**3. Results**

The participants fully engaged with the RP method and drew (and later added) their knowledge about food systems within the allotted time (for examples of RP pairs, see Figure 1). The number of overarching categories represented in the RPs increased from an average of 3.73 (±0.15 SE) in the pre-course pictures to 5.31 (±0.11 SE) in the post-course pictures (Table 2). The six categories of the food system were not equally represented in the RPs, particularly in those before the course. The value chain was the only category represented in all the pre-course pictures, i.e., the pre-course pictures all contained at least one sub-category of the value chain. Also actors were almost always represented (overall mean of 96%; Table 2). All other categories of the food system were represented less in the pre-course pictures. While system elements and boundaries were represented in considerably more than 50% of the pre-course pictures (in approximately 70%), specific topics were represented in approximately one-quarter of the pre-course pictures, and outcomes were only represented in 18% of the pre-course pictures. On average, all six categories were better represented in the post-course pictures (except for the value chain and the actors, which were already well represented in all the pre-course pictures; Table 2).

Over all three courses, 18.96 (±0.57 SE) sub-categories were represented in the post-course pictures, while only 11.02 (±0.51 SE) sub-categories were depicted in the pre-course pictures. Thus, course participants added on average 7.94 sub-categories (±0.45 SE), which is a significant increase in sub-categories represented in the picture after taking the summer school course compared to before the course (*p* = 0.019; Table 2). Neither summer school cohort, nor gender, highest degree or field of study of the participants had a significant e ffect on the di fference in the number of sub-categories represented in the pre- and post-course pictures (*p* > 0.09).

**Figure 1.** Pairs of rich pictures (RPs) drawn by participants in the World Food System Summer School. (**<sup>a</sup>**,**c**,**<sup>e</sup>**) were drawn before the start of the course (pre-course), while (**b**,**d**,**f**) were drawn at the conclusion of the course (post-course). Note that pre- and post-course RPs were scanned with different devices, leading to some discrepancies in colors.

The average representation of the di fferent sub-categories in the pre-course pictures varied between 0 for the representation of actors from international organizations and 98% for the agricultural part of the value chain (Table 3). The majority of the summer school participants (i.e., more than 50%) included the value chain elements agriculture, transport and storage, processing and packaging, retail, consumption and resource inputs in their pre-course pictures. Farmers and consumers (from the actor category) were also represented in the majority of the pre-course pictures as well as some representation of interactions (from the system elements category). All other sub-categories were represented in less than 50% of the pre-course RPs. We observed very limited representation of special topics (like organic agriculture, diversity or labels), certain actor sub-categories (like non-governmental organizations and international organizations) and food system outcomes in the pre-course pictures (Table 3).

As a result of the course, the proportion of RPs containing a particular sub-category increased between 2% for the agricultural part of the value chain (that already had the highest representation in the pre-course pictures) and 57% for the food system outcome environmental sustainability. A number of sub-categories were substantially better represented in the post-course RPs, with representation increasing by more than 30% compared to the pre-course RPs. These were in particular sub-categories of the food system outcomes and the food system boundaries but also workers (from the actor category) and waste and losses (from the value chain category; Table 3).

**Table 2.** Average number of categories (±standard error) and proportion of categories represented in pre- and post-course pictures (±standard error) as well as average number of sub-categories represented in pre- and post-course pictures (±standard error) over all three summer school courses and for each of the three courses separately. N = 17, 18, and 16 in 2016, 2017 and 2018, respectively. SE: standard error.



**Table 3.** Average representation of sub-categories in the pre- and post-course pictures as well as the average proportion of participants adding the sub-categories to their pictures after the course (i.e., di fference in post- minus pre-course picture). SE: standard error.
