Review #2 –Data coding Lee 1/6
Miha Lee
Professor Rivas
SED 600
14 March 2007
Review on the data coding:
Method of Analyzing my Qualitative Data
My tentative action research topic is “the role of the well-structured lab activities in the process of learning chemistry concepts.”The purpose of this paper is to examine the role of the well-structured experiments in chemistry and to describe conceptual change that will take place during lab activities. The questions that I want to examine are the followings:
- What are the learners’prior knowledge regarding metal?
- How do the learners’concepts change during lab activities?
- Is students’engagement in experiments effective in promoting conceptual change?
Thus, my data will consist of the documents collected from the pretest and the posttest of learning objectives in the unit metal, survey of students’attitude toward POE method, the lab papers. Those tests and lab papers will be carefully designed based on the conceptual analysis of the textbook in order to probe the students’prior knowledge and to monitor their conceptual change over the lab activities. Hoepfl (1997) points out that qualitative methodology is a powerful tool for enhancing our understanding of teaching and learning, and uses a naturalistic approach that seek to understand phenomena in content specific settings. Therefore, for my action research, qualitative approach will be taken by employing an array of open-ended questions in the documents to gain better insights about students’learning in a classroom setting. According to Johnson (2007), open-ended questions allow respondents unlimited choices, and provide a more accurate sense of what respondents are actually thinking.
However, problems with open-ended questions are the analysis of them. For the beginner researcher, qualitative data seems so messy that the researcher can be puzzled and frustrated with the undertaking to analyze it. This is the same reason why some teachers avoid using open-ended questions in their assessments. Nevertheless, I prefer to take the qualitative approach to my action research in order to get a wide range of information and insights from the data.
Bogdan and Biklen define qualitative data analysis as “working with data, organizing it, breaking it into manageable units, synthesizing it, searching for patterns, discovering what is important and what is to be learned, and deciding what you will tell others.”(1982, p 145) Qualitative researchers tend to use inductive analysis of data, which means that some generalizations emerge out of the data. (Johnson, 2007; Hoepfl, 1997; Mertler, 2006) The process of coding and interpreting data is a critical part of qualitative research because this process requires the researcher’s creativity. (Foss & Waters, 2003) It is challenging to place the raw data into logical and meaningful categories, to examine them in a holistic manner, and to find a way to communicate this interpretation to others. (Hoepfl, 1997)
For inductive data analysis, there seem to be several steps to follow. Coding data, developing themes, and interpreting the themes (Foss & Waters, 2003); organization, description, and interpretation (Mertler, 2006); open coding, axial coding, and translating (Hoepfl, 1997). The point is whether these steps are in a linear fashion or not. The answer to this question is ‘No’. The research may do these steps simultaneously and repeatedly. (Hoepfl, 1997; Mertler, 2006; Gough & Scott, 2000)
However, the most important thing to take into consideration before those steps is developing quality research questions. Research questions reflect the researcher’s ideas about “what I am looking for in the data?”(Foss & Waters, 2003) Thus, the specificity and validity of the research questions must be considered in advance. For example, in order to probe students’prior knowledge about metal, I will make a question for the pretest of the unit Metal, asking “What is the definition of metal in terms of chemistry?”instead of “What is metal?”The question “What is metal?”is too broad and ambiguous for the students to realize what they are asked. The answers to the question, thus, may be too wide and various to be categorized into a few groups. They could be a description of appearances, physical properties, chemical properties, and uses in everyday life. My imagination limits the possible categories for this question, but the students’imaginations will have no limit in the research. In contrast, the other question,“What is the definition of metal in terms of chemistry?”seems to be difficult for the students to answer in the pretest, and thus they may generate only few answers. However, in the posttest, I expect the students to give me more specific and correct answers to the question, which is the sign of learning. That’s what I am looking for in my research.
For the first step of the data analysis, the researcher has to read and reread writing down the emerging categories in a form of a paraphrase, phrase, heading or label that describes what the respondents try to say and the researcher think of important. (Foss & Waters, 2003; Mertler, 2006) This process is called data coding. Basically, data coding is the process of data reduction.(Mertler, 2006) Qualitative data are made up of a wide variety of text passages or segments. In itself, it is hard to provide useful information to the readers and even to the researcher. It requires the researcher to reduce the volume of information collected in order to identify and organize the data into important patterns and themes. (Mertler, 2006) Patterns are the same or similar ideas or expressions in data. Themes may include knowledge, beliefs, experiences, or opinions that the respondent was trying to communicate in response to the researcher's questions. (CDC EZ-Text, 2000) Thus, the identification of key themes and patterns depends on data coding process.
After finding some categories through reading data, the researcher should develop a coding scheme that is used to group data. (Mertler, 2006) And then read the data again, and code them carefully. The researcher should code the data not with his/her assumptions or preconceptions but with his/her observations and insights. The researcher should try to code data as a complete novice—someonewith no extra knowledge, assumptions, or valuesbesides what's in the data so that the research makes sense to others. (Foss & Waters, 2003) Also, a code scheme should identify data according to their speaker and the context. (Hoepfl, 1997)
The next stage of analysis involves reexamination of the categories identified to find out links or relationships among them. Those links help the researcher figure out ‘the big picture.’The purpose of coding is to not only describe but, more importantly, to explain a phenomenon of interest. (Hoepfl, 1997) This interpretation of data is the most important step to make sense of data. The ‘grounded theory’(Glaser & Strauss, 1967) provides a framework for the interpretation of qualitative data. According to Gough and Scott (2000), a ground theory is a method in which categories for the data coding are derived from the data itself, and in which emphasis is placed on the discovery and elucidation of links between categories. The procedure of ground theory consists of several steps as followings. (Shwartz, Ben-Zvi,and Hofstein, 2005)
1. Segmenting each document into units, and categorizing every unit by itscontent. Primary categories emerged from the collected data.
2. Developing more general domains by integrating related categories into a domain in terms of the research topics.
3. Mapping all documents according to the chosen domains.
4. Looking for the foci: reorganization of the data according to the chosendomains.
5. Proposing assertions based on the accumulated data, which will hopefullycontribute to answering the research questions.
Yet, many qualitative research problems have two distinct aspects, each of which has different implications for the data coding. One of these aspects is likely to be emic in nature, to be centrally concerned with the discovery the meanings attributed by respondents, and to focus on the specific research context. The other is likely to be epic in character, to concern itself with the interpretation and presentation of data in ways likely to be found meaningful by audiences outside the immediate research context, and to focus on more generalization. (Gough & Scott, 2000)
Both two aspects should be considered in any qualitative research analysis, but for my action research, which is a case study, the former aspect seems to be more focused on because the task of case study is to produce ordered reports of experience which invite judgment and offer evidence to which judgment can appeal. (Stenhouse, 1988) Therefore, I will take two approaches toward the analysis of my action research data.
One is focused on the objective nature of the research. After coding the data of each question in the documents, I will create a table of code scheme to organize the categories of each question to find out the overall tendency across the students. The tables consist of categories, brief description of them, dominant examples, and the frequency (the number of data in each category) to provide audience with meaningful information about the students’prior knowledge concerning metal. This process can be seen as a transformation of qualitative data into quantitative data. Quantitative analysis facilitates both the researcher and audience in obtaining an overview or flavor of densely packed qualitative data. (Gough & Scott, 2000) The tables will show the distribution among the categories of qualitative data. The lists of prior knowledge may have some meanings in itself, but when the frequency is provided, the teacher and audience can find out the dominant obstacles to the promoting conceptual change in teaching practices. Furthermore, with the comparison between the pretest result and the posttest result, or between predicts and explains in the lab papers, I can determine the effectiveness of the lab activities in promoting conceptual change.
On the other hand, I will generate ‘a loose network’, which is suggested by Gough & Scott (2000), to represent the flow of thought for each student. Miles & Huberman (1994, p.94) define ‘a network’as ‘a collection of “nodes”or points connected by lines.’The use of network in the analysis of qualitative data are extended and refined by Bliss at al (1983).
Networks can usefully be regarded as an extension of the familiarbusiness of putting things into categories. To categorize is to attach alabel to things; in effect to place them in boxes. A network can be seenas a map of the set of boxes one has chosen to use, which shows howthey relate to one another. (Bliss et al., 1983, p. 8)
One of my research purposes is to monitor the students’conceptual change over time. To achieve this goal, the requirement is to record and display ideas that will arise in students’written data, and links between them. Concept mapping is useful to detect students’prior knowledge and follow the conceptual change, but it takes a great deal of time to get used to drawing them. Thus, the loose network will be concept maps that are drawn by the researcher based on the data, in which coded data will be signified by a word or words the students actually use. If a student’s own words are compressed into acategory signified by a word supplied by the researcher, care will be taken to:
ensure the appropriateness of the chosen signifier (by, for example,comparing that student’s words with those used by other students whoemploy the chosen signifier);
employ as many other categories as seem necessary to convey the fullcontent of that student’s words;
ensure consistency in the making of such decisions by critical selfexamination.
(Gough & Scott, 2000)
The analysis of qualitative data named data coding seems to be complicated and challenging, but the inductive analysis and other methods provide the process I can use in my action research. Coding data and interpretation of links among categories are a simultaneous and complement work, and decide the quality of the research. It will take a lot of time and endeavor to finish the analysis. To save time and effort, I need to work on refining the research questions and developing appropriate documents for collecting data. And I need to find more literature that suggests good ideas for my data analysis.
Reference
Bliss, J., Monk, M. & Ogborn, J. (1983) Qualitative Data Analysis for Educational Research, London; Croom Helm
Bogdan, R. C., & Biklen, S. K. (1982) Qualitative research for education: Anintroduction to theory and methods. Boston, MA: Allyn and Bacon, Inc.
CDC EZ-Text, (2000) chapter 6. Overview of data coding in EZ-Text, User's Guide: CDC EZ-Text developed for Centers for Disease Control & Prevention, Atlanta, Georgia, Conwal Incorporated, McLean, Virginia Retrieved March 8, 2007 from Updated: August 14, 2000
Foss, S, K. Waters, W. (2003) Coding & Analysis of Qualitative Data (February 6, 2003), All-but-dissertation guide, Retrieved March 8, 2007 from
Glaser, B.G. & Strauss, A.L. (1967) The Discovery of Grounded Theory, Chicago, IL. Aldine
Gough, S. & Scott, W. (2000). Exploring the purpose of qualitative data coding in educational enquiry: Insights from recent research. Educational Studies, 26(3), 339-354.
Hoepfl, M. (1997) Choosing qualitative research: A primer for technology education researchers, Journal of technology education, 9(1), 47-63
Johnson, A, P. (2007) A Short Guide to Action Research (3rd edition), Boston, MA: Pearson Allyn and Bacon.
Mertler, C. A., (2006) Action research: Teachers as researchers in the classroom. Thousand Oaks, CA: Sage.
Miles, M.B. & Huberman, A.M. (1994) Qualitative Data Analysis: an expanded sourcebook, 2ndedn, Thousand Oaks, CA: Sage
Shwartz, Y., Ben-Zvi, R., and Hofstein, A. (2005) The importance of involving high-school chemistryteachers in the process of defining the operationalmeaning of ‘chemical literacy’, International Journal of science education, 27(3), 323–344
Stenhouse, L. (1988) Case study methods, in: J.P. KEEVES (Ed.) Educational Research, Methodology,and Measurement: an international handbook, pp. 49–53 , Oxford, Pergamon