“…Also, we perform Topic Modelling to assist our interpretation, and in this case 'there is no statistical test for the optimal number of topics' (DiMaggio et al, 2013:582). Therefore, we preferred to value interpretability of the model (Grimmer and Stewart, 2013, p. 20) and, following a standard procedure (Ferri et al, 2020), we elicited models with 8, 10,12,14,16,18,20,22,24 and 26 topics and qualitatively analysed them. We ended up selecting the 14 topics solutions.…”