Why not collect similar data from each type of participant by using open-ended questions in the survey? In particular, the self-completed questions could match as closely as possible the ones that you are asking in the semi-structured interviews. That way, you could say that you used a combination of in-person and self-completed semi-structured interviews.

The problem with using close-ended survey questions for "triangulation" is that it appears you would have only a small number of close-ended surveys. If so, that data would be very "thin" in comparison to the more detailed responses from the semi-structured interviews. Hence, it would be difficult to get meaningful comparisons for the triangulation.

DAVID, thanks for reply. My participants are too busy people. Several of them already told me that just give me a call and ask what you want to ask. And many of them said that they are not available for interview but will be glad to fill any survey at their convenience. What I am thinking is exactly same set of open ended questions for survey and exactly same set of questions for interview (though many times it would not be possible). That essentially means that the participant who has filled survey would not be available for interview and the one I interviewed will not like any survey before or after. Hope this clarifies my issues. Now if I get data from same set of questions - survey and interview, I can analyze it separately. But can I use survey data to triangulate or confirm findings of interviews or vice-versa? If literature support such mechanism (or if you know any research study which has done so), please link me to the study. What else do I need to do to make sure any variability or bias or anything is taken out?

The usual meaning of triangulation is something like cross validation, where the findings from one method show a similar pattern to the findings from another method. In essence, you ask the same research questions twice, and answer it with two different methods, with the goal of having more faith in your results if the methods agree. It does not require that each research participant replies to both methods, only that you have comparable data from each method.

The source of the idea of triangulation is from navigation, where two separate "lines of observation" meet at the point of a triangle. I have attached a figure to illustrate this

But triangulation doesn't sound like what you really want, because your goal is to collect similar data from everyone, using two different methods -- which isn't the same as using each set of results to validate the other. Instead, I think you basically want to "merge" two sources of data into one.

Of course, if you do collect the data using two different methods, you will need to assess the extent to which they agree, before you can be confident in merging them.

David, wonderful explanation. You are right that I am trying to use two different methods for data collections due to the challenge I have. I am planning to use same set of questions therefore the responses might follow same structure.

You have captured my intention very well. Since I am using two different methods, I could potentially analyze the data from each method separately and use results from one method to triangulate the results from another to see the degree of agreement. If I do so, will it be considered triangulation? From your reply I guess it is.

I was trying to find an example or some support from literature where they ran into same issue that some participants preferred one method and some other a different. I have seen literature indicating it from planning angle but none of them seem to address this 'participant preference or availability' concern. Do you know any such study?

Given that you are working with corporate executives with busy schedules, I don't think people will question the fact that some people wanted to do their own self-completed survey rather than doing an interview.

Simply say that the participants had a choice about whether to do the interview one way or the other.

Thanks David. Nice guidance. Appreciate your time.

David, i got to find an interesting article supporting my challenge.

E-Mail Interviewing in Qualitative Research: A Methodological Discussion. JOURNAL OF Lokman I. Meho (2006) THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY, 57(10):1284–1295

It is indeed a very good information on using the method of e-mail interviewing (sending questionnaires via e-mail) in lieu of face to face interview. The article concluded that e-mail interview actually may be better quality responses than face to face interviews because participants got the time to think through.

I am thinking I should use this approach of e-mail interview and F2F interview as two different methods to collect data and then analyze those separately

To some questions, I am using purposive sampling and experts are pre-filtered using certain criteria. I have close to 30-40 worldwide experts (mostly well known) who are known to have done work in this domain. Not a whole lot. Therefore even missing out 5 of them will impact my results or may get short of saturation (not sure at this point). I might be adding my work by employing an additional method of online open-ended questionnaire which will result in another set of massive data. However, I am concerned about missing out some experts simply because they are not available for face to face interview and sometimes some experts simply are unreachable for interviews even if they consented for interviews. In that event I can offer than alternate method which they might be able to complete even while in the airplane (hoping so).

I was looking for case studies and examples where by some journal articles would have supported this parallel mode of data collection using two different methods from different set of participants. At end of the day examiner and university will see the soundness and rigor of my data collection method as well as how I do the analysis. I wanted to make sure this kind of problem is supported by literature. I appreciate your time.

Do feel free to add any additional inputs you may have.

Raman K. Attri, I have the same problem. Did you find any literature to support your parallel data collection methods?

Mdumiseni: Here is what I learned from my research experience. If you have bounded case, then you could use combination of several different data collection methods to gather data within that case. If you have only one case, then you cross-check the outcomes from one method vs. other. This will help you establish the truthfulness of the findings. If you have multiple cases, then you could combine the data form different sources within a case to make a full picture fo the case before you compare it with next case. In such a case, you would not need to do data analysis by source, but you could just do data analysis by case.

If however, you are not using bounded cases, then the data quality and depth is different in interviews and in questionnaires even if you have same set of questions. In such a case, you may want to keep data sets from these two methods separately and analyze those seperately. If findings from one match with other set, you could use it as 'data triangulation' to establish validity of your research findings.

I will have to go back to my database for references and literature. I will try to find and provide you.