Regression Data

Our primary dataset, <b>'Tweet_Data_for_Share.csv'</b>, contains #Mahsa_Amini tweet metrics (retweets, timestamps, etc.) and image sentiment labels (e.g., compassion, indignation). This was merged with ACLED protest data and additional covariates (<b>'Complete_Regression_Variables.csv'</b>) for panel data regression. To ensure robustness, we replicated the analysis using the CTP dataset's city-level daily protest counts (<b>'Averaged_City_Protests.csv'</b>). Finally, to further support our causal inference about image emotions, we analyzed their relationship with comment linguistics using <b>'Unique_Comments.csv'</b>."

我们的主要数据集<b>'Tweet_Data_for_Share.csv'</b>包含#Mahsa_Amini推文指标（转发量、时间戳等）以及图像情感标签（如同情（compassion）、愤慨（indignation））。该数据集与ACLED抗议活动数据及额外协变量（covariates，<b>'Complete_Regression_Variables.csv'</b>）合并，用于面板数据回归（panel data regression）分析。为确保结果的稳健性，我们使用CTP数据集的城市级每日抗议活动计数（<b>'Averaged_City_Protests.csv'</b>）重复了该分析。最后，为进一步支持我们关于图像情感的因果推断（causal inference），我们使用<b>'Unique_Comments.csv'</b>分析了图像情感与评论语言学特征之间的关系。