• Tidak ada hasil yang ditemukan

TAKING ADVANTAGE OF MULTIPLE PARADIGMS IN PUBLIC ADMINISTRATION RESEARCH

specification of research question, review of relevant literature, and choice of data source. The second phase, data analysis, also contains three substages: design of analytical process, coding, tabulation, or exhibition of analyses. Thefinal phase, theory building, has three substages as well:

development and test of propositions, theory development, and evaluation of resulting theory.

For each of the nine substages, Lewis and Grimes (1999) propose corresponding inductive activities that are based on multiple paradigms. The purpose of these activities is outlined; the activities are exemplified and applied to the study of advanced manufacturing technology (AMT).

3.3 TAKING ADVANTAGE OF MULTIPLE PARADIGMS IN PUBLIC

3.3.1 RESEARCHQUESTION

Transforming broad topics to researchable questions is a starting point for most quantitative studies.

An original research question is the foundation of the house of inquiry. A good and significant question should have the potential to contribute to knowledge and theory development, as well as managerial and policy practices. It is not surprising to see that most social science studies anchor their research question to a single paradigm, reflecting what Morgan (1980) calls puzzle-solving activities that‘‘seek to operationalize the detailed implications of the metaphor defining a particular school of thought.’’Such activities, as Kuhn (1962) sees it, are characteristic of‘‘normal science’’

where the legitimacy of a question depends on whether it fits or contributes to the dominant paradigm. In a postnormal science with competing paradigms, the legitimacy of the question may vary across paradigms. As a result, rather than defining the research question from a narrow perspective based on a single paradigm, multiparadigm inquiry concentrates on defining the phenomenon of interest so that interpretativeflexibility can be enabled. This is particularly import- ant for public administration as an appliedfield where helping understand and improve adminis- trative performance regardless of theoretical perspectives is, at the very least, as important as validating or extending a single perspective.

For example, the issue regarding the performance of public service networks can be approached in different ways. A comprehensive list of factors that may help explain such performance can be identified such as the informal relational structure of the network (e.g., centrality and density), institutional arrangements (e.g., performance monitoring and oversight rules), network stability (e.g., network participant stability and personnel stability), network learning, social capital, managerial strategies (e.g., networking and managing upwards and outwards), and network contexts (e.g., resources and environments). It would be nice if all those aspects could be included and tested in a comprehensive model, but public administration scholars are often constrained in their ability to conduct such a large-scale project. Depending on their theoretical perspectives, they may narrow the topic in different ways. A positivist researcher who is more interested in examining the effects of managerial strategies would probably use network contexts as control variables, assuming that public managers can deliberately choose among alternative strategies based on rational calculations (e.g., Agranoff and McGuire, 2003; O’Toole and Meier, 1999). An interpre- tivist researcher, in comparison, would probably take network contexts more seriously, adopt qualitative designs, and examine the contexts affecting the emergence of managerial strategies and the social construction of performance (e.g., Granovetter, 1985). In the words of Berry et al.

(2004), the contexts can be treated‘‘not as noise that is incidental to the purposes of the network, but as everyday sources of meaning that guide and define the actions of the participants’’(p. 549). These two lines of research, of course, can be integrated into a larger project. Public administration researchers need to be aware of the fact that how a question should be defined depends on their paradigmatic assumptions. When it is feasible, asking several questions from different perspectives would help us gain a more complete understanding of the phenomenon of interest.

Again, it is worth acknowledging that framing research questions from a multiparadigm perspective is not always more preferable than from a single-paradigm perspective. The background and purpose of the research matter in the choice. Nevertheless, one should remember that paradig- matic assumptions determine the big questions one thinks worth pursuing. For example, from the traditional managerial perspective that focused on organizational and behavioral analysis (thefirst model in Table 3.1 or the Ideas II and IV in Holzer, Gabrielian, and Yang (2006)), Behn (1995) identifies three big questions of public management: How can public managers break the micro- management cycle? How can public managers motivate people? How can public managers measure achievement? In comparison, Neumann (1996) starts with the open systems theory and specifies three different big questions: What is the nature of a ‘‘public’’ organization? How is the public organization related to its environment? What does it mean to manage or to administer the public organization? Taking a different stance, Kirlin (1996) points out that Behn’s argument is rooted in

the managerial perspective of public administration and it ignores other perspectives including the political and legal ones (see Table 3.1). Rooted instead in a perspective of democratic polity, Kirlin (1996) puts forward seven big questions:

. What are the instruments of collective action that remain responsible both to democratic- ally elected officials and to core societal values?

. What are the roles of nongovernmental forms of collective action in society, and how can desired roles be protected and nurtured?

. What are the appropriate tradeoffs between governmental structures based on function (which commonly eases organizational tasks) and geography (which eases citizenship, political leadership, and societal leaning)?

. How shall tensions between national and local political arenas be resolved?

. What decisions shall be‘‘isolated’’from the normal processes of politics so that some other rationale can be applied?

. What balance shall be struck among neutral competence, representativeness and leadership?

. How can processes of societal learning be improved, including knowledge of choices available, of consequences of alternatives, and of how to achieve desired goals, most importantly, the nurturing and development of a democratic polity? (p. 417)

3.3.2 LITERATUREREVIEW

In multiparadigm inquiry, literature in different paradigms should be identified, compared, end evaluated. This is particularly useful when the researcher attempts to develop a new theory or perspective. For example, in explaining and advocating the New Public Administration (NPA), Frederickson (1980) reviewedfive public administration models that can be viewed as paradigms as loosely defined: classic bureaucratic model, neobureaucratic model, institutional model, human relations model, and public choice model. He used a bracketing technique by showing how thefive models are different in terms of their empirical focus, unit of analysis, characteristics, and values to be maximized. However, the review also addressed bridging among the models as they overlap in various ways. Both the classic bureaucratic model (e.g., Gulick and Urwick) and the neobureau- cratic model (e.g., Simon and March) emphasize the value of efficiency and economy, although the latter pays specific attention to rationality and productivity. Both the neobureaucratic model and the institutional model (e.g., Downs, Lindbloom and Thompson) adopt the positivist perspective and focus on decision making, but they emphasize different values—the latter stresses increment- alism, pluralism, and criticism instead of rationality, efficiency and economy.

Moreover, in comparing New Public Administration with traditional public administra- tion models, Frederickson (1980) uses a technique similar to‘‘bridging’’that attends to transition zones between paradigms. Frederickson considers the new public administration and traditional models as two extremes, in between which are transitional positions. For example, although traditional models advocate for ‘‘leadership by authority,’’ the New Public Administration pre- scribes‘‘leadership by change facilitation,’’and the transitional position indicates ‘‘leadership by consent’’(p. 55). For traditional models,‘‘the problem is basically one of reform or change’’; for the New Public Administration,‘‘the problem is one of institutionalizing change procedures’’; and for the transitional position,‘‘change and reorganization should be encouraged’’(p. 55). The book also offers similar comparisons with regard to the nature of change, uncertainty, responsiveness, and the relationship between technology and politics. In another well-articulated study, Frederickson (1996) compares the reinventing government movement and the new public administration. He sees the similarities between the two movements in terms of management and organization:‘‘Both move- ments have as their impetus the need for change. Both are committed to responsiveness but in different ways. In new public administration, it is a professional public service dedicated to both efficiency and social equity; in reinventing government, it is the empowerment of individual

customers to make their own choices’’(p. 269). In the meantime, he points out the two movements differ in significant ways. New Public Administration, in his view, is more concerned with issues of rationality, epistemology, methodology, institutions, and politics than reinvention is. In addition, the two movements have a sharp difference in the values they espouse. Although reinvention empha- sizes individual choice, incentive structure, competition, and market,‘‘new public administration is concerned more with humanistic and democratic administration, concerned more with institution building and professional competence, concerned more directly with issues of politics and with matters of justice and fairness—broadly under the label of social equity’’(p. 269).

Another excellent example is provided by Denhardt and Denhardt (2000) in outlining a theory of new public service and comparing it to old public administration and New Public Management.

Denhardt and Denhardt demonstrate clearly how the three perspectives differ with regard to ten aspects: primary theoretical and epistemological foundations; prevailing rationality and associated models of human behavior; conception of the public interest; targets of public service responsive- ness; role of government; mechanisms for achieving policy objectives; approach to accountability;

administrative discretion; assumed organizational structure; and assumed motivational basis of public servants and administrators. For instance, in old public administration, the role of govern- ment is rowing (designing and implementing policies focusing on a single, politically defined objective) and the approach to accountability is hierarchical (administrators are responsible to elected officials); in New Public Management, the role of government is steering (acting as a catalyst to unleash market forces) and the approach to accountability is market-driven (adminis- trators are responsible to customers); and in new public service, the role of government is serving (negotiating and brokering interests among citizen and community groups, creating shared values) and the approach to accountability is multifaceted (administrators are responsible to law, community values, political norms, professional standards, and citizen interests).

Reviewing the literature as suggested above may be a difficult and arduous task because authors rarely state their paradigm and often make the choice unconsciously. As Berry et al. (2004) comment on network research, they identify three traditions rooted in different disciplines: social network analysis (sociology), policy networks (political science), and public management networks (public administration). They point out that‘‘until the last few years, there has been relatively little cross-fertilization across the research traditions, and it is often unclear whether authors in each research stream have intentionally disregarded or are simply unaware of the complementary research streams’’ (p. 540). Multiparadigm inquiry requires researchers to evaluate publications carefully, attending specifically to the use of terms and metaphorical language in the publications (Willmott, 1993). This also suggests that it may be beneficial to the discipline if researchers could make conscious a choice about paradigms and then state it explicitly in their articles.

3.3.3 HYPOTHESES

With multiple paradigms available for use, researchers should make explicit the paradigmatic assumptions of their hypotheses. At the least, they should be aware of the assumptions and associated limitations even when it is clear that they are dealing with a single-paradigm study.

Whenever feasible, it is also beneficial to use competing or complementary hypotheses based on different paradigms so that a more complete picture can be drawn about the phenomenon of interest.

This can be applied not only to quantitative studies, but also to qualitative studies that include qualitative propositions.

One classic example is offered by Allison (1971) in explaining the Cuban Missile Crisis with three models of decision making, which are used as three hypotheses or propositions to be tested.

The first model, rational actor, views government decisions as the product of a single actor in strategic pursuit of self-interest. This model supports hypotheses that are related to the problem, the alternatives, the strategic benefits and costs, shared values and axioms, and internal pressures. The second model, organizational process, conceptualizes decision making as highly structured through

standard operating procedures (SOP). It relates to hypotheses that are directed to concepts such as organizational implementation, organizational options, incremental change, long-range planning, administrative feasibility, and directed change. The third model, bureaucratic politics, argues that government actions are the product of bargaining and compromise among political players.

This model entails hypotheses that are related to concepts such as power, parochial priorities, rules of the game, action-channels, and stakes and stands. In drawing the conclusion, Allison hinted that future studies should explain decision making with multiple models and associated hypotheses because the models can complement each other:‘‘The best analysts of foreign policy manage to weave stands of each of the three conceptual models into their explanations . . . By drawing complementary pieces from each of these styles of analysis, explanations can be significantly strengthened’’(p. 259). In this view, Model II is embedded in Model I, as the former ‘‘fixes the broader context, the larger national patterns, and the shared images,’’whereas the latter‘‘illuminates the organization routines that produce the information, alternatives, and action’’(p. 258). Similarly, Model III can be seen as embedded in Model II, as Model III ‘‘focuses in greater detail on the individual leaders of a government and the politics among them that determine major governmental choices’’(p. 258).

In empirical quantitative studies, researchers often include hypotheses derived from different schools of thought to improve their model’s explanatory power without explicitly acknowledging or realizing the multiparadigm nature of the hypotheses development. For example, Quinn and Rohrbaugh’s (1983) competing-values framework gives rise to four organizational culture types:

rational culture (rational goal model), group culture (human relations model), developmental culture (open systems model), and hierarchical culture (internal process model). These four types of organizational culture are all included as independent variables by Moynihan and Pandey (2005) in assessing how environmental factors and organizational factors affect government performance.

In addition to the culture variables, their model includes other variables that reflect different models of the competing-value framework such as goal clarity (rational goal model), centralization (internal process model), and elected official support of agency (open systems model). Another example is offered by Yang (2006) in examining the conditions under which public managers will honestly communicate performance measurement results to other stakeholders. Yang (2006) develops hypotheses that are based on two contrasting perspectives of bureaucratic behavior:

agency theory and culture theory. More generally, scholars have borrowed a variety of perspectives and approaches from other disciplines to study behavior of bureaucrats and bureaucracies. These perspectives fall into at least two categories: the rational choice school of which agency theory is a part and the human relations school of which culture theory is a part. Any explanation of bureaucratic behavior will necessarily involve hypotheses from both schools of thought.

3.3.4 METHODOLOGY

Given a multiparadigm reality, researchers should realize that it may be beneficial to adopt a mixed methodology, to collect data via multiple sources, and to code the data from multiple perspectives.

The argument for mixed methodology is not new in the public administration literature. Qualitative methods such as case study, ethnography, hermeneutics, action research, and grounded theory are usually associated with interpretivist or constructivist paradigms. Quantitative designs such as experiments, quasi-experiments, and cross-sectional designs are often associated with positivist paradigms. The debate of qualitative versus quantitative methods has evolved in a way that corresponds to what we depict in Figure 3.1. Taking the incommensurability position, the segrega- tionists argue that quantitative and qualitative paradigms are incompatible due to their mutually exclusive epistemological and ontological assumptions. Therefore, using different methods will lead to essentially different observations of the same phenomenon. Taking the second position in Figure 3.1, the integrationists, however, reject the dichotomy between the quantitative and the qualitative, attempting to integrate both methodologies for the same study. The quantitative and

qualitative methods will support each other and enhance the credibility of the research. Taking the third or the fourth position in Figure 3.1, the eclectics acknowledge the two approaches have distinct assumptions, but both are valid and the situation will dictate which one should be used or whether both should be used as complementary elements.

We contend that qualitative and quantitative designs are complementary. Researchers can use a parallel strategy, simultaneously applying both designs. This is particularly useful when researchers first develop or explore a new paradigm because the parallel strategy minimizes the chances of confusion between paradigms (Schultz and Hatch, 1996). Alternatively, they can take a sequential route, applying the two types of designs one after another. A common sequence is that one starts with objective paradigms and follows with subjective paradigms, obtaining generalizable overviews first and then seeking more localized meanings. For example, in studying whether and how training is provided to local government employees with regard to citizen participation, Callahan and Yang (2005) used a national survey to collect quantitative data. The survey analysis led to some interesting questions, so they decided to purposefully select ten jurisdictions for follow-up inter- views based on the level of training they provided and the nature of the comments they made in their open-ended responses. They randomly selectedfive large andfive small jurisdictions from a group of 49 jurisdictions that indicated in the survey that they provided a lot of training or developed partnerships with citizens. They then conducted telephone interviews with the city managers and coordinators of neighborhood relations programs. One can also reverse the order, seeking subjective understandings of a new topicfirst and then designing quantitative studies. This sequence is often assumed in quantitative studies, but authors rarely show whether they went though the arduous qualitative research process before testing the quantitative models.

Public administration may benefit greatly from more conscious efforts to apply multiple paradigms in a research project. Take performance measurement research for example. One stream of studies is functionalist, describing how performance measurement systems should be established and what factors affect their effectiveness (e.g., De Lancer Julnes and Holzer, 2001; Wang and Evan, 2000). Another stream is critical or postmodern, arguing that performance measurement is a control gimmick employed by politicians as a discourse that can be misused (Fox, 1996; Radin, 2000). It would advance performance measurement research if the two approaches can be more carefully aligned or even integrated. In organizational studies, it is not rare that one uses a functionalist paradigm to highlight managerialist understandings and then follows with more critical views to expose fragmentation and conflict (Martin, 1992). The sequence can be reversed to start with critical views followed by functionalist perspectives.

Paradigms relate to data collection and coding. As Lewis and Grimes (1999) contend,

‘‘choosing a source of data for multiparadigm inquiry is controversial, since the question of what constitutes data is paradigm laden (Gioia and Pitre, 1990). Managing this dilemma requires collecting a metatheoretical sample: data interpretable from multiple-paradigm perspectives’’

(p. 679). For example, Martin (1992) collected extensive and unstructured interview data about organizational culture, which were analyzed through divergent theoretical lenses. In the meantime, with multiple paradigms in mind one is more likely to ‘‘see’’ the data with greater analytical depth. Multiparadigm coding typically involves a two-stage process: becoming intimate with the data and then imposing alternative spins. During thefirst step, researchers take detailed notes, developing first impressions of nuances and patterns of the data. During the second step, researchers ‘‘read’’ the data via each paradigm, enabling construction of the different insights (Lewis and Grimes, 1999). The application of each paradigm is a reconstruction process and a recoding process.

3.3.5 RESULTS

With a richer understanding of various paradigms, a researcher is more likely to develop insightful analysis because he or she is more likely to capture certain features and subtleties in the data and to