September 27, 2007 Chapter 2: Examining Preliminary Considerations This chapter looks at mixed methods paradigms, elements of quant and qual research, and types of research problems best addressed by mixed methods. The terms “worldview” and “paradigm” address how we look at the world and therefore how we conduct our research. Different Worldviews or Paradigms “…[B]ehind each study lies assumptions the research makes about reality, how knowledge is obtained, and the methods of gaining knowledge.” Four worldviews are shown in Table 2.1. I used blue text to highlight elements that might apply to the test error analysis research. Table 2.1 Four Worldviews Used in Research Postpositivism - Determination - Reductionism - Empirical observation and measurement - Theory verification Constructivism - Understanding - Multiple participant meanings - Social and historical construction - Theory generation Source: Creswell (2003) Advocacy and Participatory - Political - Empowerment and issue oriented - Collaborative - Change oriented Pragmatism - Consequences of actions - Problem centered - Pluralistic - Real-world practice oriented They say the worldviews “have common elements, but take different stances on” them in terms of the ontology (nature of reality), the epistemology (how we learn what we know), the axiology (roles of values in research), the methodology (process of research), and the rhetoric (language of research). Here, “methodology” is a process rather than a “philosophical framework”. The methodology associated with constructivism later in this chapter is severely limited and indicates a lack of understanding of a constructivist view of the research process. Postpositivist research takes a “top down” approach from theory to an experiment that confirms or denies it. Constructivism takes a “bottom up” approach using the participants’ views to build broader themes and generate a theory interconnecting them. Clearly, my study is on the constructivist side of this continuum. I agree. In fact, you and I had assumed that my study stemmed from a constructionist epistemology. (Are you intentionally distinguishing between “constructionist” and “constructivist” or was this a typo?) Oops! Typo! Thanks for catching it! This Markup setting is weird and will take some getting used to! However, based on what they’re saying, the pragmatism seems to fit as well, with hints of the advocacy/participatory paradigm. Quoting this excerpt from earlier issues of my study plans, I figured that …[the] test error analysis tool should help students place meaning on their graded tests and turn the assessment process into a learning process and meta-cognitive aid. The tool allows them to interpret their testing errors, analyze the results, and make plans to minimize error recurrence. In turn, the tool’s documentation of the students’ cumulative assessment could help them learn about their strengths and their areas for improvement and use that knowledge in a powerful way. I used the theoretical framework shown below to provide a model for the research: Student Intra-personal Interaction Student Parents Teacher Student Meta-cognitive Loop with Triangle of Student/Parents/Teacher Interaction This framework and its interpretivist perspective reflect the meaning the students place on their analysis and their development of personal solutions to promote their own improvement. This is represented by the student’s metacognitive loop at the top. The triangle at the bottom represents the communication patterns between students, parents, and teachers. The test analysis tool supports student analysis and communication at a detailed level to and from parents and to and from the teacher as well as between parents and the teacher. Secondarily, meta-cognitive loops could be added for the parents and teacher showing any guidance gained by the tool for use in parenting and teaching processes. Note that the highlighted components supporting constructivism in Table 2.2 are: understanding, multiple participant meanings, social and historical construction, and theory generation. It seems to me that from a constructivism perspective, the tool should promote student understanding of patterns and trends in addition to teacher and parent understanding of student performance. The very foundation of test error analysis is that teachers can’t always determine the root causes of student errors without additional student input, thereby recognizing possible multiple participant meanings. Students are reflecting back on errors they made, in an exercise of historical construction. Student brainstorming sessions to provide advice on overcoming errors allow for social construction. Furthermore, the entire study might evolve into a bit of theory generation, if I correctly understand the meaning of any of these phrases. (If not, please enlighten me!) However, then I look at the advocacy/participatory column, emphasizing political, empowerment and issue oriented, collaborative, and change oriented. Although I don’t see political ramifications to the research, the process empowers students to analyze their errors and learn from them, which may not be a major issue in the grand scheme of things, but perhaps a minor one. It’s collaborative in three ways: (1) the students collaborate to come up with ideas for preventing the errors, (2) they can tailor the tool to their needs with “other” error types, and (3) the study may involve other teachers who choose to use the tool. If the tool proves to be effective, it could encourage change. Furthermore, consider the pragmatism paradigm. Its bullets are: consequences of action, problem-centered, pluralistic, and real world practice oriented. Using the tool, the students analyze consequences of actions (points lost for errors). It is problem-centered in that each error is a problem and the problems are analyzed by the students to educate the teacher and parents. It seems pluralistic in that different students can make the same errors for different reasons and the tool helps to distinguish between those differences (for example, if the student left a question on a test blank, was it because they accidentally omitted the question, ran out of time, didn’t understand the question, or what?). Finally, it seems to be real-world and practice oriented because the testing is very real to the students and the teacher. So, if I interpret the components of these worldviews correctly, it spans constructivism, advocacy/participatory (though maybe not as much), and pragmatism. At first I was thinking, am I missing something here? Is the study really straddling these three worldviews? Do I have to pick one? Or can the study be mixed in this way, too? Read on to see how we can resolve this discomfort. I think the aspects of pragmatism and advocacy/participatory paradigms you associate with your proposed research design can fit within a constructivist world-view. In other words, referencing aspects of these three world-views can be consistent with an overarching view of how knowledge is constructed that would fit within the “constructivist” framework. Thank you for making sense of this stuff! Table 2.2 on the next page further clouds my thinking. I can link elements of all the blue text rectangles to my study, and arguments could be made for some of the others. Worldview Element Ontology (What is the nature of reality?) Postpositivism Singular reality (e.g., researchers reject or fail to reject hypotheses) Epistemology (What is the relationship between the researcher and that being researched?) Distance and impartiality (e.g., researchers objectively collect data on instruments) Axiology (What is the role of values?) Unbiased (e.g., researches use checks to eliminate bias) Methodology (What is the process of research?) Deductive (e.g., researches test an a priori theory) Rhetoric (What is the language of research?) Formal style (E.g., researchers use agreed-on definitions of variable) Constructivism Multiple realities (e.g., researchers provide quotes to illustrate different perspectives) Advocacy and Participatory Political reality (e.g., findings are negotiated with participants) Pragmatism Singular and multiple realities (e.g., researchers test hypotheses and provide multiple perspectives) Closeness (e.g., Collaboration Practicality researchers visit (e.g., (e.g., participants at researchers researchers their sites to actively involve collect data by collect data) participants as “what works” collaborators) to address research question) Biased (e.g., Biased and Multiple researchers negotiated (e.g., stances (e.g., actively talk researchers researches about their negotiate with include both biases and participants biased and interpretations) about unbiased interpretations) perspectives) Inductive (e.g., Participatory Combining researchers start (e.g., (e.g., with researchers researchers participants’ involve collect both views and build participants in quantitative and “up” to all stages of the qualitative data patterns, research engage and mix them) theories, and in cyclical generalizations) reviews of results) Informal style Advocacy and Formal or (e.g., change (e.g., informal (e.g., researchers researchers use research may write in a language that employ both literary, will help bring formal and informal style about change informal styles and advocate of writing) for participants) So…what do you think? I feel like this theory stuff really bogs me down. Is that normal?!!! You have to keep in mind that the above categorization is the interpretation that these particular authors give to these different world-views. Okay! Other researchers may differ in the way they would interpret how each world-view relates to each of these elements! I guess reading the other book next semester will give me a broader perspective. Okay, so let’s move on to the four elements basic to any research process. They reprinted Crotty’s (1998) famous table in Table 2.3 and recommended adding mixed methods research as a methodology as shown. Table 2.3 The Four Elements Basic to Any Research Process Epistemology Objectivism Theoretical Perspective Positivism (and postpositivism) Methodology Experimental research Methods Sampling Constructivism Interpretivism - Symbolic interactionism - Phenomenology - Hermeneutics Survey research Measurement and scaling Subjectivism (and its variants) Critical inquiry Ethnography Questionnaires Feminism Phenomenological research Observation - Participant - Non-participant [parents] Postmodernism, etc. Grounded theory Interview Heuristic inquiry Focus group Action research ? Case study Discourse analysis Life history Feminist standpoint research, etc. Narrative Add: Mixed methods research Visual ethnographic methods Statistical analysis Data reduction Theme identification Comparative analysis Cognitive mapping Interpretative methods Document analysis Content analysis Conversation analysis, etc. I really take issue with this table. To list “survey research” and “measurement and scaling” as the only methodology and methods related to constructivism indicates a total lack of understanding of constructivism. I would argue that constructivism could embrace every methodology listed in this table, as well as some that are not listed (e.g. teaching experiment, classroom teaching experiment). I would also argue that a constructivist research project could make use of almost all of the methods listed except for possibly the “measurement and scaling” (listed under constructivism!) and possibly the “sampling” technique, if what is meant by this is the “randomized assignment of individuals to experimental and control groups.” Thanks, again for your voice of sanity. I have found this table confusing since I took that first qual course a few years ago, and what you’re saying really helps me to find peace with these issues! Worldviews and Mixed Methods Research The authors suggest that I might convey my stances on my worldviews in a section titled “Philosophical Assumptions” or in the methods section of the dissertation. You probably can advise me better about UGA’s expectations, and I’m sure they’ll come clear to me next year in EMAT 9630 and 9640. (I hope so!) There are three different stances discussed in the mixed methods literature. Stance 1. There is one “best” worldview that fits mixed methods research. Tashakkori and Teddlie indicate that at least 13 authors dub that to be pragmatism. Reasons are: 1. Both quant and qual research methods may be used in a single study. 2. The research question should be of primary importance --- more important than either the method or the philosophical worldview that underlies the method. 3. The forced-choice dichotomy between postpositivism and constructivism should be abandoned. 4. The use of metaphysical concepts such as “truth” and “reality” should also be abandoned. 5. A practical and applied research philosophy should guide methodological choices. However, Tashakkori and Teddlie also mention one other “best” worldview: the transformative-emancipatory paradigm, another term for the advocacyparticipatory approach. That may explain why I have four table entries from that column highlighted in Table 2.2, further muddying the water. Stance 2. Researchers can use multiple paradigms or worldviews in their mixed methods study. This “dialectical” perspective (Greene & Caracelli, 1997, 2003) recognizes that contradictions and tensions reflect different ways of knowing about and valuing the social world and says you can use multiple paradigms. Ah, I’m starting to like these people. Stance 3. Worldviews relate to the type of mixed methods design and may vary depending on the type of design. I could also agree with this stance. …further adding to the nebulosity of all this theory! Okay, so what are you thinking so far? Does it look like my study mixes constructivism, advocacy/participatory, and pragmatism paradigms? Yes, but see my remarks about this mixing being consistent with a more encompassing view of constructivism. I like the way you think! The Basics of Quantitative and Qualitative Research Table 2.4 provides a nice set of continua showing elements where quant and qual research differ in the basic intent and implementation of the research. Table 2.4 Elements of Qual and Quant Research in the Process of Research Elements of Qualitative Elements of Quantitative Process of Research Tend Toward… Research Tend Toward… Research - Understanding meaning - Test a theory deductively to Intent of the individuals give to a support or refute it research phenomenon inductively - Minor role - Major role How literature is - Justifies problem - Justifies problem used - Identifies questions and hypotheses - Ask open-ended questions - Ask closed-ended questions How intent is - Understand the complexity of focused - Test specific variables that a single idea (or phenomenon) form hypotheses or questions - Words and images - Numbers How data are - From a few participants at a - From many participants at collected few research sites many research sites - Studying participants at their location - Text or image analysis - Themes - Larger patterns or generalizations - Identifies personal stance - Reports bias - Uses validity procedures that rely on participants, the researcher, or the reader How data are analyzed Role of the researcher How data are validated - Sending or administering instruments to participants - Numerical statistical analysis - Rejecting hypotheses or determining effect sizes - Remains in background - Takes steps to remove bias - Uses validity procedures based on external standards, such as judges, past research, statistics Notice the blue text on each side indicating the tendencies of the test error analysis research. I’m thinking that the intent of my research is about understanding the meaning individuals give to their error causes and the tool/process itself (qual). There’s not a lot of closely related lit, so I see it as having a minor role (qual). The tool and its related process are a combination of open- and closed-ended questions (both). Data will be from administered instruments, as well as journals, questionnaires, and informal interviews: numbers and words from 80 – 130 students at one site (both). Data will be analyzed numerically and with themes, looking for large patterns and generalizations (both). I’ll do my best to reveal my biases and try to remove them (both). Data will be evaluated by the participants and the researcher as well as stats (both). So I figure it starts out looking “qual”-ish, but finishes with a strong “both”. Does that make sense? Yes! I would add, that from the qualitative aspect of your proposal, you might want to consider the notion of “viability” rather than “validity” (see von Glaserfeldt for a discussion of “viability” versus “validity”). I’m looking at a 2001 von Glasersfeld article that “suggests the substitution of ‘viability’ or ‘functional fit’ for the notions of Truth and objective representation of an experiencer-independent reality.” I’m not sure how this idea affects the study. Please say more! Table 2.5 matches different types of research problems with methods designs. Table 2.5 Types of Research Problems and Matching Methods or Designs Types of Methods (Designs) Type of Research Problem Suited to Studying the Problem Need to see if a treatment is effective Experimental design Need to see what factors influence an outcome Correlation design Need to identify broad trends in a population Survey design Need to describe a culture-sharing group Ethnography design Need to generate a theory of a process Grounded theory design Need to tell the story of an individual Narrative research It looks like we’re dealing with an experimental design. I’m still not sure how you are going to indicate “effective” with your proposed research. Are you tracking errors made over several test-taking sessions, and looking for a reduction in errors in the group using your TEA as compared with a (similar?) group not using TEA? I wasn’t really planning on a control group since there are so many uncontrolled variables between classes and students. What is the “treatment” in your experimental design? I was thinking that all of the students could perform test error analysis and we (students, teacher, and parents) would give our opinions of the process and its helpfulness. Error types are categorized into testing process errors and content errors. I’d like to see if there’s a reduction in testing process errors. I’m not sure it’s fair to look at the same think with content errors since the content is ever-changing and growing in complexity. Is “effective” the wrong word to use? If you are not intending to conduct a “comparison between treatment and control groups” then I would hesitate to classify your research problem as “needing to see if a treatment is effective.” It could still be “experimental” in design, but with a different kind of research problem (in other words, the above matches are over-simplified and restricting). Yes, yes, yes! In some ways I think you are trying to generate a theory of a process. I basically want to know if I would recommend it to other teachers or not and under what circumstances. What happens when students are given the opportunity to analyze their test errors in a systematic way? (grounded theory) How do students (and parents and teachers) react to this process? (ethnographic and/or narrative). You do intend to conduct an “experiment” in the sense that you will be trying out your TEA with a large number of students, … yes … but will you also collect data from a similar group of students NOT using TEA? I don’t think so. The following are situations that they say lead to mixed methods designs: 1. A need exists for both quant and qual approaches. The combination…provides a more complete picture by noting trends and generalisations as well as in-depth knowledge of participants’ perspectives. One form of [data]… might contradict the other. [Together they] can clarify subtleties, cross-validate findings, and inform efforts to plan, implement, and evaluate intervention strategies. 2. A need exists to enhance the study with a second source of data. 3. A need exists to explain the quant results (e.g., understand social interactions). 4. A need exists to first explore qualitatively. Multiple reasons might …exist, and we recommend that investigators first ask themselves what all the reasons are for using mixed methods research and then specifically state these reasons clearly in their study. I think my study falls in the first situational category. I’d like to know quantitatively the frequencies of error types and qualitatively what the participants [students, teacher(s), parents] think of the process. I agree. So this is not going to be a “treatment-control” comparison experiment. True.