Our respondents might think that any given issue is the most important facing the country. As such, we allow them to answer this question using an open text box. The problem, however, is that these data can be difficult to analyse. To aid our users, we categorise our respondents’ text-responses into a set of consistent closed categories.
The categorisation process involves computer-assisted trained human coders. Our coders receive one-on-one training and then use a reference guide to ensure that they code the open text responses consistently. For some waves, our coders also used machine learning tools to help speed up the process. In these cases, the software would learn which words tended to be associated with which categories and make suggestions to the coder to speed up the coding process. Discretion always lay with the human coder.
The original open text responses are available in the “Strings” files that accompany each panel release. The coded categories are available as “mii_cat” (where responses are coded into one of fifty closed categories) and “small_mii_cat” (where they are coded into one of twelve much broader categories). We strongly recommend that users verify that the open-ended responses match their interpretation of the category, especially when relying on categories with smaller numbers of responses.