Q Methodology

Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 721 28.8.2007 8:51pm Compositor Name: deshanthi 37 Q Methodology Steven R. Brown, Dan W. Durning, and Sally C. Selden CONTENTS 37.1 Introduction ........................................................................................................................ 722 37.1.1 The Essential Elements of Q Method.................................................................. 722 37.1.2 Research with Q and R Methods: A Practical Example ..................................... 724 37.1.2.1 The R Method Approach.................................................................... 724 37.1.2.2 Using Q Methodology ........................................................................ 725 37.1.3 Summary: Key Differences in Q and R Methodologies ..................................... 726 37.2 Understanding and Using Q Methodology: A Detailed Explanation................................ 726 37.2.1 History and Intellectual Foundations of Q Methodology.................................... 726 37.2.1.1 Brief History ....................................................................................... 726 37.2.1.2 The Quantum Connection .................................................................. 727 37.2.1.3 Concourse Theory .............................................................................. 728 37.2.2 Problem Clarification with Q Methodology: A Case Study ............................... 730 37.3 Applications of Q Methodology in Public Administration: A Bibliographical Guide and Research Agenda.............................................................................................. 745 37.3.1 General Public Administration and Public Management .................................... 745 37.3.1.1 How Bureaucrats View Issues Affecting Their Jobs and Management Decisions................................................................ 745 37.3.1.2 How Public Sector Professionals Understand Their Jobs and Carry Out Their Work? ............................................................... 746 37.3.1.3 Public Management Topics Such as Organizational Culture and Leadership.................................................................................... 746 37.3.1.4 Potential Topics for Further Research................................................ 747 37.3.2 Public Administration Fields, Including Personnel Administration.................... 748 37.3.2.1 Researching Public Personnel Administration ................................... 748 37.3.2.2 Q Research in Other Fields of Public Administration ....................... 749 37.3.3 Researching Decision Making and Public Policy ............................................... 750 37.3.3.1 Understanding Influences on Decision Makers.................................. 750 37.3.3.2 Identifying the Perspectives (Decision Structures) of Stakeholders ................................................................................... 751 37.3.3.3 Studying the Marginalized and Less Powerful, and Giving them Voice.......................................................................................... 752 37.3.3.4 Facilitating the Search for Compromise Solutions to Difficult Problems .......................................................................... 753 37.4 Conclusion ......................................................................................................................... 754 Endnotes........................................................................................................................................ 755 References ..................................................................................................................................... 755 721 Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 722 28.8.2007 8:51pm Compositor Name: deshanthi 722 Handbook of Research Methods in Public Administration 37.1 INTRODUCTION Q methodology resembles in many ways the other public administration research methods described in this book. It requires the collection and manipulation of data, followed by the analysis of this data using sophisticated statistical techniques. And, as with other methods, Q methodology can be used to explore a phenomenon of interest to gain insight into it and to generate and test hypotheses. Despite these common traits, Q methodology differs from the usual body of statistical techniques employed in public administration research in ways that have profound implications for its use. In fact, the designation of this method as ‘‘Q’’ is intended to differentiate it from ‘‘R’’ methodology, the statistical methods used for ‘‘objective’’ or ‘‘scientific’’ research in the social sciences. The differences between Q and R methods are not simply a matter of technique; they reflect different philosophies of inquiry that encompass competing epistemologies and different understandings of what constitutes sound scientific practice. Although some researchers are attracted to Q methodology by its philosophical underpinnings, others value it for the insights it provides. The first set of researchers view Q methodology as an alternative to R methods, which they consider to be inadequate tools of a discredited positivism, whereas the second group is drawn for practical reasons: it yields information that often differs from that obtainable through R methods. For these researchers or public administration professionals, Q methodology is a new research or analytic tool to add to their repertoire. It provides them with another lens to investigate an issue or research topic. Whatever the motivation for its use, Q methodology is both accessible to novice researchers and a challenge to the most experienced. It is an intensive methodology that maps how individuals think about an event, issue, or topic of research interest. Depending on the focus of the study, Q can provide deeper understanding of the opinions, beliefs, perspectives, decision structures, frames, or narratives of individuals on any topic that has a subjective component. In fact, it has been described as ‘‘the best-developed paradigm for the investigation of human subjectivity’’ (Dryzek and Holmes, 2002, p. 20). This chapter introduces Q methodology to researchers who know little about it but might like to employ it in their work. First we summarize the essentials of conducting Q-methodology research and compare research using Q and R methodologies. The second section provides a short history of Q methodology and its foundations, then illustrates in detail how a Q method study is carried out. In the third section, we review public administration-related studies that have been carried out using Q methodology and suggest future projects. A number of frequently asked questions (FAQs) concerning Q methodology are included in the Teachers Guide accompanying this text, along with the data file that provides the basis for the demonstration presented in Section 37.2. 37.1.1 THE ESSENTIAL ELEMENTS OF Q METHOD Q methodology is best understood as a type of research that identifies the operant subjectivity of individuals in regard to a particular subject. The methodology encompasses a broader philosophy of how subjectivity can best be studied, an inherent epistemology, and a method that includes a series of well-defined steps or phases. In this section, we introduce these steps or phases, and then show how they are applied. The major steps or phases of Q method research include identifying the concourse (what is the flow of communication—that is, what is being said or written—about the topic of interest), creating a sample of the concourse that captures its diversity (the Q sample), selecting the people of interest to carry out the sort (the P sample), administering the sort, conducting a statistical analysis of the completed sorts, and interpreting the Q factors that emerge from the analysis. In carrying out these steps, the researcher’s role is manifest and transparent, beginning with the identification of the concourse to the interpretation of the factors. Unlike R methods, which obscure or hide the importance of researcher judgment and a priori choices in the technicalities of technique, Q methodology clearly Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 723 28.8.2007 8:51pm Compositor Name: deshanthi Q Methodology 723 engages researchers in each step of the research process. These phases are summarized as follows and are elaborated more fully in Section 37.2: (1) Identifying the concourse: Any topic of interest to people in general or to individuals in specific roles, such as policy maker or bureaucrat, generates conversation about it. This conversation occurs in natural language, and it may appear in discussions or arguments, in e-mails and blogs, in newspapers and magazines, in books, and in other forms of communication. This communication of facts, information, beliefs, opinions, and feelings about a topic comprises a concourse. In the first Q method phase, the researcher identifies—to the extent possible—the communication on the topic of interest. To do so, the researcher may interview people likely to be engaged in communicating about the topic; alternatively, the researcher may collect statements from written sources. Depending on the topic, the researcher will collect dozens to hundreds to thousands of expressions of opinions, assertions, and arguments related to the topic. These expressions—usually in the form of statements—comprise the concourse. (2) Sampling the concourse: When the concourse has been thoroughly documented, the researcher extracts a representative sample from it. This sample is not randomly drawn from the concourse, but is selected carefully by the researcher with the goal of capturing the diversity and complexity of the different views contained within the concourse. Usually, the researcher is guided in the selection of a sample by a framework that has been formulated to model the important elements of the topic. This framework is practical in that it is designed to insure that the main viewpoints are fully represented in the sample, and it is theoretical in that, instead of being ad hoc, it draws on models explicated in previous research. The sample must include enough statements to fully represent the diversity of the concourse, but must not have so many statements that it cannot be used effectively in the sorts to be administered. (3) Q sorting: A Q sort results when the researcher asks a selected person to place the statements comprising the Q sample in rank order. The researcher provides (a) a Q sort deck, which consists of all Q-sample statements written on separate cards that have been randomly numbered, (b) instructions on how the cards should be ranked; for example, the sorter may be asked to place the cards along a nine-point continuum (beginning with 4 and ending with þ4 with 0 as midpoint) following a quasi-normal distribution, and (c) instructions on the conditions governing the sort; for example the sorter may be asked to rank the statements from agree or disagree, placing two statements with which the sorter disagrees the most in the 4 category and the two statements with which the sorter agrees the most in þ4 category, then to put three of the remaining statements with which the sorter most disagrees in the 3 category and the three that are agreed with the most in the þ3 category, and so on. If the sort is administered in person, the researcher can observe the sorting process, record comments, and ask questions about the decisions involved in placing certain statements in the extreme categories. (4) Selecting the sorters (the P sample): Because Q methodology is an intensive methodology, the selection of the people to complete a sort (the P sample) is an important element of the method. The selection depends highly on the topic that is being investigated. If the study focuses on a topic of concern largely to a specific organization, every person of interest can be included. If the study addresses a broader topic affecting a larger group of people and interests, the selection of participants should be designed to make sure that the full range of opinions and positions are represented in the P sample. (5) Analyzing the Q sorts: After the sorting is complete, the researcher analyzes the completed Q sorts. The statistical analysis begins by correlating the Q sorts, followed by factor analysis of the correlation matrix and factor rotation. This statistical analysis takes into account a distinct feature of the Q sort data: the statements comprising the Q sample are the Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 724 28.8.2007 8:51pm Compositor Name: deshanthi 724 Handbook of Research Methods in Public Administration observations of the study and the individuals completing the Q sorts are the variables. In other words, in a statistical analysis, the statements are the dependent variables and the sorters are the independent variables. Thus, when factor analysis and factor rotation are completed, the Q factors are made up of groups of sorters who have similar views on the topic of interest. (6) Interpreting the factors: The final research step is to interpret the factors that have been identified, including how they differ and how they are similar. To carry out this step, the researcher examines the weighted average sort of each factor and compares that sort to the weighted average sorts of the other factors. This type of analysis uses abductive logic, which involves reasoning from observed effects to plausible causes (Brown and Robyn, 2004; Wolf, 2004). From this comparison, the researcher can describe the structure of thought that exists for each factor, and can identify how the factors resemble each other and how they differ. 37.1.2 RESEARCH WITH Q AND R METHODS: A PRACTICAL EXAMPLE How does Q method work in practice? Suppose a researcher is interested in how managers of public agencies in another country, say Ukraine, view their jobs, specifically the organization of their work; the methods of control and discipline they use; their attitudes toward change; and their relationships with subordinates, superiors, and the public. As part of this study, the researcher would like to investigate whether younger managers have perspectives different from those of older managers. Also, the investigator might wish to compare the attitudes of Ukrainian public managers with those of public managers in the United States. 37.1.2.1 The R Method Approach The common scientific approach to this type of research would be to formulate hypotheses about the different types of managers, and how the types of managers would vary by location and age. Then, the hypotheses would be tested with data collected from application of a survey instrument that would contain questions or statements with scales allowing the respondents to indicate their degree of agreement or disagreement. The survey instrument would seek to measure the dimensions of the public manager’s job addressed in the hypotheses. For example, to test the hypothesis that older managers in Ukraine are more likely than younger managers to have an autocratic managerial style, the researcher would create a set of questions and statements designed to measure how the manager relates to subordinates (e.g., whether the manager gives orders or invites participation in decisions). This survey might be sent to a random sample of public managers in Ukraine, or perhaps a stratified random sample to insure that both younger and older managers are fully represented. Most likely, however, because of the difficulties in identifying the whole population of public administrators and extracting a random sample, the target population would be the public employees in a particular city or region, and the sample would be drawn from that population. To render it a cross-national comparison, a random sample of public managers in the United States, or more likely, managers in a comparable government, would be selected and sent to the survey. After receiving the completed surveys, the researcher would enter the data into a spreadsheet or statistical program, creating a huge matrix of variable responses by observation. Then, different statistical analyses would be performed. Likely, the responses to different questions or statements would be used to construct dependent variables, which would represent different views of the elements of public management under study (for example, more or less autocratic managerial style). Then, these dependent variables could be included in regression analyses to test hypotheses (for example, with managerial style as the dependent variable, the independent variables might be the age of the respondent, years of experience, educational background, nationality).1 Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 725 28.8.2007 8:51pm Compositor Name: deshanthi Q Methodology 37.1.2.2 725 Using Q Methodology The researcher could also use Q methodology to investigate this research question and to explore many of the same hypotheses. However, the process of research and the results would look much different and, we would argue, they would more accurately reflect the richness and complexity of the views of the different managers. The researcher would follow the steps or phases described above, beginning by identifying the communication (what is being said) about the topic of interest, the jobs of public managers. This communication could be identified by interviewing several public managers about their jobs, focusing particularly on the job dimensions of research interest, or it might be done by reading accounts of managers in the two countries about their jobs. From this concourse about the jobs of public managers, a sample would be selected that represented, as far as possible, the diversity of communication in all of its important dimensions. This sample, in the form of 30 to 60 statements, would comprise the Q sort to be administered to managers. The researcher would then ask selected managers to complete the Q sorts. The selection of managers would be intended to insure that those most likely to have different views would be included. A key goal in selecting managers to complete the Q sorts would be to obtain the largest possible diversity of views. Also, the researcher would purposely select a group of younger and older managers to insure that information would be obtained that would help explore whether the two groups have different views toward their jobs. For a comparative perspective, the sorts would of course be administered to groups of Ukrainian and American public administrators. The managers would complete the Q sort by placing the statements in a quasi-normal distribution from ‘‘most agree’’ to ‘‘most disagree.’’ The sort would be forced, in that the number of cards to be placed in each category—say, from þ4 (most agree) to 4 (most disagree)—would be specified. In this case, there would be nine categories, with the largest number of cards to be placed in the 0 category and the fewest to be placed in the þ4 and 4 categories. As the sort was being completed, the researcher could engage in a dialogue with the sorter, noting questions that were raised, comments that accompanied the placement of statements, and reactions to certain statements. Then, the sort could be followed up by inquiring into the manager’s reasons for placing statements in the most extreme positions. When all of the Q sorts were completed, they would be analyzed by first correlating them, then using the correlation matrix for factor analysis. However, because the factor analysis would treat the sorters as variables and the statements as observations, the resulting factors would represent the cluster of managers whose views of public management are quite similar. Using information about how the different clusters of managers completed their sorts, the researcher would then identify and discuss the different views about public management among the managers who completed the sorts. The analysis of the Q sorts would provide insight into how public managers understand their job. These views would not necessarily conform to any models that were specified a priori, nor would they be forced into categories based on responses to any particular statements whose meaning was specified in advance by the researcher. In fact, before carrying out the analysis of the sorts, the number of different perspectives and their nature would be unknown. Thus, the managers participating in the Q study would determine the number and nature of the perspectives of public management by revealing through their Q sorts their operant subjectivities. In some cases, a researcher might want to use the Q sort results to explore the extent to which sorters with specific characteristics are clustered in the same factors or are spread out among different factors. For example, with this study of Ukrainian and American managers it is possible to examine the characteristics (e.g., age, experience, nationality) of the managers in different factors to see if managers with specific characteristics have similar or different understandings of the job of public manager. In this way, it is possible to explore hypotheses, e.g., that older managers will display more autocratic management styles than will younger managers. To the extent that older managers grouped together in a factor that could be characterized as more autocratic, and younger Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 726 28.8.2007 8:51pm Compositor Name: deshanthi 726 Handbook of Research Methods in Public Administration managers clustered together in another factor with a less autocratic orientation, the Q study would provide support for acceptance of the hypothesis. (Of course it is possible that a factor is not systematically associated with any demographic variables; such a factor could therefore never be predicted in advance, yet through Q methodology it could be demonstrated to exist.) It should be emphasized that researchers must be cautious when using Q sort results to investigate the distribution of (R-type) populations among different factors. This type of investigation should be acknowledged as only suggesting a pattern of common or different viewpoints related to certain demographic characteristics because Q methodology is intended to identify subjectivities that exist, not to determine how those subjectivities are distributed across a population. 37.1.3 SUMMARY: KEY DIFFERENCES IN Q AND R METHODOLOGIES This comparison of how Q and R methodological studies would be carried out on the same topic provides some insight into the key differences in the methodologies.2 These include: . . . Q methodology seeks to understand how individuals think (i.e., the structure of their thoughts) about the research topic of interest. R methodology identifies the structure of opinion or attitudes in a population. Thus, the results of Q method will identify how an individual, or individuals with common views, understand an issue; the results of R methods describe the characteristics of a population that are associated statistically with opinions, attitudes, or behavior (e.g., voting) being investigated. Although R methods are intended for the ‘‘objective’’ analysis of research issues, Q methodology is designed to study subjectivity. R methodology is found on logical positivism in which the researcher is an outside objective observer. In contrast, Q methodology is more closely related to postpositivist ideas (Durning, 1999) that reject the possibility of observer objectivity and question the assumption that the observer’s vantage point, if not objective, is in some sense superior to that of any other observer, including the person being observed. Thus, Q methodology is in tune with phenomenological (Taylor et al., 1994), hermeneutic (McKeown, 1990, 1998), and quantum theories (Stephenson, 1983).3 Q methodology is an intensive method that seeks in-depth understanding of how at least one person thinks about the topic of investigation. As an intensive method, Q methodology requires a small number of well-selected subjects to complete the Q sort. R methods are extensive methods designed to extract an understanding of populations through representative samples of them; thus, they require—depending on the population size and sampling techniques—data from a certain percentage of the population of interest. 37.2 UNDERSTANDING AND USING Q METHODOLOGY: A DETAILED EXPLANATION In this section of the chapter, we discuss in more detail the history and some of the intellectual foundations of Q methodology; then we present a detailed case study of the use of Q methodology. This case study is intended to be a methodological guide for researchers and practitioners who would like to conduct their own Q method research. Included in the accompanying Teachers Guide is a collection of frequently asked questions (FAQs) about Q methodology, which we hope will address the issues and concerns of potential users of Q methodology. 37.2.1 HISTORY 37.2.1.1 AND INTELLECTUAL FOUNDATIONS OF Q METHODOLOGY Brief History William Stephenson, the inventor of Q methodology, was the last graduate assistant to Charles Spearman, who in turn is best remembered as the inventor of factor analysis. Spearman’s main Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 727 28.8.2007 8:51pm Compositor Name: deshanthi Q Methodology 727 interest, however, was in unlocking the creative potential of the mind, and factor analysis was merely his way of mathematically modeling the processes of thinking in which he had interest. Spearman once referred to Stephenson as the most creative statistician in psychology, but like his mentor, Stephenson was likewise interested in the mind’s potential, and the mathematics of his Q methodology are to a large extent secondary to that interest. Stephenson’s innovation can be traced to an August 24, 1935 letter to the editor of the British science journal Nature (Brown, 1980, pp. 9–10; Good, 1998, 2005a,b) in which he drew attention to the possibility of ‘‘inverting’’ conventional factor analysis. In R factor analysis, as the conventional approach is often called, traits are correlated across a sample of persons, where ‘‘trait’’ is taken to mean any quantitatively measurable characteristic: the factor analysis of the trait correlations points to families of similarly covarying traits. In Q factor analysis, by way of contrast, persons are correlated across a sample of statements which they have rank-ordered, the ranking being called a Q sort: the correlations reflect the degree of similarity in the way the statements have been sorted, and factor analysis of the person correlations points to families of like-minded individuals. A detailed illustration of what is technically involved is presented later in this part of the chapter. Stephenson’s innovation was misunderstood practically from the start and by such eminent University of London colleagues as Sir Cyril Burt, R.B. Cattell, and Hans Eysenck, and at the University of Chicago by L.L. Thurstone, so that even today his name is often associated with a statistical development which was not only not his, but also was one which he strongly opposed. Virtually without exception, texts which address Q and R factor analysis regard the two as simply the transposed equivalents of one another—that R consists of correlating and factoring the columns of a data matrix, and that Q consists of correlating and factoring the rows of the same matrix. In fact, Stephenson’s most sustained articulation of Q methodology, his book The Study of Behavior (Stephenson, 1953, p. 15), is typically cited in support of this position despite his clear assertion, repeated often, that ‘‘there never was a single matrix of scores to which both R and Q apply.’’ In this connection, Miller and Friesen (1984, pp. 47–48) are far from alone when, in their factor-analytic studies of organizations, they confidently assert that ‘‘Q-technique is merely R-technique using a transposed raw-data matrix. It treats similarities between companies, rather than between variables. . . . Discussion of Q-technique in particular can be found in Stephenson. . . . ’’ And more recently, Waller and Meehl (1998), with equal confidence, state that ‘‘mathematically, there is nothing remarkable about how Q correlations work. They can be computed . . . by simply transposing a persons 3 variables matrix into a variables 3 persons matrix prior to calculating the correlations’’ (p. 80). And Stephenson, they go on to say, is ‘‘one of the developers and most vociferous advocates of this technique’’ (p. 81), when in fact he was a life-long critic of it. 37.2.1.2 The Quantum Connection Miller and Friesen’s (1984) book is entitled Organizations: A Quantum View, which is fortuitous inasmuch as factor analysis and quantum mechanics (as elaborated by Werner Heisenberg and Max Born in particular) are virtually identical mathematically, both relying on matrix algebra and sharing much of the same nomenclature. Originally trained as a physicist, Stephenson was aware of this parallel in the 1930s, as was Burt (1940), who wrote extensively about it in his The Factors of the Mind, which is as fundamental for R methodology as Stephenson’s book is for Q. But Burt and Stephenson (1939) parted company over what it was that was to be measured. Burt was locked into the study of variables, such as intelligence, assertiveness, temperament, and the thousands of others with which social and psychological science is familiar. Variables have known properties, but they are largely categorical and owe much to logic and to ‘‘operational definitions,’’ hence their quantum character is nothing more than an analogy: Quantum theory is not about variables as such, but about states (of energy). Although they may enter into dynamic relations with other variables, the variables of R methodology are not in themselves dynamic and are typically expressed in a single number, usually an average score across a number of responses. Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 728 28.8.2007 8:51pm Compositor Name: deshanthi 728 Handbook of Research Methods in Public Administration The situation is quite different in Q methodology. Suppose that a person ranks a set of statements (say, from agree to disagree) to represent the person’s own point of view about the organization. The statements do not measure anything a priori, i.e., their meaning is indeterminate; they are simply assertions that have been made about the organization (e.g., that ‘‘it is a pleasant place to work’’). Meaning and significance are imposed on the statements by the person in the course of Q sorting them, hence the inseparability of measurement and meaning. The process reflects a dynamic state (of mind) in a relationship of complementarity to other states (i.e., to Q sorts by others in the organization). The final product (the completed Q sort) is not an average, nor is it subject to any external norms; rather, it is a dynamic pattern of interrelationships. The selfreferentiality of the Q sorter is obviously central to the situation, and the way in which the statements will be understood and scored is solely in the hands of the Q sorter; it can never be known in advance, therefore, how many factors will emerge, nor what their form and content will be. Everything is indeterminate, and the parallel with quantum theory is made the more remarkable by virtue of the fact that it is a function not of analogy, but of the ‘‘sovereignty of measurement’’ (Stephenson, 1989). In sum, there is considerably more to the difference between R method and Q method than a simple decision whether to analyze the relationships between the columns of a data matrix or the rows of the same data matrix. A much more fundamental difference is between the study of the variables of R methodology, conceived as objective and subject to classical conceptions of cause and effect; and the study of a data matrix of a wholly different kind, one thoroughly saturated with self-referentiality and probabilism. 37.2.1.3 Concourse Theory As was noted previously, Stephenson, like his mentor Charles Spearman, was interested in the creative potential of the mind. What is the source of creativity and how do we liberate it for the social good—more specifically, for administrative advantage? In an instructive book on word origins, C.S. Lewis (1960) devotes a chapter to ‘‘conscience and conscious,’’ both of which derive from a common Latin antecedent in which the prefix con means with—hence conscio means sharing what one knows with someone, which includes sharing with oneself, as in musings, daydreams, and mulling things over. But there was also a weaker sense in which conscientia connoted simply awareness, as in being conscious of something, epitomized in Descartes’s awareness of his own thinking (cogito ergo sum) and in introspectionism. Needless to say, this weaker sense of conscientia, upon which modern cognitive psychology is based, has virtually replaced the stronger sense of knowledge as shared, thereby elevating the individual thinker while removing almost all traces of the social context within which thinking takes place. Yet most of ordinary life is based on shared knowledge, and it was on this account that Lewis introduced the term consciring (Stephenson, 1980). Administrators and policy makers spend much of their time exchanging views in committee meetings or around the drinking fountain, and in reading and responding to one another’s memos and reports. Under more managed conditions, ideas are shared via Delphi or nominal group techniques in which each person’s idea-sharing may release new thoughts in others, the total being greater than what could be produced by the same participants working in isolation. The opposite is equally true that a single individual can often produce ideas that elude groups due to the isolate’s relative freedom from conformity and the sometimes suffocating effects of collegiality. Administration in general has been characterized as ‘‘the activities of groups cooperating to accomplish common goals’’ (Simon et al., 1950, p. 3), and as a group assembles and begins to cooperate, a vast array of possibilities begins to appear, typically in linguistic form. We may assume that the impetus for an assemblage is the existence of a task or problem—e.g., how to pare the budget, what to do about a councilman’s complaints concerning the streets in his ward, when to mount a campaign for the school levy, determining who would make the best executive director, Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 729 28.8.2007 8:51pm Compositor Name: deshanthi Q Methodology 729 etc.—and that the fruits of the group’s cooperative efforts take the form of proposed solutions. Hence, using the budget as illustrative, we might hear proposals such as ‘‘The easiest thing to do is to slash 3.2 percent across the board,’’ or ‘‘We can’t really cut back further on social services,’’ or ‘‘It’s been a mild winter so there should be some fat in the street maintenance line,’’ which may prompt the service director’s warning that ‘‘We haven’t even completed all the repairs from last winter’s freeze,’’ and so forth. All such communicability is inherently contestable, infinite in principle, ubiquitous in character, and inescapably subjective. In quantum theoretical terms, it is also unpredictable, paradoxical, and erratic. No one knows in advance what someone else is going to say or suggest, or how what one person says is going to impact on what others say or think. In Q methodology, such communicability is referred to as a concourse (Stephenson, 1978, 1980), a concept traceable to Cicero, but which takes its most modern form in Peirce’s (1955) ‘‘Law of Mind’’—that ideas spread and affect other ideas and eventually combine into a system, or schema. Concourse is therefore at the foundation of a society and provides lubrication for all its parts, and it constitutes the very stuff of which decisions are made and problems solved. And it is concourse that supplies the elements of Q methodology. Concourse is present in the loftiest of philosophical discourse to the simplest coos and gurgles of the nursery, as almost any random examples will easily show. In a recent posting on an Internet list devoted to quantum theory and consciousness, for instance, the following assertions were made: The universe is simple at fundamental levels. . . . A unified approach requires complementary modes of description. . . . Non-locality is firmly established. . . . Self-organization and complexity are prevalent at all scales in the universe. . . . All human experience is connected to the universe. Of such things volumes have been written, many by Nobel laureates and others of repute, each idea midwifing yet other ideas in endless profusion. And although supported in part by facts, communicability of this kind is thoroughly subjective and typically goes beyond known facts, with the facts as such colloidally suspended in the subjectivity. Such is the character of concourse. Or consider Sasson’s (1995) study of the way in which citizens construct crime stories, based on the shared communicability found on op-ed pages of newspapers; or Finkel’s (1995) study of jurors’ common sense understandings of justice; or Roe’s (1994) volume, which is brimming with narratives on such diverse social problems as budgets, global warming, animal experimentation, and so forth. With the contemporary move away from government and in the direction of governance, more emphasis is being given to deliberation (Hajer and Wagenaar, 2003) and to actionable political talk (Irwin, 2006), but apart from their narrative richness, the common limitation of efforts such as these is methodological in character—i.e., once Sasson and the others have gathered their mountains of discourse material, they typically end up sorting it into various more-or-less logical categories. What began on a sound footing of naturalism, therefore, ends up being sectioned according to the categorical proclivities of the analyst. Perhaps of more direct pertinence to decision-making settings of the kind more familiar to administrators is the concourse that emerged from a meeting of an international commission established to assist in encouraging Central American development (Brown, 2006a). The organizing question concerned the commission’s role, and the following observations were among those which were added to the concourse as each commissioner stepped to the microphone: The Commission must attend not only to economic growth, but to social and cultural growth. . . . We must begin with those elements which bring us together, and only later address those issues which divide us. . . . It is incumbent upon the Commission to recognize the central role of agriculture in the development of the region. . . . It must be adopted as an operating principle that the problems to be considered are autonomous to the region, and not reflections of East-West security issues. . . . The process which the Commission works out must, through its structure, recommend that aid be conditioned on measurable and swift progress toward genuine democracy. Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 730 28.8.2007 8:51pm Compositor Name: deshanthi 730 Handbook of Research Methods in Public Administration The commission spent the better part of a day and a half in this vein, which continued during meals, breaks, over drinks, and in walks on the hotel grounds—perhaps even into some members’ dreams. Such is the nature of ideational spreading and affectability characteristic of Peirce’s Law of Mind. Concourse proliferates not only around problems, but in terms of the interests of problem solvers: Ask a manager and a worker how best to improve the organization and different suggestions will be offered. This can be seen concretely in the comments obtained from a group of sixth grade students when asked what might be done to improve their school: Have more plays or assemblies. . . . Show more movies. . . . Make the halls more colorful and interesting by decorating them with students’ work. . . . Have the PTA hold more activities like the carnival. . . . Plant more flowers, bushes, and trees around the building. Aesthetics, desire for pleasurable activities, and something of a communal spirit predominate, as does the kind of dependency upon adults still characteristic of students at this age. When the same question is asked of a group of young eleventh grade policy makers, however, some of the same issues remain at the surface, but private interest and a desire for autonomy are now more prominent: Increase the variety and amount of food for lunch. . . . Do away with assigned seats in classes. . . . Don’t assign homework on Friday or the day before a vacation. . . . Add vending machines to the cafeteria (such as candy and pop machines), and put a juke box in the cafeteria. . . . Add a course for fourth year French and Spanish. (Ad infinitum) And in equal volume the solutions were proffered by a group with different interests—graduate students and faculty who were queried as to how best to improve their graduate program: Establish a rule or procedure whereby faculty are required to specify clearly and precisely the criteria for grading. . . . Structure course offerings so that students can choose between quantitative and nonquantitative approaches. . . . Increase the minimum grade-point requirement for newly admitted graduates. . . . Place more pressure on the faculty to do research and to publish. Assembling a concourse relative to a particular decisional situation is part of what Simon (1960) designated as the design phase of the process, in which alternative courses of action are collected. Mere archiving is not the goal, of course, but a new starting point from which alternatives are then appraised, developed, and eventually adopted or discarded, and it is at this point in the process that some of the qualitative methods (e.g., narrative, discourse, and ethnographic analysis) often falter and sometimes never fully regain balance. Given a welter of textual material, the qualitative analyst must find some method of categorization so as to bring order to the enterprise, and this almost inevitably means the superimposition onto verbal protocols of a logical scheme of one kind or another. The conscientious analyst will of course exercise as much care as possible in an effort to assure that the categories used are also functional and not logical only, but there is no cure for the nagging doubt that the categories belong to a greater or lesser extent to the observer rather than the observed. Q methodology alleviates these doubts to a considerable extent by revealing the participants’ own categories and establishing these as the basis for meaning and understanding (Brown, 1996, in press; Watts and Stenner, 2005). How this is accomplished is best grasped in the context of a more extended example, which will also serve to illustrate the quantitative procedures that are involved. 37.2.2 PROBLEM CLARIFICATION WITH Q METHODOLOGY: A CASE STUDY This study, which has been reported in more detail elsewhere (Mattson et al., 2006), focuses on the problem-identification phase of the policy process. In this particular case, a group of stakeholders Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 731 28.8.2007 8:51pm Compositor Name: deshanthi Q Methodology 731 concerned with large carnivore conservation in the Northern Rocky Mountains—ranchers, environmentalists, animal activists, government officials, and academics—met for a two-day workshop to explore one another’s perspectives and to search for a common-interest alternative to legal confrontation. The first day of the workshop was devoted to examining different understandings of the underlying problems and the second day to proposing possible solutions. The process began by inviting the approximately two dozen assembled participants silently to contemplate ‘‘what are the various aspects of the problems associated with carnivore conservation?’’ Participants anonymously jotted down freely associated ideas on a sheet of paper until it became apparent that few if any new ideas would be forthcoming, at which point the sheets were placed in the middle of the table and each participant picked up a sheet containing ideas generated anonymously by some other participant. Exposure to others’ ideas typically generated new ideas, which were added to the sheets. After sheets had been widely circulated, the number of new ideas gradually dwindled. The facilitator then proceeded to guide the group through a round-robin process in which one of the solutions was nominated in turn from the sheet in each participant’s possession. Each nominated solution was discussed, modified through group discussion, and finally added to a list that was taped to the wall for all to see. (Ultimately, second and third round-robin phases assured that no good ideas were omitted from the final sample of problem elements.) Group members then copied each item on a 3’’ 3 5’’ card that had been provided, using the same wording as on the wall. The items were numbered serially, and the item numbers were also recorded on each of the cards. Eventually, the participants collectively generated N ¼ 51 problem elements, and each member was in possession of a pack of 51 cards on which those problems were written. Before proceeding to technicalities, it is important to note that the participants had been purposely selected so as to represent the diverse interests associated with large carnivore conservation in the Northern Rocky Mountain region; consequently, the 51 propositions generated were of wide scope, and there was not a single one among the 51 that all participants did not immediately understand as a matter of shared knowledge. All of the items are reported in Table 37.6, but a small sampling will give a sense of the problems confronting the region: 1. Not all citizens have equal say. 4. Some stakeholders have purposely attempted to accentuate the uncertainty in carnivore science to forestall land management decisions. 8. Increased anxiety among the general public regarding carnivore attacks. 13. Habitat fragmentation and isolation are reducing management options. Tributaries into this concourse of communicability obviously emanate from values, political commitments, specialization expertise, and other social forces, and it is a virtue of Q methodology that it sharpens and clarifies the form and substance of such forces. It should be noted in passing that the wording of some of the items may sound odd and lacking in the niceties of conventional prose, but participants were given ample time in which to offer editorial amendments and to clarify meaning, and many items underwent alteration before the final version was collectively approved. However, unusual or ambiguous the phrasings might appear to an outsider, therefore, there is little reason to doubt that the insiders themselves understood each and every statement. The purpose of this initial phase of item generation was, in this instance, to gather as comprehensive a set of problems as possible, i.e., to render manifest the problems at the top of each individual’s and group’s agenda. The next phase involved distinguishing important from relatively less important problems, and this was accomplished by instructing each participant to Q sort the 51 items along a scale from þ4 (agree) to 4 (disagree), as shown in Table 37.1 for one of the participants. It was first recommended that participants divide the items into three groups (agree, unimportant, and disagree) and that from the agreeable items they then select those five deemed most agreeable: these were placed under the þ4 label, with the five next-most agreeable Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 732 28.8.2007 8:51pm Compositor Name: deshanthi 732 Handbook of Research Methods in Public Administration TABLE 37.1 Q Sort for Person 1 Most Disagree 4 12 19 22 26 33 Most Agree 3 2 1 0 11 12 13 14 7 21 41 43 51 1 14 24 44 45 47 6 8 11 37 46 50 2 9 16 23 25 40 48 18 34 36 38 42 49 15 17 30 31 32 39 10 13 20 28 35 3 4 5 27 29 being placed under the þ3 label. (The labels were arrayed across the tabletop in front of each participant, to serve as an aid in the sorting.) The participants then examined the stack of problem statements with which they most disagreed and selected the five most disagreeable of all (for 4), and then the five next-most disagreeable ( 3). Eventually all 51 items were arrayed in front of the participant, from those most agreeable on the right down to those most disagreeable on the left. The statement numbers were then entered on score sheets so as to preserve the way in which each of the participants prioritized the problems. It deserves mention in passing that the so-called forced distribution pictured in Table 37.1, although somewhat arbitrary in shape and range, is nevertheless recommended for theoretical and practical reasons. Theoretically, a quasi-normal distribution models the Law of Error and is backed by a hundred years of psychometric research indicating that Q sorting and other quasi-ranking procedures typically result in distributions of this kind (Brown, 1985). From a practical standpoint, a standard symmetrical distribution of this kind constrains responses and forces participants to make decisions they might otherwise conceal (e.g., by placing all statements under þ4 and 4), thereby increasing the likelihood that implicit values and priorities will be rendered explicit and open to view. However, because the shape of the sorting distribution has little statistical impact on the subsequent correlation and factor analysis, recalcitrant respondents (who might otherwise refuse to cooperate) can be permitted to follow their own inclinations while being encouraged to adhere as closely as possible to the distribution specified. Given 51 possible problems to consider, there are literally billions of different ways—in fact, more than a billion billion—in which they could be prioritized, and yet the participants individually work through the complexities involved within just a few minutes, which is testimony to the adaptability and efficiency of the human mind. Each person is guided in this enterprise by the values, interests, and principles that are brought to the task, and what these values and interests are can usually be inferred through inspection of the person’s Q sort. Consider, for instance, the Q sort provided by participant 1 (Table 37.1), who singled out the following problems as most important (score þ4): 3. People tend to talk to members of their own group and reinforce their own prejudices. 4. Some stakeholders have purposely attempted to accentuate the uncertainty in carnivore science to forestall land management decisions. 5. Costs of carnivore conservation are disproportionately accrued by local citizens. 27. Lack of trust among participants. 29. Necessary incentives for appropriate behavior by all parties are absent. Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 733 28.8.2007 8:51pm Compositor Name: deshanthi Q Methodology 733 To maximize candidness, participants were not required to place names on their score sheets, and so in many instances it was not possible to associate specific responses with particular individuals, although the identities of many of the respondents eventually came to be known. Personal identities are generally unimportant in a study such as this, however, for what is of interest are the subjective views themselves—i.e., the perspectives that exist within the group—and not the associated characteristics of those who espouse them. In this connection, the first participant’s choice of important issues indicates no allegiance to any side; rather, a sensitivity in particular to process factors such as trust, reinforcement of prejudices, and lack of incentives on all sides. The problem elements with which this person most disagrees (score 4) are also illuminating: 12. 19. 22. 26. 33. Recovery zones for carnivores are arbitrary. Worldview of locals does not include the scarcity of large carnivores. People whose livelihoods are hindered by carnivores are clinging to outmoded ways of life. Wolves are reducing hunting opportunity, and therefore hurting the economy. Humans are intolerant of carnivores. This Q sorter was open about his identity as a Canadian government official whose duties brought him into contact with both environmentalists and ranchers threatened in particular by expanding grizzly populations, and who was therefore aware both that recovery zones are not arbitrary and that the worldviews of local people were not opposed to issues of wildlife scarcity. The Q sort is part of the technical accouterment of Q methodology, to which it bears a relationship analogous to the telescope to astronomy or the galvanometer to electricity; i.e., it brings into view for direct and sustained inspection those structures of subjectivity and preference that suffuse political and administrative life. The decisional situation facing the participants in this study contains objective realities—e.g., the existence of wolf packs and grizzlies, disconnected habitats, farming and ranching operations, and public lands, as referred to in the statements above—but what is felt to constitute an important as distinguished from an unimportant problem is a phenomenon of another kind: it is the subjective medium within which the facts as known by each participant are suspended, and in terms of which they are given meaning and salience, as rendered manifest by the þ4= 4 subjective scale. The concourse of problems that the participants produced is fed from such subjective currents, and yet these sources remain obscure until transformed by the mechanics of Q sorting. Among the features of Q methodology which distinguish it from many other quantitative procedures is that the elements comprising the Q sort, unlike those in a rating scale, are not constrained by prior meaning; consequently, there can be no correct way to do a Q sort in the same sense as there is a right way to respond to an IQ test or to the Graduate Record Examination. What a particular rancher or environmentalist considers the most important problems facing society is simply that person’s judgment, which may or may not be in agreement with others’ appraisals. It is this absence of norms—not just in practice, but in principle—that renders the issue of validity of such little concern in Q methodology (Brown, 1992, 1993). Validity aside, however, we can nevertheless proceed to compare subjective appraisals, and to determine the varieties of perspectives in existence and the shape and character of each, and it is at this point that quantification is brought to bear. Q sorts are conventionally intercorrelated and factor analyzed, and the correlation phase is fairly elementary, as shown in Table 37.2 for the Q sorts provided by participants 1 and 2. The scores in column 1 are the same as those shown in the previous table (for the 51 issues and problems which were sorted), whereas those in column 2 were given during the same sitting by participant no. 2. The numbers in column d2 are the squared differences in score between the two performances. Were the two participants’ views absolutely identical, there would of course be no differences in score between the two for any of the statements, in which case the squared differences would also be zero, as would the sum of squared differences: In this extremely rare and virtually impossible instance, Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 734 28.8.2007 8:51pm Compositor Name: deshanthi 734 Handbook of Research Methods in Public Administration TABLE 37.2 Correlation between Q Sorts 1 and 2 Item (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) (25) (26) (27) (28) (29) (30) (31) (32) (33) (34) (35) (36) (37) (38) (39) (40) (41) (42) (43) (44) (45) (46) (47) (48) 1 2 0 4 4 4 1 3 1 0 3 1 4 3 2 2 0 2 1 4 3 3 4 0 2 0 4 4 3 4 2 2 2 4 1 3 1 1 1 2 0 3 1 3 2 2 1 2 0 2 2 1 2 0 3 4 4 1 4 1 0 3 3 4 3 4 1 0 0 0 4 3 2 2 3 4 0 1 4 3 4 2 1 2 2 1 2 1 1 2 1 3 1 3 4 0 2 3 d2 0 1 36 16 49 25 1 0 16 16 1 49 0 36 1 16 1 1 16 9 1 1 4 0 9 0 16 16 64 25 36 0 9 1 25 4 9 0 1 4 16 4 16 1 36 1 16 9 Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 735 28.8.2007 8:51pm Compositor Name: deshanthi 735 Q Methodology TABLE 37.2 (Continued) Correlation between Q Sorts 1 and 2 Item (49) (50) (51) 1 2 1 1 3 r12 ¼ 1 2 2 1 640 ¼ 620 d2 1 9 16 Sd 2 ¼ 640 0:03 the correlation would be r ¼ 1.00; were the two views diametrically opposite, the correlation would be r ¼ 1.00. In the instant case, there are differences of a greater or lesser extent among the statements, as recorded in column d2, the sum amounting to 640. When Q sorts follow the same distribution (hence have the same mean and variance), a convenient formula for correlation is as shown in Table 37.2, where 640 is the sum of squared differences and 620 is the combined sum of squares of the scores in the two Q sorts. The correlation between these two Q sorts is therefore r ¼ 0.03. Q-sort correlations are rarely of any interest in and of themselves and typically represent only a phase through which the data passes on the way to being factor analyzed. It is worth noting, however, that the correlation coefficients are subject to standard error formulae. For example, we can assume related if they exceed pffiffiffiffi pro tem that two participants’ views are substantially pffiffiffiffi 2:58(1= N ) ¼ 0:37 (for N ¼ 51 Q statements), where sr¼0 ¼ 1= N is the standard error of a zero-order coefficient, and z ¼ 2.58 is the number of standard errors required to incorporate 99 percent of the area under the normal curve. The above correlation of 0.03 for participants 1 and 2 (which is less than the requisite 0.37) therefore indicates that their respective appraisals of problems associated with carnivore conservation share little in common. Of the 26 participants originally involved in this problem-clarification process, only seven have been included in the following analysis so as to keep calculations and tabular displays within manageable limits for illustrative purposes. The intercorrelations among the seven participants are displayed in Table 37.3. Note that the correlation between respondents 1 and 2 is r ¼ 0.03, as calculated above. The correlation matrix has a certain dynamic to it, just as did the group from which the Q sorts were taken: hence a rancher may have been concerned about the threat to his herds and pets from wolf packs, and might therefore have been less sympathetic to other participants’ expressed concerns about endangered species and habitat fragmentation. Which issues each person places at the top, middle, or bottom of the Q sort can therefore be influenced by myriad considerations, from personal motivation to the climate created by national and local politics, each force being explicitly or sometimes only implicitly weighed by the Q sorter and ultimately having its impact on the final statement arrangement. The correlation matrix is therefore a thick soup of dynamic forces of chaotic proportions that nevertheless summarizes the balance of influences and the way in which the various participants have worked their way to separate conclusions about what are important versus unimportant problems related to carnivore conservation. Q methodology was conceived in the context of factor analytic developments as they were unfolding in the 1930s, which was late in the heyday of Charles Spearman and the ‘‘London School.’’ Stephenson’s (1935) psychometric innovation of ‘‘correlating persons instead of tests’’ consisted of applying the mathematics of factor analysis to correlation matrices of the above kind, in which one person responses were correlated with other person responses. The result was typically a typology of response, with one subgroup of similar Q sorts constituting one factor, another group Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 736 28.8.2007 8:51pm Compositor Name: deshanthi 736 Handbook of Research Methods in Public Administration TABLE 37.3 Correlations among Seven Q Sorts Ps 1 2 3 4 5 6 7 (1) (2) (3) (4) (5) (6) (7) — 03 — 39 36 — 15 59 30 — 19 36 45 10 — 05 84 27 42 34 — 09 22 50 33 39 32 — Note: Decimals to two places omitted. constituting another factor, and so forth. Q factors therefore have the status of separate attitudes, perspectives, or understandings, or, in the extant case, of problem priorities. By whatever substantive terminology (attitudes, perspectives, value orientations, problem priorities, etc.), the factors in Q methodology consist of conglomerates of convergent subjectivity as determined by the concrete operations of the persons themselves as they perform the Q sorts—hence the term operant subjectivity (Stephenson, 1977). The number and content of the factors, despite their thoroughly subjective character, are therefore emergent and purely empirical features of the thinking and feeling of the persons who provided the Q sorts: Had the group members felt differently about the issues, their Q sorts would have been different, and so would the structural features of the correlation matrix and so, as a consequence, would the factors, which in their turn summarize those structural features. The role of factor analysis in this instance is to document the current state of thinking within this diverse group with respect to the issues at the group’s focus of attention. The technicalities of factor analysis are addressed elsewhere in this volume, and relatively simplified introductions are available for those wishing to achieve a conceptual grasp (e.g., Adcock, 1954; Brown, 1980, pp. 208–247; Kline, 1994); we will therefore bypass the detailed calculations involved in extracting the unrotated loadings for centroid factors (a) through (g), as shown in Table 37.4. Suffice it to say that from a statistical standpoint, the seven unrotated factors represent a partial decomposition of the previous correlation matrix. This can be illustrated in terms of any two Q sorts (say, 2 and 6), which are correlated in the amount 0.84 (see Table 37.3). The sum of the cross products of the unrotated factor loadings for these two Q sorts is (0.75)(0.71) þ ( 0.58)( 0.39) þ


þ ( 0.05)(0.12) ¼ 0.84, which indicates that all of the original correlation of 0.84 can be composed from these seven factors. (Factor loading cross products rarely sum up to exactly the original correlation, as they did in this instance, due to residual correlations that typically remain after the seven factors have been extracted.) The factor loadings indicate the correlation of each Q sort with the factor; hence, Q sort 2 correlates with the first factor in the amount f ¼ 0.75; factor loadings are therefore subject to p the ffiffiffiffi same standard error estimates as noted previously for correlation coefficients, i.e., sr¼0 ¼ 1= N ¼ 0:14, where N ¼ 51 statements. Factor loadings in excess of 2.58(0.14) ¼ 0.37 are significant (p < .01), which means that the Q sort of the second person is significantly associated with the first factor. Had there been only a single viewpoint shared by all participants, then all of the correlations would have been large and positive, only one significant factor would have been in evidence, and there would have been no trace of significant loadings on the other factors. As Table 37.4 shows, however, at least the first two of the unrotated factors contain significant loadings, and some of the loadings on the fifth and sixth factors are also substantial; we would therefore anticipate that there are at least two and perhaps three or four separate points of view within this group of participants. Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 737 28.8.2007 8:51pm Compositor Name: deshanthi 737 Q Methodology TABLE 37.4 Unrotated and Rotated Centroid Factor Loadings Unrotated Factors Rotated Ps a b c d e f g X Y (1) (2) (3) (4) (5) (6) (7) 14 75 72 46 55 71 56 39 58 43 31 22 39 17 21 18 03 34 18 28 22 16 30 22 29 04 12 13 22 04 26 02 04 12 37 10 09 11 15 05 07 51 04 05 11 15 15 12 24 14 97 22 61 21 77 25 39 13 82 08 54 22 67 Note: Factor loadings in boldface significant ( p < .01). Although there are some occasions in which the factor analyst might rest content with the unrotated loadings as these have been extracted from the correlation matrix, in the overwhelming number of cases unrotated loadings do not give the best view of what is transpiring; it is typically the case, therefore, that the unrotated factor loadings are superseded by an alternative set of loadings which give a more focused view. This transformation process from the unrotated to an alternative set of loadings is accomplished through the process of factor rotation. The most conventional scheme for factor rotation is to rely on the Varimax routine found in all software packages (such as SPSS) containing factor analysis, and it is the statistical goal of Varimax to rotate the factors in such a way that each variable (or Q sort) is maximized on a single factor and minimized on all other factors, a solution referred to as ‘‘simple structure.’’ If a researcher is totally in the dark about the topic under examination, as is sometimes the case, then leaving factor rotation to Varimax or some other algorithm may be as good a strategy as any other. However, it is unlikely that there is a single set of mathematical rules, such as Varimax, which is apt to provide the best solution to problems under any and all conditions. In particular, when an investigator has some knowledge or even vague hunches about a situation, then it is often wise to permit that information to play some role in the experimental setting. It is for situations such as these that room was made in Q methodology for ‘‘theoretical rotation,’’ a judgmental procedure that is explicitly built into the PQMethod freeware program (Schmolck and Atkinson, 2002) and into the PCQ Windows-based program (Stricklin and Almeida, 2004). Space precludes going into great detail concerning theoretical rotation (see Brown and Robyn, 2004; Knamer and Gavina, 2004), but what is essentially at issue can be demonstrated in terms of Figure 37.1, which graphically displays the location of each Q sort in terms of unrotated factors (a) and (b) in Table 37.4. The pairs of loadings for each of the seven Q sorts are duplicated in Figure 37.1 where it is shown that Q sort 1 has a loading on factor (a) in the amount 0.14 and on factor (b) in the amount 0.39, and these two loadings serve to locate Q sort 1 in the two-dimensional space in Figure 37.1. Similarly, Q sort 6 is located 0.71 on factor (a) and 0.39 on (b). The relative proximity of each of the Q sorts is a spatial expression of their degree of similarity with respect to these two factors, the nature of which are undefined at this point. Seven unrotated factors were originally extracted, which is the default in the PQMethod program. Of these seven, factors (a) and (b) were initially chosen for rotation due to an interest in Q sorts 4 and 6, the former a rancher and the latter an environmental activist, who were expected to be at loggerheads vis-à-vis most of the issues associated with large carnivores. As indicated in Table 37.4, Q sorts 4 and 6 have substantial loadings on factor (a) ( 0.46 and 0.71, respectively) and also on (b) (0.31 and 0.39), which means that they are to a considerable extent at opposite poles of both factors. Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 738 28.8.2007 8:51pm Compositor Name: deshanthi 738 Handbook of Research Methods in Public Administration 1.0000 2 6 Factor a 3 7 0.5000 (a) (b) 1 14 39 2 75 −58 3 72 43 4 −46 31 5 55 22 6 71 −39 7 56 17 5 1 0.0000 4 −0.5000 −1.0000 −1.0000 −0.5000 0.0000 0.5000 1.0000 Factor b FIGURE 37.1 Location of each Q sort in terms of unrotated factors (a) and (b), illustrating spatially the degree of similarity of the Q sorts with respect to factors (a) and (b). Factor loadings in boldface significant ( p < .01). Figure 37.2 displays the relationships among the Q sorts when the original vectors are rotated counterclockwise by 378. Also shown are the new factor loadings which indicate that Q sort 4 is now saturated on the new factor (a) in the amount 0.56 (up from 0.46) and on (b) at 0.03 (down in magnitude from 0.31). Q sort 6 is likewise more focused on factor (a) to the extent of 0.80. 1.0000 2 6 (a) (b) 1 −12 40 2 93 −01 3 31 78 4 −56 −03 5 30 51 6 80 11 7 35 47 0.5000 7 Factor a 5 3 0.0000 1 −0.5000 4 −1.0000 −1.0000 −0.5000 0.0000 0.5000 1.0000 Factor b FIGURE 37.2 Displays each Q sort after rotating the original vectors counterclockwise by 378. Factor loadings in boldface significant ( p < .01). Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 739 28.8.2007 8:51pm Compositor Name: deshanthi 739 Q Methodology These two sets of coefficients, unrotated, and rotated, are mathematically equivalent, as can be seen when the respective loadings are squared and summed, as follows (to four decimal places, for accuracy): Q sort 4 Q sort 6 ¼ 0.3093 unrotated: ( 0.4615)2 þ 0.31032 rotated: ( 0.5553)2 þ ( 0.0300)2 ¼ 0.3093 unrotated: 0.70512 þ ( 0.3908)2 rotated: 0.79832 þ 0.11222 ¼ 0.6499 ¼ 0.6499 These sums are identical in both cases as they would be for any of the infinite number of other rotations. Another method of verification is to examine the cross products of factor loadings for any two Q sorts, e.g., 4 and 6: Unrotated: ( 0.4615) (0.7051) þ (0.3103) ( 0.3908) ¼ 0.4467 Rotated: ( 0.5553) (0.7983) þ ( 0.0300) (0.1122) ¼ 0.4467 which, again, are identical. The cross product sums of 0.4467 indicate that of the original correlation of 0.42 between Q sorts 4 and 6 (see Table 37.3), factors (a) and (b), both the rotated and unrotated versions, account for 45 percent of that amount. It is scarcely necessary for an investigator to comprehend the mathematics underlying the rotational phase of a Q study because the PQMethod program (Schmolck and Atkinson, 2002) displays the data configuration pictorially, similar to Figure 37.1: the investigator then simply directs the vectors to a new location based on theoretical considerations, e.g., by specifying that factors (a) and (c) be rotated 378 counterclockwise so as to locate Q sort 4 on a single factor. The selection of any particular Q sort response as the focus for factor rotation can be due to any considerations that might arise in the course of a study—for instance, the person’s status (e.g., as an environmental activist, as in the case of participant 4, or as rancher, as in the case of participant 6), something a person said in the course of an interview, others’ attitudes toward a specific person or group, and so forth. It is at the stage of factor rotation that an investigator’s hunches, guesswork, and hypothetical interests enter into the analysis. The rotation described above was only the first of several that resulted in the final two-factor solution shown in Table 37.4. Space precludes detailing the thinking that went into these rotations, but the two rotated factors (X and Y ) indicate that the participants were, like Gaul, divided three ways in their understanding of the problems associated with large carnivore conservation. The figures show that participants 2 and 6, both environmental activists, define the positive pole of factor X, which means that these two participants share a common view about the problems before them; that participant 4, a rancher, takes a position that is diametrically opposed to participants 2 and 6 (hence tends to score at 4 what the other two score at þ 4, and vice versa); and that participants 1, 3, 5, and 7 define factor Y, which represents a third perspective that is orthogonal to the other two. As noted previously, the seven Q sorts featured above (in Tables 37.4 and 37.5 and Figures 37.1 and 37.2) were abstracted from a larger group of 26 Q sorts for purposes of demonstrating the mechanics of correlation, factor analysis, and factor rotation. The complete analysis therefore involved a 26 3 26 correlation matrix that condensed into three factors (A, B, and C), as shown in Table 37.5. (As a technical sidelight, factors X and Y in Table 37.4 were based on a centroid factor analysis with theoretical rotation, whereas factors A, B, and C in Table 37.5 were based on a principal components analysis with Varimax rotation.) Factor A in the larger study represents the same polarity between environmentalists and ranchers that was captured in factor X (Table 37.4), and factor B is equivalent to the previous factor Y and represents the view mainly of academic policy scientists. Factor C is yet a fourth view and one adopted by many participants occupying government positions. This latter set of factors (A, B, and C) provides the basis for factor interpretation, which relies on the arrays of factor scores. Before moving on to the interpretative phase of Q methodology, it is important to emphasize once again the operant character of these three factors. The Q sample of problems concerning Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 740 28.8.2007 8:51pm Compositor Name: deshanthi 740 Handbook of Research Methods in Public Administration TABLE 37.5 Rotated Factor Loadings Ps A B C (1) (2) 3 4 5 6 (7) (8) 9 10 11 (12) 13 14 15 16 17 18 (19) 20 21 (22) 23 24 25 26 23 87 71 26 64 25 15 60 29 29 72 23 28 53 25 05 43 85 79 72 15 25 41 62 08 16 41 21 03 05 11 52 75 11 38 56 08 71 66 33 42 75 08 20 28 00 63 66 02 34 07 32 61 05 42 56 04 16 06 13 40 29 15 04 34 31 38 20 61 02 00 27 11 06 66 20 64 53 Note: Participants (Ps) in parentheses are the same as 1–7, respectively, in Table 37.4. Loadings in boldface significant ( p < .01); decimals to two places omitted. carnivore conservation was generated solely by the 26 participants themselves, the Q technique rankings of the problems emanated from their own subjective perspectives, and the three factors are therefore natural categories of thought and sentiment within the group itself; i.e., they are of functional and not merely categorical significance (Brown, 2002, 2006b). This is not to assert methodological individualism, nor is it to deny that the views expressed in the group could have been socially constructed or in other ways a function of forces in the social and economic order. It is simply to acknowledge that the factors are a function of the lived experiences of individuals, and that they are purely inductive in the sense that their number and character have been induced from those individuals who produced them. Table 37.5 shows that participants 2, 3, 5, 11, 18, 19, and 20 (those in boldface) share a common outlook (i.e., the seven of them have sorted the Q-sample statements in essentially the same pattern), and what is common to their perspective can be approximated by melding their separate responses into one. This is accomplished by calculating factor scores, which are the scores (from þ4 to 4) associated with each statement in each of the factors. The seven responses first have to be weighted to take into account that Q sort 2 (with a loading of 0.87) is a closer approximation to factor A than is Q sort 3 (with a loading of 0.71). Weighting proceeds according to the formula w ¼ f=(1 f 2), Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 741 28.8.2007 8:51pm Compositor Name: deshanthi 741 Q Methodology where f is the factor loading. The weight of Q sort 2 is therefore w(2) ¼ 0.87=(1 0.872) ¼ 3.58, and by the same calculations w(3) ¼ 1.43, hence the former is magnified 3.58=1.43 ¼ 2.50 times the former when the two responses are merged. The same process is repeated for factors B and C. Computational details are to be found in Brown (1980, pp. 239–247). The end products of the above calculations, as shown in Table 37.6, are three composite Q sorts (one each for factors A, B, and C), with each factor Q sort being composed of the weighted individual Q sorts which define the factor. (All computations are built into the PQMethod and PCQ TABLE 37.6 Factor Score Arrays Factors A B C Q-Sample Statements 0 1 1 2 2 4 4 1 2 2 3 3 1. 2. 3. 4. 2 3 4 1 4 3 1 3 4 4 3 2 2 4 0 2 3 0 3 4 1 1 3 1 2 2 2 0 2 4 4 1 1 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 4 1 0 2 3 1 0 1 2 16. 17. 18. 0 0 3 2 2 2 3 4 1 2 3 4 19. 20. 21. 22. 4 2 2 23. 3 3 1 1 3 1 24. 25. 4 1 2 1 3 4 4 3 1 3 4 4 0 2 4 26. 27. 28. 29. 30. Not all citizens have equal say. Agencies use rigid ‘‘planning’’ that excludes people in genuine problem solving. People tend to talk to members of their own group and reinforce their own prejudices. Some stakeholders have purposely attempted to accentuate the uncertainty in carnivore science to forestall land management decisions. The costs of carnivore conservation are disproportionately accrued by local citizens. Elected and appointed officials are captured by industry. There are too many grizzly bears on private land. Increased anxiety among the general public regarding carnivore attacks. Carnivore population goals are set politically and not biologically. National groups fail to acknowledge local support for conserving carnivores. Over consumption creates demands for gas and oil production in carnivore habitat. Recovery zones for carnivores are arbitrary. Habitat fragmentation and isolation are reducing management options. Too few carnivores; not enough carnivore habitat, in quantity and quality. Known trends in human population and development are ignored in management plans. Carnivore management falls short of what science shows—a need for recovery. Fragmented efforts and jurisdictions. Participants in carnivore conservation do a poor job of distinguishing facts from values. The worldview of locals does not include the scarcity of large carnivores. Communication, especially listening, is poor or insufficient. Current federal regulations promote deadlock and conflict more than collaboration. People whose livelihoods are hindered by carnivores are clinging to out-moded ways of life. The listing of the species through litigation only is having a negative impact on conservation. The Endangered Species Act doesn’t define what recovery is. Public lands are managed to protect business and political interests instead of maintaining biodiversity, including carnivores. Wolves are reducing hunting opportunity, and therefore hurting the economy. Lack of trust among participants. Local affected interests are alienated by current processes. The necessary incentives for appropriate behavior by all parties are absent. Economic incentives aren’t playing a large enough role in carnivore conservation. (continued ) Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 742 28.8.2007 8:51pm Compositor Name: deshanthi 742 Handbook of Research Methods in Public Administration TABLE 37.6 (Continued) Factor Score Arrays Factors A B C 4 4 1 2 2 4 0 2 3 0 1 0 0 3 3 0 1 2 2 2 2 1 1 0 0 2 0 4 4 0 1 2 1 3 0 3 4 3 3 4 1 1 4 2 2 1 1 3 2 4 1 1 3 3 3 2 3 1 0 0 1 1 1 Q-Sample Statements 31. Carnivore conservation is in the context of declining populations when the opposite is happening. 32. Human population growth and consumption patterns contradict large carnivore conservation. 33. Humans are intolerant of carnivores. 34. The role of carnivores in the ecological system is undervalued and misunderstood. 35. The human=carnivore interface challenges individual values concerning living with carnivores. 36. Insufficient funding for conservation, research, and management. 37. Federal managers fail to respect and include independent researchers. 38. Past and ongoing forest practices. 39. Too much motorized access. 40. Carnivore conservation interferes with economic interests. 41. Dependence on hunting license revenue compromises the integrity of game and carnivore management. 42. Critical voices are denied access to the process. 43. Inadequate understanding and appreciation of laws directing the conservation of carnivores and their purpose. 44. Too much data is required to implement management actions. 45. Lack of skill and leadership within the agencies. 46. Carnivores are emotionally charged political symbols. 47. Too many carnivores killed by people. 48. A federal compensation program funded by public–private investment does not exist. 49. Lack of critical thinking about management choices before decisions are made. 50. Too much deference is given to paranoia or the implicit threat of violence. 51. Lack of good information and education. software packages; see Schmolck and Atkinson, 2002; Stricklin and Almeida, 2004.) In the case of factor A, for instance (see Table 37.6), the five highest scores (þ4) are associated with statements 9, 13, 14, 16, and 47, the five next highest (þ3) with statements 6, 12, 15, 25, and 42, and so forth in the same distribution as employed in the original Q sorts (see Table 37.1). What started off as 26 Q sorts is therefore reduced to 3, which subsume the others, and it is these 3 Q sorts that provide the basis for factor interpretation. In most circumstances, factor interpretation proceeds best by (a) physically laying out the factor Q sort and describing its manifest content, giving special attention to those statements at the positive and negative extremes of the distribution, (b) establishing the underlying theme, which is the glue holding the Q sort together, and (c) comparing and contrasting the factor with whatever other factors have also been revealed. Factor interpretation is as much hermeneutical art as science, and necessarily so, but the interpreter is not given free rein in this respect because an interpretation is constrained by the factor scores from which it cannot stray too far. Interpretation is further complicated by the fact that the statements in a Q sort can assume various meanings depending on their position relative to other statements, and this places Q methodology in the same league with literary theory in the sense that it has to address ‘‘the ways in which the form that transmits a text . . . constrains [or facilitates] the production of meaning’’ (Chartier, 1995, p. 1). Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 743 28.8.2007 8:51pm Compositor Name: deshanthi Q Methodology 743 Space again precludes going into detail, but something of what is involved can be seen in reference to a cursory look at factor A, which, as Table 37.6 reveals, gave highly positive scores to the following statements: Most Positive (þ4, þ 3): (9) Carnivore population goals are set politically and not biologically. (12) Recovery zones for carnivores are arbitrary. (14) Too few carnivores; not enough carnivore habitat, in quantity and quality. (15) Known trends in human population and development are ignored in management plans. (16) Carnivore management falls short of what science shows—a need for recovery. (25) Public lands are managed to protect business and political interests instead of maintaining biodiversity, including carnivores. (47) Too many carnivores killed by people. There were other statements receiving high positive scores, but the statements displayed above were selected because they also serve to distinguish factor A from B and C; i.e., the factor scores associated with the above statements in A are significantly higher than the scores for the same statements in factors B and C, as Table 37.6 shows (as a rule of thumb, differences in score of 2 or more are significant, p < .01; for details, consult Brown, 1980, pp. 244–246). The statements given positive emphasis suggest that the persons comprising factor A, many of whom are environmental activists, are prepared to advocate on behalf of carnivores as well as the proconservation clientele groups that many of their agencies serve, consequently they complain about too little habitat, that carnivores still need to be protected, and that decisions concerning control of animal populations are made on the basis of politics rather than science and are designed to protect economic and political interests. As Figure 37.2 and Table 37.5 show, factor A is bipolar and therefore represents two diametrically opposed points of view. In marked contrast to the environmentalists at the positive pole of factor A, participants 8 and 14 are ranchers who have tended to agree most with those statements with which the environmentalists have disagreed most. Consequently, the ranchers’ perspective can be approximated by reversing the signs of the factor A scores in Table 37.6, which reveals the following distinguishing statements: Most Positive Statements for Factor A-negative (þ4, þ 3): (7) There are too many grizzly bears on private land. (10) National groups fail to acknowledge local support for conserving carnivores. (23) The listing of the species through litigation only is having a negative impact on conservation. (24) The Endangered Species Act doesn’t define what recovery is. (26) Wolves are reducing hunting opportunity, and therefore hurting the economy. (31) Carnivore conservation is in the context of declining populations when the opposite is happening. (44) Too much data is required to implement management actions. Unlike the environmentalists, therefore, who strongly disagree with the above, the participants involved in ranching think that there are too many large carnivores, that these animal populations are not in decline, and that the listing of endangered species is ambiguous. They also object to national groups characterizing local ranchers as somehow opposed to conservation, and complain about excessive bureaucratic demands (requiring ‘‘too much data’’) associated with management implementation. The polarity between the environmentalists and agricultural interests as documented in factor A is thrown into sharper relief when it is contrasted with the different commitments of the other two factors. Consider first a few of those statements to which factor B gives high scores and which distinguish B from factors A and C (see scores in Table 37.6): Distinguishing Statements for Factor B (þ4, þ 3): (2) Agencies use rigid ‘‘planning’’ that excludes people in genuine problem solving. (17) Fragmented efforts and jurisdictions. (21) Current federal regulations promote deadlock and conflict more than collaboration. (28) Local affected interests are alienated by current processes. (46) Carnivores are emotionally charged political symbols. Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 744 28.8.2007 8:51pm Compositor Name: deshanthi 744 Handbook of Research Methods in Public Administration Factor B is not partisan in the way in which the polarized groups comprising factor A are; rather, B regards the problems surrounding carnivore conservation (including the factor-A polarization itself) as more a function of jurisdictional complexities that have resulted in fragmented authority and decision making, the exclusion of local participation, and the inadvertent stirring of emotions and conflict. As it turned out, almost all of the participants serving to define factor B—nos. 6, 7, 12, 13, 16, 21, and 22 (see Table 37.5)—were primarily academic policy specialists with substantive interest in conservation biology, and who therefore saw problems more as a consequence of structural defects one level above the fray. Factor C seemed least well defined of the factors, but of those few Q sorters significantly loaded on that factor only (see Table 37.5), most were state and federal agency employees. As might be expected, therefore, factor C deflects attention from government and points a figure at other problematic (and especially economic) sources: Distinguishing Statements for Factor C (þ4, þ 3): (4) Some stakeholders have purposely attempted to accentuate the uncertainty in carnivore science to forestall land management decisions. (30) Economic incentives aren’t playing a large enough role in carnivore conservation. (32) Human population growth and consumption patterns contradict large carnivore conservation. (35) The human=carnivore interface challenges individual values concerning living with carnivores. (36) Insufficient funding for conservation, research, and management. In fact, statements that place blame specifically on government officials are among those that factor C rejects to a much greater degree than do factors A and B (see Table 37.6): Distinguishing Negative Statements ( 4, 3): (37) Federal managers fail to respect and include independent researchers. (42) Critical voices are denied access to the process. (45) Lack of skill and leadership within the agencies. Before concluding, it is important to note that just as there are statements that can serve to distinguish a factor from all others, so are there often statements on which all factors converge as a matter of consensus, and these provide the basis for locating common understandings of the core problems and possible ways out of the conceptual quagmire. In the instant case, there is precious little common ground on which the three factors might stand together, especially given the polarity of factor A, but more detailed inspection of the perspectives at issue reveals several points of virtual consensus, among which are the following (for details, consult Mattson et al., 2006, p. 401): (3) People tend to talk to members of their own group and reinforce their own prejudices. (13) Habitat fragmentation and isolation are reducing management options. (27) Lack of trust among participants. (32) Human population growth and consumption patterns contradict large carnivore conservation. (36) Insufficient funding for conservation, research, and management. Despite the distinct and often incompatible views of the four groups, there remains at least mild agreement and no strong disagreement with the perceptions that all parties tend toward intransigence and mistrustfulness, and that habitat fragmentation, human settlement patterns, and insufficient funding are aggravating the situation. The scores for these statements range from þ4 to no lower than 1, which gives rise to the possibility that the four factor groupings could begin to cooperate by endeavoring to effect outcomes in the directions in which these statements point. In this regard, Harold Lasswell (1963) once remarked that ‘‘a continually redirected flow of events can be progressively subordinated to the goals conceived in central decision structures’’ (p. 221), and those structures which he conceived are not unlike these factors, which are clearly aimed at directing the flow of events. Whether any one or all three of them can succeed in bending future events to their preferences depends at least in part on the kind of self clarification which Q methodology can provide, as the preceding illustration demonstrates. Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 745 28.8.2007 8:51pm Compositor Name: deshanthi 745 Q Methodology 37.3 APPLICATIONS OF Q METHODOLOGY IN PUBLIC ADMINISTRATION: A BIBLIOGRAPHICAL GUIDE AND RESEARCH AGENDA Q methodology is valuable for public administration research, including the investigation of topics related to general public administration and management and to fields such as budgeting, public personnel administration, and public policy. In this section, we review the use of Q methodology for public administration research and summarize some representative studies. Also, we point out other possible avenues for future research using Q. Of course, this section only suggests the potential contribution of Q methodology to the study of public administration and is not exhaustive: the possibilities for the application of Q methodology are boundless wherever subjectivity is implicated. 37.3.1 GENERAL PUBLIC ADMINISTRATION AND PUBLIC MANAGEMENT Q methodology has been used to explore many larger issues in public administration and public management, including the attitudes and perceptions of bureaucrats, administrative behavior, organizational culture, and leadership. These issues can be divided into three categories: how bureaucrats view different issues affecting their jobs, decisions, and behaviors; how public administrators and other public sector professionals perceive their jobs and their work; and how public organizations function, including their culture and leadership. 37.3.1.1 How Bureaucrats View Issues Affecting Their Jobs and Management Decisions Q methodology can identify the opinions or viewpoints of individuals on relevant events or issues and is consequently well suited to investigate the viewpoints and attitudes of bureaucrats, which have received much attention in public administration. In fact, researchers have used Q methodology to assist in understanding the views of bureaucrats on issues such as affirmative action programs (Decourville and Hafer, 2001), bureaucratic discretion (Wood, 1983), administrative ethics (Hiruy, 1987), the impact of diversity on government policies (Wolf, 2004), and public service motivation. This latter topic was addressed by Brewer et al. (2000), who examined why people choose to work in the public sector. They administered Q sorts on public sector motivation to 69 public employees and students of public administration in six different states. In their analysis of the Q sorts, they identified four Q factors—labeled Samaritans, Communitarians, Patriots, and Humanitarians—each representing a different motivation for engaging in public service. Among other things, they found that ‘‘the desire for economic rewards’’ was not a defining feature of any of these motivations. Related studies have ventured into more remote territory. For example, studies by Sun (1992) and Sun and Gargan (1999) examined the opinions of public administration scholars and practitioners regarding the practical use of scholarly research in Taiwan, and Vajirakachorn and Sylvia (1990) investigated the influence of traditional Buddhist culture and modern bureaucracy on Thai administration attitudes. They asked 94 Thai administrative elites in the Ministry of Interior to sort 54 statements,4 and from this process four administrative types emerged: the first group endorsed modern and bureaucratic values most strongly; the second group was characterized by mixed attitudes and held values that were midway between modern bureaucratic principles and Buddhist traditional ideas; and the third and fourth groups expressed a slight mix in opinion, but in general expressed more agreement with values that correspond to traditional bureaucratic practices, such as planning in management, rationalism and scientific reasoning in making decisions, and merit based civil service. This research illustrates the utility of Q in discerning attitude differences among public administrators. Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 746 28.8.2007 8:51pm Compositor Name: deshanthi 746 37.3.1.2 Handbook of Research Methods in Public Administration How Public Sector Professionals Understand Their Jobs and Carry Out Their Work? Q methodology has been applied in research to understand better how different public administrators perceive their work and their roles, including their relationships with superiors, subordinates, colleagues, and the public (Bidwell, 1957; Johnson, 1973; McDale, 1989; Scheb, 1982; Selden et al., 1999; Vroman, 1972). Typically these studies have identified varying perspectives that managers have toward their jobs and work, and they show that managers differ greatly in how they approach and carry out their work. Although most of these Q studies have focused on American public managers, at least two have included comparisons of American public managers with managers in other countries. Durning and Gajdamaschko (1998, 2000) examined how Ukrainian public managers perceive their jobs and roles in society, and they compared those perspectives to those of a group of American public managers. Gough et al. (1971) compared the managerial perspectives and preferences of American to those of Italian public managers. The Q sorts of Italian administrators were analyzed, and the following administrative types were identified: innovator, mediator, actionist, the moderate, achiever, and the realist. Then, the authors collected Q sorts from 110 American administrators to analyze typological variations among the two cultures. The addition of American public administrators revealed some stylistic variations in perspectives and practices among the two groups of administrators. The result of the Q sorts showed that American administrators perceived interpersonal relationships and career opportunities to be highly important, whereas Italian administrators were more concerned with job security and structure. In terms of the administrative types identified in the analysis of Italian administrators, American bureaucrats identified more with the mediator role; that is, they perceived themselves as tolerant, modest in demands, and generous in relationships. American administrators were least likely to assume the role of actionist, a manager who is tough minded, decisive, and indifferent to the feelings of subordinates. In related studies using Q methodology, Yarwood and Nimmo (1975) examined different stakeholders’ perceptions of bureaucratic images and how these images differed between academicians and other stakeholders. Also, Yarwood and Nimmo (1976) explored the ways in which administrators, legislators, and citizens orient themselves to bureaucracy and the accuracy of these groups in estimating the attitudes of each other toward the bureaucracy. 37.3.1.3 Public Management Topics Such as Organizational Culture and Leadership Public management research is closely linked to general public administration and public personnel scholarship. It includes topics related to organization development and leadership, and scholars have employed Q to research both. For example, organization culture has been the focus of studies by Chatman (1989), O’Reilly et al. (1991), and Sheridan (1992), and in Romania by Iliescu (2005, pp. 174–196). Dunlap and Dudley (1965), Thomas and Sigelman (1984), and Wong (1973) have researched topics related to leadership, and Tolymbek (2007) has examined emerging leadership patterns and expectations in Kazakhstan. Beyond these uses of Q methodology, Q could be valuable to explore public management topics such as public–private distinctions. For example, scholars could explore patterns among public and private managers on a wide range of subjects related to organizational tasks, functions, and characteristics. Overtures to this effect from the private side have recently been made by Gauzente (2005). Another potential application of Q would be to assess the trust of managers, an increasingly important topic in public management. Most of the existing research on trust is based on attitude surveys and personal interviews of public managers (Carnevale, 1995). Alternatively, Q methodology could be used to explore public managers’ worldviews toward trust. A concourse of Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 747 28.8.2007 8:51pm Compositor Name: deshanthi Q Methodology 747 statements could be constructed that captures dimensions that are fundamental in understanding managerial philosophies (Wong, 1973, p. 35). These include 1. Degree to which individuals perceive that people are trustworthy or untrustworthy 2. Degree to which individuals believe that people are altruistic or selfish 3. Degree to which individuals believe that people are independent and self reliant or, alternatively, dependent and conformist 4. Degree to which individuals believe that people are simple versus highly complex The two dimensions, trust (1 and 2) and control (3 and 4), suggest something about the kinds of management methods, personnel policies, and work procedures to which an individual would respond or elect to use as a manager (Carnevale, 1995). With a good understanding of one’s own perceptions and the perceptions of others who work in the organization, a manager could shape agency practices and culture to maximize performance by facilitating views congruent with the mission and goals of the agency. It is worth noting in passing that many of the tasks that often fall under the management heading have begun to receive independent attention by scholars employing Q methodology, such as conflict management (Dayton, 2003; Wood, 2004), informatics (Gottschalk, 2002; Kendall and Kendall, 1993; Kim, 2002; Morgado et al., 1999; Thomas and Watson, 2001; Valenta and Wigger, 1997), operations (Nelson et al., 2003; Wright and Meehling, 2002), quality assessment (McKeown et al., 1999; Wright et al., 1998), risk management (Karasz, 2006; McKeown et al., 1999; Simmons and Walker, 1999), and strategic planning, which is addressed in more detail below. 37.3.1.4 Potential Topics for Further Research Beyond the use that has already been made of Q methodology to research general public administration topics, it could be employed to study a myriad of other public administration issues. For example, a Q study of bureaucratic responsibility would provide a new approach for examining an enduring and important issue. Most of the empirical research on bureaucratic responsibility has relied primarily on R methodology and has frequently concentrated on a single method of ensuring bureaucratic responsibility. However, with Q methodology, different conceptions of bureaucratic responsibility could be considered and evaluated simultaneously. In this regard, a useful framework for such a study would be Gilbert’s (1959) two-by-two typology of administrative responsibility. The concourse of statements would represent the four categories that emerge from dividing the horizontal axis into internal and external categories of responsibility and splitting the vertical axis into formal and informal means of control. The researcher could specify different conditions for sorting the statements. For example, subjects could be asked to order the statements to reflect to whom the administrator feels most responsible or, alternatively, to indicate to whom the subject perceives that he or she should be responsible. Moreover, bureaucrats’ perceptions of responsibility could be compared to other important stakeholders, such as elected officials and the public, to identify the extent to which views of bureaucratic responsibility converge among these groups. As suggested by research described above, Q methodology has been utilized in the study of administrative roles (e.g., Bidwell, 1957; Van Dusen and Rector, 1963) and could be applied to study other frameworks. For example, a Q study could be based on competing values model of managerial roles of Faerman et al. (1990), which has been used extensively to study public management (see, for example, Giek and Lees, 1993). So far, the empirical research using this model has relied on extensive closed-ended questionnaires that address numerous managerial tasks (Ban, 1995). An alternative approach would be to use Q methodology to allow public managers to define operantly and formally model their perspectives towards the managerial roles set forth in the Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 748 28.8.2007 8:51pm Compositor Name: deshanthi 748 Handbook of Research Methods in Public Administration competing values framework. The competing values model is based on a two-dimensional scheme: the horizontal axis is split into two categories of focus, internal versus external, and the vertical axis is divided into two managerial approaches, control versus flexibility. The quadrants that emerge reflect existing models from the classic management literature: (A) human relations model, (B) open systems model, (C) internal process model, and (D) rational goal model. Each quadrant or model includes two specific roles. Hence, A is identified with (a) facilitator and (b) mentor; B with (c) innovator and (d) broker; C with (e) monitor and (f ) coordinator; and D with (g) director and (h) producer. Administrators would be asked to weight the alternative roles and to sort the statements to reflect their role orientations. 37.3.2 PUBLIC ADMINISTRATION FIELDS, INCLUDING PERSONNEL ADMINISTRATION Q methodology has been employed not only to research general issues of public administration and public management, but also to investigate issues in different fields within the discipline, including public personnel administration, budgeting, and evaluation. 37.3.2.1 Researching Public Personnel Administration Just as Q methodology has provided insights into how public managers view their work, it can also help to identify the personnel-related concerns—such as job satisfaction and motivation—of public sector employees. In fact, some studies have used Q methodology for this purpose (e.g., Chinnis et al., 2001; Gillard et al., 2005). As an illustration, Sylvia and Sylvia (1986) studied job satisfaction and work motivation by collecting 43 Q sorts on these topics from a randomly selected sample of midlevel rehabilitation managers. Analyzing and interpreting the Q sorts, they identified factors representing different viewpoints. The first, ‘‘the positive concerned supervisor,’’ identified managers as being concerned for subordinates, with achievement, recognition, and work as sources of satisfaction. The second characterized job satisfaction as stemming from a positive attitude toward advancement, work, and coworkers. Individuals loading on the third factor experienced feelings of job satisfaction for a number of the same reasons as found in the first two factors, as well as from the freedom they were granted to try new ideas and programs. Job satisfaction is widely studied using R methodology and, as illustrated above, some effort has been made to use Q to study job satisfaction. Despite this research, no coherent framework of factors that determine job satisfaction has surfaced. According, to Rainey (1991, p. 146), this absence of a coherent framework is not surprising ‘‘because it is obviously unrealistic to try to generalize about how much any single factor affects satisfaction.’’ Nevertheless, progress might be made toward identifying the relative importance of different factors for job satisfaction through the application of Q methodology, as in the study by Shah (1982), who showed that some administrators’ rewards were intrinsic to the job whereas others’ were extrinsic. Studies such as this could provide insights into the facets of the job such as supervision, pay, and promotion that contribute to an individual’s job satisfaction, which could then be used to shape agency practices, training, and development. Another personnel-related area of research that has been studied via Q technique is work motivation. Gaines et al. (1984) explored the relationship between perceptions of promotion and motivation in two Connecticut police departments. Specifically, they investigated the need structures of police officers and the extent to which promotion fulfilled those needs. Applications of Q methodology in the study of work motivation could provide insights into what needs, motives, and values are most important to public sector employees. Exemplary in this regard is the study by Davis (1997), which examined the value structures operable among managers in a Canadian agency involved in research and development funding. Further research of this kind could draw from a number of existing typologies such as Murray’s (1938) List of Basic Needs, Maslow’s (1954) need hierarchy, Alderfer’s (1972) ERG model, Lasswell’s (1963) value categories, and Rokeach’s (1973) value survey. Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 749 28.8.2007 8:51pm Compositor Name: deshanthi Q Methodology 749 In addition, Q could be used to study the types of incentives that induce public sector employees to contribute positively to their agency. The following frameworks are suitable for studying this using Q methodology: Herzberg et al. (1957), Locke (1968), and Lawler (1973). Another potential application of Q would involve identifying methods and techniques that motivate employees. A concourse could be constructed to represent various methods and techniques employed to facilitate high performance in public organizations, such as performance appraisals, merit pay, pay for performance, bonuses, job redesign, job rotation, flex time, and quality circles. The completed sorts would show how employees would view and value alternative efforts to improve organizational performance. Q would also be an appropriate technique to use to develop a self-assessment performance tool. Employees would be asked to sort a group of statements pertaining to their performance and the resulting Q sorts would represent the employees’ own constructions of their performance strengths and weaknesses. As part of a performance appraisal system, Q could facilitate and structure feedback discussions and suggest employee skills and knowledge that need further development. Also, as suggested by Chatman (1989), Q could be used to assess the extent to which an individual ‘‘fits’’ into a specific public agency setting or job. 37.3.2.2 Q Research in Other Fields of Public Administration Q methodology could help in the budgeting process to identify spending priorities of decision makers, stakeholders, and others whose opinion is important. However, researchers have made limited use of Q for studies of budgeting. The best example is research by Cooper and Dantico (1990), who used Q methodology to determine if groups of employees in an organization had similar or different perspectives on budget expenditure or taxing priorities. In another brief study, Dick and Edelman (1993) demonstrate how Q technique was used as a way to prioritize journal subscriptions fated for cancellation in light of budget reductions. As noted previously, Q methodology can also be a valuable tool to aid the process of strategic planning; that is, in exploring different understandings of the organization’s past, its present problems, and its paths to the future. For example, Fairweather and Swaffield (1996) employed Q methodology to identify preferences among stakeholders for land use in the Waitaki Basin in New Zealand, and Popovich and Popovich (2002) described how Q methodology assisted the strategic planning process of a hospital. Users of Q methodology have themselves recently engaged in a selfstudy appraisal designed to clarify past achievements and future threats and opportunities (Hurd and Brown, 2004, 2005). In using Q methodology to assist with strategic planning, Gargan and Brown (1993) found that Q methodology facilitates clarification of ‘‘the perspectives of decision makers’’ and, in conjunction with other procedures, can ferret out ‘‘prudent courses of action’’ (p. 348). According to them, Q methodology helps overcome the limitations of the mind in dealing with complexity and serves to locate elements of consensus (if they exist) that might otherwise go unnoticed in the emotional turmoil of political debate. Beyond budgeting and planning, Q methodology has strong potential to be a valuable tool for policy evaluation, both to identify better how different participants and stakeholders perceive the impacts of a program and to measure the effects of a program on individuals. The first use is exemplified by Kim (1991), who studied the design and implementation of community development block grants (CDBG). Kim asked citizens, General Accountability Office (GAO) administrators, and other employees of federal, state, and local agencies to provide their viewpoints through Q sorts. Other evaluations of this type have been carried out by Corr, Phillips, and Capdevila, (2003), Fang (2006), Garrard and Hausman (1986), Meo et al. (2002), and Oring and Plihal (1993). Examples of the second use—to determine how a program affected individual attitudes—are found in Pelletier et al. (1999) and Nobbie (2003). Both measured the impacts of programs by administering pre- and postintervention Q sorts to persons in the target audience. Pelletier et al. (1999) Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 750 28.8.2007 8:51pm Compositor Name: deshanthi 750 Handbook of Research Methods in Public Administration examined the effects of two-and-a-half day participatory planning events in six rural counties in northern New York. The researchers were interested in finding out whether participant views of the local food system—the topic of the events—would change following attendance at the events. They found that participation in a participatory planning event may cause some participants to alter their viewpoints in ways that appear contrary to their values and interests as expressed prior to the deliberative event. Nobbie (2003) measured the effects of a program by the Georgia Development Disabilities Council (GDDC) to ‘‘empower people with disabilities and their families to obtain appropriate services, develop leadership potential and influence policy change in this arena.’’ As part of her evaluation, she administered Q sorts to the participants in this program at the beginning of the program and after the eight-month program was completed. The Q sort measured perspectives on advocacy, citizenship, and empowerment. Both the pre- and postprogram sorts were completed by 21 of 34 participants. Nobbie’s analysis of the completed Q sorts showed three Q factors, which were labeled as Beginning=Dependent Advocate, Excluded Advocate, and Unfolding Advocate, the latter most resembling the type of advocate envisioned by the GDDC. Nobbie found that seven participants changed their perspectives after participating in the program. They loaded on the first two factors in the preprogram sort, but were in the third factor in the postprogram sort. Another ten participants had mixed factor loadings in the first sort, but were solidly part of the third factor in the second sort. These changes indicate that the program helped change attitudes of most of the program participants in the direction desired by the GDDC. Combes et al. (2004) provide another illustration of persons with disabilities being involved, via Q methodology, in the evaluation of planning aimed at this particular stakeholder group. 37.3.3 RESEARCHING DECISION MAKING AND PUBLIC POLICY The use of Q methodology in the study of decision making traces back to an early paper by Stephenson (1963), which was given scant attention, but in the past decade, an increasing number of policy analysts and policy researchers have turned to Q methodology for leverage in understanding decisions within their own policy domains (see Durning and Brown, 2007, for a detailed overview). The various uses of Q methodology can be divided into four major categories, although individual projects may well fit into more than one of the categories. These categories are the use of Q methodology (1) to research influences on decisions that were made in the past, (2) to understand better the perspectives of stakeholders and decision makers on decisions that will be made in the future, (3) to provide a mechanism for marginalized or powerless groups to make their views known, and (4) to facilitate the search for compromise solutions to difficult policy issues. 37.3.3.1 Understanding Influences on Decision Makers Some researchers have turned to Q methodology to help understand why certain decisions were made by identifying the ‘‘decision structures’’ of the people who made them. Decision structures—a term borrowed from Harold Lasswell (1963) and incorporated into Q methodology by Stephenson (1987)—are the configuration of values, beliefs, interests, and information that influence the position taken by a decision maker (see Durning and Brown, 2007). Examples of Q-methodology studies designed for this purpose have been conducted by Donahue (2004), who identified the factors affecting the decisions of fire chiefs about the level of fire protection their local fire departments would provide; by Van Exel et al. (2004) and Steg et al. (2001), who examined motivations behind automobile use and other modes of transportation; and by Webler et al. (2001), who explored the question of what factors influenced New England local government officials to participate or refuse to participate in regional collaborative environmental policy making on watershed management planning (see also Webler and Tuler, 2001; Webler et al., 2003). Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 751 28.8.2007 8:51pm Compositor Name: deshanthi Q Methodology 751 To carry out their research, Webler et al. (2001) created a sample of 52 statements based on their interviews with 45 local officials on the topic of why they did or did not participate in a watershed management planning process. This Q sort was administered to 39 local government officials in three states (New Hampshire, Massachusetts, and Connecticut) in which these planning processes were active. The authors identified five factors—which they called ‘‘coherent narratives’’—that reflected the perspectives of the local officials toward the opportunity to participate in a watershed management planning process. The authors described these narratives (or decision structures) as follows: ‘‘One centers on strategic calculations of influencing outcomes. A second weighs interest and available time. A third looks at how the community would benefit. A fourth is rooted in one’s personal environmental ethics. And a fifth attempts to match skills and experiences with the needs of the policy endeavor’’ (p. 105). The authors found that local government officials made their decision to participate or not based on three general considerations: ‘‘They feel they can help make a positive difference; they see working on the problem as consistent with their environmental ethic; and it is in their community’s interest that they participate in the process’’ (p. 118). Related to these studies of specific decisions, another Q study explored the attitude climate among legislators and administrators on issues of policy making and policy implementation: Cunningham and Olshfski (1986) focused their investigation on the context of policy decisions rather than on specific decisions. Looked at the other way around, Bedrosian (2004) showed how the decision-making styles of preschool teachers impacted the teaching environment. 37.3.3.2 Identifying the Perspectives (Decision Structures) of Stakeholders Policy decisions are made within specific contexts; that is, every policy issue is unique in that it is located within a particular policy domain and involves a set of decision makers and stakeholders whose interests and beliefs are engaged. Over time, controversial issues generate narratives that link the issue and proposed solutions to symbols, history, and imagined futures. Q methodology can provide insights into these policy domains and narratives by identifying the competing decision structures and narratives of stakeholders on particular issues, showing the values and beliefs underlying their positions. In recent years, many researchers have employed Q methodology to clarify competing stakeholder perspectives on particular policy issues or broader issues areas. These studies typically seek to identify and understand the different decision structures (values, beliefs, interests, and knowledge) or narratives that underlie the positions of decision makers, stakeholders, and the public on issues to be decided. Studies of this type have addressed many different issues, including forests (Clarke, 2002; Steelman and Maguire, 1999), coastal zone management (Shilin et al., 2003), nature restoration and sustainability (Addams and Proops, 2000; Barry and Proops, 1999, 2000; Colorado Institute of Public Policy, 2006; Meo et al., 2002; Peritore, 1999; Walton, 2006; Wooley and McGinnis, 2000; Wooley et al., 2000), conservation biology (Byrd, 2002; Mattson et al., 2006; Rutherford and Gibeau, 2006), biotechnology in Mexico (Galve-Peritore and Peritore, 1995), agricultural practices (Brodt et al., 2004; Kramer et al., 2003; Wilkins et al., 2001), health-related issues (Baker, 2006; Baker et al., 2006; Barbosa et al., 1998; Mrtek et al., 1996; Stainton Rogers, 1991; Valenta and Wigger, 1997) land use (Coke and Brown, 1976), city-county consolidation (Durning and Edwards, 1992), institutional development (Lindsey and Rafferty, 2006; Ramlo, 2005), corrections policy (Baker and Meyer, 2002, 2003), school violence (Greene, 1999; Wester and Trepal, 2004), the redevelopment of brown fields (Deitrick, 1998), and the impact of diversity on New Zealand’s government (Wolf, 2004). A good example of these types of studies is Hooker’s (2001) research to understand the competing concepts of preservation, conservation, and development that influence individual perspectives on forest policy. Specifically, she was interested in how different stakeholders perceive the relationship of individuals to society and to forests. In her study, Hooker selected 60 statements from approximately 400 taken from literature on forest and natural resource policy, and administered Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 752 28.8.2007 8:51pm Compositor Name: deshanthi 752 Handbook of Research Methods in Public Administration these in a Q sort to a diverse group of 189 participants, including forest landowners, government officials, forest industry representatives, trade association representatives, scientists, leaders of conservation groups, academics, and public interest group representatives. From her analysis of the completed sorts, Hooker identified five factors (labeled New Stewards, New Conservationists, Individualists, Traditional Stewards, and Environmental Activists), each representing a different perspective or decision structure influencing the positions that people take on forestry issues, and each explained in terms of different, but sometimes overlapping, views on policies regarding the use of forests. Hooker suggested that knowledge of these five different perspectives, especially information on the views they have in common, would be a good starting point for structuring beneficial interactions among factors. She wrote, ‘‘Conversations among analysts and members of the public who are interested in forest policy can use the new framework of beliefs identified in this study to redefine a policy agenda as well as commence facilitating dialogue’’ (p. 174). She also suggested that the results of her study could be used to assist an effort to ‘‘structure a more effective public involvement strategy’’ (p. 174), and argued that citizen participation should be set up so that all of the four perspectives were represented in the discussions. By including people with the four main perspectives in public hearings and advisory groups, policy makers could make sure that all of the competing views are heard. 37.3.3.3 Studying the Marginalized and Less Powerful, and Giving them Voice Many advocates of Q methodology suggest that it is an excellent research tool to identify the attitudes and perspectives of underrepresented, ignored, marginalized, or less powerful groups on issues that are important to them. Some argue that Q methodology also has a practical value in this regard: it can allow these groups to have a voice—to have their opinions heard—when decisions are being made or their interests are at stake. If used for this purpose, Q methodology can make the decision-making process more democratic by providing a structure to enable the widest range of attitudes and opinions to be heard. According to Brown (2005, 2006a), Q methodology is perfectly suited for clarifying the perspectives of the marginalized on issues of interest to them, and he identifies different research that has addressed issues of concern to the less powerful—among them, race (Abrams, 2002; Benoit, 2001; Carlson and Trichtinger, 2001; Smith, 2002; Stowell-Smith and McKeown, 1999), gender (e.g., Anderson, 2004; Beaver, 2002; Gallivan, 1994, 1999; Jacobson and Aaltio-Marjosola, 2001; Kitzinger, 1999; Oswald and Harvey, 2003; Snelling, 1999; Wrigley, 2006), the disabled (Combes et al., 2004; Goodman et al., 2002), patients, and clients who pose special difficulties (Jones et al., 2003; Lister and Gardner, 2006), farm workers (Kramer et al., 2003; Warnaars and Pradel, 2007; Züger Cáceres, 2003), and the poor (Brown, 2005). Peace et al. (2004) turned to Q methodology to investigate the views of two large client groups, the recipients of Sickness Benefits and Invalid Benefits (SB=IB) in New Zealand, about ‘‘well being, employment, and interdependence.’’ They administered Q sorts to 20 clients, and their analysis of the sorts identified five different factors, showing that ‘‘the client community holds a range of different views about what constitutes well-being in the context of independence and employment.’’ In related research, Combes et al. (2004) employed Q methodology to involve people with intellectual disabilities in the evaluation of a planning process. Both of these studies documented innovative uses of Q methodology to obtain the opinions and perspectives of a less powerful group of people. Donner (2001) has provided guidelines for using Q methodology as a ‘‘participatory exercise’’ in international development to help clarify views concerning proposed local development projects, and suggests Q methodology as a way to involve a wide range of people and groups in a process to obtain consensus on local development investments. In a worked application, he demonstrates the variety of Q factors that exist among adopters of mobile phones among Rwandan businessmen (Donner, 2005). Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 753 28.8.2007 8:51pm Compositor Name: deshanthi Q Methodology 37.3.3.4 753 Facilitating the Search for Compromise Solutions to Difficult Problems Q methodology is being used by many policy analysts and researchers to do more than attempt to understand why decisions were made or to identify different stakeholder positions. They are using information about stakeholder positions (decision structures) to assist the search for compromise solutions to policy issues. Q methodology can aid the search for compromises because it not only identifies different stakeholder perspectives, but it also shows those statements upon which many or most of the different factors agree. These statements can be the focus of negotiations concerning which actions could be taken that would gather enough support to be adopted. Good examples of this type of study are Maxwell’s (2000) use of Q methodology to solve a conflict within a county government and two studies by Focht (2002) and Maxwell and Brown (1999). Focht (2002) used Q methodology to identify different perceptions of water management problems in the Illinois River Watershed (Oklahoma) and their proposed solutions. He administered two Q sorts to 120 stakeholders representing the diversity of interests at issue, the first gauging the perceptions of the current problems created by the impacts of different activities (e.g., agriculture, tourism) on the river, and the second addressing viewpoints toward different ways to manage negative impacts on the river by different activities. From his analysis of these sorts, Focht discerned major differences in views, but also some areas of agreement that, he concluded, could help formulate a strategy for the management of conflict when writing a policy to regulate use of the river. Based on common views across factors, Focht wrote that for a management plan to have broad support, it should ‘‘(1) initially [focus] on the illegal dumping of trash, (2) [commission] unbiased scientific studies of the magnitude and sources of threats to water quality, and (3) [examine] how potential impact management alternatives will impact economic development, land use rights, and traditional lifestyles and freedoms’’ (p. 1134). Maxwell and Brown (1999) described the use of Q methodology as a technique that consultants can use to help solve difficult disputes within organizations. Their case concerned a middle school in which faculty members disagreed about how best to deal with increasing levels of student misconduct. The consultants were brought in to help the school manage the conflict and to find solutions that would be widely accepted. They began by conducting a study to determine how members of the organization understood the problem of student misconduct. They interviewed the teachers, staff members, and administrators in the middle school, and from those interviews they compiled a list of 44 problems. Using these as a Q sort, they asked teachers and staff members to sort them according to their importance (i.e., most important problems to least important). The factor analysis of the sorts showed that most participants loaded on one of two factors, which the consultants labeled as (1) the resentment factor and (2) the differentiating factor. The first factor strongly identified with teachers and staff and complained about students, parents, administrators, and the school board, all of whom were viewed as ‘‘placing them in an untenable position’’ (p. 38). The second factor was more concerned with intragroup relations, and differentiated among students (those needing support as well as those needing discipline) and among teachers (those who punish effectively and those who do not). The consultants presented the results of this study to the participants, showing them the statements that represented the sources of disagreement about the issue, but also pointing out the statements on which they agreed. They likened this process to ‘‘holding a mirror up to the teachers and staff . . . so that they might see themselves and co-workers more clearly’’ (p. 40). In the second part of the study, the same participants were interviewed to elicit their proposals about how to address the problems that had been identified. The faculty produced a list of 35 potential solutions, which became the Q sort that was completed by 28 faculty and staff members. Analysis revealed three different perspectives, which were described as (1) punishment (solve the discipline problem through harsher punishment), (2) quarantine (use special programs to segregate problem children from others), and (3) coordination (get teachers and staff to work together more effectively through cooperation and coordination). Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 754 28.8.2007 8:51pm Compositor Name: deshanthi 754 Handbook of Research Methods in Public Administration In addition to these competing approaches to solving the problem, the Q sorts identified actions that all three groups agreed should be implemented or should be avoided: (a) Establish a parental contract regarding the rules of conduct and the consequences for misconduct that would apply to their child and (b) consistently follow rules and regulations already in existence, such as the Student Conduct Code. Based on their findings, the consultants informed school administrators about the differences of opinion concerning the causes of student misconduct and the differing preferences for actions to address the problems, and identified the actions that were agreeable to all three factors as well as options that were unanimously opposed. Other studies report a similar approach to identifying competing problem definitions and solutions for other policy issues. For example the study by Mattson et al. (2006), which is presented in detail above (see Tables 37.1 and 37.6), describes a two-day workshop that had as its goal ‘‘to gain insight into perspectives and narratives of participants in large carnivore conservation, and to identify areas of potential common ground’’ (p. 393). And Brown et al. (2004) describe a similar workshop designed to identify challenges and potential solutions for efforts to plan for conservation from Yellowstone to the Yukon (Y2Y). At both workshops, stakeholders were asked what problems they perceived, and the list of problems generated was used as a Q sort. After the different perspectives of the problems were identified, the stakeholders were asked how they would address these problems and these potential solutions made up a second Q sort, which was used to identify different perspectives on solutions. From these sorts, it was possible to identify both differences in perspectives and agreements across perspectives on specific statements concerning problems and solutions. These mutually accepted statements can be used to start the effort to find agreements on what should be done concerning the issues being considered. Two other researchers have demonstrated that Q methodology can be valuable in helping to find solutions to ‘‘intractable’’ policy problems. Dayton (2000) proposed that Q methodology be used to identify the competing frames (models, metaphors, understandings) of elites on major controversial issues. According to Dayton, by understanding the competing frames, it is possible to gain insight into the origins of policy conflicts, the beliefs that underlie the intractability of the issue, and possible ways to build a consensus to address the issue. He used the issue of global warming as an example of how Q methodology could identify frames and focus dialogue on areas of agreement. Van Eeten (2001) demonstrates the utility of Q methodology in helping to ‘‘recast intractable problems’’ so as to make them more amenable to solution, using as a case study efforts to determine the future policy for expanding Amsterdam’s Schipol Airport. A controversial expansion of the airport (building a fifth runway) had been approved in 1995, and future expansion was due to come up again soon for consideration. Through a Q-methodology study, Van Eeten identified five different perspectives (which he called ‘‘policy arguments’’) of the relevant stakeholders on the proposed expansion and the problem that it would address. He concluded that only two of the five factors were bipolar—these represented the public arguments that were magnified by media accounts—but that the others had overlapping elements. Van Eeten suggested that the search for policy should de-emphasize the two bipolar factors, which had made the issue intractable, and to give more attention to elements of the nonpolar factors, which defined the problem in tractable ways so that a compromise could be found. 37.4 CONCLUSION In this chapter, we have discussed the motivations for using Q methodology and described in some detail how to carry out Q studies. We have suggested that some researchers and practitioners will use Q methodology pragmatically to answer important research and practical public administration questions from a perspective that differs from the usual R method approach. Other researchers and practitioners may turn to Q methodology in reaction to the shortcomings of R methods, which are founded on a barren positivism, an epistemology increasingly challenged by theorists in all social science disciplines (e.g., Morçöl, 2002). Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 755 28.8.2007 8:51pm Compositor Name: deshanthi Q Methodology 755 Whatever the motivation for using Q methodology, public administration researchers and practitioners will find that this method can be valuable for their work. As we have described in the chapter, researchers have used Q method to investigate important issues in general public administration, public personnel administration, public management, and public policy, and this methodology is well suited for exploration of other key issues in these research areas. Q methodology can also be a valuable tool for public managers and policy analysts to identify and understand conflicting values, preferences, and opinions concerning organizational and policy issues. Also it has been used, and should be used further, in policy evaluations. In addition, it can contribute to the democratization of management and policy making by allowing the voices of stakeholders and the interested public to be more fully articulated and understood. A researcher or practitioner who wishes to conduct a Q-methodology study can do so by following the general procedures described in the case study described in Section 37.2 of this chapter, and many questions that might arise in such a study are addressed in the ‘‘frequently asked questions about Q methodology’’ contained in the accompanying Teachers Guide. We believe that the value of Q methodology, both as a pragmatic tool and as a nonpositivist or postpositivist research method, is more than sufficient to reward public administration researchers and practitioners fully for their efforts to master it. ENDNOTES 1. The procedures for carrying out Q method research are discussed in more detail in Section 37.2. The most useful and complete methodological guides are Brown (1980, 1993) and McKeown and Thomas (1988). After mastering the basics of Q methodology, researchers may want to read Stephenson (1953), The Study of Behavior, which laid the foundation for Q methodology as the science of subjectivity (Brown, 1994–1995). 2. See Dryzek (1990), chapters 8 and 9, for an in depth comparison of the use of survey research (an R methodology) and Q methodology to investigate public opinion. 3. We should note that sensible positivists have been aware of the limits of science. For example, Stephenson, the creator of Q methodology and a physicist as well as a psychologist, regarded himself as a positivist, yet would not have taken the extreme position attributed to positivists by some postmodernists. Stephenson’s view of science was not defended on the basis of objectivity, but of detachment; that is, trying to establish conditions under which the observer might have reason to believe that what was being observed was independent of the self and its desires. Ironically, it is precisely in terms of detachment that R methodology falters: the measuring device (e.g., an autocratic management scale) carries the observer’s undetached meaning along with it; in Q, on the other hand, the respondent’s meaning is permitted to reign, as detached from whatever meaning the investigator might have given to the same statements. It is Q, not R, that comes closest to achieving detachment (as opposed to ‘‘objectivity’’). On the relationship of Q methodology to postpositivist assumptions, consult Durning (1999). 4. Statements about the two main concepts were collected for the following nine issues: source of authority, dominant values in work, decision making patterns, recruitment, placement, transfer and promotion, superior-subordinate relationships, work performance, accountability, and group orientation. REFERENCES Abrams, M.E. 2002. Like a tree planted by the water: Black journalism students face the job market, Negro Educational Review, 53: 13–29. Adcock, C.J. 1954. Factorial Analysis for Non-Mathematicians, Melbourne University Press, Melbourne. Addams, H. and Proops, J. (Eds.) 2000. Social Discourse and Environmental Policy: An Application of Q Methodology, Edward Elgar, Cheltenham, United Kingdom. Alderfer, C.P. 1972. Existence, Relatedness, and Growth: Human Needs in Organizational Settings, Free Press, New York. Anderson, D.R. 2004. The Application of Q Methodology to Understanding the Experiences of Female Executives in Biotechnology Companies in Massachusetts: A Contextual Perspective. Unpublished Ph.D. dissertation, Regent University. Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 756 28.8.2007 8:51pm Compositor Name: deshanthi 756 Handbook of Research Methods in Public Administration Baker, R. and Meyer, F. 2002. Women and support for correctional reform. Paper presented at the meeting of the American Political Science Association, Boston, Massachusetts. Baker, R. and Meyer, F. 2003. A community economic elite and support for correctional reforms. Paper presented at the meeting of the Midwest Political Science Association, Chicago, Illinois. Baker, R.M. 2006. Economic rationality and health and lifestyle choices for people with diabetes, Social Science & Medicine, 63: 2341–2353. Baker, R.M., Thompson, C., and Mannion, R. 2006. Q Methodology in Health Economics, Journal of Health Services Research & Policy, 11: 38–45. Ban, C. 1995. How Do Public Managers Manage? Bureaucratic Constraints, Organizational Culture, and the Potential for Reform, Jossey Bass, San Francisco, California. Barbosa, J.C., Willoughby, P., Rosenberg, C.A., and Mrtek, R.G. 1998. Statistical methodology: VII. Q-methodology, a structural analytic approach to medical subjectivity, Academic Emergency Medicine, 5: 1032–1040. Barry, J. and Proops, J. 1999. Seeking sustainability discourses with Q methodology, Ecological Economics, 28: 337–345. Barry, J. and Proops, J. 2000. Citizenship, Sustainability, and Environmental Research: Q Methodology and Local Exchange Trading Systems, Edward Elgar, Cheltenham, United Kingdom. Beaver, A.K. 2002. Women leaders alter medical training and practice, Women in Higher Education, 11(2): 32–33. Bedrosian, W.K. 2004. Teacher Decision Making in the Stage Manager Role: A Q Method Study. Unpublished Ph.D. dissertation, Kent State University. Benoit, M.A. 2001. I want to participate in the go girls program because . . . using Q-methodology to assess overweight African-American girls’ motivation to participate in a physical activity and nutrition education program. Unpublished M.A. paper, Emory University. Bidwell, C.E. 1957. Some effects of administrative behavior: A study in role theory, Administrative Science Quarterly, 2: 162–181. Brewer, G.A., Selden, S.C., and Facer, R.L. 2000. Individual conceptions of public service motivation, Public Administration Review, 60: 254–264. Brodt, S., Klonsky, K., Tourte, L., Duncan, R., Hendricks, L., Ohmart, C., and Verdegaal, P. 2004. Influence of farm management style on adoption of biologically integrated farming practices in California, Renewable Agriculture and Food Systems, 19: 237–247. Brown, S.R. 1980. Political Subjectivity: Applications of Q Methodology in Political Science, Yale University Press, New Haven, Connecticut. Brown, S.R. 1985. Comments on ‘‘The search for structure,’’ Political Methodology, 11: 109–117. Brown, S.R. 1992–1993. On validity and replicability, Operant Subjectivity, 16: 45–51. 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. 2002. Structural and functional information, Policy Sciences, 35: 285–304. Brown, S.R. 2005. Applying Q methodology to empowerment, Measuring Empowerment: Cross-Disciplinary Perspectives, D. Narayan (Ed.), The World Bank, Washington, DC, pp. 197–215. Brown, S.R. 2006a. A match made in heaven: A marginalized methodology for studying the marginalized, Quality and Quantity, 40: 361–382. Brown, S.R. 2006b. Q methodology and naturalistic subjectivity, Modern Perspectives on J.R. Kantor and Interbehaviorism, B.D. Midgley and E.K. Morris (Eds.), Context Press, Reno, Nevada, pp. 271–276. Brown, S.R. (in press). Q methodology. In L.M. Given (Ed.), The SAGE Encyclopedia of Qualitative Research Methods. Thousand Oaks, California: Sage. Brown, S.R., Byrd, K.L., Clark, T.W., Rutherford, M.B., Mattson, D.J., and Robinson, B. 2004. Challenges and strategies in large-scale conservation: The case of the Yellowstone-to-Yukon conservation initiative. Paper presented at the meeting of the Policy Sciences Annual Institute, Yale University School of Law, New Haven, Connecticut. Brown, S.R. and Robyn, R. 2004. Reserving a key place for reality: Philosophical foundations of theoretical rotation, Operant Subjectivity, 27: 104–124. Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 757 28.8.2007 8:51pm Compositor Name: deshanthi Q Methodology 757 Burt, C. and Stephenson, W. 1939. Alternative views on correlations between persons, Psychometrika, 4: 269–281. Burt, C. 1940. The Factors of the Mind, University of London Press, London. Byrd, K. 2002. Mirrors and metaphors: Contemporary narratives of the wolf in Minnesota, Ethics, Place, and Environment, 5: 50–65. Carlson, J.M. and Trichtinger, R. 2001. Perspectives on entertainment television’s portrayal of a racial incident: An intensive analysis, Communication Review, 4: 253–278. Carnevale, D.G. 1995. Trustworthy Government, Jossey Bass, San Francisco, California. Chartier, R. 1995. Forms and Meanings, University of Pennsylvania Press, Philadelphia, Pennsylvania. Chatman, J.A. 1989. Improving interactional organizational research: A model of person-organization fit, Academy of Management Review, 14: 333–349. Chinnis, A.S., Summers, D.E., Doerr, C., Paulson, D.J., and Davis, S.M. 2001. Q methodology: A new way of assessing employee satisfaction, Journal of Nursing Administration, 1: 252–259. Clarke, A.H. 2002. Understanding sustainable development in the context of other emergent environmental perspectives, Policy Sciences, 35: 69–90. Coke, J.G. and Brown, S.R. 1976. Public attitudes about land use policy and their impact on state policy makers, Publius, 6: 97–134. Colorado Institute of Public Policy 2006. Water in 2025: Beliefs and values as a means for cooperation, Colorado Institute of Public Policy, Fort Collins, Colorado. Available http:==www.cipp.colostate. edu=pdf=CIPP_Water_2025_www_Final2.pdf Combes, H., Hardy, G., and Buchan, L. 2004. Using Q-methodology to involve people with intellectual disability in evaluating person-centred planning, Journal of Applied Research in Intellectual Disabilities, 17: 149–159. Cooper, D.R. and Dantico, M.K. 1990. Priority setting techniques for the budgetary process: An experimental comparison to determine what works, Experimental Study of Politics, 9(1): 94–117. Corr, S., Phillips, C.J., and Capdevila, R. 2003. Using Q methodology to evaluate a day service for younger adult stroke survivors, Operant Subjectivity, 27: 1–23. Cunningham, R. and Olshfski, D. 1986. Interpreting state administrator-legislator relationships, Western Political Science Quarterly, 39: 104–117. Davis, C.H. 1997. Value structures in international development research management: The case of a Canadian R&D funding agency, Science and Technology in a Developing World, T. Shinn, J. Spaapen, and V. Krishna (Eds.), Kluwer Academic, Dordrecht, The Netherlands, pp. 263–296. Dayton, B.C. 2000. Policy frames, policy making and the global climate change discourse, Social Discourse and Environmental Policy, H. Addams and J. Proops (Eds.), Edward Elgar, Cheltenham, United Kingdom and Northampton, Massachusetts, pp. 71–99. Dayton, B.C. 2003. Policy dialogues under conditions of uncertainty: Using Q methodology as an aid to dispute resolution. Paper presented at the meeting of the Western Political Science Association, Long Beach, California. DeCourville, N.H. and Hafer, C.L. 2001. Attitudes toward affirmative action programs: A Q methodological study, Operant Subjectivity, 24: 183–200. Deitrick, S. 1998. Examining community perceptions of brownfields revitalization in Pittsburgh, Pennsylvania. Paper presented at the meeting of the Association of Collegiate Schools of Planning, Pasadena, California. Dick, M.J. and Edelman, M. 1993. Consequences of the budget crunch: Using Q-methodology to prioritize subscription cancellations, Journal of Nursing Education, 32: 181–182. Donahue, A.K. 2004. Managerial perceptions and the production of fire protection, Administration & Society, 35: 717–746. Donner, J.C. 2001. Using Q-sorts in participatory processes: An introduction to the methodology, Social Analysis: Selected Tools and Techniques (Social Development Papers 36), R.A. Kreuger, M.A. Casey, J. Donner, S. Kirsch, and J.N. Maack (Eds.), Social Development Department, The World Bank, Washington, DC, pp. 24–29. Available http:==siteresources.worldbank.org=INTRANETSOCIAL DEVELOPMENT=873467–1111741676294=20502176=SDP-36.pdf Donner, J. 2005. Microentrepreneurs and mobiles: An exploration of the uses of mobile phones by small business owners in Rwanda, Information Technologies and International Development, 2: 1–21. Dryzek, J.S. 1990. Discursive Democracy, Cambridge University Press, Cambridge, United Kingdom. Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 758 28.8.2007 8:51pm Compositor Name: deshanthi 758 Handbook of Research Methods in Public Administration Dryzek, J.S. and Holmes, L.T. 2002. Post-Communist Democratization: Political Discourses across Thirteen Countries, Cambridge University Press, Cambridge, United Kingdom. Dunlap, M.S. and Dudley, R.L. 1965. Quasi Q-sort methodology in self evaluation of conference leadership skill, Nursing Research, 14: 119–125. Durning, D. 1999. The transition from traditional to postpositivist policy analysis: A role for Q-methodology, Journal of Policy Analysis and management, 18: 389–410. Durning, D.W. and Brown, S.R. 2007. Q methodology and decision making, Handbook of Decision Making, G. Morçöl (Ed.), CRC Press (Taylor & Francis Group), New York, pp. 537–563. Durning, D. and Edwards, D. 1992. The attitudes of consolidation elites: An empirical assessment of their views of city county mergers, Southeastern Political Review, 20: 355–383. Durning, D. and Gajdamaschko, N. 1998. Are the managerial perspectives of Ukrainian public administrators changing? An exploratory study using Q methodology, Proceedings of the Conference on Organizational Psychology and Transition Processes in Central and Eastern Europe, September 1998, Dubrovnik, Croatia. Durning, D. and Gadamaschko, N. 2000. The managerial perspectives of local public administrators in Ukraine, Proceedings, Public Administration and Local Government in Ukraine: Ways of Reforming, Uzhhorod, Ukraine, pp. 18–31. Faerman, S.R., Quinn, R.E., Thompson, M.P., and McGrath, M.R. 1990. Supervising New York State: A Framework for Excellence, Governor’s Office of Employee Relations, Albany, New York. Fairweather, J.R. and Swaffield, S. 1996. Preferences for scenarios of land-use change in the Mackenzie= Waitaki Basin, New Zealand Forestry, 41: 17–26. Fang, K.H. 2006. Taiwan’s Officials’ Perceptions of Fiscal Decentralization: An Analysis Using Q Methodology. Unpublished Ph.D. dissertation, University of Pittsburgh. Finkel, N. 1995. Commonsense Justice: Jurors’ Notions of the Law, Harvard University Press, Cambridge, Massachusetts. Focht, W. 2002. Assessment and management of policy conflict in the Illinois River watershed in Oklahoma: An application of Q methodology, International Journal of Public Administration, 25: 1311–1349. Gaines, L.K., Van Tubergen, N., and Paiva, M.A. 1984. Police officer perceptions of promotion as a source of motivation, Journal of Criminal Justice, 12: 265–275. Gallivan, J. 1994. Subjectivity and the psychology of gender: Q as a feminist methodology, Women, Girls, and Achievement, J. Gallivan, S.D. Crozier, and V.M. Lalande (Eds.), Captus University Publications, Toronto, Canada, pp. 29–36. Gallivan, J. 1999. Gender and humor: What makes a difference? North American Journal of Psychology, 1: 307–318. Galve-Peritore, A.K. and Peritore, N.P. 1995. Mexican biotechnology policy and decision makers’ attitudes toward technology policy,’’ Biotechnology in Latin America: Politics, Impacts, and Risks, N.P. Peritore and A.K. Galve-Peritore (Eds.), Scholarly Resources, Wilmington, Delaware, pp. 69–95. Gargan, J.J. and Brown, S.R. 1993. ‘What is to be done?’ Anticipating the future and mobilizing prudence, Policy Sciences, 26: 347–359. Garrard, J. and Hausman, W. 1986. The priority sort: An empirical approach to program planning and evaluation, Evaluation Studies Review Annual, 11: 279–286. Gauzente, C. 2005. La méthodologie Q et l’étude de la subjectivité (Q Methodology and the Study of Subjectivity), Management des Ressources Humaines: Méthodes de Recherche en Sciences Humaines et Sociales, P. Roussel and F. Wacheux (Eds.), De Boeck, Brussels, Belgium, pp. 177–206. Giek, D.G. and Lees, P.L. 1993. On massive change: Using the competing values framework to organize the educational efforts of the human resource function in New York state government, Human Resource Management, 32: 9–28. Gilbert, C. 1959. The framework of administrative responsibility, Journal of Politics, 21: 373–407. Gillard, L.A., Lewis, V.N., Mrtek, R., and Jarosz, C. 2005. Q methodology to measure physician satisfaction with hospital pathology laboratory services at a midwest academic health center hospital, Labmedicine, 36: 361–365. Good, J.M.M. 1998. William Stephenson and the quest for a science of subjectivity, Revista de Historia de la Psicología, 19: 431–439. Good, J.M.M. 2005a. Stephenson, William (1902–1989), Dictionary of Modern American Philosophers, J.R. Shook (Ed.), Thoemmes Press, Bristol, United Kingdom, pp. 2322–2323. Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 759 28.8.2007 8:51pm Compositor Name: deshanthi Q Methodology 759 Good, J.M.M. 2005b. Stephenson, William, Encyclopedia of Statistics in Behavioral Science, Vol. 4, B.S. Everitt and D.C. Howell (Eds.), John Wiley, Chichester, United Kingdom, pp. 1898–1900. Goodman, G., Tiene, D., and Luft, P. 2002. Adoption of assistive technology for computer access among college students with disabilities, Disability and Rehabilitation, 24: 80–92. Gottschalk, P. 2002. Key issues in IS management in Norway: An empirical study based on Q methodology, Advanced Topics in Information Resources Management (Vol. 1), M. Khosrowpour (Ed.), Idea Group, London, United Kingdom, pp. 127–140. Gough, H.G., Misiti, R., and Parisi, D. 1971. Contrasting managerial perspectives of American and Italian public administrators, Journal of Vocational Behavior, 1: 255–262. Greene, M.W. 1999. Redefining violence in Boulder Valley, Colorado, Peacebuilding for Adolescents, L.R. Forcey and I.M. Harris (Eds.), Peter Lang, New York, pp. 57–88. Hajer, M.A. and Wagenaar, H. (Eds.) 2003. Deliberative Policy Analysis: Understanding Governance in the Network Society, Cambridge University Press, Cambridge, United Kingdom. Herzberg, F., Mausner, B., Peterson, R.O., and Capwell, D.F. 1957. Job Attitudes: Review of Research and Opinion, Psychological Service of Pittsburgh, Pittsburgh, Pennsylvania. Hiruy, M. 1987. Exploring the Perspectives of Ethics: The Case of Public Administrators in the United States. Unpublished Ph.D. dissertation, Kent State University. Hooker, A.M. 2001. Beliefs regarding society and nature: A framework for listening in forest and environmental policy, Operant Subjectivity, 24: 159–182. Hurd, R.C. and Brown, S.R. 2004–2005. The future of the Q methodology movement, Operant Subjectivity, 28: 58–75. Iliescu, D. 2005. Metodologia Q, Comunicare.ro, Bucureşti, României. Irwin, A. 2006. The politics of talk: Coming to terms with the ‘new’ scientific governance, Social Studies of Science, 36: 299–320. Jacobson, S.W. and Aaltio-Marjosola, I. 2001. ‘Strong’ objectivity and the use of Q-methodology in crosscultural research: Contextualizing the experience of women managers and their scripts of career, Journal of Management Inquiry, 10: 228–248. Johnson, P.E. 1973. Functional Typologies of Administrators: A Q sort Analysis. Unpublished Ph.D. dissertation, University of Iowa. Jones, S., Guy, A., and Ormrod, J.A. 2003. A Q-methodological study of hearing voices: A preliminary exploration of voice hearers’ understanding of their experiences, Psychology and Psychotherapy, 76: 189–209. Karasz, H.N. 2006. Anti-Corporate Collectivists, Capable Individualists, and Relativists: A Q-Methodological Exploration of Audiences for Health Communication about Contaminated Soils. Unpublished Ph.D. dissertation, University of Washington. Kendall, J.E. and Kendall, K.E. 1993. Metaphors and methodologies: Living beyond the systems machine, MIS Quarterly, 17: 149–171. Kim, S.E. 1991. Factors Affecting Implementation Effectiveness: A Study of Community Development Block Grant Implementation. Unpublished Ph.D. dissertation, Kent State University. Kim, S.E. 2002. An application of Q-methodology in selecting evaluation criteria for information projects, Korean Local Government Studies, 5: 207–226. [Korean] Kitzinger, C. 1999. Commentary: Researching subjectivity and diversity: Q-methodology in feminist psychology, Psychology of Women Quarterly, 23: 267–276. Kline, P. 1994. An Easy Guide to Factor Analysis. London: Routledge. Kramer, B., de Hegedus, P., and Gravina, V. 2003. Evaluating a dairy herd improvement project in Uruguay: Testing and explaining Q methodology, Journal of International Agricultural and Extension Education, 10(2), 41–50. Kramer, B. and Gravina, V. 2004. Theoretical rotation as a tool for identifying points of leverage in people’s perspectives for program improvement, Operant Subjectivity, 27: 125–144. Lasswell, H.D. 1963. The Future of Political Science, Atherton, New York. Lawler, E.E. 1973. Motivation in Work Organizations, Brooks=Cole, Pacific Grove, California. Lewis, C.S. 1960. Studies in Words, Cambridge University Press, Cambridge, United Kingdom. Lindsey, J.L. and Rafferty, T.J. 2006. Organizational change research using the Q method: Developing a doctoral program. Paper presented at the meeting of the Midwest Educational Research Association, Columbus, Ohio. Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 760 28.8.2007 8:51pm Compositor Name: deshanthi 760 Handbook of Research Methods in Public Administration Lister, M. and Gardner, D. 2006. Engaging hard to engage clients: A Q methodological study involving clinical psychologists, Psychology and Psychotherapy, 79: 419–443. Locke, E.A. 1968. Toward a theory of task motivation and incentives, Organizational Behavior and Human Performance, 3: 157–159. Maslow, A.H. 1954. Motivations and Personality, Harper and Row, New York. Mattson, D.J., Byrd, K.L., Rutherford, M.B., Brown, S.R., and Clark, T.W. (2006). Finding common ground in large carnivore conservation: Mapping contending perspectives, Environmental Science & Policy, 9: 392–405. Maxwell, J.P. 2000. Managing conflict at the county level: The use of Q methodology in dispute resolution and strategic planning, Public Administration Quarterly, 24: 338–354. Maxwell, J.P. and Brown, S.R. 1999. Identifying problems and generating solutions under conditions of conflict, Operant Subjectivity, 23, 31–51. McDale, S.E. 1989. The Identification of Employee Types through Q Methodology: A Study of Parttime and Seasonal Recreation and Parks Employees. Unpublished M.A. thesis, Pennsylvania State University. McKeown, B. 1990. Q methodology, communication, and the behavioral text, Electronic Journal of Communication, 1. McKeown, B. 1998. Circles: Q methodology and hermeneutical science, Operant Subjectivity, 21: 112–138. McKeown, B. and Thomas, D. 1988. Q Methodology, Sage Publications, Newbury Park, California. McKeown, M., Hinks, M., Stowell-Smith, M., Mercer, D., and Forster, J. 1999. Q methodology, risk training and quality management, International Journal of Health Care Quality Assurance, 12: 254–266. Meo, M., Focht, W., Caneday, L., Lynch, R., Moreda, F., Pettus, B., Sankowski, E., Trachtenberg, Z., Vieux, B., and Willett, K. 2002. Negotiating science and values with stakeholders in the Illinois River Basin, Journal of the American Water Resources Association, 38: 541–554. Miller, D. and Friesen, P.H. 1984. Organizations: A Quantum View, Prentice Hall, Englewood Cliffs, New Jersey. Morçöl, G. 2002. A New Mind for Policy Analysis: Toward a Post-Newtonian and Postpositivist Epistemology and Methodology, Praeger, Westport, Connecticut. Morgado, E.M., Reinhard, N., and Watson, R.T. 1999. Adding value to key issues research through Q-sorts and interpretive structured modeling, Communication of the Association for Information Systems, 1, art 3. Mrtek, R.G., Tafesse, E., and Wigger, U. 1996. Q-methodology and subjective research, Journal of Social and Administrative Pharmacy, 13(2): 54–64. Murray, H.A. 1938. Explorations in Personality, Oxford University Press, New York. Nelson, B.D., Simic, S., Beste, L., Vukovic, D., Bjegovic, V., and VanRooyen, M.J. 2003. Multimodal assessment of the primary healthcare system of Serbia: A model for evaluating post-conflict health systems, Prehospital and Disaster Medicine, 18: 6–13. Nobbie, P.D. 2003. Evaluating partners in policymaking: Using Q-sort to assess changes in perception, attitudes, and efficacy in advocacy and empowerment. Paper presented at the meeting of the International Society for the Scientific Study of Subjectivity, Athens, Georgia. O’Reilly, C.A., Chatman, J.A., and Caldwell, D.F. 1991. People and organizational culture: A profile comparison approach to assessing person-organization fit, Academy of Management Journal, 34: 487–516. Oring, K.E. and Plihal, J. 1993. Using Q-methodology in program evaluation: A case study of student perceptions of actual and ideal dietetics education, Journal of American Dietetic Association, 93: 151–157. Oswald, D.L. and Harvey, R.D. 2003. A Q-methodological study of women’s subjective perspectives on mathematics, Sex Roles, 49: 133–142. Peace, R., Wolf, A., Crack, S., Hutchinson, I., and Roorda, M. 2004. Well-being, employment, independence: The views of sickness and invalids’ benefit clients. Working paper prepared for the Centre for Social Research and Evaluation, New Zealand Ministry of Social Development, Wellington, New Zealand. Peirce, C.S. 1955. The law of mind, Philosophical Writings of Peirce, J. Buchler (Ed.), Dover, New York, pp. 339–353. Pelletier, D., Kraak, V., McCullum, C., Uusitalo, U., and Rich, R. 1999. The shaping of collective values through deliberative democracy: An empirical study from New York’s north country, Policy Sciences 32: 103–131. Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 761 28.8.2007 8:51pm Compositor Name: deshanthi Q Methodology 761 Peritore, N.P. 1999. Third World Environmentalism: Case Studies from the Global South, University Press of Florida, Gainesville, Florida. Popovich, H. and Popovich, M. 2002. Use of Q methodology for hospital strategic planning: A case study, Journal of Health Management, 45: 405–414. Rainey, H.G. 1991. Understanding and Managing Public Organizations, Jossey Bass, San Francisco, California. Ramlo, S.E. 2005. An application of Q methodology: Determining college faculty perspectives and consensus regarding the creation of a school of technology, Journal of Research in Education, 15: 52–69. Roe, E. 1994. Narrative Policy Analysis, Duke University Press, Durham, North Carolina. Rokeach, M. 1973. The Nature of Human Values, Free Press, New York. Rutherford, M.B. and Gibeau, M. 2006. Interdisciplinary problem solving for grizzly bear conservation in the Banff-Bow Valley region of Alberta, Canada. Paper presented at a meeting of the Society for Conservation Biology, San Jose, California. Sasson, T. 1995. Crime Talk: How Citizens Construct a Social Problem, Aldine de Gruyter, New York. Scheb, J.M. 1982. Merit Selection, Role Orientation and Legal Rationalization: A Q Technique Study of Florida State District Courts. Unpublished Ph.D. dissertation, University of Florida. Schmolck, P. and Atkinson, J. 2002. PQMethod (2.11). Computer program, available at http:==www.rz.unibwmuenchen.de=~p41bsmk=qmethod= Selden, S.C., Brewer, G.A., and Brudney, J.L. 1999. Reconciling competing values in public administration: Understanding the administrative role concept, Administration and Society, 31: 171–204. Shah, S.M. 1982. Work Orientation of Middle-Level Managers from the Public Sector. Unpublished Ph.D. dissertation, Kent State University. Sheridan, J.E. 1992. Organizational culture and employee retention, Academy of Management Journal, 35: 1036–1056. Shilin, M.B., Durning, D., and Gajdamaschko, N. 2003. How American ecologists think about coastal zone environments, Values at Sea: Ethics for the Marine Environment, D. Dallmeyer (Ed.), University of Georgia Press, Athens, Georgia, pp. 239–259. Simmons, P. and Walker, G. 1999. Tolerating risk: Policy principles and public perceptions, Risk Decision and Policy, 4: 179–190. Simon, H.A., Smithburg, D.W., and Thompson, V.A. 1950. Public Administration, Knopf, New York. Simon, H.A. 1960. The New Science of Management Decision, Harper, New York. Smith, J. 2002. Fear as a barrier?: African American men’s avoidance of counseling services, Journal of African American Men, 6(4): 47–60. Snelling, S.J. 1999. Women’s perspectives on feminism: A Q-methodological study, Psychology of Women Quarterly, 23: 247–266. Stainton Rogers, W. 1991. Explaining Health and Illness, Harvester Wheatsheaf, London, United Kingdom. Steelman, T.A. and Maguire, L.A. 1999. Understanding participant perspectives: Q-methodology in national forest management, Journal of Policy Analysis and Management, 18: 361–388. Steg, L., Vlek, C., and Slotegraaf, G. 2001. Instrumental-reasoned and symbolic-affective motives for using a motor car, Transportation Research, Part F: Psychology and Behaviour, 4: 151–169. Stephenson, W. 1935. Correlating persons instead of tests, Character and Personality, 4: 17–24. Stephenson, W. 1953. The Study of Behavior: Q-Technique and Its Methodology, University of Chicago Press, Chicago, Illinois. Stephenson, W. 1963. Applications of communication theory: III. Intelligence and multivalued choice, Psychological Record, 23: 17–32. 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, Communication Yearbook 4, D. Nimmo (Ed.), Transaction Books, New Brunswick, New Jersey, pp. 7–36. Stephenson, W. 1983. Quantum theory and Q-methodology: Fictionalistic and probabilistic theories conjoined, Psychological Record, 33: 213–230. Stephenson, W. 1987. How to make a good cup of tea, Operant Subjectivity, 10: 37–57. Stephenson, W. 1989. Quantum theory of subjectivity, Integrative Psychiatry, 6: 180–195. Stowell-Smith, M. and McKeown, M. 1999. Locating mental health in black and white men: A Q-methodological study, Journal of Health Psychology, 4: 209–222. Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 762 28.8.2007 8:51pm Compositor Name: deshanthi 762 Handbook of Research Methods in Public Administration Stricklin, M. and Almeida, R. 2004. PCQ for Windows (academic edition). Computer program, available at http:==www.pcqsoft.com= Sun, T.W. 1992. Indigenization of public administration knowledge in Taiwan, Asian Thought and Society, 17: 97–112. Sun, T.W. and Gargan, J.J. 1999. Public administration scholar-practitioner differences: A Q study of theorypractice connections in Taiwan, Handbook of Comparative Public Administration in the Asia-Pacific Basin, H-K. Wong and H.S. Chan (Eds.), Marcel Dekker, New York, pp. 141–161. Sylvia, R.D. and Sylvia, K.M. 1986. An empirical investigation of the impacts of career plateauing, International Journal of Public Administration, 8: 227–241. Taylor, P., Delprato, D.J., and Knapp, J.R. 1994. Q-methodology in the study of child phenomenology, Psychological Record, 44: 171–183. Thomas, D.B. and Siegelman, L. 1984. Presidential identification and policy leadership: Experimental evidence on the Reagan case, Policy Studies Journal, 12: 663–675. Thomas, D.M. and Watson, R.T. 2001. Q-sorting and MIS research, CAIS: Communications of the Association for Information Systems, 7. Available http:==cais.isworld.org=articles=8–9=default.asp?View ¼ Journal& x ¼ 46&y ¼ 12 Tolymbek, A. (2007). Political Leadership Style in Kazakhstan. Unpublished Ph.D. dissertation, Kent State University. Vajirakachorn, S. and Sylvia, R.D. 1990. Administrative attitudes of elite officials in a buddhist polity, Operant Subjectivity, 13: 163–174. Valenta, A.L. and Wigger, U. 1997. Q-methodology: Definition and application in health care informatics, Journal of the American Medical Informatics Association, 4: 501–510. Van Dusen, W. and Rector, W. 1963. A Q sort study of the ideal administrator, Journal of Clinical Psychology, 19: 244. Van Eeten, M. 2001. Recasting intractable policy issues: The wider implications of the Netherlands civil aviation controversy, Journal of Policy Analysis and Management, 20, 391–414. Van Exel, N.J.A., de Graaf, G., and Rietveld, P. 2004. Getting from A to B: Operant approaches to travel decision making, Operant Subjectivity, 27: 194–216. Vroman, H. 1972. Types of Administrators and Some of Their Beliefs—A Q Factor Analysis. Unpublished Ph. D. dissertation, University of Iowa. Waller, N.G. and Meehl, P.E. 1998. Multivariate Taxometric Procedures: Distinguishing Types from Continua (Advanced Quantitative Techniques in the Social Sciences series, No. 9), Sage Publications, Thousand Oaks, California. Walton, M.L. (2006). An Investigation of the Effect of Deliberative Discussion on Environmental Preferences, Using the Coastal Plain of the Arctic National Wildlife Refuge as a Case Study. Unpublished Ph.D. dissertation, Rensselaer Polytechnic Institute. Warnaars, M. and Pradel, W. (2007). The Perceptions of Urban and Rural Farmer Field School Participants in Lima: A Q-Method Approach, Urban Harvest, Centro Internacional de la Papa, Lima, Peru. Watts, S. and Stenner, P. 2005. Doing Q methodology: Theory, method, and interpretation, Qualitative Research in Psychology, 2: 67–91. Webler, T. and Tuler, S. 2001. Public participation in watershed management planning: Views on process from people in the field, Human Ecology Review, 8(2): 29–39. Webler, T., Tuler, S., and Krueger, R. 2001. What is a good public participation process? Five perspectives from the public, Environmental Management, 27: 435–450. Webler, T., Tuler, S., Shockey, I., Stern, P., and Beattie, R. 2003. Participation by local governmental officials in watershed management planning, Society and Natural Resources, 16: 105–121. Wester, K.L. and Trepal, H.C. 2004. Youth perceptions of bullying: Thinking outside the box, Operant Subjectivity, 27, 68–83. Wilkins, J.L., Kraak, V., Pelletier, D., McCullum, C., and Uusitalo, U. 2001. Moving from debate to dialogue about genetically engineered foods and crops: Insights from a land grant university, Journal of Sustainable Agriculture, 18, 167–201. Wolf, A. 2004. The bones in a concourse, Operant Subjectivity, 27, 145–165. Wong, H.S. 1973. Utilizing group Q-technique to index leadership behavior in task-oriented small groups, Unpublished M.A. thesis, University of Hawaii. Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 763 28.8.2007 8:51pm Compositor Name: deshanthi Q Methodology 763 Wood, J. 2004. Mediation styles: Subjective description of mediators, Conflict Resolution Quarterly, 21: 437–450. Wood, R.A. 1983. Perspectives on Prosecutorial Discretion: A Q-Methodology Analysis. Unpublished Ph.D. dissertation, University of Missouri-Columbia. Woolley, J.T. and McGinnis, M.V. 2000. The conflicting discourses of restoration, Society & Natural Resources, 13: 339–357. Woolley, J.T., McGinnis, M.V., and Herms, W.S. 2000. Survey methodologies for the study of ecosystem restoration and management: The importance of Q-methodology, Integrated Assessment of Ecosystem Health, K.M. Scow, G.E. Fogg, D.E. Hinton, and M.L. Johnson (Eds.), Lewis, Boca Raton, Florida, pp. 321–332. Wright, C.M. and Meehling, G. 2002. The importance of operations management problems in service organizations, Omega, 30: 77–87. Wright, C.M., Riggle, C.G., and Wright, B.G. 1998. Technique for pre-implementation assessment in total quality programs, International Journal of Quality & Reliability Management, 15: 414–430. Wrigley, B.J. 2006. Bumping their heads: A study of how women in public relations and communications management view variables contributing to the glass ceiling,’’ Journal of Human Subjectivity, 4(1): 33–47. Yarwood, D.L. and Nimmo, D.D. 1975. Perspectives for teaching public administration, Midwest Review of Public Administration, 9: 28–42. Yarwood, D.L. and Nimmo, D.D. 1976. Subjective environments of bureaucracy: Accuracies and inaccuracies in role-taking among administrators, legislators, and citizens, Western Political Quarterly, 29: 337–352. Züger Cáceres, R. 2003. Do participatory interventions empower people? Paper presented at the meeting of the International Society for the Scientific Study of Subjectivity, Canton, Ohio. Miller/Handbook of Research Methods in Public Administration AU5384_C037 Revise Proof page 764 28.8.2007 8:51pm Compositor Name: deshanthi