Fig 2. Effect plots displaying results from GLMMs at the bout level of four parameters varying according to the social role of callers (SR: aggressor vs. victim) and conflict severity (Sev: mild in orange vs. severe in blue).

a) number of screams, b) average scream duration (s), c) scream rate (number of screams/s) and d) percentage of NLP (%). While dots represent the predicted mean of the parameters (previously transformed when necessary and scaled), bars represent standards errors. Significant results are represented by * when adjusted p-values < 0.05, ** <0.01 and *** < 0.001.