Introduction
The Research Problem
The purpose of this paper is to offer one vision of developing a methodological theory of mixed methods research co-equal with that of quantitative and qualitative research. I use a case study of the US Army Command & General Staff College engaged in a redesign of its curriculum, its teaching practices and its design process itself in a period of revolutionary change while supporting a nation at war. I describe circumstances and worldviews in which I argue that only mixed methods research may be employed to simultaneously develop a deep appreciation of uncertainty, improve decision making through an appropriate gathering, mixing and analyzing of quantitative and qualitative data, and applying “learning in action” as a strategy to manage success. I contrast the view of research as a process of increasing knowledge for control with a worldview of research as a learning-in-action that allows for deep appreciation of complexity but without the assertion that appreciation and research can lead to prescriptive measures of control. I examine the merging feedback system of the CGSC curriculum redesign as a mixture of qualitative and quantitative data. The concept of “Voice” that emerges from the CGSC action research process will be described, along with a multi-phased, multi-year research plan that demonstrates the practical development of an interactive dynamic research plan that is also adaptive to interim and periodic results. The paper reflects a pragmatic worldview as it focuses on practical outcomes inside an organization concerned with real-world results, but acknowledges the importance and utility of the other 3 worldviews described by Creswell (2007, p.6), namely advocacy/participatory, post-positivist and constructivist.
Waldrop (1992, 2008) described the emerging science of complexity in a rich description of the inter-disciplinary work developing at The Sante Fe Institute. Sixteen years later (Waldrop, 2008) he found that the pioneer days of complexity research had evolved into a rich diversity of programs in major and minor universities worldwide, with lines of business and cognitive domains each finding ways to apply the ideas of emergence, uncertainty and complexity in new and profound ways. What remained unchanged from the origins of the research were the questions of what next and so what and how much more is there and what does it mean to apply an appreciation of complexity to everyday problems and opportunities. The field of education is only beginning to appreciate how complexity and uncertainty may change the dynamics, structure, content and practice of adult education (Siemens, 2004). Professions in particular will be challenged by educating for complexity, since deep, profound, and reliable bodies of knowledge are at the center of professional practice. Educators, themselves members of a profession, are examining what it means to educate, teach and instruct in light of an emerging awareness of complexity.
The US Army Command & General Staff College (CGSC) is a self-described “learning organization” (Senge, 2000), engaged in a revolutionary re-design of curriculum and teaching practice, with a mission to educate 1500 US Army Majors for uncertainty and complexity, while engaged in a global war on terror and in direct combat in Iraq and Afghanistan (Long, 2009). This provides an opportunity to examine reflective learners and practitioners in action (Schon, 1987) using mixed methods research and using multiple worldviews (Creswell, 2009; Creswell & Plano Clark, 2007).
Past Research on the Problem
Edmondson and McManus (2003) propose a structured approach to selecting research methods that fit the state of theory in a given field. Their 3 archetypes of the state of theory: nascent, intermediate and mature are connected to qualitative, mixed methods, and quantitative research methods, resulting in an appropriate methodological fit that aims to meet the needs of researchers worldwide. Creswell(2009) offers a systematic approach to analyzing: researcher worldview, research purpose, research questions, the state of theory, data collection, populations and situations to be studied, and data analysis in order to further refine the methodological fit and better connect purpose with practice across all 3 methods. Creswell and Plano Clark (2007, p.8) offers a functional working description of the state of mixed methods research, which proceeds from a deep review of current field practice, establishes a superb framework for classifying current choices of mixed methods research design and the means by which methodological fit may be refined, but stops at the boundary of developing a deep theory of mixed methods of research.
Deficiencies in past research and Need for Mixed Methodology
Conventional professional education processes have been adapting at an increasingly frequent rate as a consequence of Army senior leader directives and direct field feedback. The adaptive processes and decisions to date have been single issue, single iteration problem solving exercises inherited from an environment in which incremental change was the norm and most appropriate. These processes are less and less suitable as rates of required change increase and the relevance of existing processes and curriculum are increasingly called into question (Long, 2009)
Audience
The audience for this research include: staff, faculty and students of CGSC; educators of military professional schools; curriculum developers in graduate schools and organizations engaged in preparing leaders for uncertainty; scholar-practitioners of mixed methods looking to adapt practical field methods of for mixing qualitative and quantitative data; scholars examining the deep theory of the methodology of mixed methods.
Purpose
Purpose of the study, and reasons for a mixed methods study
The purpose of this study is to examine the CGSC curriculum redesign project and the emerging feedback system that guides design decisionmaking, which incorporates both quantitative and qualitative data. The project can be described as intermediate theory in the Edmondson and McManus ontology, and therefore suitable for mixed methods research since I am introducing an emerging concept (“Voice”), focusing on the exploration of theoretical propositions (the theory of mixed methods), the availability of sets of rich theory that inform the research (adult learning, decisionmaking, complexity, design, learning organizations, narrative inquiry and action research), and incorporating multiple data types and analysis (Edmondson& McManus, 2007, p.1165).
Research Questions and Hypotheses
Quantitative Questions
H1: Student satisfaction measured on the Noel-Levin Adult Learner Satisfaction Survey is not different than their reported overall satisfaction
H2: Student education priorities measured on the Noel-Levin Adult Learner Satisfaction Survey are not different than those of faculty and college leadership as measured on the same instruments
H3: Student education priorities measured on the Noel-Levin Adult Learner Satisfaction Survey do not vary through time in the course of the academic year
H4: Student education priorities measured on the Noel-Levin Adult Learner Satisfaction Survey do not vary after graduation and reassignment to field units
H5: Student satisfaction measured on the Noel-Levin Adult Learner Satisfaction Survey do not vary from satisfaction as measured by existing CGSC Quality Assurance surveys
Qualitative Questions
What are the dominant and subordinate narratives that emerge from focus group discussions on educational priorities and practice and environment within CGSC?
How does the curriculum design decisionmaking process respond to similarities and differences in narratives that emerge from groups of students, faculty and senior leaders?
Describe the development, emergence of the construct of “Voice” from the CGSC PAR cycles, and how this prototype construct is evolving and being applied by various sub-groups within and associated with CGSC, by applying various interpretive methods of the narrative inquiry tradition.
Mixed Methods Questions
1. To what extent are qualitative insights generated from PAR cycles, focus groups, and individual interviews supported by quantitative data generated from surveys and actual use data of digital communication and collaboration mediums?
2. How are various organizational narratives constructed by sub-groups within the CGSC curriculum design process in order to make sense of quantitative data?
3. What insights are offered by the application of various narrative inquiry traditions? Which traditions are favored or overlooked or rejected by curriculum design decisionmakers?
4. What happens within CGSC when students and faculty are given opportunities to exercise “Voice”?
Philosophical Foundations for Mixed Methods Research
Quantitative research literature review
Student satisfaction surveys built on consumer theory (Watkins, 2009) are broadly applied in colleges and universities, and treat students as free-willed individuals that choose between alternatives for an educational institution and particular fields of study. They are seen as rational actors with definite expectations about what they want in their educational experience, and that satisfaction occurs when their expectations are met or exceeded. Smart, Feldman and Ethington (2006) note a decline in the attention being paid to the attitudes and behaviors of faculties, administrators, and the college and university environments as contributors to student success.(p.2.). These insights are related to the “college impact” model of student success. Applying the Noel-Levitz Adult Student Priorities Survey leverages a robust, nationally recognized, validated research instrument whose dimensions reflect the areas of importance emerging from the CGSC Participatory action research (PAT) study (Long, 2009) and enables quantitative research into the existing database of historical satisfaction measures currently applied in the college’s curriculum design process.
Qualitative research literature review
The James, Milenkiewicz and Buchnam (2008) application of Participatory Action Research (PAR) develops measureable action steps that can lead to revolutionary transformations within educational institutions. The use of measureable qualitative and quantitative data gives power and legitimacy to the insights it generates inside an organization that values rigor and validity, while respecting the intuitive insights of qualitative research. Prasad describes many techniques of Narrative Inquiry that offer many techniques for interpreting and making sense of qualitative and quantitative data. Reason & Bradbury, (2008) and Clandinin, (2007), describe these disciplines and crafts of action research and narrative inquiry as having a relatively mature foundation of theory and best practices, with enough variation between sub-disciplines as to create real and significant choices for researchers . Various methodologies in each discipline can be characterized according to their own logic that connects their particular world view (Creswell, 2009), ontology, research technique, data requirements, classification and analysis protocols, and strategies for sense-making of the results of inquiry. The combination of PAR and narrative inquiry offer a robust set of strategies for generating insightful qualitative data with connections to quantitative data sets, which make them especially useful and practical to mixed methods researchers.
Mixed methods research literature review
Creswell and Plano Clark (2007) provide a broad yet detailed overview of the current state of the art of mixed methods research. They offer a working definition of the field derived from a survey of practice which proceeds from a deep understanding of high quality methods of practice, through choices of design and point to potentials for the development of deep methodological theory. They offer mixed methods as an appropriate research strategy as a way to improve on the use of either qualitative or quantitative research alone. In their view, mixed methods are more comprehensive, can answer more types questions, encourages collaboration and the deliberate incorporation of more than 1 worldview and is especially well suited for situations where practicality and pragmatism are prized (Creswell & Plano Clark 2007, p8-11).
Methods
A definition of mixed methods research
Creswell and Plano Clark (2007, p.5) define mixed methods research in the following way:
“As a method, it focuses on collecting, analyzing and mixing both quantitative and qualitative data in a single study or series of studies. Its central premise is that the use of quantitative and qualitative approaches in combination provides a better understanding of research problems than either approach alone”.
The type of design used and its definition
In this section I will briefly describe the theoretical shortcoming of treating mixed methods merely as a practical solution to improving upon either the qualitative or quantitative approach alone, and why a broader and deeper theory of mixed methods is appropriate for developing deep appreciation of complexity and uncertainty. I will briefly describe two different designs that would pass the test of the Creswell and Plano Clark ontology of mixed methods designs. With appropriate development, either would be approved for research within CGSC.
Comparison table (adapted from Creswell & Plano Clark (2007)
| Design 1 | Design 2 | |
| Choices | Explanatory | Exploratory |
| Theoretical Description | A 2 phase design, where qual helps explain or build upon initial quan results (p.71) | 2 phase design where qual results help develop or inform 2d phase quan inquiry |
| Description of application to CGSC | Round 1: the Noel-Levitz Adult Learner Satisfaction Survey is applied to a population, and results are tabulated, analyzed, compared against national graduate student norms and in a time series from the beginning, midpoint and endpoint of the academic year. Insights are developed
Round 2: A series of focus groups and individual interviews are used to develop qualitatitive insights to make sense of the quantitative findings |
Round 1: a set of participatory action research cycles identify areas of pressing concern to leaders, faculty and students within the college. A grounded theory is developed and constructs are defined by the community of practice, informed by theory from PAR outsiders.
Round 2: A quant survey is developed to explore deeply into issues and constructs developed by the PAR teams to cross check for validity, to confirm or deny, to support or modify the emerging grounded theory and provide the basis for future inquiries as selected by PAR teams. (Note: this is s summary of the actual process used at CGSC as the basis for this case study. The various tangents deriving from the initial rounds of inquiry generated my epistemological concerns with the pragmatic assumptions of mixed emthods) |
| Design notation | QUAN->QUAL | QUAL->QUAN |
| Justification | Needs qual to help explain significant, , non-significant, outlier or surprising quant results | Exploration is needed because:
1. no existing instrument 2. unknown variables 3. immature theory or framework Well suited for exploring a phenomenon or when researcher wants to generalize to other populations, test emerging theory or classifications (p. 75) |
| Variants | 1. Follow-up explanations (quan results, insights need additional explanation)
2. Participant selection (where a sampling of representative outliers are selected for follow-on inquiry) |
1. instrument development model
2. taxonomy development model |
| Strengths | Straightforward implementation
Feasible for single researcher 2 section report of results Supports both single and multiple phase studies Appealing to quan researchers |
1. easy to design, describe, implement and report
2. although initial emphasis is on qual, the quan phase makes it easier to appeal to quan audience 3. both variants supports multiphase studies well |
| Challenges | 1. Time consuming
2. Decisions on which individuals to use by phase w/justification 3. Difficulties with IRBs 4. Deciding which results to explore 5. Specifying criteria for follow-on inquiry (before or after results?) |
1. time consuming
2. difficult to specify phase 2 construct for IRB prior to phase 1 results 3. deciding up front which individuals to use in phase 2 4. which data to use in phase 2 instrument 5. deciding relevancy of phase 1 results for phase 2 taxonomy |
| Timing | 2 phase sequential model | 2 phase sequential model |
| Weighting | I think the QUAL(quan) model is more likely. This design relies on at least a mature enough state of theory to allow for initial quan inquiry, but we are more concerned with the interpretation and application of insights than in model or theoretical validation | the equal weighted choice is more logical; the desired outcome is an improvement to state of theory (quan) by either a better instrument or by an improved taxonomy (ontology). Yet the reliance on initial qual inquiry as a guide makes it at least co-equal to quan. |
| Mixing the data | Either Merging or Connecting is more likely than embedding. Embedding implies a single phase, whereas this is defined as a 2phase design. The improved explanation of initial quan findings is how the design could be “connected”. If the interpretation or meaning making is intended to create “rich description” then either variant of merging is logical | Connecting is by far the most logical design choice, as the 2 phases are explicitly linked; quan follows qual and the connection is either an instrument or a taxonomy.
There is a distinct “manufacturing or processing” aspect to this design, which does not seek to produce a rich description that is a blend, but rather produces a better quan framework as guided by the initial qual inquiry |
Diagrams:
Research model 1: Explanatory:
Research model 2: Exploratory:
Analyzing the data:
In both models of mixed methods design the quantitative data would be subjected to power analysis, tests for relationship and causality. The quantitative hypotheses are framed in the form of null hypotheses in order to determine if there were differences that could be attributed to a difference in instruments and what they are measuring (existing survey vs the Noel-Levitz survey); through time series tests to see if there is a treatment effect, and with the samples subjected to control variables to examine the effects of demographic, career experience, educational goals and faculty specific effects on the measures of satisfaction and importance.
Qualitative data would be subjected to thematic analysis according to the practices of the grounded theory, narrative inquiry and PAR traditions/ Narrative inquiry traditions are especially important here as outlined in Boje (2001).
Analyzing the mixed data would be drivcefn by the specific design selected as noted above.
Theoretical analysis of the consequences of choice in mixed methods design
Both designs would be interpretable as providing a deeper insight and understanding than a study restricted to either of their individual qualitative or quantitative components. The functional definition of mixed methods (Creswell & Plano Clark, 2007, p.5) would lead decisionmakers, particularly of the pragmatic worldview, to ‘receive the wisdom of experts” and seek to straightforwardly apply the insights based on a justifiable belief that they now knew more about what was going on, and had in some fashion reduced the amount of uncertainty about the world around them. My central argument is that there are situations so complex and uncertain that no amount of research and conclusions drawn from best practices of the traditions of both quantitative and qualitative inquiry, and the best practices emerging from mixed methods as described by Creswell and Plano Clark. In fact my use of the word “situation” in the preceding sentence, is an implied assumption that there is such a thing in the real world as a definable “situation” or problem set which may be bounded and contained by a problem solving, decisionmaking entity. While this construct is the basis for the post positivist tradition, which has endlessly proven its utility in countless settings, it is normal for pragmatists to conflate utility with reality.
It can be argued that there is could be a tacit agreement between constructionists and post-positivists to allow each other the primacy of method and interpretation based on typical problems, and indeed much work is being done to increase cross-discipline understanding, cooperation and integration. The common assumption between these two worldviews is that the end product of such effort is a measureable increase in our knowledge of the world as it is, from which we may exert more control and prediction by having reduced the amount of uncertainty by some amount. This tacit shared assumption I submit is expressed through the research practice of pragmatists who are “naturally” drawn to the mixed methods designs and practices described so well by Creswell and Plano Clark. Given the fertile and as yet only partially explored areas best suited for mixed method research it would be natural for the deeper philosophical theory or theories of mixed methods to be postponed, much as Smart, Feldman and Ethington (2006) found a willingness for researchers to revert to their preferred and more easily measured research domains and begin to neglect the messy and challenging issues of environmental factors affecting student success. One is reminded of the story of the man who’d lost his key in a dark alley but was searching for it under the streetlight because the light was better there.
I am arguing that there are situations where even mixed methods are properly and rigorously applied, and interpreted in best professional practice, that the insights may serve only to help decision-makers appreciate the vastness of what they do not understand, and better act within an uncertain environment, humble in their ignorance, yet moved to action from values and on the basis of principles informed by the best practice of inquiry.
It is my contention that in those situations described so aptly as “wicked problems” by Rittel & Webber, (1973) that a deep theory of mixed methods may be developed that is co-equal to that of qualitative and quantitative methods. I argue that mixed methods not only are useful in solving less-than-wicked problems, as described by Creswell and Plano Clark, but most appropriate to engage with uncertainty and complexity for the express purpose of appreciating deeply the current situation. The deep theory of mixed methods I anticipate would require explicit inclusion of all 4 world views, since there is no a priori basis for excluding any of the 4. I thinkit quite likely that a reasonable assumption of a deep theory of mixed methods in fact could require an explicit inclusion of the best practices of each world view in some fashion, details to be determined, of course.
The shift in epistemological perspective seems important to me, and which should be developed in tandem to the directions for improvements in design and pure method described by Creswell and Plano Clark. Checkland’s application of soft systems methodology, artfully describes “learning towards success” in a satisfying way (Checkland & Poulter, 2006).
The best expression of the theoretical stance towards irreducible complexity intersecting the human need for the state of nature or through any objective criteria (Boje, 2001) and in the work of Hayden White (1987) concerning the relationship between narrative discourse and the historical representation. These insights are causing me to reflect deeply on my own essentially pragmatic worldview and its underlying assumptions, and lead me inevitably back to the proposition that we need the methodological theory of mixed methods developed simultaneously with that of its design choices and specific methods.
References
Argyris, C. (2008). Teaching smart people how to learn. Boston, Harvard Business Press.
Astin, A. (1999) Student involvement: A developmental theory for higher education. Journal of College Student Development (Sep/Oct, 19990 (Vol 40, No 5)
Beltyukova, S. & Fox, C. (2002) Student satisfaction as a measure of student development: Towards a universal metric. Journal of College Student Development (Mar/Apr 2002)
Checkland, P. & Poulter, J. (2006). Learning for action: A short definitive account of soft systems methodology and its use for practitioners, teachers, and students. West Sussex, England, John Wiley & Sons Ltd.
Clark, M. & Rossiter, M. (2008). Narrative learning in adulthood. In S. Merriam (Ed) Third Update On Adult Learning Theory (pp 61-70). San Francisco, Josey-Bass.
Creswell, J. (2009). Research design: Qualitative, quantitative and mixed methods approaches (Third ed.). Thousand Oaks, CA: Sage Publications, Inc.
Dorner, D. (1996). The logic of failure: Recognizing and avoiding error in complex situations. New York, Metropolitan Books.
Edmondson, A., and McManus, S. (2007). Methodological fit in management field research. Academy of Management review 32: 1159-1176
Fenwick, T. (2008). Workplace learning: Emerging trends and new perspectives. In S. Merriam (Ed) Third Update On Adult Learning Theory (pp 17-28). San Francisco, Josey-Bass.
James, E. A., Milenkiewicz, M., Buchnam, A., (2008). Participatory action research: Data driven decision making for school leadership. Thousand Oaks, CA: Sage Publications
Kearsley, G. (1997). The Virtual Professor: A Personal Case Study. Retrieved from http://home.sprynet.com/~gkearsley/virtual.htm
Lincoln, Y. (1983) Expectancy theory as a predictor of grade-point averages, satisfaction and participation in the college environment. Annual meeting paper, Association for the Study of Higher Education, 1983.
Long, K. (2009). Participatory Action Research pilot study notes. Ft Leavenworth, KS: CGSC (unpublished).
March, J. (1994). A primer on decision making: How decisions happen. New York, The Free Press (Simon & Schuster Inc).
Prasad, P. (2005). Crafting qualitative research: Working in the postpositivist traditions. Armonk, New York, M. E. Sharpe.
Rittel, H., and M. Webber (1973). Dilemmas in a general theory of planning. Policy Sciences, Vol. 4, pp 155-169.
Schon, D. (1987). Educating the reflective practitioner. San Francisco, Jossey-Bass.
Senge, P., McCabe,N., Lucas, T., & Kleiner, A. (2000). Schools that learn: A fifth discipline fieldbook for eductaors, parents and everyone who cares about education. New York, Doubleday.
Siemens, G., (2005). Connectivism: A learning theory for the digital age. International Journal of Instructional Technology and Distance Learning 2 (10), 2005.
Smart, J., Feldman, K., & Ethington, C. (2006). Holland’s theory and patterns of college student success. Commissioned report for the national symposium on postsecondary student success: Spearheading a dialog on student success
Taylor, K. & Lamoreaux, A. (2008). Teaching with the brain in mind. In S. Merriam (Ed) Third Update On Adult Learning Theory (pp 49-60). San Francisco, Josey-Bass.
TRADOC Pam 525-5-500. (2008). Commander’s Appreciation and Campaign Design.
Trochim, W. Research methods knowledge base (2006). Types of designs. Retrieved from http://www.socialresearchmethods.net/kb/desdes.php
Waldrop, M. (1992,2008). Complexity: The emerging science at the edge of order and chaos. New York, Simon & Schuster, Inc.
Watkins, T, (2009). Consumer theory in economics. San Jose State University Economics Department website. Retrieved from http://www.sjsu.edu/faculty/watkins/constheo.htm.
White, H. (1987). The content of the form: narrative discourse and historical representation. Baltimore, The Johns Hopkins University Press.