Welcome to the Coin Toss Project repository accompanying the "Fair Coins Tend to Land on the Same Side They Started: Evidence from 350,757 Flips" manuscript (https://arxiv.org/pdf/2310.04153). The repository consist of multiple componets: - Data - Preregistration - Instruction - Analysis The Data component contains links to repositories with raw data (including video footage). The Preregistration component contains the projects preregistration (primarily preregistration of the focal prior distribution). The Intruction component contains instructions given to the participants. The Analysis component contains processed data, cleaned and formated data, and analysis scripts. The Analysis repository is structured followingly: - analyses - audit - data-raw - figures - functions You can reproduce the analyses via scripts in the analyses folder: - analysis-simple.R performs simple IID tests - analysis-simple-robustness.R performs robustness checks with respect to the prior distributions of simple IID tests - analysis-simple-trimmed.R performs robustness checks with respect to outlier removal of simple IID tests - analysis-hierarchical.R performs tests accounting for the between-person and between-coin heterogeneity - analysis-hierarchical-robustness.R performs robustness checks with respect to the prior distributions of tests accounting for the between-person and between-coin heterogeneity - analysis-hierarchical-trimmed.R performs robustness checks with respect to outlier removal of tests accounting for the between-person and between-coin heterogeneity - analysis-time performs the state-space model to assess the practice effects - analysis-variablity.R performs checks of whether the same side bias varies by the data collection location The folder also contains the following datasets (each can be constructed via merge-data.R and merge-data-add.R) - df_long.csv complete data set with each coin flip on a new row - df_long-add.csv additional data from Larwood and Ku in the same format - df_agg.csv aggregated format of the data benefitial for fitting hierarchical models - df_agg-add.csv additional data from Larwood and Ku in the same format - df_time.csv complete data set with each coin flip on a new row with better formating for practice effect analysis - df_time-add.csv additional data from Larwood and Ku in the same format - df_time-agg.csv aggregated format of the data benefitial for practice effect analysis (aggregating based on person-flip categorization) - df_time-agg-add.csv additional data from Larwood and Ku in the same format - df_time-agg2.csv aggregated format of the data benefitial for practice effect analysis (aggregating based on 100 flips categorization) - df_time-agg2-add.csv additional data from Larwood and Ku in the same format The functions folder contains additional helper functions and stan models code. The data-raw folder contains processed and cleaned data as well as the data cleaning scripts. The figures folder contains output figures. The audit folder contains scripts and data from the data audit.