanalytic Inquiry, 9, 169-192.

Eysenck, H.J. (1954). [Review of the book The study of behavior]. Journal of Educational Psychology, 45, 374-376.

Febbraro, A.R. (1995). On the epistemology, metatheory, and ideology of Q methodology: A critical analysis. In I. Lubek, R. Van Hezewijk, G. Pheterson, & C.W. Tolman (Eds.), Trends and issues in theoretical psychology (pp. 144-150). New York: Springer.

Felkins, P.K. & Goldman, I. (1993). Political myth as subjective narrative: Some interpretations and understandings of John F. Kennedy. Political Psychology, 14, 447-467.

Fisher, R.A. (1935). The design of experiments. London: Oliver & Boyd.

Gallivan, J. (1994). Subjectivity and the psychology of gender: Q as a feminist methodology. In J. Gallivan, S.D. Crozier, & V.M. Lalande (Eds.), Women, girls, and achievement (pp. 29-36). Toronto: Captus University Publications.

Gershenson, C.P. (1955). [Review of the book The study of behavior]. Social Service Review, 29, 333.

Gillespie, J.D. (1993). Politics at the periphery: Third parties in two-party America. Columbia: University of South Carolina Press.

Glueck, B. (1954). [Review of the book The study of behavior]. American Journal of Psychiatry, 111, 479-480.

Grosswiler, P. (1997, October). Subjectivity or the death of the subject: Q theory vs. postmodern theory. Read at a meeting of the International Society for the Scientific Study of Subjectivity, Syracuse University.

Kantor, J.R. (1959). Interbehavioral psychology: A sample of scientific system construction (2nd ed.). Granville, OH: Principia Press.

Kerlinger, F. (1964). Foundations of behavioral research. New York: Holt, Rinehart and Winston.

Kleine, S.S., Kleine, R.E., III, & Allen, C.T. (1995). How is a possession "me" or "not me"? Characterizing types and an antecedent of material possession attachment. Journal of Consumer Research, 22, 327-343.

Knight, M. & Doan, R. (1994). The stories we tell ourselves: ISpi: A technique for narrative assessment. Orlando, FL: Harcourt Brace.

Logan, R.A. (1991). Complementarity, self and mass communication: The contributions of William Stephenson, 1902-1989. Mass Comm Review, 18, 27-39.

McNemar, Q. (1954). [Review of the book The study of behavior]. Psychological Bulletin, 51, 527-528.

Miller, M.J., Prior, D., & Springer, T. (1987). Q-sorting Gloria. Counselor Education and Supervision, 27, 61-68.

Mrtek, R.G., Tafesse, E., & Wigger, U. (1996). Q-methodology and subjective research. Journal of Social and Administrative Pharmacy, 13(2), 54-64.

Parker, K.R. (1994-1995). Storylistener subjectivity in response to Mark's Passion-Resurrection narrative. Operant Subjectivity, 18, 51-69.

Peritore, N.P. (1990). Socialism, communism, and liberation theology in Brazil: An opinion survey using Q-methodology. Athens: Ohio University Press.

Senn, C.Y. (1993). Women's multiple perspectives and experiences with pornography. Psychology of Women Quarterly, 17, 319-341.

Skinner, B.F. (1938). The behavior of organisms: An experimental analysis. New York: D. Appleton-Century.

Smith, N.W. & Smith, L.L. (1996). Field theory in science: Its role as a necessary and sufficient condition in psychology. Psychological Record, 46, 3-19.

Spearman, C. (1927). The abilities of man: Their nature and measurement. New York: Macmillan.

Stainton Rogers, Rebekah (1993). Deconstructing "the spectator": An examination of reading strategies in relation to Edward Scissorhands. Bachelor's dissertation (History of Art), University of Leeds, U.K.

Stainton Rogers, R. (1995). Q methodology. In J.A. Smith, R. Harré, & L. Van Langenhove (Eds.), Rethinking methods in psychology (pp. 178-192). Thousand Oaks, CA: Sage.

Stainton Rogers, W. (1991). Explaining health and illness: An exploration of diversity. London: Harvester Wheatsheaf.

Stenner, P. & Marshall, H. (1995). A Q methodological study of rebelliousness. European Journal of Social Psychology, 25, 621-636.

Stephenson, W. (1935a). Technique of factor analysis. Nature, 136, 297.

Stephenson, W. (1935b). Correlating persons instead of tests. Character and Personality, 4, 17-24.

Stephenson, W. (1936a). The foundations of psychometry: Four factor systems. Psychometrika, 1, 195-209.

Stephenson, W. (1936b). The inverted factor technique. British Journal of Psychology, 26, 344-361.

Stephenson, W. (1936c). A new application of correlation to averages. British Journal of Educational Psychology, 6, 43-57.

Stephenson, W. (1950). A statistical approach to typology: The study of trait-universes. Journal of Clinical Psychology, 6, 26-38.

Stephenson, W. (1953). The study of behavior: Q-technique and its methodology. Chicago: University of Chicago Press.

Stephenson, W. (1954a). Comments on Cronbach and Gleser's review of: The study of behavior: Q-technique and its methodology. Psychometrika, 19, 331-333.

Stephenson, W. (1954b). Q-technique and the Rorschach test. In S.J. Beck, The six schizophrenias (Research Monograph No. 6). Chicago: American Orthopsychiatric Association.

Stephenson, W. (1967). The play theory of mass communication. Chicago: University of Chicago Press.

Stephenson, W. (1977). Factors as operant subjectivity. Operant Subjectivity, 1, 3-16.

Stephenson, W. (1978). Concourse theory of communication. Communication, 3, 21-40.

Stephenson, W. (1980). Consciring: A general theory for subjective communicability. In D. Nimmo (Ed.), Communication yearbook 4 (pp. 7-36). New Brunswick, NJ: Transaction.

Stephenson, W. (1982). Q-methodology, interbehavioral psychology, and quantum theory. Psychological Record, 32, 235-248.

Stephenson, W. (1984). Methodology for statements of problems: Kantor and Spearman conjoined. Psychological Record, 34, 575-588.

Stephenson, W. (1985). Perspectives in psychology: Integration in clinical psychology. Psychological Record, 35, 41-48.

Stephenson, W. (1986). William James, Niels Bohr, and complementarity: II--Pragmatics of a thought. Psychological Record, 36, 529-543.

Stephenson, W. (1988). Quantum theory of subjectivity. Integrative Psychiatry, 6, 180-187.

Stephenson, W. (1991). Ulysses and Finnegans Wake: A Q-methodological look at profundity (Part I: Ulysses). Operant Subjectivity, 14, 89-105.

Stephenson, W. (1995-1996). Quantum theory media research: I. Reception analysis. Operant Subjectivity, 19, 1-11.

Sun, T-W. (1992). Indigenization of public administration knowledge in Taiwan. Asian Thought and Society, 17, 97-112.

Taylor, P., Delprato, D.J., & Knapp, J.R. (1994). Q-methodology in the study of child phenomenology. Psychological Record, 44, 171-183.

Thomas, D.B. & Baas, L. (1994). Reading the romance, building the bestseller: A Q-technique study of reader response to The Bridges of Madison County. Operant Subjectivity, 17, 17-39.

Thomas, D.B., McCoy, C., & McBride, A. (1993). Deconstructing the political spectacle: Sex, race, and subjectivity in public response to the Clarence Thomas/Anita Hill "sexual harassment" hearings. American Journal of Political Science, 37, 699-720.

Thompson, J.W. (1962). Meaningful and unmeaningful rotation of factors. Psychological Bulletin, 59, 211-223.

Turner, R.H. (1955). [Review of the book The study of behavior]. American Journal of Sociology, 61, 167-169.