Software and Qualitative Analysis
This overview of the qualitative data analysis process has followed, in general terms, the outline first presented here in Figure 1. Here is a slightly different summary of the process, which some readers may find more helpful.
The first step is coding the data. But don’t be misled by the phrase “the first step.” Coding should begin before all the data is collected, and will usually be returned to for more refinement as analysis proceeds and ideas develop. In fact, none of the following “steps” are strictly sequential.
Next, the analyst will want to start summarizing themes. In order to do this, it is often important to partition the dataset, for example to be able to look for variation in the theme across different demographics, different outcome groups, etc. Further, the analyst will often need to get different sorts of overviews of the data, for example by counting the numbers of cases that fall into different categories, building matrix displays, and so on.
In addition to summarizing themes, the analyst will often want to summarize cases. Again, descriptive matrices can be extremely helpful here.
With both the cases and the major themes summarized, the researcher will usually proceed to looking for patterns and relationships. There are several different ways to find them. First, and perhaps most important, is noticing them as you read. In addition, you can do clustering, whether by case or by variable. And further, you can build all sorts of matrices to identify patterns and relationships.
Finally, it is vital if the study is going to be reliable and credible, that you go back and verify your conclusions. There are a range of strategies here, including triangulating across data sources or data types, searching for negative evidence, checking for representativeness, checking the meaning of outliers, and so on. Again, matrices are indispensible tools.
Careful application of the techniques discussed here can both help the researcher stay focused and on-track, and help to ensure that the findings produced are sound and persuasive.