There are basically two types of categories: formal or registry categories, which identify the units of study, and content categories, which are the ones that will be used for the analysis. The latter must be homogeneous, pertinent (related to the research objectives), clear and productive. The categories can be classified into groups, which will correspond to the variables. The categories should cover all the variables and aspects reserved in our design for resolution by means of the content analysis technique, and should therefore be exhaustive. They should also be exclusive; there should not be two categories to define the same trait or phenomena. And they must be, therefore, precise, and therefore relate to the explicit objectives of the research. We summarize it in this table:

Source: Own elaboration

When designing the sample and identifying the units of analysis, a comparison could be made between a survey questionnaire and the analysis sheet for content analysis, since in both cases the appropriate questions must be asked to obtain answers that are useful for the research objectives, whether to question individuals (surveys) or communicative products (content analysis). First of all, it is necessary to differentiate the questions between identification categories (name of the medium, date, section, genre, page, size occupied, etc.) and topical categories (referring to the contents of the messages, responding to the research objectives). In order to be productive when collecting data, it is also necessary to delimit the scope of each category.

Categories are closely related to some interpretative techniques, such as grounded theory, designed to generate inductive theories (or thick descriptions, actually) about our object or case(s) of study and the universe it may represent. It is based on the systematic analysis of data and, simultaneously, the analysis of the same around previously established categories that are adjusted, modified and reordered as the analysis is carried out on the universe or corpus until theoretical saturation occurs and the application of the categories no longer yields new results. The objective is to discover precisely which are the central categories.

As for the procedure, the idea is to draw up a category tree, an analysis sheet and instructions for data mining. Therefore, a codebook will also be generated.

The analysis sheet is the instrument to be implemented on the selected sample. On the card, the various categories are arranged, which should also contain several response options (as in a questionnaire). In this case, the response options are called codes and must show all possible manifestations of the analytical category.

The codes must meet exclusion criteria (they respond correctly to the category and do not overlap with others) and sensitivity criteria (they are productive and serve to extract data, eliminating the rest).

The codebook is the researchers’ instruction manual to ensure that the conditions of replicability and triangulation are met. The codebook prevents researchers from making subjective decisions on their own. It will therefore establish rules and formulas for data reliability and validity.