README ------ The files in this module represent all data analyzed to produce the statistics and figures for "Does Communication Help People Coordinate?". The data is captured in the following files: - **balanced_color_preference_games.csv** The treatment parameters and outcomes for each valid game with balanced assignment of different color preferences (the number of participants who receive higher payout for 'red' consensus equals that of participants who are paid more for 'green' consensus. (Fig. 2, 5) - **marginal_info_data** The files within this folder contain data about the marginal information conveyed by messages over time, in different communication settings and over different network types. (Fig. 3) - **categorized_unconstrained_messages.csv** Records of messages sent in games with unconstrained communication, categorized by content. (Fig. 4) - **minority_power_data** The files within this folder contain data related to the experiments examining the power of a well-connected and globally-communicating minority over the consensus choice. (Fig. 6, 7, 8, 9) ---------- **balanced_color_preference_games.csv** - **experimentId**: identifier for experiment session - **gameId**: identifier for game within the corresponding experiment session - **communicationMode**: indicates the communication capabilities of participants ('mixed' indicates that some participants can communicate locally, while others can communicate globally; 'none' indicates that participants cannot communicate) - **globalCommunicators**: the number of participants who can send messages not only to their neighbors, but to all network nodes - **communicationType**: in games where communication is allowed, indicates whether participants can send only pre-defined messages, or can type and send custom messages - **networkType**: the network model used to generate the underlying network - **incentives**: indicates whether participants receive higher payout when consensus is reached on a particular color, and lower payout on the other - **consensusReached**: whether the players reached a consensus on one of the two colors before 60 seconds have expired - **timeToConsensus**: the number of seconds it took players to reach consensus (set to 60.0 for games in which consensus was not reached) ---------- Files within the **marginal_info_data** folder - **timeStep**: identifies a time interval - **global**: the marginal information conveyed in the corresponding time period averaged over games with global communication - **local**: the marginal information conveyed in the corresponding time period averaged over games with local communication \* Currently, only the data for degree 3 nodes is uploaded. We are working on formatting the complete data set to a more manageable form, avoiding a nested file structure and will upload it as soon as possible. ---------- **balanced_color_preference_games.csv** - **experimentId**: identifier for experiment session - **gameId**: identifier for game within the corresponding experiment session - **sendTime**: the recorded time of when the message was sent, within the corresponding game - **message**: original message, as typed by the participant - **category**: one of four message categories, based on the message content \* This file contains all unconstrained messages, including those sent in games with mixed communication, whereas Fig. 4 only presents the message category counts for games with local or global communication. Fig. 4 can be reproduced by using the game parameter values in balanced_color_preference_games.csv. ---------- Files within the **minority_power_data** folder **color_counts.csv** (Fig. 6, 7, 9) - **experimentId** and **gameId** have the same meaning as in *balanced_color_preference_games.csv* - **time**: the recorded time of when the a node color was changed and the overall color count was updated, within the corresponding game - **red-green**: the number of network nodes with red and green color, respectively **configurations.csv** (Fig. 6, 7, 9) - **experimentId**, **gameId**, **incentives**, **communicationMode**, and **communicationType** have the same meaning as in *balanced_color_preference_games.csv* - **networkModel**: specifies the value of reqiring *q* parameter used to generate the underlying network - **minorityNodes**: the number of network nodes forming the well connected minority; also the global communicators The **networks** folder contains files with the adjacency matrices of the underlying networks, listed in game order within the corresponding experiment session. (Fig. 8) \* The outcome and end time of a game can be recovered by looking at the last entry for the corresponding game in the *color_counts.csv* file. For each adjacency matrix in the *networks* file, the 5 nodes indexed 0-4 are considered to be the well-connected minority of globally-communicating nodes - this is needed to reproduce Fig. 8.