Data from: Social disappointment and partner presence affect long-tailed macaque refusal behaviour in an "inequity aversion" experiment ## Summary of dataset contents The test conditions of our experiment were run using a 2x2 factor design. The first factor we manipulated was distributor type - whether a monkey received a reward from a human or from a machine. Distributor was a between-subject factor. The second factor we manipulated was partner presence - whether a monkey received a reward in the presence or absence of a well-rewarded partner monkey. Partner presence was a within-subject factor. The control conditions in our experiment were run subsequent to the test conditions. In a control condition, each subject received low-value food in the presence of a partner monkey who was equally poorly rewarded. All test and control conditions were video-recorded. The videos were coded by an experimenter using the Mangold interact software and the data logged during this coding process was exported to an excel file: '2021-06-02_social-disappointment_data.csv'. Eighteen columns (columns A - R) in `2021-06-02_social-disappointment_data.csv' contain data. Each of these eighteen variables is briefly explained below: A. 'Duration_Time': A '0.04' in this column indicates that the duration of a behaviour was not of interest i.e., the behaviour was logged as an event. Values above 0.04 log time durations associated with a behaviour (unit: seconds). B. 'date': When the session took place (format: dd/mm/yyyy) C. 'subject': Name of subject D. 'sex': Sex of subject (m/f) E. 'age': Age of subject (in years) at start of experiment F. 'condition': `inequality' indicates that the session was a test condition and `equality' indicates that the session was a 'control' condition. G. 'session': Session number - subjects experienced 8 inequality (test) sessions and 4 equality (control) sessions H. 'trial': Trial number - subjects experienced 12 trials per session I. 'distributor': half of the subjects received a food reward from a 'human' and half from a remote controlled apparatus ('machine'), this column details the distributor group to which the subject was assigned. J. 'partner presence': Whether a conspecific was present in a neighbouring test cage (y/n) K. 'partner_identity': The identity of the partner monkey in the partner present test conditions and in the control condition. We used two partner monkeys in the experiment - "ilja" and "linus" L. 'dyad': Information in this column takes the form of "subject name_partner name" (in partner present conditions) or "subject name_ghost" (for partner absent conditions) M. 'behavior': All events coded by the experimenter. The information in this column takes the form of "P_event" or "S_event". Events prefixed with "P_" record the actions of the partner, and events prefixed with "S_" record the actions or events associated with the subject. N. 'counterbalance': Whether the subject experienced the partner present or partner absent test condition first. Note that this variable is also sometimes referred to as 'order' ('order' and 'counterbalance' are interchangeable here) O. 'refusal': whether the monkey deigned to take the food that was offered (y/n/NA). An NA indicates that the trial did not count or that the behavior was not related to a food-refusal decision. P. 'number_refusals_in_session': A summary variable documenting the total number of times that a monkey refused the food on offer in a session (maximum refusals possible: 12) Q. 'number_trials in session': In general there were 12 trials per session but for a number of reasons (documented in tables S2 and S3 in the electronic supplementary material) in a limited number of sessions fewer than 12 trials counted in a session (i.e. trials were discounted). R: 'discount_trial': The experimenter logged whether the trial counted (y/n) based on a number of set criteria (these criteria are documented in section 3 of the electronic supplementary material). ## Description of the data and file structure As outlined above, the raw data was stored in an excel file: 2021-06-02_social-disappointment_data.csv Four data subsets were created from this raw data: 1. model1A.csv 2. model1B.csv 3. model2.csv 4. model3.csv The R-script used to create these four subsets has been provided (subsets.R) The data in the first three subsets (model1A.csv, model1B.csv and model 2.csv) were used to run three GLMMs (Model 1A, Model 1B and Model 2) that investigated the food refusal behaviour of our subjects. The data in the fourth subset (model3.csv) were used to run a survival analysis (Model 3). The survival analysis examined the pull-latency behaviour of our subjects - whether the length of time (in seconds) taken to carry out a task (pull a lever) differed systematically between the four test conditions. The R-scripts that were written for each analysis have a name that corresponds to the respective .csv file i.e., the R-script used to run Model 1A is `Model1A.R' and `Model1A.R' imports data subset 'model1a.csv'. All of the R-scripts have been comprehensively annotated. We source six R functions (written by statistician Roger Mundry) at various stages of our analyses. These six functions are: 1. diagnostic_fcns.r 2. glmm_stability.r 3. boot_glmm.r 4. coxme_stab.r 5. coef_from_coxme.r 6. drop1_para_coxme.r Roger Mundry has made these R functions available on Zenado: https://doi.org/10.5281/zenodo.7670524 ## Sharing/Access information Links to other publicly accessible locations of the data: In addition making the data subsets and r-scripts available here on Dryad, these files are also viewable on OSF at the following link: https://osf.io/fu6j4/?view_only=f076ef4272464947abfd52bf8b9dab63 Data was derived from the following sources: The raw data stored in 2021-06-02_social-disappointment_data.csv were derived from an experimenter coding video-recorded test sessions in Mangold interact (software). An .act Mangold interact file was exported to excel. ## Code/Software All four analyses were run in R (version 4.1.1). Citation: R Core Team. (2021). R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing. URL https://www.R-project.org/. The packages used in each analysis are listed at the beginning of each R-script.