Line by line coding is definitely the standard for initial or open coding in Grounded Theory. As your coding develops in a more focused direction, you still need to pay just as much attention to the data as you do in the beginning, but with a different set of goals in mind (i.e., the creation of a relatively small set of theoretical categories).

Also, if you use your initial analytic insights to modify either the questions you ask or the people choose to interview (both of which are highly recommended in GT), then you should use the same initial or open coding procedures that you would with any "new" source of data.

Line by line coding is definitely the standard for initial or open coding in Grounded Theory. As your coding develops in a more focused direction, you still need to pay just as much attention to the data as you do in the beginning, but with a different set of goals in mind (i.e., the creation of a relatively small set of theoretical categories).

Also, if you use your initial analytic insights to modify either the questions you ask or the people choose to interview (both of which are highly recommended in GT), then you should use the same initial or open coding procedures that you would with any "new" source of data.

David is correct Diya. Also - if you want to claim theoretical saturation then, for rigour, you would have to state that you have used the same coding process throughout.

I agree, however, I would collect empirical material during a longer time period and might also collect additional interviews in rather late stages of my analysis. I might code inductively and create analytical memos for a longer period. The when I get more focused (have gone through several rounds of coding) I may also direct my sampling towards very specific objects of observation - in an attempt to find boundaries and confirmation of my findings. Therefore, the coding approach can also change. May I suggest readings by Charmaz as a compliment to what you study.

In regard to line-by-line coding, the illustrations/guidance as per the following publications may further help:

• Bryant, A. and Charmaz, K. (2007) The SAGE Handbook of Grounded Theory. London: SAGE Publications Ltd.
• Charmaz, K. (2006) Constructing Grounded Theory: A Practical Guide Through Qualitative Analysis. London: SAGE Publications Ltd.
• Glaser, B. G. and Holton, J. (2004) Remodeling grounded theory, Forum: Qualitative Social Research, 5, 2, Art. 4, pp. 1-22.
• Holton, J. A. (2008) Grounded Theory as a General Research Methodology, Grounded Theory Review, 7, 2, pp. 67-93.

Hi Diya, as Dr. Morgan and Dean said line by line is a necessary step in a grounded theory approach for having the opportunity to elaborate a theory due to each step enrich the data processing.

Glaser's prescription of line-by-line coding, in order to keep interpreted concepts relevant to one's data, is probably the principal criterion.

I don't agree with David at all. Whether line-by-line coding is necessary depends - as all other things - on the research question and thus on the purpose of coding. The open nature of interviewing and the fact that documents we might code were written for particular purposes in the field under study make it possible that statements in texts are not data at all but noise. Coding noise is not only inefficient but might even distort your analysis if you are going to develop themes from data.

For example, if you code interviews in order to build a theory of how people give meaning and respond to certain events in their lives, you would certainly want to do line-by-line coding because you need to identify latent meaning, which requires you to analyse everything you have. Any utterance and any pause constitutes data. If, however, you code data for some kind of causal analysis, you would not want to code an interviewee's rant about something entirely different, which you let happen for the sake of the interview to continue on good terms. More generally speaking: If your interviewees are your research object, everything about them matters. If they are informants, some things they said might not constitute data depending on the purpose of your investigation.

Thus, even if we use coding for the narrower aim of buidling grounded theory from data, we must take into account that these theories can take different forms andhave different purposes, and your handling of data depends on how clear you can formulate your aim prior to data analysis. You still have to look at all your material in order to decide what is relevan tand what isn't but if you decide something does not constitute data for your investigation, you do not need to do line-by-line coding.

There is an inherent tension between the aspiration of the inventors of the Grounded Theory approach for the approach to be a general model of qualitative research and some rather narrow methodological prescriptions. This tension becomes even stronger for all the cases in which coding is borrowed from GT and used for purposes other than 'building theory from data'.

Jochen

I sort of agree with you; my conception of LbyL coding is simply to look at all your data for what seems relevant to a particular study; there may indeed be a lot of 'noise' which on consideration, you don't code.

Jochen Glaser, my answer was entirely specific to the issue of line-by-line coding in Grounded Theory, which was the subject of the original question. The role that such coding plays in qualitative research in general is another matter, but I certainly stand by what I said as it applies to the explicit standards in Grounded Theory.

Thanks for all those valuable insight, I am also a novice in qual research using GT. At first I did not use the line by line coding but I think I need to revisit my pilot interviews and do it to get more clear initial codes. Is this advisable?

As I said in my answer to Jochen, above, just code what seems, on consideration, relevant to your study as you look at all your data.

Remember, the essence of coding in GT is its literal relevance to the data.

Hisham Yaacob Can you say why you feel a need to revisit your earlier data?

I did the initial coding by paragraph using Nvivo and got around 30 codes from 5 interviews...I have not merged or restructure the codes yet, but I thought by paragraph then I might missed some important themes...anyway, it pretty fast with the Nvivo, so I dont mind to redo it. What do you say? Thanks in advance.

Hi Hisham Yaacob ,

It is also important take care with the differences among the data analysis considering the different methodological perspectives of the Grounded Theory.

I have this article comparing and discussing the differences: http://www.scielo.br/scielo.php?script=sci_arttext&pid=S0080-62342018000100600&lng=en&nrm=iso&tlng=en

Good luck with your research!

Thirty codes from five interviews would be very low for the usual kind of open coding associated with GT.

As it seems you have received a bit of different responses, you can see how grounded theory (along with other designs) require some in-depth thought and description. As you move forward, you then have the chance to explain to your readers (and other audiences) why you made each of your decisions and in this way, you provide a rationale throughout justifying your thoughtful approach. Good luck!

Dear A. S. CohenMiller David L Morgan and José Luís Guedes dos Santos

many thanks for wonderful ‘eye-opening’ tips...thanks 4 the wishes 2...

Some of the issues here involve the extent to which you have done "authentic" grounded theory coding. As A. S. CohenMiller notes, this can vary quite a bit from one version of grounded theory to another, but I think Charmaz provides some of the most useful examples in her book on Constructing Grounded Theory (2nd ed., 2014). So, take a look at her illustrative coding and think about how useful it would be for your purposes.

Along with that, think about whether you really need to use grounded theory as a label for you are doing. In particular, it sounds like you collected all your data before you started analyzing it, which is a better fit with Braun and Clarke's (2006) thematic analysis. So, take a look at the six steps proposed by Braun and Clarke and see how well that would fit would your analysis.

Ultimately, the goal is to develop a set of themes that communicate with your readers about what you learned from your analysis, and it doesn't really matter all that much which technical method you used to accomplish that.

Thanks for all the insightful thoughts!

As my coding going further, I gradually find that this question is becoming less confusing. I recognized several important concepts and have been trying to narrow my focus during the process, and I found it easier to tell noises from important information now. But to fully overcome this problem, I may need to take a break and read some methodological instructions again which I can't understand pretty well before. Thank David L Morgan for sharing the helpful materials.

It's always helpful to take a break...'chill'...mull things over.