How we change what others think, feel, believe and do
Coding is an important technique in qualitative research such as anthropology, ethnography and other observer and participant-observer methods. In summary:
The basic objective of coding is to extract key words and other 'meaningful chunks' from the haystack of data that will allow a grounded theory to be developed. The individual codes thus provide both inspiration and verification.
A grounded theory is connected to the real world and gives explanation for patterns of behavior that are relevant and important for the people involved. Through discovery of significant categories and many relationships it becomes dense and saturated and hence theoretically complete.
The process of identifying codes, categories and theories is usually long and exhaustive and can include repeated word-by-word analysis.
Note that the overall process should be a slow one. Leaping to meaning, categories and conclusions can block out much more meaningful discoveries that may emerge if given the time and open opportunity. Chance discovery can give great insight if you are ready for it at every turn.
Data that provides the material from which codes are extracted is often largely based on observer notes, logs, diaries, etc. Additional data may also be found in items such as published and unpublished documents, papers, books, public records, letters, photographs, videos and assorted artefacts.
The problem with data is that the more you have the more effort it required to analyze, and with time increasing sharply with the amount of data. Yet more data leads to better categories, theories and conclusions. What is 'enough' data is subject to debate and may well be constrained by the time and resource the researcher has available.
Deciding when and where to collect data can be a critical decision. A deep analysis at one point may miss others, whilst a broad brush may miss critical minutiae. Several deep dives can be a useful method.
Data can be difficult to access, for example when political forces oppose potentially critical investigations or where rituals are considered secret. Ethics, confidentiality and determination can all play a part here.
There is often less division of activity phases in qualitative research, and the researcher may be memoing and coding as they go.
Theoretical sampling is an approach to identifying what data is required next, based on the analytical findings so far. In this way, data collection is managed and controlled by the emerging theory.
Selective sampling occurs where the researcher decides to sample in a particular locale or seek particular types of people. Again, this is driven by rational thought rather than convenience or bias.
Selection of data and cases for exploration can be based on one or more of three purposes:
The types of data selected in theoretical sampling often needs to be be varied and is based on what is called 'slices of data' -- samples of many different kinds and sources rather than a focus on one area to the exclusion of many others.
Coding starts with open coding, in which codes are identified without any restrictions or purpose other than to discover nuggets of meaning.
The main secret of open coding is a mental openness that allows for the discovery of the unexpected along with a curiosity that does not allow for final closure, even after texts have been read and codes identified from it. Coding is thus a very questioning activity (see the Kipling Method).
Open coding is particularly about labeling and categorizing of phenomena. This must be a careful activity as names come with many connotations.
The constant comparative method may be used by constantly comparing each piece of data with codes and notes already identified. Comparison helps identify distinct characteristics and ordinal position on any relevant scale.
It is a trap to worry about the 'real meaning' of words, as this is a form of closure; open coding is about opening up lines of inquiry.
Theoretical saturation is achieved when no further new codes or categories are being identified. Further analysis then only goes to test and support the identified theory.
Coding can be quite a tedious activity but it requires expertise. The quality of categories and theories depends on the quality of the coding. Employing others to help coding can very much speed the activity, but they do need to know what they are doing.
Memos are theoretical notes that occur to the researcher as they are coding and may at some time lead to the discovery of categories and may cause the researcher to go back to the data to explore more.
Memos may identify concepts, half-formed ideas, action notes and other thinking that is a first step towards making cohesive sense from the data.
Note that ideas and meaning can be identified at any time, including when the work is not formally under way. When the mind is saturated with data, it can come up with ideas at the most inopportune moments. The researcher thus always carries note-taking equipment everywhere.
Memos, categories and codes may be sorted at any time, looking for relationships between them and priorities of the people involved when they need to make choices. This is also called data ordering.
Priorities often emerge, much like other information, in an unstructured way as the subconscious realizes patterns after a long wading through the data.
Items may for example be written on post-it notes or cards and moved around to form clusters or 'clouds' that can turn into categories or allow new relationships to be found.
'Category folders' may also be kept, containing clippings and codes on items that support a single category.
A critical aspect of coding is the identification and naming of categories, such as 'greeting people' or 'vehicle breakdown'.
Codes that lead to discovery of a 'greeting' category might come from observation of encounters with other people in which particular rituals and significance is identified.
Categories can also include sub-categories, such as 'shaking hands' or 'removing the wheel'.
Categories can include such as:
Naming of categories is important as this gives a handle by which the category can thought about and discussed. Names come with previous meanings, so their choice is very important. For example, 'rain', 'raining' and 'precipitation' have a different impact on thoughts. Additional words may create a more precise phrase, such as 'heavy rain' or 'sudden showers'.
The core category in a coding exercise is the central code or principle around which other codes cluster. There is often one core category, although there can be more.
For example a study of commuters might conclude the core category is 'going to work' and a study of teenagers could settle on 'growing up'.
A core category should:
The core category gives central meaning to the conclusions of the research and is often the 'holy grail' that the researcher is seeking. It is the main theme of the situation and may represent a central problem or issue for the people involved.
Once the core category is identified, then other related categories can be linked to it, leading to an integrated and coherent explanation of the subject of research.
Axial coding occurs where there is a strong focus on discovering codes around a single category, for example looking for interactions, strategies and so on that relate to the category. For example in a category of 'greeting', there may be a search for encounters with others, talk about previous encounters and emotional impacts from meeting others.
Axial coding can also be used to develop categories, seeking relationships that will expose a category. Where open coding is about identification and naming, axial coding is about links and relationships.
Strauss and Corbin (1990) identify a Paradigm Model by which coding looks for:
Axial coding may be done at any time, even before firm categories are identified, for example when a code of 'rain' is first encountered, then an exploration of the impacts and importance of rain may ensue.
Axial coding also helps identify relationships between categories and the links that create a web of meaning for the people under study.
Selective coding is even more focused as it works around the core category, looking specifically for links to it and how it may or may not be the heart of the matter. This particularly helps with integration of categories.
When theory is being developed, 'triangulation' is finding a third element outside of the cause and effect items to corroborate apparent relationships. Where possible, quantitative data may be sought to triangulate qualitative findings.
Glaser, B. and Strauss, A. (1967). The discovery of grounded theory. Chicago: Aldine
Strauss, A. and Corbin, J. (1990). Basics of Qualitative Research, Newbury Park, CA: Sage