Qualitative content analysis – combining categories

Categories: Qualitative


Truth to tell, I had not exactly planned to write another blog entry today. Easter is almost here – time to walk outside in the sunshine, sip a coffee and watch people walking by, chat with friends, maybe paint an egg or two… But: The sky is grey, it is freezing cold, and the last thing I would want to do is sit outside. Much better to finish telling you about how to assess symbol visibility using qualitative content analysis – before that first post has slipped everybody’s mind, including my own…

Ok, so just to take you back to what this was all about: I was telling you about a student who wanted to assess the visibility of symbols in TV news reports about the EU, and we had agreed that she would use the following categories to code all the symbols she found in her material: type of symbol (with many subcategories such as: EU politicians, EU flag, EU buildings, the Euro coin, etc.) foregrounding (with symbol in the foreground and symbol in the background as subcategories) and distance (close, medium, and long shot as subcategories). She also wanted to assess for how long each symbol was shown (duration). The question was: How can she get to assessing overall symbol visibility on the basis of her codings?

Well, she will have to create what I call a combined category or a meta-category. If you are familiar with quantitative research, you may have come across the corresponding quantitative procedure: creating an index. To create an index, you take values on a number of variables and combine them into one single value, for example by adding them all up. To create a meta-category, you create meaningful combinations of subcategories across different main categories – and make these meaningful combinations the subcategories of your new meta-category.

Oh no – this sounds impossibly complicated :-(. Let me try again, using the example this time round. To code visibility, she may want to distinguish between symbols with high, medium, and low visibility. Visibility will become her meta-category (main category), and high, medium, and low visibility will be the subcategories of this category. Nothing new so far, just like any other main category and its subcategories. The difference (that makes this a meta-category rather than any other ‘plain’ main category) comes in at the point when she defines what she means by high, medium, and low visibility. Visibility is a meta-category because the values of the subcategories will be dependent on the codings for the categories foregrounding, distance, and duration.

By the time the student is done with her coding, she will have many symbols classified according to type. For each symbol, she will also know what type of shot it is, whether it is displayed in the foreground or the background, and for how long it has been shown. She may now define high visibility as: to be used when a symbol is shown from close-up, in the foreground, and for 15 seconds or more. Medium visibility might be coded when a symbol is shown in a medium shot, regardless whether it is in the foreground or the background, and when it is shown for a duration of 5 to 14.59 seconds. And low visibility would be coded for symbols that are shown using a long-distance shot, in the background, and for a maximum of 4.59 seconds. Chances are that an additional residual subcategory ‘indeterminate’ will be needed. This will be used to code the visibility of symbols that represent yet other combinations, such as long-distance shots of symbols in the background that are shown for longer than 5 seconds. Of course she may also want to classify such combinations as cases of medium visibility. This is just a question of definition and completely up to her. The number of seconds used in this example are completely arbitrary, by the way. In fact she will have to wait and see what the actual duration range turns out to be and define the values for high, medium, and low visibility accordingly.

Hopefully it is a bit clearer now what meta-categories are all about. They are just like other categories in that they combine main categories, subcategories, and definitions. But they are different from other categories in three ways. First, the subcategories are defined based on the subcategories of other categories. Second, they are coded only once you are done with your coding for the other categories. Third, there is typically little room for interpretation. The combination of values that make up the subcategories make it very clear which subcategory is to be used.

Did I really have to focus on something so complicated for my second post on this topic? Surely it is quite unusual to use something like meta-categories in qualitative content analysis? But it actually isn’t. I have come across this lots of times. Do not let the ‘duration’ in the above example lead you into thinking that this is a ‘quantitative thing’ to do. It isn’t. The fact that a metric variable was involved here was pure coincidence and has nothing to do with the basic idea of combining categories. And don’t believe that this is something specific to coding visuals. It is not – meta-categories are useful regardless of the type of your material. You may surprise yourself and find that you will need this some day.

Don’t forget: If you have questions yourself, you are welcome to write them down in a comment.

And for now: Happy Easter to those of you who celebrate. And a hopefully sunny and warm weekend to everyone!