Describing The approach of representational to coding data
The aim here is to represent the surface content fairly so you often go for key words in what has been said as the core concepts in the coding categories. A representational code list is produced.
Representational approach to coding data just tries to summarise the manifest content of replies.
Constructing a code list more closely resembles art than science – it is a matter of interpreting what is there, systematically but not rigorously – and a great deal of personal judgement and preference is brought to bear on it.
This is another area where validity would be best served if we used more than one judge to devise the code list and assign the items to it, so that it does not reflect only the personal preferences of one researcher.
Irrespective of the mode of coding adopted, you may be able to use a machine to help you ‘qualitative analysis’ computer software works by assigning numeric codes to themes or topics, at the instruction of the researcher, and some packages (for example, NVivo) will allow you to copy off the indexing system as a file of number which may be merged with your file of precoded variables.
This does not in any way lessen the element of personal judgement involved – it is still the researcher, not the machine, who makes the choices – but where large data files are involved it can substantially decrease the amount of time and effort involved in making full use of the ‘office-coded’ variables and allow you to a more through job of the classification.