Dear Johan, my advice to you is to choose one of those software applications you've mentioned and learn how to use it. I think the skill you would acquire from learning the software will be eventually worth it since it will help in completing your phd thesis.

Stating that you'd never use the application after completing your phd may not be true. You may end up establishing a career in research where analysis applications can be of tremendous help.

I've been personally using NVivo since my PhD days. To date, I still find the software very helpful either in carrying out qualitative analysis or in conducting literature review.

Hi Abubakar

Thank you for the feedback, which is helpful.

The reason why I say I won't use it again, is because I am already 71 years old and do not see myself making a career of this. But, I suppose, you never know.

Regards

Johann

The first thing to say is that all of these programs offer a lot of complex options, but they also offer detailed tutorials for beginners. So, if all you want to do is code the data and then work with your codes, pick a program that has basic introductory material that target your needs.

As for expense, you might look at a free program such as QDA Miner Lite or Taguette. Another option would be to reconsider Dedoose, because you pay by the hour and only for the hours that you use it.

Thank you Anastasia.

The fallacy that many newbies fall into is to believe that qualitative software is the black box that can do wonders. Simply put, the software is more to support or a guide in the coding e.g. themes and patterns process. I suggest you look into --> Miles, M. B., & Huberman, A. M. (1994). Qualitative Data Analysis (2nd ed.). London: Sage.

Thank you for the feedback and suggestions.

you can learn thematic analysis of qualitative data in theme there are various codes appear such as categorize the paragraphs.

you can search explore YouTube

I use NVivo and find it very user-friendly. I would suggest online tutorials to get familiar with it but especially learning by doing - trialing it with one or two interviews just to practice.

Hi Johann,

This must have been very frustrating for you. The best way will be to do manual coding. Take each transcript, read it slowly, sentence by sentence, and identify the main idea in each sentence. Assign labels to the ideas, these become your category labels. The bales can be taken from the text (Nvivo codes), or from previous research (theoretical code). For each paragraph, identify the core issue being talked about by the participant. Do this throughout the transcript. Read the transcript again, this time around try to see if some of the category labels can be grouped into a single main idea as a theme, based on the purpose of your research. Make short notes on the printed transcript reflecting on the participant's experience, meanings, interpretations, and questioning your own assumptions. If you are working on word documents, use the review and comment functions to make notes and reflective remarks along the way, as you code the transcripts. You may have to read and code one transcript several times. Keep all the codes (category and theme labels) you make and the page and line numbers of the text that these labels can be found in the transcript. Repeat these processes for each transcript. Collect all your labels and their indexing details. Work with the labels and be open to making changes by reorganizing, recategorizing, and renaming them as you immerse yourself in the data.

Manual coding may be slow, but it's the best way to code and analyze data. It gets you to read and understand each transcript or data material.

I hope this helps. Good luck with your analysis

Abukari Kwame I think 20 interviews is pushing the limit where you can do this kind of detailed manual analysis. Not only will the original coding take a long time, it will also be difficult to keep track of that coding when it comes time to do the actual interpretation of the data.

Good day.

Thank you for your insights. I have come to realise that there is no shortcut here. However, I am going to use the coding software. At the moment it is between Nvivo and Atlas.ti. Also, I found a very good academic article, which explains the coding of qualitative coding and analysis in very simple terms. This is the article: Skjott Linneberg, M. and Korsgaard, S. (2019). Coding qualitative data: a synthesis guiding the novice. Qualitative Research Journal, Vol. 19 No. 3, pp. 259-270. https://doi.org/10.1108/QRJ-12-2018-0012

Best regards

Johann

Here is a link to the Linneberg & Korsgaard article here on ResearchGate.

Article Coding qualitative data: a synthesis guiding the novice

After reading it, I have only one concern, which is with their first cycle of descriptive coding, that they apply to "small segments" to capture "what the data is about," which leads to a "categorized inventory of the data." This approach is also known as coding for "manifest content," in contrast to "latent content" which codes for more interpretive material.

The potential problem in starting with descriptive codes is creating too many codes, especially if you are trying to "inventory" the contents of your data. Earlier in their article, Linneberg & Korsgaard do warn about that qualitative software makes it so easy to create codes that you may end up with too many, but they do not repeat this warning with regard to doing descriptive coding.

To give a realistic (although possibly extreme) example, I once had a graduate student come to me and tell me that he had completed his initial coding, and he wanted to know what to do next, given that he had 224 codes.

Of course, there is no way to determine how many codes are too many or too few for any given project. But if your initial code list is getting very large, it will be a lot of work to manage it for during any further "cycles" of coding.

Hi David from what you are writing, it seems that you are very knowledgable on this topic.

I also read some parts of Johnny Salanha's Coding Manual (2016). In there Friese (2014) recommends between 50 and 300; Lichtman (2013) between 80 and 100 for educational research; and MacQueen, et al. 30 to 40. On the other hand Creswell (2013) starts with 5 to 6 'provisional codes' in what he calls 'lean coding'.

As a non-academic myself, I believe in 'less is more' and would, therefore, limit the number of codes. Personally, I would not go for line by line coding, but rather find the gist in the interview transcripts, from the interviewees perspectives.