Qualitative Data Analysis


having completed the data collection and transcription of 30 in-depth interviews, I am now embarking on the data analysis. I am using Atlas ti. software. Although I expect to have a total of 8 overall themes with some sub-themes in the writing up, I find that after coding 4 interviews in Atlas, I already have more than 220 codes.

Can any one tell me if this is the norm, or am I creating work for myself? s it right that I should have so many codes, and am I right in assuming that after initially coding all 30 interviews and having maybe having about 1,500 codes, that I must then narrow all of those down further to suit my themes?

As usual, any comments or advice would be most welcom - and more particularly if you have used Atlas ti. Kindest regards, Murt


Hi Murt,

I have not used Atlas ti, yet the amount of codes you have already seems excessive to me, you probably will loose oversight very quickly!

Like you suggest, I would try and make sense of what you have got solar. Have a look at one interview and check whether the categories, as per Atlas, actual do make sense. Perhaps it is an idea to analyse one interview "by hand" (just highlighting (in different colours) yourself what you think the key areas are and which remarks belong together). That would indicate quickly whether you're on the right track or not.

Good luck,



Hello, I just had a training session on atlasti this very week! 220 codes does seem an awful lot, have you checked that you have not used different codes for very similar data? I did that when undertaking my MSc. That said, I did my data analysis by hand. Good luck.


Many thanks for that. I am re-examining the codes I used and whittling them down to a manageable proportion and checking, in the process, whether I am , in fact, duplicating for similar pieces of info.

Best regards and thank you again.


Mny thanks for your helpful advice and sharing of your experience, Rick. I am re-examining my codes to eliminate duplication of codeing of similar information.

thanks again.


I used Grounded Theory which also relies on codes. What I said in my thesis was that I was aware that some codes were near duplications of others but I was reluctant to interpupt my initial coding by tackling this. At a more advanced stage I mentioned that I examined the codes with a view to teasing out duplications and that I conflated my large number of codes to a far smaller number! It was a retrospective 'analytic' technique on my part but my supervisor never queried it and tbh, in hindsight that is exactly what I did - although I wasn't aware of it at the time!!