Responsibilism

Code and data associated with the Responsibilism project: [Google Doc].

Figure 1. Constructing a State-Level Collectivism Index Using Reddit Data

Code

The code contains the following parts which readers may run to reproduce our findings.

1. Train_Word2Vec.ipynb - Scrape each US states' subreddit using Convokit and train separate a separate Word2Vec model for each state.
2. Construct_Dimensions.ipynb - Construct state-level collectivism index.
3. Construct_Dataframe.ipynb - Link index to other state-level covariates.
4. Regression.ipynb - Test convergent validity of index and use index to predict state-level happiness.

External Data & Packages

• Convokit - scrape subreddits.
• Gensim Word2vec (Rehurek & Sojka, 2011) - Train word embedding models
• roBERTa, VADER, Evaluative Lexicon (Rocklage et al., 2017) - Sentiment analysis on each state.
• Gallup Wellbeing Index - Convergent validity for sentiment scores.
• Vandello Cohen Collectivism Index (Vandello & Cohen, 1999) - Convergent validity for our collectivism index.