Methodology
The methodology was based on a master’s dissertation about developing heuristics in the context of education (Silva, 2017) and divided into four stages: exploration, analysis, synthesis and verification.
The analysis stage aims to group the existing heuristics into categories relevant to the proposal (Silva, 2017). The preliminary results follow.
Qualitative content analysis
In order to fulfil what was proposed for the analysis stage, the heuristics drawn from existing sets were analysed qualitatively. The result of this stage is presented in a systematic table, available below, in which 30 sets of heuristics were assessed in full.
As set out in the methodology, care was taken over defining the categories that would represent the heuristics analysed, namely: 1) "makes a great deal of sense"; 2) "makes little sense"; 3) "makes no sense" for the work. For these reasons, the process of defining categories and categorising the heuristics analysed, adopted at this stage, drew on a colour scheme, since colour lends emphasis and clarity and organises information visually (Menezes, 2017). The colours selected were green for "makes a great deal of sense", yellow for "makes little sense" and red for "makes no sense".
Table 9 — Qualitative content analysis reproduces exactly the same listing of the 30 sets of heuristics presented in Table 8 (Exploration tab), now put through the process of categorisation by level of relevance to the project. So as not to repeat the listing, we refer the reader to Table 8.
Table 9, "Qualitative content analysis", repeats the listing of the 30 sets of heuristics analysed (Authors, Year and Heuristics in detail), showing the scope of the material put through the process of assessment by level of relevance to the project, classifying them according to their thematic and practical relevance to the research.
Grouping by similarity
At the start of the process it was not clear which categories would be used to group the heuristics; the structure of a mind map, however, revealed itself along the way, on the basis of a careful appraisal of the heuristics analysed. Although the process was not linear, by the end of the analysis it was possible to observe that the categories defined emerged in essentially two ways.
The first considered nine recurring themes: "feedback", "students", "teacher–student communication", "tutoring", "equity", "content", "material", "usability" and "replicability". That process gave rise to the second category, in which two new dimensions emerged, named "material" and "communication", which grouped the nine earlier themes.
A notable feature of the categorisation process was the inconsistency of pattern identified across the sets analysed. Some heuristics belonging to the same set had to be split and associated with more than one category. There were also situations, however, in which a single category grouped several heuristics from the same set.
Finally, as expected, some of the recommendations analysed were not fully aligned with the subject and were therefore removed from the categorisation. In this sense, only the sets by Sidney Smith and Jane Mosier (1986), Ben Shneiderman (1986), Donald Norman (1988), Soren Lauessen and Houman Younessi (1988), Jakob Nielsen and Rolf Molich (1990), Jan Dul and Bernard Weerdmeester (1991), Alan Dix (1993), Christien Bastien and Dominique Scapin (1993), Jeffrey Liker and Ann Majchrzak (1994), Preece et al. (1994), Apple Computer, Inc. (1995/2014), Gregg Vanderheiden (1997), Theo Mandel (1997), Patrick Jordan (1998), ISO 9241:10 (1998), Neto et al. (2000), Simone Diniz Junqueira Barbosa (2002), Lari Karkkainen and Jari Laarni (2002), Kent Norman (2003), Bruce Tognazzini (2003), Brigitte Borja de Mozota (2003), Elisabeth Fátima Torres and Alberto Angel Mazzoni (2004), Jeniffer Ferreira (2005), Tuncer Oren and Levent Yilmaz (2005), Chorianopoulos (2008), Pinelle et al. (2008), Walter Cybis (2010), Frederick van Amstel and Paulo Jorge Sousa (2012) and Neema Moraveji and Charlton Soesanto (2012) contained heuristics that fitted the scope of this work in full.