Contents - Previous - Next

DATA ANALYSIS FORM 7.2
Ranking of Practitioners (sample)

Illness Term: Safed Paani

DATA ANALYSIS FORM 7.3
Tabulation Sheet for Practitioner Characteristics
(sample)

Illness Term: Safed Paani

Unit 8. Preliminary Analysis and Devising Coding Scheme for Textual Data

Purposes

This unit provides trainees with the opportunity to evaluate data collected in the first seven units. It is intended:

1. To continue the process of analysis of the data set.

2. To develop a coding system for the textual data.

3. To identify areas of further research and areas which are no longer fruitful within each of the three topic areas.

4. To evaluate which research methods worked well and which should either be dropped or modified.

Suggestions for Preliminary Analysis

1. Read through the field notes and tabulated data several times thinking of one topic area each time. Keep a notebook and make notes about anything that strikes you as important, being sure to include the interview date and number.

2. Which of your initial hypotheses or ideas have been confirmed? Are you sure they have been confirmed? During the next phase, constantly search for examples which disprove these hypotheses. Which have not been confirmed?

3. What new hypotheses have come to mind? Think about how they could be tested in the next phase.

4. Does one data source contradict another? Are there any clues as to why there is a discrepancy? Keep careful notes on these discrepancies as they must be addressed in the next phase.

5. Which methods worked well and which did not? Of those that did not work well, can they be modified or should they be abandoned? If abandoned, what other methods could be used to address the same topics?

Microcomputer Application (optional)

Ethnographic field notes can be coded rapidly using a word processing program such as Word Perfect 5.1, through the creation of macros (Appendix C-2).

Coding consists of reviewing field notes to search for key cultural concepts and then inserting a code into the text to facilitate retrieval.

Suggested Process for Manual Coding (Using a Set of Preliminary Codes)

1. Develop a preliminary list of 20 to 30 topic headings (codes). (These may be partially completed during the first few days of training.)

2. Prepare a system of short labels for the topics (may be in either mnemonic (e.g., HS DO) or numeric format).

3. Read through field notes from Units 1 to 7 (Units 2 and 6 in particular), and make modifications to the preliminary coding system as required. Avoid adding too many codes; 50 should be the upper limit.

For future units, coding should be a continuous process as you write out your expanded fieldnotes.

As an example, here are some of the codes developed by a NGO conducting ethnographic studies of slum women in Bombay, India:

4. Prepare code list with commonly agreed and accepted rules and definitions for use by coders.

5. Number every page of the expanded notes with a unique number. This will aid you later in the rapid and efficient retrieval of information.

6. Code every page using your modified version of the codes. Codes may be written in the margin of the written expanded field notes. next to the portion of text they refer to.

7. Coded pages must then be photocopied and one copy of coded pages should be put in the master file. One should also fill out a summary sheet and affix it to the folder.

Some Things to Remember When Coding

1. Do not collapse "emic" (local) terms into "etic" (medical) terms. That is, local cultural vocabulary should be preserved in coding, instead of putting those words/phrases into biomedical terminology (e.g., "camomile tea" should be coded as "home remedies" or "home treatments").

2. As you code the data, make a list of questions for the next phase of field work.

3. Mark "Quotables" (statements by informants which express a point or concept very well) for easy retrieval. These will prove useful in the future when you are writing reports.

4. If you have added new codes, you may need to go back and revise previous coding. This is possible when a code you have developed and been using appears to be covering a very wide set of ideas or concepts. You may wish to break the code into sub-categories.

5. Codes should be at a level of abstraction higher than the original text they refer to.

Multi-level Coding

A multi-level coding system will assist in developing coding that is both efficient and effective. Multi-level coding means developing a hierarchy, or a tree of codes to represent the different items or concepts that exist in the data and is based on the questions that you are interested in answering from the data. When examining the textual data, think about using a model and representative codes. One useful example is the taxonomic model that Scrimshaw and Hurtado (1987) illustrate to describe various forms of diarrhea treatment as illustrated below.

In this example, the component elements of the particular code indicate its general classification. When there is a number of different layers of information (usually in the form of cognitive domains) to be coded, it is useful to portray these domains' layers in the actual coding scheme. Use the symbol "_" between the particular mnemonics or acronyms representing the cognitive domains, and the depth of the layering. In this example TR_HO_DR, represents a code for the domain of drinks within the domain of home remedies within the domain of treatment of diarrhea. Note that the diagram above represents one "tree" of codes. It is not necessary for all your codes to fit in a single tree.

Additional Analytic Steps to be Done During Coding

1. Construct a dictionary of key local terms used by informants.

2. Create a visual taxonomy of your coding structure, this is especially important when your working scheme is stratified into multiple levels.

Contents - Previous - Next