Table 1.
Different model specifications for rank in dyadic interaction models tested in this study.
+ indicates that main effects were entered independently in the model, * indicates a statistical interaction term.
Fig 1.
Results of model comparison for aggression interactions.
The y-axis portrays delta LOO-ICs (distance to best model)—the best model is therefore set to 0, and higher scores indicate poorer performance. Points indicate models (split by the respective category), while lines connect otherwise comparable models (e.g., the main effects ordinal scale models for both Elo ratings and David’s Scores).
Table 2.
Interpretation of Elo results for aggression.
Model interpretations are displayed for all models using Elo ratings only to facilitate interpretation. Downwards arrows indicate that lower-ranking individuals show higher rates, upward arrows indicates that higher-ranking individuals show higher rates. Interactions can show Down The Hierarchy (DTH; Targeting lower-ranking individuals), Closely Ranked Receiver (CRR; Targeting individuals with similar rank); Up The Hierarchy (UTH; Targeting higher-ranking individuals), or a mix of those.
Fig 2.
Results of model comparison for grooming interactions.
The y-axis portrays delta LOO-ICs (distance to best model)—the best model is therefore set to 0, and higher scores indicate poorer performance. Points indicate models (split by the respective category), while lines connect otherwise comparable models (e.g., the main effects ordinal scale models for both Elo ratings and David’s Scores).
Table 3.
Interpretation of Elo results for grooming.
Model interpretations are displayed for all models using Elo ratings (similar results for David’s Scores). Downward arrows indicates that lower-ranking individuals show higher rates, upward arrows indicates that higher-ranking individuals show higher rates. Interactions can show Down The Hierarchy (DTH; Targeting lower-ranking individuals), Closely Ranked Receiver (CRR; Targeting individuals with similar rank); Up The Hierarchy (UTH; Targeting higher-ranking individuals), or a mix of those.
Fig 3.
Results of model comparison for proximity interactions.
The y-axis portrays delta LOO-ICs (distance to best model)–the best model is therefore set to 0, and higher scores indicate poorer performance. Points indicate models (split by the respective category), while lines connect otherwise comparable models (e.g., the main effects ordinal scale models for both Elo ratings and David’s Scores).
Table 4.
Interpretation of Elo results for proximity.
Model interpretations are displayed for all models using Elo ratings (similar results for David’s Scores). Downward arrows indicate that lower-ranking individuals show higher rates, upward arrows indicate that higher-ranking individuals show higher rates. Interactions can show Down The Hierarchy (DTH; Targeting lower-ranking individuals), Closely Ranked Receiver (CRR; Targeting individuals with similar rank); Up The Hierarchy (UTH; Targeting higher-ranking individuals), or a mix of those.
Fig 4.
Results of model comparison for supplant interactions.
The y-axis portrays delta LOO-ICs (distance to best model)—the best model is therefore set to 0, and higher scores indicate poorer performance. Points indicate models (split by the respective category), while lines connect otherwise comparable models (e.g., the main effects ordinal scale models for both Elo ratings and David’s Scores).
Table 5.
Interpretation of Elo results for supplants.
Model interpretations are displayed for all models using Elo ratings (similar results for David’s Scores). Downward arrows indicate that lower-ranking individuals show higher rates, upward arrows indicate that higher-ranking individuals show higher rates. Interactions can show Down The Hierarchy (DTH; Targeting lower-ranking individuals), Closely Ranked Receiver (CRR; Targeting individuals with similar rank); Up The Hierarchy (UTH; Targeting higher-ranking individuals), or a mix of those.