Synthetize. Once the loop (1)-(2)-(3) has been completed because the ICA measures reached an acceptable threshold, we can rely in the output of the codification process and start drawing conclusions. At this point, there exists a consensus about the meaning, applicability and limits of the codes and semantic domains of the codebook. Using this processed information, this phase aims to provide a description of higher-order themes, a taxonomy, a model, or a theory. The first action is to determine how many times each domain appears in the data in order to estimate its relevance (grounded) and to support the analysis with evidences through quotations from the interviews. After that, the co-occurrence table between semantic units should be computed, that is, the table that collects the number of times a semantic domain appears jointly with the other domains. With these data, semantic networks can be created in order to portray the relationships between domains (association, causality, etc.) as well as the relationship strength based on co-occurrence. These relations determine the density of the domains, i.e. the number of domains you have related to each domain. If further information is needed, it is possible to repeat these actions for each code within a domain. In the case study considered in this paper, the co-occurrence tables and semantic networks were computed for each semantic domain related with RQ1 (problems) and RQ2 (results). In addition, for the most grounded codes of each research question this analysis was also repeated to straighten the conclusions. Finally, problems and results were analyzed by case (organization) and the correlation relationships between problems and results was discussed,
