If you have produced countable codes, then you can use just about any form of statistical analysis. The obvious problem, however, is the non-random nature of your sample, so I would worry about using tests of statistical significance.

In general, the concept of "validation" raises complex issues in qualitative research, especially in terms of using quantitative evidence to validate qualitative results. So, I would think about who will be the audience for your results, and whether they will consider this kind of analysis praiseworthy or unacceptable.

Can you say more about the variables you are using and what you want to do with them?

You could analysis your questionnaire according to validity and reliability process. I recommended my Paper entitled "Persian version of the Moorong Self-Efficacy Scale:

psychometric study among subjects with physical

disability"

Dear David,

Thanks for sharing valuable insight and suggestions. For this study, I have developed a list of open codes by using inductive content analysis. Hence, I have a list of open codes with frequency. of occurrence, Can I use this for any further statistical test?

For example, for eWOM content (theme), I have found 30 open codes divided into 4 categories. Similarity, I have found codes for other themes. Now challenge is how to relate them. In fact , I used axial and selecting coding to develop model. At the same time I am looking for simple statistical model where two antecedents (themes)  affects an outcome variable. 

I have done some review of related papers. I have found few studies used log linear analysis to link themes (generated from inductive content analysis). 

Please let know your view. 

I also extend my sincere thank other members for sharing their views. 

Regards,

Gobinda

Log-linear analysis is a multi-variate version of chi-square, so unless you have a complex set of hypotheses, I would stick to the simpler statistic. Note that this format does not look at how often any given person gave a code; instead, it counts how many people gave a code, based on their score on another variable. (The simplest version of this model is a two-by-two table, where every one is score either 1 or 0 on some variable).

My own approach is to look for "obvious" patterns in the qualitative data, and then to use them as a basis for investigating the qualitative data more deeply, to ask why and how those patterns arose. I personally would not use statistics to do that, but if you want to "test" the existence of a pattern, you will need to justify that choice (and don't expect a reviewer who is a qualitative researcher to like this choice).

Here is a copy of my work in this area.

Article Qualitative Content Analysis: A Guide to Paths not Taken