Is it possible to determine the relationship between variables arising out of analyzing qualitative data?

Usually the relationship between variables is determined via quantitative research i.e. it is a correlational research.  Another words, after the qualitative research is completed whereby the researcher is suspecting there is a relationship between variables e.g. in this case it is between participants' perceptions of the innovation and their levels of use of it, then the researcher can develop some hypotheses to illustrate the relationships between the variables.  Then s/he can operationalize the quantitative research with an instrument, collect data & perform the needed data analysis to find out the testing results.  This qualitative then followed by a quantitative method in mixed method research is called Sequential Exploratory Design.

Is it possible to determine the relationship between variables arising out of analyzing qualitative data?

Usually the relationship between variables is determined via quantitative research i.e. it is a correlational research.  Another words, after the qualitative research is completed whereby the researcher is suspecting there is a relationship between variables e.g. in this case it is between participants' perceptions of the innovation and their levels of use of it, then the researcher can develop some hypotheses to illustrate the relationships between the variables.  Then s/he can operationalize the quantitative research with an instrument, collect data & perform the needed data analysis to find out the testing results.  This qualitative then followed by a quantitative method in mixed method research is called Sequential Exploratory Design.

Very interesting topic & question.

My answer would be both YES and NO, depending on the following brief considerations: 

If we Google e.g., this string "the differences between qualitative vs. quantitative research", we'll only get confused because, simply speaking, there is no any universal consensus on these two types of researches. For example, many would argue as soon as we've based our research on numerical data and analyzed it by using any types of statistical methods and techniques such as factor analysis (this technique is widely used to what is known as reducing the number of variables in any say Likert scale based questionnaires), we've thus done a quantitative research vs. if we might analyze the same data by addressing it from two perspectives of  knowing (1) why, and (2) how,  then we've as a result done an qualitative research, etc.      

But non of those statements above is true because there is no pure either qualitative or quantitative research, specially in the social sciences. Why? Because before doing any types of researches, we indeed have a lot of preconceptions, theories, questions, and you name it. In other words, what makes a research to become quantitative or qualitative depends also on the types of questions asked... and the analyses used. We can say we conduct 100% quantitative research but draw qualitative conclusions, or vice versa.

Additionally, seeing the whole issue from a philosophical perspective, one might say measuring e.g., any user's perception of "Actual System Use" is first of all a subjective/partial measurement (read opposed to impartial measurement which proponents of quantitative research love to hear it) and thus when it's subjective, therefore such finding cannot be generalized, which automatically will be opposed to the quantitative research in most cases which is widely believed to be "grounded," etc

However, it is not easy to only say whether "yes" or "no". Whereas, we must explain in detailed what we have done, why, and how.   

Dear @Debra. My quick response is a YES. Various instruments such as SERVQUAL (developed by Parasuraman, Berry and Zeithaml, 1985), weighted SERVQUAL, SERVPERF (developed by Cronin and Taylor, 1992) and weighted PERVPERF are available to measure customer perceptions. In this regard, we define service quality as the ability of the organization to meet or exceed customer expectations. It is the result of the comparison that customers make between their expectations about a service and their perception of the way the service has been performed. Service quality comprises attributes (tangibles, reliability, responsiveness, assurance and empathy) that are both measurable and variable.

A tool kit is proposed, May be you need to customize 

https://www.google.co.in/url?sa=t&source=web&rct=j&url=http://www.intrac.org/data/files/resources/629/INTRAC-Advocacy-and-Campaigning-Toolkit.pdf&ved=0ahUKEwjGtazM4p3MAhUGMKYKHV2yCJg4PBAWCCQwAg&usg=AFQjCNGYBzttDCCoqjBvnYvyG58nK25W7A

SERVQUAL is designed to measure service quality as perceived by the customer. It uses gap analysis results (difference between expectations and perception of the received service) in measuring service quality. The SERVPERF scale is one of the important variants of the SERVQUAL scale. In SERVPERF, measurement of service quality is based only on perception. Since service quality attributes are not expected to be equally important across various (service) industries, it has been suggested to include importance weights in the service quality measurement scales. The use of weighted scales (weighted SERVQUAL and weighted PERVPERF) is also proposed.

SERVQUAL is popular and has found wide applications. It is used to measure service quality in healthcare (Kilbourne, Duffy, Duffy & Giarchi, 2004), hotels (Ramsaran-Fowdar, 2007), tourism (Nowacki, 2005), public transports (Liong, Mustafa, Mohamed & Lee, 2007), library (Ahmed & Shoeb, 2009), banking (Othman & Ismail, 2008), courier service (Zisis, Garefalakis & Sariannidis, 2009), higher education (Tan & Kek, 2004) and police service (Donnelly, Kerr, Rimmer & Shiu, 2006).

The attached link is an application of these instruments:

Seyed Yousef Hejazi · Fatemeh Rajabiyan Gharib · Mahmoud Omid (2015) Comparison of services quality assessment models between students from teaching and learning process quality by using artificial neural network.

Article Comparison of services quality assessment models between stu...

Qualitative research can be used to explore variables, get a sense of what relationships may exist. Findings usually give an indication of what might be going on with the variables, and from there you carry on with a quantitative research to find support (or reject) your "hunch" derived from that previous qualitative work. 

It also works the other way around. You did a quantitative study and found unexpected relationships. You want to then look into it more deeply to untangle the variables, gain a better understanding, so you can draw more solid conclusions.

That's the beauty of using mixed methods. However, you will need time and resources to do both well, so it is important to give the reader an understanding of your research goals, methods and also limitations. Neither method, to my mind, is absolutely conclusive when new concepts, relationships, or areas of research are being pursued. 

I would also say that it does not look good when one asks a research question like ‘is there a correlation between these two’ and then uses qualitative methods. What kind of question would the student ask from the data, does it make sense to use qualitative data to answer it?

One could say that the qualitative data suggest there could be a relation between the two, but the data does not prove the relation.

Many thanks to everyone for your very insightful and instructive responses in answering my question. I am now better informed about what is the best possible approach to answering the research question.

Best regards,

Debra

Some years ago, qualitative and quantitative research were almost "antagonic". Nowadays both may be used in data analysis. Therefore, no problem in terms of statistics applied to qualitative results.

It depends on what are you considering as "qualitative" or "quantitative". I consider that these terms refer to kinds of analysis, instead kinds of research.

Having in mind that assumption, there are statistics for qualitative and quantitative data. In terms of bivariate correlation, Pearson, Spearman and contingency tables are useful, depending on the kind of data that you have. Is the same for multivariate analyses: there are factor analysis and an equivalent for categorial data, the correspondence analysis. You can also think on analysis of open response (i.e., interviews) analyzed with grounded theory, where grounded and density measures indicate different kind of relation among both emergent and predefined categories.

Therefore, to find relations among variables is posible in both quinatitative and qualitative ways.

All of the major qualitative data analysis packages now contain what they call a "mixed methods" mode, where the codes from a qualitative analysis can be paired with quantitative data about each of the participants. With this kind of data, the thematic codes can be treated as variables, where actual use of an innovation could be a quantitative measure.

Beyond that, if all of the data was qualitative, one might still use the idea of correlation as a metaphor to say that those who were more likely to do one thing were also more likely to do the other. Over the years, there have been any number of mentions of the idea that qualitative researchers apply this kind of quasi-numeric reasoning when they say that one subgroup does something "more often" than another group. So, describing something as a correlation would just be an extension of the same kind of thinking.

Qualitative research can identify key variables in which regards as findings. Such findings may put into a conceptual framework or a research framework, later stage, that research framework can be tested via various statistical methods such as correlation, structural equation modelling ect. Because, qualitative research do not prove any significant relationship among variables, but quantitative research do. However, based on data gathered in qualitative research, the researcher can show that there may have some relationships among variables through graphically or by using NVivo qualitative data analysis software.