Fig. 4.
Heatmaps of inter-rater reliabilities measured with Krippendorff’s alpha. External annotations were aggregated by majority voting. The first 4 rows of each heatmap show alpha coefficients across four different combinations of annotation perspectives: (1) SP = self vs. partner, (2) SE = self vs. external, (3) PE = partner vs. external, and (4) All = self vs. partner vs. external, while the last row (Diff [SE - SP]) shows the difference between self vs. external agreement and self vs. partner agreement. The columns show those for each participant.
