Christopher - option 2 is the most conventional. Statistical reporting should not usually be interspersed with narrative. If it was just a few open-ended questions and not much narrative - you could simply use a few examples in the discussion section to note 'comments of interest' supporting/not supporting the descriptive findings. Depending on how you structured the study, from the beginning, you could also present it as a mixed methods study. In this case - it would look very much as if it were a sequential QUANT - qual design (assuming that you have enough narrative data to perform content analysis - and that it would not be 'equal to' the statistical data).

Christopher - option 2 is the most conventional. Statistical reporting should not usually be interspersed with narrative. If it was just a few open-ended questions and not much narrative - you could simply use a few examples in the discussion section to note 'comments of interest' supporting/not supporting the descriptive findings. Depending on how you structured the study, from the beginning, you could also present it as a mixed methods study. In this case - it would look very much as if it were a sequential QUANT - qual design (assuming that you have enough narrative data to perform content analysis - and that it would not be 'equal to' the statistical data).

Thank you Dean. Yes, I am leaning towards option 2. I have never written a mixed-method article before, only published quant/qual data separately. I will look more into the free-text comments to see if I have enough narrative data. Best, Chris

In the field of mixed methods research, this would be treated as an issue of integration. Probably the most effective form of integration for the kind of data you have (and for the purpose of evaluation) would be an "explanatory sequential" design (QUANT --> qual), where you first establish a set of quantitative results, and then use the qualitative data to help understand those results. Because this design assigns a specific purpose and goal to the qualitative data, it is essentially integrated by definition.

Beyond, that you have proposed two alternative approaches to what I would call a "complementarity" design, where each method makes its own separate contribution to the overall project. These kinds of designs do not have any pre-determined way of integrating the results, and you are considering two of the most common strategies.

One notable advantage of organizing around the quantitative results and illustrating them with quotations is that it reflects the original balance toward the quantitative in your study. In particular, if you did not design your qualitative questions to generate rich, systematic data, then you may have high non-response rates and relatively brief, thin responses when people do fill out the open-ended items. Further, this strategy provides a direct route for integrating the qualitative data into the framework provided by the quantitative results.

The alternative approach of analyzing the qualitative data by themselves can be particularly problematic with regard to integration, because it often leads to two separate reports of the results -- one section for the statistical outcomes and a second section for the summary of the themes. With regard to integration, the typical solution is to use the Discussion section to say something about how the results do or do not fit together. But because the two sets of results were produced independently, this can be difficult to accomplish.

A different approach to the qualitative analysis would be to begin with a codebook that was matched to the original research questions and the accompanying quantitative measures. The goals of this analysis would be to address the findings from the quantitative data, through this "deductively" generated set of codes. But you would not need to be limited to that. In addition, you could develop new codes for content from the analysis of the qualitative data that would be generated "inductively." This second set of codes would allow you address issues that you discover through the analysis of the qualitative data. Taken together, this combination of deductive and inductive analysis is known as a "hybrid" approach.

As this long reply shows, you have entered a complex territory -- a least from the perspective of mixed methods research. A simpler reply, would be that there is no one right answer to your question, and that instead, you need to think about the kinds of results (and "Results Section") that you want to generate, as well as the resources that it would take to reach that goal.