You can transform the qualitative data (coded) into quantitative data, I haven't tried it but it seems to be well cited. Let me know if it is not accessible. I can download it through our University Library.

• Auer-Srnka, Katharina J., and Sabine Koeszegi. "From words to numbers: how to transform qualitative data into meaningful quantitative results." Schmalenbach Business Review 59 (2007).

Now the the challenge there is to make the measurements:

• Scale -  http://stattrek.com/statistics/measurement-scales.aspx
• Data Type  - http://en.wikipedia.org/wiki/Level_of_measurement

So that  you can statistically compare both data sets, otherwise it is impossible. So maybe best to consult a statistician about this. It should be possible, just that you need to be careful with measurement error that might occur during the data transformation.

However if your evaluators are amenable to  utilizing mixed methods (this is after all the essence of triangulation). I would say:

• Statistically analyze the survey. Assuming that the survey is basically assumed themes or focus that you have used for the semi -structured interviews
• The interviews can provide more details on possible explanation of the Themes of the phenomenon that you are investigating.

You might also want to think on the third axis... 1)survey, 2)interview, 3)?literature?. 

Goodluck with your research,

Harry

Your question appears to include some assumptions. I would suggest that you should focus on your research questions rather than your data collection techniques or the software you have used to analyse them. The philosophy behind semi structured interviews is usually inductive whereas the philosophy behind closed questionnaires is usually deductive. If this is the case with your research then I would suggest focusing upon the complementary roles of your findings from your analysis rather than trying to measure the level of corroboration.

Good research is about selecting and using appropriate methods rather than the level of significance or corroboration of the findings.

Hi,

Your question: How do I compare data between survey questionnaires and semi-structured interviews?

Kindly see the following link:

http://pareonline.net/getvn.asp?v=15&n=1

http://www.colorado.edu/ibs/pubs/pop/pop2009-0008.pdf

http://www.qualitative-research.net/index.php/fqs/article/view/1450/2946

http://www.rand.org/content/dam/rand/pubs/technical_reports/2009/RAND_TR718.pdf

The main issue with triangulation is getting the data into some kind of directly comparable format, and as Harry says, the most common approach to doing so is to convert the qualitative data into something like counts of codes. In mixed methods, this is often know as "quantizing" the qualitative data.

I have to say, however, that this generates only a limited comparison because it eliminates most of the reasons why one would use qualitative data in the first place, including interpreting the meaning in that data. In general, doing triangulation through a strict comparison of two such different types of research turns out to be an idea that makes more sense in theory than it does in practice.

Thank you for the various feedback. I found an article recently (Practical Assessment, Research & Evaluation e-journal)  indicating that, according to Stemler (2004), if both methods can achieve a consensus score of 70% or better that could indicate results that are corroborative. An empirical research I read compared focus group results with SPSS statistical results, i.e., using Chi-square and P values. Not sure how they were able to "quantizize" the focus group data. 

You might look at Cline et al. "Factors Influencing RNs’ Perceptions of Quality Geriatric Care in Rural Hospitals." Western Journal of Nursing Research. They created a cross-tabulation between the significant relationships in the quantitative data and the main themes in the qualitative data.

That approach will not give you anything you can count, but it will produce a systematic comparison between the two sets of results.

Hau , as mentioned before the triangulation is used to ensure that the data gathered is valid and supported .Traditionally using a mixed method approach to a research topic the qualitative datinformation is arranged into thems and a thematic analysis done to pull the relevant factors that can form the independant and dependant variables can be identified . From this  analysis the quantitative data gathering instrument , the survey can be developed and tested .

The triangulation will be validated if the same themes  in the qualitative instrument prove significant in the quantitative analysis.

Dear Michael,

Yes, that was how I have developed the survey instrument - the questionnaire was based on themes captured from the pilot qualitative research. I ran the initial data collected (50%) from the survey through the SPSS software and the results were very positive. Based on skewness, factor analysis and reliability tests, there was corroboration between the qualitative and quantitative. But is that sufficient? Do I need to do hypothesis testing on the quantitative data as well? Sounds like a no-brainer here but I have reasons for asking the latter.

The notion behind Qualitative research is to use themes {NOT NUMBERS}, thus, I disagree with the opinions which say to transfer the qualitative results through coding to numbers.

The reason behind your using of triangulation was to gather different information; and not to find the similarities!! between Qant. and Qual.

Using qualitative methods to create a survey instrument is not generally considered to be triangulation. In true triangulation, the two sets of data need to be collected independently, so that their separate results can be compared.

The design you have used to develop a survey instrument from from a preliminary qualitative study is quite well recognized in mixed methods research, but it is not triangulation. In particular, Creswell and Plano-Clark, following the original work of Greene et al. 1989, call this as "exploratory design." The goal in this case is to explore the phenomenon qualitatively, and then use those results to help develop a set of quantitative measures.

 Thank you. I stand corrected. I have used mixed methods and not triangulation. An important difference. Good learning.