3.3.4. Study Methods Used within the Research Area Groups

The next stage of our work was to verify, based on the texts that were analyzed, which methods and strategies of studying CI in policymaking were used in the research areas. Figure 9 visualizes the number of research articles, in which the specific methods and strategies used for studying CI in policymaking were used, broken down by research areas, in total for the period 2012–2020.

**Figure 8.** The number of studies on collective intelligence in policymaking published yearly within the RAGs.

**Figure 9.** Number of research articles in which the methods and strategies used for studying CI in policymaking were used, broken down by research area groups, in total for the period 2012–2020. The assignment of particular methods and strategies to the labels numbered from 1 to 15, as described in Table 2.

We also compared the percentage of method usage (MU) in particular research areas to the percentage of MU in all the reviewed studies. This allowed us to see which methods and strategies were used more frequently and which were used less frequently in the examined research areas. Below, in Figure 10, we present the visualization of this comparison. The visualized difference between MU in the whole sample and in particular research areas, from this point forward referred to as the difference in percentage points (DPP). The source data are presented in Table A1 in Appendix B.

**Figure 10.** Method usage within the research area groups compared to the reviewed studies. The assignment of particular methods and strategies to the labels numbered from 1 to 15 as described in Table 2.

The mean absolute error (MAE) analysis has shown that computer science and political sciences are the most characteristic areas of research for issues related to CI and policymaking. As can be seen, in the field of computer science, the most important methods that were used most often in the entire analyzed sample were the analysis of created values (the difference in percentage points, or DPP: +9.09) and the analysis of e-participation process (DPP: +5.68). In turn, the most underrepresented methods were analysis of organizational structure (DPP: −7.10), analysis of impact on policymaking (DPP: −4.83) and state-of-the-art review (DPP: −4.55). On the other hand, in the field of political sciences, as if in opposition to the previous group, an increased interest in analysis of organizatonal

structure (DPP: +6.88) was observed, as well as in analysis of collaboration model (DPP: +5.50), whereas low interest in analysis of decision-making (DPP: −6.46) was observed. It is also noticeable that in this group, as in the entire study sample, the analysis of impact on policymaking method is relatively rarely used, which is surprising. When it comes to the research area of the social sciences and humanities (other than political science), we noticed the great popularity of the analysis of participants' behavior (DPP: +18.18) and the analysis of innovation process (DPP: +17.05), with a complete lack of interest in the analysis of created values. On the other hand, the research conducted within the natural sciences and mathematics was characterized by the little use of the analysis of organizational structure (DPP: −22.73) and the analysis of the e-participation process (DPP: −19.32), but a significantly increased use of the analysis of decision-making process (DPP: +28.41). However, it should be remembered that the studies assigned to areas no. 3 and no. 4 constituted a much smaller sample than those grouped in other areas. Finally, the last presented group of disciplines are applied sciences. In this group, as in computer sciences, the increased use of the analysis of created values (DPP: + 12.50) is observed, and at the same time we see the smaller than in the entire sample, use of the analysis of participants' behavior (DPP: −9.09), and the analysis of collaboration model (DPP: −9.09).
